Untangling the Octopus Vulgaris Species Complex Using a Combined Genomic and Morphological Approach

Untangling the Octopus vulgaris species complex using a combined genomic and morphological approach

Michael Douglas Amor

BSc (Hons)

Submitted in total fulfilment of the requirements for the degree of Doctor of

Philosophy

Department of Ecology, Environment and Evolution

School of Life Sciences

La Trobe University, Victoria, Australia

October, 2016

Contents

Untangling the Octopus vulgaris species complex using a combined genomic and morphological approach ...... i Contents...... ii Abstract ...... iv Statement of authorship ...... v Statement of co-authorship ...... v Statement of material included for the award of another degree ...... viii Acknowledgements ...... ix 1. Introduction ...... 1 1.1 Cryptic speciation ...... 1 1.2 Cryptic speciation among cephalopods ...... 3 1.3 The Octopus vulgaris group ...... 4 1.4 Implications for unresolved taxonomy ...... 9 1.5 Species tree inference ...... 9 1.6 Recent advancements in sequencing technology ...... 11 1.7 Thesis overview ...... 13 2. Allopatric speciation within a cryptic species complex of Australasian octopuses .. 16 2.1 Abstract ...... 16 2.2 Introduction ...... 17 2.3 Methods ...... 21 2.3.1 Molecular methods ...... 21 2.3.2 Morphological methods ...... 24 2.4 Results ...... 28 2.4.1 Molecular analyses ...... 29 2.4.2 Morphological analyses ...... 32 2.5 Discussion ...... 36 2.6 Supplementary information ...... 45 3. Morphological assessment of the Octopus vulgaris species complex evaluated in light of molecular-based phylogenetic inferences...... 58 3.1 Abstract ...... 58 3.2 Introduction ...... 59 3.3 Methods ...... 63 3.3.1 Sampling ...... 63 3.3.2 Molecular analyses ...... 66 3.3.3 Morphological analyses ...... 67 3.3.4 Comparative analyses ...... 69 ii

3.4 Results ...... 70 3.4.1 Phylogenetic relationships ...... 70 3.4.2 Morphological relationships ...... 72 3.5 Discussion ...... 81 3.6 Supplementary information ...... 87 4. Genome-wide sequencing uncovers cryptic diversity and mito-nuclear discordance in the Octopus vulgaris species complex...... 101 4.1 Abstract ...... 101 4.2 Introduction ...... 102 4.3 Methods ...... 106 4.4 Results ...... 113 4.5 Discussion ...... 120 4.6 Supplementary information ...... 126 5. Reconstructing the biogeographic history of speciation within the Octopus vulgaris species group...... 132 5.1 Abstract ...... 132 5.2 Introduction ...... 133 5.3 Methods ...... 136 5.4 Results ...... 143 5.5 Discussion ...... 150 5.6 Supplementary information ...... 155 6. Discussion ...... 157 6.1 Thesis overview ...... 157 6.2 Implications ...... 158 6.3 Limitations ...... 160 6.4 Future research ...... 161 6.5 Summary ...... 165 7. References ...... 166

iii

Abstract

Benthic octopuses (family Octopodidae) are a highly valuable global fisheries resource (worth >$US1/2 billion per annum). However, taxonomic relationships within the Octopodidae remain unresolved, which impedes their appropriate management. The type species of the genus Octopus, Octopus vulgaris, has historically been considered cosmopolitan. Recent studies, however, suggest O. vulgaris may represent a complex of morphologically similar, yet genetically distinct, species (the “Octopus vulgaris complex”). My thesis aims to investigate the species-level relationships within the O. vulgaris complex. In chapter two, I found congruence between morphological and mitochondrial DNA-based evidence, suggesting that allopatric populations of O. tetricus from the east and west coasts of Australia are distinct species. In addition, Asian O. vulgaris formed a monophyletic clade with both Australian species, which was distinct from other O. vulgaris populations. In chapters three and four, I investigate the species-level diversity within the O. vulgaris complex using unprecedented levels of morphological and genomic data, respectively, resulting in the most comprehensive global-scale investigation of this group to date. Discrete differences in morphology successfully delimited three species within the O. vulgaris complex; (1) Mediterranean/NE Atlantic and South Africa (2) southern

Brazil and (3) Asia. Genome-wide evidence obtained via next generation sequencing provided greater species-level resolution in comparison to those using mtDNA, supporting the distinction of South African O. vulgaris, and highlighting the limitations of mtDNA for resolving relationships within the group.

In chapter five, I investigated the ancestral biogeography of the O. vulgaris group and found the ancestor was a widespread species. Furthermore, speciation of

iv extant taxa was most likely driven by the sequential isolation of the eastern-most populations over the past 8 Ma. These results have significant taxonomic implications for the O. vulgaris group. As the majority of octopus fisheries worldwide are in decline, the findings of this thesis may inform the implementation of appropriate management strategies.

Statement of authorship

This thesis includes work by the author that has been published as described in the text. Except where reference is made in the text of the thesis, this thesis contains no other material published elsewhere or extracted in whole or in part from a thesis accepted for the award of any other degree or diploma. No other person's work has been used without due acknowledgment in the main text of the thesis. This thesis has not been submitted for the award of any degree or diploma in any other tertiary institution.

Statement of co-authorship

The following authors contributed to publication of work undertaken as part of this thesis:

BSc(Hons) Michael. D. Amor, La Trobe University, Australia

Associate Professor Jan M. Strugnell, La Trobe University, Australia

Dr Mark Norman, Parks Victoria, Australia

BSc(Hons) Hayley E. Cameron, Monash University, Australia

Dr Alvaro Roura, La Trobe University, Australia

Dr Tatiana Leite, Universidade Federal do Rio Grande do Norte (UFRN), Brazil.

Dr Ian G. Gleadall, Tohoku University, Japan

Dr Amanda Reid, Australian Museum, Australia

Dr Catalina Pearales-Raya, Instituto Español de Oceanografía, Spain

Dr Chung-Cheng Lu, Museum Victoria, Australia

BSc(Hons) Colin J. Silvey, Museum Victoria, Australia

Dr Erica A. G. Vidal, Universidade Federal do Paraná (UFPR), Brazil

Dr Xiaodong Zheng, Ocean University, China

Dr Stephen R. Doyle, Wellcome Trust Sanger Institute, United Kingdom

Dr Andrew Robinson, La Trobe University Australia

Dr Nathan E. Hall, La Trobe University, Australia

Author details and their roles:

Chapter two is published in PLOS ONE. Michael D. Amor contributed intellectually to the study design, collected the majority of data, performed all analyses and was the primary author of this chapter. Jan M. Strugnell (primary) and Mark D. Norman acted as supervisors for this publication and contributed

vi intellectually to the study design and manuscript preparation. Hayley E. Cameron assisted with morphological data collection and manuscript preparation.

Chapter three is published in Zoologica Scripta. Michael D. Amor contributed intellectually to the study design, collected the majority of data, performed all analyses and was the primary author of this chapter. Jan M. Strugnell (primary) and Mark D. Norman acted as supervisors for this publication and contributed intellectually to the study design and manuscript preparation. Alvaro Roura provided analysis advice and assisted with data collection and manuscript preparation. Colin J. Silvey assisted with the data collection and proofread the manuscript. Tatiana Leite, provided samples, data and proofread the manuscript.

Ian G. Gleadall, Amanda Reid, Catalina Pearales-Raya, Chung-Cheng Lu, Erica

A. G. Vidal, and Xiaodong Zheng provided samples and proofread the manuscript.

Chapter four is in preparation for publication. Michael D. Amor contributed intellectually to the study design, collected all data, performed all analyses and was the primary author of this chapter. Jan M. Strugnell (primary) and Mark D.

Norman acted as supervisors for this publication and contributed intellectually to the study design and manuscript preparation. Steven R. Doyle assisted with data collection and manuscript preparation. Nathan E. Hall and Andrew Robinson assisted with the development of bioinformatics/analysis pipelines. Alvaro Roura provided samples and proofread the manuscript.

Chapter five is in preparation for publication. Michael D. Amor contributed intellectually to the study design, collected all data, performed all analyses and was the primary author of this chapter. Jan M. Strugnell acted as the primary supervisor for this publication and contributed intellectually to the study design

vii and the manuscript preparation. Mark D. Norman contributed to the initial study design in a supervisory role.

Statement of material included for the award of another degree

Chapter two, includes molecular and morphological data collected during the candidate’s completion of his BSc (Hons) degree at La Trobe University, 2011.

Additional morphological data collection and all analyses, writing and manuscript preparation were conducted during the candidature for the current degree of

Doctor of Philosophy.

Michael D. Amor 18/10/2016

viii

Acknowledgements

My PhD allowed me to travel around the world collecting octopuses. I would like to thank my supervisors Jan Strugnell and Mark Norman for this great opportunity and the guidance and support that they provided. I am extremely grateful to Steve Doyle and Alvaro Roura for mentoring me throughout my PhD. I am grateful to Nathan Hall, Andrew Robinson and Ira Cooke for bioinformatics support and Shannon Hedtke for her support with phylogenetic analyses.

I owe several people thanks for assisting with collection or providing octopus samples; Alvaro Roura, Catalina Perales-Raya, CC Lu, Chia-Hui Wang, Chih-

Shin Chen, Colin Silvey, Eduardo Alamansa, Erica Vidal, Eric Hochberg, Ian

Gleadall, Jorge Ramos, Mandy Reid, Manuel Haimovici, Maria Cecilia Pardo

Gandarillas, Michelle Guzik, Stephen Leporati, Tatiana Leite, Vladimir

Laptikhovski, Xiaodong Zheng and Yuanyuan Ma. Many thanks to Julian Finn,

Chris Rowley, Melanie Mackenzie, CC Lu, Tim O’Hara, Dave Staples and

Genefor Walker-Smith of Museum Victoria’s Marine Invertebrate department for their assistance. I am also grateful to Stella Claudius and her ability to bring people together. I would also like to thank the ABRS and La Trobe University for providing research funding.

A very big thanks to my loving family for their support during my time at university. Finally, thank you to Hayley Cameron for her support, encouragement and love over the years. I could not have done this without you.

1. Introduction

1.1 Cryptic speciation

Speciation does not always correspond with detectable changes in phenotype

(Mayr, 1963). As a result, cryptic diversity is widespread throughout the animal kingdom (Trontelj & Fišer, 2009), from terrestrial to marine environments and polar to tropical regions (Knowlton, 2000; Bickford et al., 2007). Many cryptic taxa rely on chemical (e.g. marine species; Knowlton, 1993) or auditory (e.g. insects; Henry, 1994) signals, which typically do not require corresponding morphological adaptations (Mayr, 1963; Bickford et al., 2007). Alternatively, strong stabilising selection can result in divergence with reduced (or absent) change in morphology (e.g. hoverflies; Schönrogge et al., 2002). Therefore, assessments of species based on morphological characters alone have the potential to underestimate diversity (Bickford et al., 2007; Pfenninger & Schwenk,

2007).

The marine environment is suggested to contain relatively high levels of cryptic diversity (Knowlton, 2000). This is, in part, due to the relative difficultly for marine researchers to investigate living specimens (Knowlton, 1993; Bickford et al.,

2007), and as a result behavioural-based delimitation is rare. Many marine taxa depend on chemical cues for mate choice and egg-sperm recognition (Knowlton,

1993). For example, shallow-water octopuses (e.g. Octopus vulgaris Cuvier,

1797) are known to possess advanced chemical recognition abilities (Graziadei,

1964; Boyle, 1983; Hague et al., 2013) and display associated adjustments in

1 reproductive behaviour (Boal, 2006; Tricarico et al., 2014). When available, chemical and auditory signals are a useful addition to species descriptions as either primary or complimentary delimiting traits (Bickford et al., 2007).

Morphological similarity among cryptic species, by definition, makes defining phenotypic species boundaries difficult (Knowlton, 1993). Further to the inclusion of behavioural and auditory traits (e.g. Henry, 1994), the fields of taxonomy and systematics have greatly benefited from the inclusion of molecular data (Bickford et al., 2007). DNA barcoding (sequencing of a single gene, commonly cytochrome c oxidase subunit I; COI) is often applied to aid in the discovery of new species or to assign individuals with unknown taxonomy to an existing species (Hebert et al., 2003, 2004). Several examples show that once species boundaries have been recognised using molecular techniques (e.g. DNA barcoding), diagnostic morphological differences, or statistically significant differences among overlapping traits are often able to be identified (for review see; Knowlton, 1993).

Once species boundaries are understood, investigating the biogeographic history of closely related species is an important step in increasing our knowledge of the present-day distributions of lineages (Ree et al., 2005), the processes that drove the emergence of new taxa (Crisp et al., 2011) and for understanding overall patterns of biodiversity. Ancestral area reconstruction (AAR) and estimates of timing of divergence are able to place diversification in context with past geological events that potentially shaped the present-day distribution of extant taxa (Crisp et al., 2011; Ronquist & Sanmartín, 2011). Unresolved taxonomic relationships, such as those among cryptic species, present a significant limitation to the reconstruction of ancestral histories (Andersson, 1996). Accurate

2 taxonomic resolution among cryptic taxa enables such inferences of past biogeographical events, which may provide insight into role of dispersal/founder events and vicariance in shaping modern day distributions of species.

1.2 Cryptic speciation among cephalopods

Cryptic speciation is common among cephalopods, especially among squids and octopuses (Pickford & McConnaughey, 1949; Söller et al., 2000; Allcock, 2005;

Leite et al., 2008; Allcock et al., 2011; Norman et al., 2014a; Norman et al.,

2014b). The characteristic soft bodies of cephalopods have few hard structures

(e.g. stylets, beak) on which to base taxonomy (Bookstein et al., 1985). Hard morphological structures, such as the beak, have proven useful for identifying species among pelagic squid (for review see; Xavier et al., 2007). Octopus beaks, however, are suggested to be relatively homeomorphic (Clarke, 1986) and are considered to be unreliable indicators at the species level (Voss, 1977), and better suited to generic identification (Ogden et al., 1998; Lu & Ickeringill,

2002; Allcock, 2005). Although, rare exceptions exist, where differences in beak morphology were reported between tropical O. cyanea Gray, 1849 and temperate/sub-tropical O. vulgaris, no differences were observed between suspected cryptic O. vulgaris taxa from South Africa and the

Mediterranean/eastern North Atlantic (Smale et al., 1993). The genus Octopus has long been considered a ‘catch all’ genus (Nesis, 1998), as few morphological characters have proven useful for delimiting closely related taxa. Resolution among closely related octopus species is, therefore, mostly reliant on soft body parts. Yet, cephalopods display high levels of morphological plasticity (Robson, 3

1929; Pickford, 1945; Voight, 1994; O'Shea, 1999) and specimens are prone to distortion and loss of pigmentation upon preservation (Robson, 1929; Pickford,

1964; Burgess, 1966; Voight, 2001), which has traditionally impeded taxonomy.

Robson (1929) identified seven morphological traits that were to form the basis of Octopus species descriptions. Roper and Voss (1983) extended this list of traits by identifying 18 morphological characters (and 22 related indices) that were suggested as a minimum requirement for descriptions of pelagic and benthic octopus species. Voight (1994) investigated the utility of the majority of these traits for delimiting shallow-water octopuses, and reported that discriminant analyses failed to differentiate among most species (although, species from rocky reefs were generally able to be delimited from species inhabiting environments with sand and seagrass substrates). A more recent assessment of the utility of these traits in the genus Pareledone (contained within closely related family Megaleledonidae) also showed they were unreliable for species-level delimitation (Allcock et al., 2008).

1.3 The Octopus vulgaris group

Recent taxonomic revisions (O'Shea, 1999; Norman et al., 2014a) and molecular-based phylogenetic studies (Guzik et al., 2005; Kaneko et al., 2011;

Acosta-Jofré et al., 2012; Lü et al., 2013) show that the genus Octopus is polyphyletic and contains a large assemblage of species groups across several genera. The species group most similar in morphology and behaviour to the type species of the genus (Octopus vulgaris Cuvier, 1797) has been identified as the

‘O. vulgaris group,’ based on general similarities in overall size, mantle shape, arm length and skin sculpture (Robson, 1929). Species in the O. vulgaris group are now considered to comprise the genus Octopus sensu stricto (O'Shea,

1999).

Historically, O. vulgaris was considered to be a cosmopolitan species (Robson,

1929; Mangold, 1983). The earliest records of this species are presumed to be from the Mediterranean Sea and eastern North Atlantic (Robson, 1929), although

Lu et al., (1995) were unable to trace type material, which may never have been designated (Mangold, 1998). Further reports include specimens from Mauritius,

India, Timor, Haiti, Cuba, Bahia (d’Orbigny, 1840, p. 30), South Africa (Krauss,

1848, p. 132; Thiele, 1915), Japan (Ortmann, 1888; Appellöf, 1886; Wülker,

1910; Berry, 1912), the Andaman Islands (Goodrich, 1896), the Red Sea

(Wülker, 1920) and Australia (Cox, 1882). More recently, Mangold (1998) suggested differences in host-specific parasites, spermatophore morphology, skin patterns and paralarval chromatophore patterns potentially indicate O. vulgaris is composed of several cryptic taxa.

To date, no attempts have been made to investigate morphological relationships within the O. vulgaris group. However, several studies utilising molecular-based data have attempted to resolve phylogenetic relationships within the group. Two cryptic species have been identified using molecular-based data; Octopus mimus

Gould, 1852 was redescribed after previously being synonymised with O. vulgaris (Guerra et al., 1999) and subsequent molecular-based analysis (partial cytochrome c oxidase subunit III [COIII] fragment) confirmed O. mimus to be distinct from O. vulgaris (Söller et al., 2000). Octopus insularis Leite and

Haimovici, 2008, a species from northern Brazil once considered conspecific with

O. vulgaris, was also recently described based on analyses of mitochondrial (mt)

DNA (partial 16S rRNA fragment; Leite et al., 2008). Further to the identification of O. mimus and O. insularis, several studies have investigated phylogenetic relationships within the O. vulgaris group.

Warnke et al., (2004) analysed partial sequence data from two mt genes (16S rRNA and COIII), obtained from individuals from South Africa, Tristan da Cunha,

Senegal, France, Japan, Taiwan, Brazil and Venezuela. The resulting topology supported the monophyletic status of O. vulgaris, however, well-supported structure within the group was recorded and considered to represent intraspecific differences. One limitation of the phylogeny presented by Warnke et al., (2004) was the exclusion of members of the O. tetricus complex, which Guzik et al.,

(2005) identified as belonging to the O. vulgaris group. Incomplete sampling is common among studies investigating the O. vulgaris group, as obtaining samples from its global distribution is challenging. However, this meant the resulting phylogenetic structure presented by Warnke et al., (2004) could not be compared to known interspecific differences among close relatives.

Guerra et al., (2010) validated the presence of a closely related species of O. vulgaris from Amsterdam and Saint Paul Islands (Indian Ocean) by independently analysing partial COI and COIII fragments. Both topologies constructed based on COI and COIII did not show congruent phylogenetic structure among O. vulgaris individuals from the Atlantic and Indian Oceans.

However, individuals from the western North Pacific (Japan and Taiwan) formed a distinct well-supported clade in both topologies, as did O. tetricus from the western South Pacific. The analysis based on COIII data showed individuals from southern Brazil were distinct from O. vulgaris and O. tetricus, however the

6 relationships among these three groups were unable to be inferred from the COI based topology as data from southern Brazilian individuals were not available.

Subsequently, the phylogenetic relationships among members of the genus

Octopus were investigated using COIII data (Acosta-Jofré et al., 2012), including the sequences from Guerra et al., (2010). Octopus vulgaris individuals from the western South Atlantic and the Caribbean Sea formed a clade that was sister to a clade containing individuals from the Mediterranean Sea, eastern Atlantic,

Indian and Pacific Oceans and O. tetricus from the western South Pacific. This study showed ‘O. vulgaris’ to be paraphyletic, and provided some evidence for the existence of multiple species currently being treated under a single species name. Within O. vulgaris, three sub-clades were supported; (i) Asia, (ii) Brazil and Caribbean and (iii) France, western Africa, South Africa, Tristan da Cunha and Amsterdam Island.

Recently, individuals from China, Taiwan and Japan were recognised to be a distinct species based on their phylogenetic relationships (Reid & Wilson, 2015).

A phylogenetic clade based on COIII data was composed of these individuals and individuals from the Kermadec Islands, a newly sampled locality in the southern West Pacific Ocean. This species was originally described as O. jollyorum Reid and Wilson, 2015, however Gleadall (2016) subsequently revalidated an earlier available name for the clade, O. sinensis d’Orbigny, 1841.

The O. vulgaris group is currently recognised to be composed of three species;

O. vulgaris (Mediterranean, Atlantic and Indian Oceans), O. sinensis (western

Pacific Ocean) and O. tetricus (western South Pacific and eastern South Indian

Oceans).

Norman et al., (2014a) proposed five species ‘Types’ based on previous molecular-based evidence, geographic isolation and lack of plausible gene flow mechanisms. Octopus vulgaris sensu stricto (s. s.) occurs in the Mediterranean and eastern North Atlantic. Type I occurs in the Caribbean and Gulf of Mexico;

Type II in the western South Atlantic along the coast of Brazil; Type III occurs in the eastern South Atlantic and the Indian Ocean, along the coast of South Africa;

Type IV (O. sinensis) occurs in subtropical to temperate eastern Asia. Further molecular and morphological evidence is required to resolve the species-level relationships among potential cryptic species within this group. Very little is known about the historical evolution of the genus Octopus, although it is estimated to have originated in the late Jurassic (Kröger et al., 2011).

Divergence between O. bimaculoides Pickford & McConnaughey, 1949 (genus

Octopus s. s.) and Hapalochlaena maculosa Hoyle, 1883 has been estimated to have taken place within the last 10-65 Ma (Strugnell et al., 2004, 2006, 2008b).

However, no studies have investigated divergence times within the genus

Octopus.

Comprehensive morphological and molecular-based investigations into the O. vulgaris species complex are required to resolve species-level relationships among each hypothesised cryptic species. A robust study design should include a global sampling effort combined with robust analysis of all available taxonomically informative morphological traits and the analysis of multiple genetic loci.

1.4 Implications for unresolved taxonomy

Improved taxonomic resolution within the family Octopodidae d'Orbigny, 1839 is particularly important in light of the growing global exploitation of octopuses as a commercial fisheries resource (Norman & Finn, 2014). Global fisheries and aquaculture production of octopuses exceeds 350,000 tonnes and has a total export value of US$1.07 billion, surpassing many valuable finfish fisheries (FAO,

2012). A major limitation of the global catch statistics reported by the FAO is the poor state of octopus taxonomy, with only five (Octopus vulgaris Cuvier, 1797, O. maya Voss & Solis, 1966, Eledone cirrhosa Lamarck, 1798, E. moschata

Lamarck, 1798 and Enteroctopus dofleini Wülker, 1910) of the estimated 100 species of commercially harvested octopuses listed in global statistics (Norman

& Finn, 2014). The ‘Octopus vulgaris species complex’ is of extremely high market value and profile (Norman et al., 2014a), particularly in north-western

Africa, the largest single-species octopus fishery in the world (FAO, 2012). Given that the majority of octopus fisheries world-wide are in decline (Norman & Finn,

2014), there is an important need for improved taxonomic resolution among octopuses to assist appropriate fisheries management strategies.

1.5 Species tree inference

Several limitations for the use of a single genetic marker to infer phylogenetic relationships have been identified (Moritz & Cicero, 2004). Individual trees based on a single gene (or linked genome such as the mitochondria) do not necessarily 9 reflect species trees (Page & Charleston, 1997). Processes such as homoplasy

(where a single base position is the site of repeat mutations), incomplete lineage sorting, horizontal gene transfer, duplication, hybridisation and recombination may also result in discordance among gene trees (Avise et al., 1994; Zink &

Barrowclough, 2008; Degnan & Rosenberg, 2009). To avoid associated biases, incorporating multiple loci, particularly nuclear (nu) DNA (Edwards et al., 2005;

Edwards & Bensch, 2009), can ensure the accuracy of phylogenetic estimates

(Liu & Pearl, 2007). However, due to the generally lower average mutation rate of nuDNA relative to mtDNA, roughly two to three times more nucleotides may need to be sequenced to obtain the equivalent number of variant sites, and therefore provide equivalent resolution (Zink & Barrowclough, 2008). However, the relatively low copy-number of nuDNA within cells makes obtaining sequence data inherently more difficult.

The most common (and simple) method for estimating species trees includes the concatenation of sequence data from multiple loci, followed by Maximum

Likelihood (ML) reconstruction where all sites are modelled to evolve identically and independently (Chou et al., 2015). This method, however, is known to allow convergence on a topology that may not be the most accurate representation of the species tree (Roch & Steel, 2015). Coalescent-based ‘summary methods’ such as BEAST (Heled & Drummond, 2010), BEST (Liu, 2008) and BUCKy

(Larget et al., 2010) are designed to estimate a gene tree on each locus alignment, which is then combined with all other gene trees to produce a species tree. Common issues associated with this approach include their computationally intensive nature, which limits their application to genome-scale inference

(Mirarab et al., 2014; Zimmermann et al., 2014), and the sensitivity to estimation error when individually informative loci do not produce well-resolved gene trees 10

(Roch & Warnow, 2015; Mirarab et al., 2016), which is common among methods that obtain short sequencing reads (e.g. Ree & Hipp, 2015).

To avoid the issue of error associated with individual gene tree estimation, a coalescent approach avoiding this process has been developed to examine the patterns at each individual site to estimate the species tree. This includes approaches that are suited to unlinked single nucleotide polymorphisms (SNPs) and complete nucleotide sequence data (Bryant et al., 2012; Chifman & Kubatko,

2014), with the latter being preferred when discarding a substantial amount of informative data can be avoided (Ree & Hipp, 2015). The least computationally intensive, and therefore most applicable method, to large genomic datasets is the recently developed quartet-based approach ‘SVDquartets’ (Chifman &

Kubatko, 2014), which estimates the most likely unrooted species tree of four taxa and combines each quartet tree into a single species tree. The relative computational speed and accurate inference (Chou et al., 2015) obtained via this approach suggests it is an appropriate method for species tree inference when using next-generation sequencing (NGS) approaches that obtain large volumes of genome-wide data such as genotype-by-sequencing (GBS) and restriction site associated DNA sequencing (RADseq) (Ree & Hipp, 2015).

1.6 Recent advancements in sequencing technology

In comparison to traditional Sanger sequencing of individual loci, NGS technology has allowed for the production of extremely large volumes of data

(Metzker, 2010). The application of NGS allows for the potential to sequence and

11 genotype thousands of markers from virtually any genome of interest (Stapley et al., 2010). Novel NGS methods, such as GBS and RADseq, also enable sequencing and genotyping with little or no previous available genetic information (Davey et al., 2011). This is particularly advantageous for obtaining orthologous sequence data and inferring relationships of non-model organisms, where no reference genome is available.

‘Reduced representation’ NGS technologies screen for polymorphisms across the genome via various sub-sampling methods, without the requirement of sequencing or analysing the entire genome. This approach to sequencing is relatively cost effective and requires less computational resources to analyse.

Reduced representation sequencing effectively sub-samples the entire genome for markers, giving greater power to resolve population genomic (Narum et al.,

2013) and phylogenomic (e.g. Peterson et al., 2012) questions (Davey et al.,

2011). Commonly used examples of reduced representation NGS protocols include GBS, where DNA of organisms with small genomes is sheared randomly, and RADseq (Baird et al., 2008), where restriction enzymes are used to digest

DNA in a repeatable manner.

Double digest RADseq (Peterson et al., 2012) expands on the original RADseq protocol (Baird et al., 2008) by using a second restriction enzyme in the digestion process. Fine scale size selection is also added to allow for recovery of a

‘tunable’ number of regions. This enables researchers to optimise the volume of

DNA fragments repeatedly obtained from orthologous genes across multiple individuals. RADseq has been shown to be suitability applied to phylogenomic studies (Peterson et al., 2012; Lemmon & Lemmon, 2013; Ree & Hipp, 2015),

12 with studies obtaining resolution with divergence time estimates up to 63 million

(Cariou et al., 2013) and even 100-360 million years (Gonen et al., 2015).

Studies increasingly recognise the importance of using multiple loci for species- level inferences. The use of a single or multiple mtDNA marker(s) has, so far, been insufficient for delimiting species within the O. vulgaris complex (Warnke et al., 2004; Guerra et al., 2010), given that O. vulgaris is understood to be paraphyletic (Acosta-Jofré et al., 2012). RADseq represents a relatively fast and cheap approach for obtaining genome-wide data from multiple loci among members of the O. vulgaris complex, despite the unavailability of a reference genome.

1.7 Thesis overview

The aim of this thesis was to provide species-level resolution within the O. vulgaris group. I incorporate multiple lines of evidence, such as mtDNA, genome- wide nuDNA loci and morphological data, to investigate the utility of each individual approach. Finally, I provide insight into the origin of the O. vulgaris group via divergence time estimation and ancestral area reconstruction.

Prior molecular-based attempts to resolve species-level relationships within the

O. vulgaris group have incorporated mtDNA. In chapter two, I investigated the validity of this marker as a taxonomic tool within the O. vulgaris group by contrasting mtDNA-based results with morphological evidence. Chapter two investigates species boundaries within the O. tetricus species complex by combining evidence based on analyses of 17 morphological traits and molecular 13 data, including ML and Bayesian Inference (BI) phylogenies of five mtDNA genes.

Chapter three contrasts the use of morphology and DNA barcoding for delimiting species within the O. vulgaris group. Multivariate analyses of up to 35 morphological traits were used to investigate species diversity among six putative species, as determined by ML and BI analyses of COI sequence data.

This chapter also contrasts the ability of male and female-based morphological data to successfully delimit species within the genus Octopus. I also investigate interspecific phenotypic variation within O. vulgaris s. s. from France, Galicia and

Mauritania.

Chapter four investigates the genome-wide phylogenetic relationships among members of the O. vulgaris species complex. Phylogenetic analyses of double digest RADseq data were performed and combined with investigations of genome-wide concordance and species tree estimation to delimit species within the O. vulgaris species complex.

The timing of divergence events and historical reconstruction of ancestral distribution was estimated for O. vulgaris species complex ancestors in chapter five. An ultrametric phylogeny was estimated via BI and known nuDNA substitution rates were applied to obtain estimates of ancestral divergence times.

Based on the BI topology, ancestral state reconstruction (ASR) models were contrasted to test hypothetical scenarios of evolution and to estimate the historical distribution of the group’s most recent common ancestor (MRCA).

Chapter six summarises the advancements in O. vulgaris species complex taxonomic knowledge presented within this thesis, the implications for fisheries

14 management and conservation, and future avenues of research within the fields of taxonomy and phylogenetic relationships.

2. Allopatric speciation within a cryptic species complex of

Australasian octopuses

This chapter is published as:

Amor, M. D., Norman, M. D., Cameron, H. E., & Strugnell, J. M. (2014).

Allopatric speciation within a cryptic species complex of Australasian octopuses.

Plos One, 9, e98982. doi:98910.91371/journal.pone.0098982.

2.1 Abstract

Despite extensive revisions over recent decades, the taxonomy of benthic octopuses (Family Octopodidae) remains unresolved. Among these unresolved groups 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 species or sub-species, and O. 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 ‘O. tetricus complex’. Phylogenetic analyses (based on five mitochondrial genes: 12S rRNA, 16S rRNA, COI, COIII and cytb) and

Generalised Mixed Yule Coalescent analysis (based on COI, COIII and cytb) distinguished east and west 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 and 16 female octopuses (based on 16 and nine morphological traits, respectively) provided further evidence of cryptic speciation between east (including New

Zealand) and west coast populations, while males proved more useful in morphological distinction among members of the O. tetricus complex. In addition, phylogenetic analyses suggested populations of octopuses currently treated under the name O. vulgaris are paraphyletic; providing evidence of cryptic speciation among global populations of O. vulgaris, the most commercially valuable octopus species worldwide.

2.2 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 taxonomy of this family remains unresolved (Carlini et al., 2001;

Guzik et al., 2005; Norman & Hochberg, 2005). The most widely studied and economically significant cephalopods world-wide is the ‘Octopus vulgaris group’ of octopods. The type species of this group is the common octopus, O. vulgaris

Cuvier, 1797. Octopus vulgaris alone accounts for >50% of the world’s total octopod fisheries catch, which in 2007 was recorded to exceed 380,000 tonnes with an international export value of >US$1 billion (FAO, 2009). The O. vulgaris 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 (Robson, 1929).

Within the subtropical waters of Australasia there is a group of morphologically, behaviourally and functionally similar Octopus species, closely related to O. vulgaris (Guzik et al., 2005; Acosta-Jofré et al., 2012). These species, currently treated under the names O. tetricus Gould, 1852 on the east and west coasts of

Australia and O. gibbsi O’Shea, 1999 in New Zealand, have been suggested to be an unresolved species complex (Guzik et al., 2005). We treat these taxa collectively herein as the ‘O. tetricus complex’, after the first formally described species within this group, O. tetricus; the common Sydney octopus.

The O. tetricus complex is composed of three geographically distinct member taxa (Fig. 1). Octopus tetricus was originally described from New South Wales, and is recorded as occurring along the east Australian coastline, ranging from

Eden in southern New South Wales to Moreton Bay in southern Queensland

(Edgar, 2000). Octopus tetricus contributes to a major portion of the small-scale commercial octopod fisheries landings in New South Wales (Nottage et al.,

2007), and is also often caught as by-catch in prawn and finfish trawls (Norman et al., 2014a). Recently O. tetricus sightings have been reported in Flinders

Island, Tasmania, significantly south of its previous known range (REDMAP,

2011), although this report has not been verified by molecular data.

Fig. 1: Known distributions (red) and sample locations (black) for Octopus tetricus, (east Australia), O. cf. tetricus (Western Australia) and O. gibbsi (New

Zealand). Location acronyms: WP= Woodman’s Point, MA= Mandurah, AL =

Albany, ES = Esperance, CG = Cape Le Grand, FI = Flinders Island, Tasmania,

WL = Wallaga Lake, NA = Narooma, PS = Port Stephens, LE = Leigh, New

Zealand.

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 O. tetricus (Joll, 1976, 1977, 1978, 1983; Roper et al.,

1984) 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-species or distinct species

(Norman, 2000; Norman & Hochberg, 2005). Joll (1983) 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 consequently is considered a pest that negatively impacts this economically important fisheries resource.

A third nominal species, O. gibbsi was coined to describe a benthic octopus of unknown relation found within the shallow coastal waters off northern New

Zealand. Prior to description, O. gibbsi had been treated under the name O. tetricus (Anderson, 1997), and more recently the validity of O. gibbsi as a distinct species has been questioned (Norman & Hochberg, 2005). 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 (Norman & Hochberg, 2005).

A phylogenetic analysis of the sub-family Octopodinae using amino acid sequences from two mtDNA markers (cytochrome c oxidase subunit III [COIII] and cytochrome b apoenzyme [cytb]) and a single nuDNA marker (elongation factor-1α) supported a sister taxa relationship between O. tetricus and O. cf. tetricus (Guzik et al., 2005). 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 a single representative from Western Australia and New South Wales were sequenced. Although, analyses of Guzik et al., (2005) were a valuable first step, sampling of only one individual per locality meant they lacked the power to detect speciation between disjunct east and west populations. Furthermore, no

20 molecular work to date has investigated the phylogenetic status of O. gibbsi, thus its taxonomy remains unresolved.

This study aims to use a combination of morphological and molecular techniques to resolve the taxonomy and phylogenetic relationships of three Australasian octopod taxa that comprise the O. tetricus complex. Due to increasing fisheries value and a lack of species-level resolution within the O. tetricus complex, taxonomic resolution within this group will aid in the management of these valuable marine resources.

2.3 Methods

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

2.3.1 Molecular methods

Sampling: Tissue samples of the ingroup (O. tetricus [n = 13], O. cf. tetricus [n =

13] and O. gibbsi [n = 5]) were sourced from collections at Museum Victoria or provided by researchers associated with The University of Adelaide, the Western

Australian Fisheries and Marine Research Laboratories Department and the

University of Tasmania (Table S1). Tissue samples (as arm or mantle tissue ~1

21 cm in length) were taken from individuals collected from the Australian mainland,

Flinders Island (Tasmania) and New Zealand (Fig. 1). All tissue samples were stored at -20oC in 70-90% ethanol until processing.

Sequencing: DNA was extracted from mantle or arm tissue using the High Salt

Method (Donnan Laboratories, 2001). Partial sequences of five mtDNA genes were targeted; including12S ribosomal RNA (12S) (Simon et al., 1990), 16S ribosomal RNA (16S), and COI (Folmer et al., 1994), COIII and cytb (Guzik et al.,

2005). 25 µL reactions were composed of 0.1 µL Taq (Onetaq, New England

Biolabs), 2.5 µL 10 x buffer (Paq5000TM), 2 µL dNTP mix (10 µM, Bioline), 0.5 µL forward primer (10 µM), 0.5 µL reverse primer (10 µM), 17.4 ddH2O and 2 µL

DNA (diluted to between 1-5 ng/µL). Reaction conditions are detailed elsewhere

(Folmer et al., 1994). PCR products were sequenced by Macrogen Inc, Seoul,

Korea.

Genetic sequence data generated in this study are accessible from GenBank

(Table S1). Octopus mimus and O. oculifer Hoyle, 1904 were selected as outgroup taxa on the basis that they share morphologically similarities, and are the closest known available relatives of the ingroup (Norman & Hochberg, 2005;

Kaneko et al., 2011; Acosta-Jofré et al., 2012). Sequence data of outgroup taxa and additional sequences of ingroup taxa from previously published work were downloaded from GenBank (Table S2). Multiple sequence alignments were performed using Geneious Muscle Alignment feature using the ClustalW default settings (Larkin et al., 2007).

Phylogenetic analyses: jModelTest v0.1.1 (Posada, 2008) was used to carry out statistical selection of best-fit models of nucleotide substitution on the concatenated alignments and a COI-only alignment. The appropriate model was

22 selected on the basis of ‘goodness of fit measure’ via the Akaike Information

Criterion (AIC) (Akaike, 1974). ML topologies were constructed using PhyML v3.1 (Guindon et al., 2010). 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 (BS) replicates. BI marginal posterior probabilities (PP) were calculated using MrBayes v3.2 (Ronquist & Huelsenbeck, 2003). 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

(Rambaut & Drummond, 2003) was used to ensure Markov chains had reached stationarity, and to determine the correct ‘burn-in’ for the analysis (the number of additional generation that must be discarded before stationarity is reached).

Genetic distance: Molecular Evolutionary Genetic Analysis (MEGA) v5.2

(Tamura et al., 2011) was used to calculate genetic distances among O. tetricus,

O. gibbsi and O. cf. tetricus using the Tamura-Nei model (Tamura & Nei, 1993).

Genetic distance was calculated using MEGA default settings (with the exceptions of the model and ‘pairwise deletion of missing data’ option). Mean values ± SE of interspecific and intraspecific variations in number of mutations per site were calculated for the barcoding mtDNA gene COI to allow comparison with published literature.

Divergence time estimation: 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; (Strugnell et al., 2012) within a generalised molecular clock.

Coalescent delimitation: Potential species delimitation between O. tetricus and

O. cf. tetricus was investigated using a Generalised Mixed Yule Coalescent

(GMYC) model (Pons et al., 2006) applied to the molecular data. Partitioned sequence data from the mtDNA genes COI, COIII and cytb were prepared into

XML files using the software program BEAUti v1.7.5 (Drummond et al., 2012).

12S and 16S regions were excluded from the analysis due to low sample representation (see Table S1). A coalescent prior and relaxed molecular clock

(Monaghan et al., 2009) were set as parameters before BI analysis was performed using BEAST v1.7.5 (Drummond et al., 2012). Each analysis was performed independently twice and log/tree files were combined using

LogCombiner v1.7.5 (Drummond et al., 2012). The data were then analysed via a single threshold model (Talavera et al., 2013) in the software package Splits

(Ezard et al., 2009) available in R v3.0.1 (R Team, 2015), whereby clades with posterior probability values greater than 0.9 were acknowledged.

2.3.2 Morphological methods

Sampling: Morphological data were obtained from preserved whole specimens sourced from Museum Victoria, the Australian Museum 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) and Flinders

Island, Tasmania (n = 11; 2011) (Table S3). All specimens were initially fixed in

10% formalin and transferred to 70-90% ethanol for preservation. Morphological data for O. gibbsi (n = 6) was sourced from O’ Shea (1999).

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) (Voight, 1995). Maturity in males was determined on the basis of the presence or absence of enlarged suckers (for mature and immature specimens, respectively) (Voight, 1991a). Maturity in females was determined by the state of egg development (Iribarne, 1991). 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 (1997) (Table 1). Dorsal mantle length (MLd), mantle width (MW), head width (HW), arm width (AW), and sucker diameter (SD) were measured using digital callipers to the nearest 0.1 mm. For males, the largest enlarged sucker diameters (SDe), the length of hectocotylus components (i.e. ligula [LL] and calamus [CL]) and, following dissection of the mantle, terminal organ length

(TOL) 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 on the third right (SCR3) and third left (SCL3) arms were counted with the aid of a dissecting microscope. When damage to arms was perceived to inhibit growth, suckers appeared damaged, or arm regeneration was evident, arm length and sucker counts were not recorded.

When sucker and arm damage was minor and 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.

Table 1: Description of morphological measurements recorded.

Abbreviations Description

MLd Dorsal mantle length

MW Greatest width of mantle

HW Greatest width of head at the level of eyes

AW Width of stoutest arm

SDn Diameter of largest non-enlarged sucker on any arm

WD Measurement of deepest web sector, from beak to midpoint of sector

ALL3/R3 Length from beak to tip of third left/right arm

SDeL2/R2* Largest enlarged sucker diameter on the second left/right arm

SDeL3/R3* Largest enlarged sucker diameter on the third left/right arm

SCL3/R3 Entire number of suckers along intact third left/right arm

LL* Length from distal most sucker to tip of hectocotylised arm

CL* Length from distal most sucker to tip of calamus

TOL* Length of male terminal organ

* Denotes morphological trait only recorded for male octopuses.

Morphology-based analyses: All analyses of morphology were performed using

Systat v13 (Systat Software, 2009). Trait differences among O. tetricus complex taxa were investigated using a multivariate General Linear Model (GLM), where location was treated as a fixed factor, morphological counts were treated as dependent variables and MLd was entered as a co-variate (Berner, 2011).

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 (R2 = 0.8467, data not shown), is more often provided in the literature compared to total body length, and is a standardised measurement when compared to body weight (which can be obtained from fresh or preserved specimens) (Voight, 1991b). The presence or absence of an interaction between locations and MLd was investigated. A

26 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 independently to allow the inclusion of male reproductive organs in morphology-based analyses. Mean scaling was performed on all dependant variables prior to analyses as per Berner (2011) using the software package R v3.0.1 (R Team, 2015), whilst the co-variate (MLd) was log transformed (male) and 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 analyses of female morphology.

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 (Quinn & Keough, 2002b). Principle components were then plotted for visualisation and canonical correlations were used to calculate eigenvalues and the proportion of variance explained by each

PC (Tables S4-S12).

The importance of each morphological character in delineating O. tetricus complex taxa was further investigated by Roy-Bargman step-down analysis (Roy

& Bargmann, 1958), 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

27 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, while higher priority traits were added as co-variates in each successive analysis. Tukey’s post-hoc tests were performed for each significant step-down 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 O. 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

(Quinn & Keough, 2002a). For all DFAs, Jackknifed correlation matrices were used as they are considered a more reliable estimator of group membership assignment (Quinn & Keough, 2002b).

2.4 Results

2.4.1 Molecular analyses

Phylogenetic analyses: The AIC indicated that TrN+G was the preferred evolutionary model for the concatenated alignment and this was utilised for ML and BI phylogenetic analyses. Topologies resulting from ML and BI analyses were identical. Both ML and BI recovered a highly supported clade containing O. tetricus from east Australia and Tasmania and O. gibbsi from New Zealand (BS =

97.6, PP = 1; Fig. 2). All individuals collected from Western Australia formed a highly supported monophyletic clade (BS = 98.6, PP = 1). A sister-taxon relationship was supported between the west Australian and east Australian

(New South Wales and Tasmania)/New Zealand clades (BS = 92.6, PP = 1).

Fig. 2: Bayesian topology depicting the phylogenetic relationships among five currently accepted species of Octopoda. Analyses are based on partial

29 sequence data of five combined mtDNA genes (12s rRNA, 16s rRNA, COI, COIII and cytb), showing BS values ≥50 below each node and PP values ≥0.7 above each node. Outgroup is composed of O. oculifer and O. mimus. Node labels reflect locations represented by individuals contributing to node (Western

Australia, 1 = Mandurah, 2 = Woodman’s Point, 3 = Albany, 4 = Cape Le Grand,

5 = Esperance; East Australia, 1 = Wallaga Lake, 2 = Port Stephens, 3 =

Narooma; South Africa, 1 = Port Elizabeth, 2 = Umhlanga, 3 = Hout Bay, 4 =

Durban).

All O. vulgaris individuals collected from the waters off Japan and China formed a highly supported monophyletic clade (BS = 97.3, PP = 1). Japanese and

Chinese O. vulgaris and the O. tetricus complex were supported as a monophyletic clade (BS = 81.2, PP = 0.95). This clade fell within a larger clade containing O. vulgaris individuals from Spain (type location; Mediterranean Sea),

South Africa, St Paul and Amsterdam Islands, thereby rendering the O. vulgaris clade to be paraphyletic.

Genetic distance: Octopus gibbsi was treated as O. tetricus during genetic distance calculations based on COI sequence data, as high phylogenetic support values suggested they were a single phylogenetic unit. Intraspecific (i.e. within

O. cf. tetricus or within O. tetricus and O. gibbsi) and interspecific comparisons of

TrN genetic distance for O. tetricus (including O. gibbsi) and O. cf. tetricus showed that mean interspecific divergence (3.34%) was approximately 17.5 times greater than mean intraspecific divergence (0.19%).

Timing of divergence: Based on TrN distances, a divergence date of ~3.2-6.9 million years ago (ma) was estimated between O. tetricus (including O. gibbsi) and O. cf. tetricus (Table S13). Furthermore, the Australian O. tetricus complex and the Japanese/Chinese O. vulgaris clade were estimated to have diverged

~5.4-11.6 ma (Table S14).

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 (O. tetricus) and New Zealand (O. gibbsi) comprised a single monophyletic clade, whilst the second monophyletic clade was composed entirely of individuals from Western Australia (Fig. 3). A third clade was supported by GMYC analysis and was composed of a single individual from

Western Australia. However, this third clade was paraphyletic and formed a monophyletic clade with all other individuals from Western Australia.

Fig. 3: Generalised Mixed Yule Coalescent (GMYC) Bayesian topology depicting the phylogenetic relationships of Octopus tetricus (east Australia and Tasmania),

O. cf. tetricus (Western Australia) and O. gibbsi (New Zealand). Analysis was based on concatenated partial sequence data of three mtDNA genes (COI, COIII and Cytb). Three species/clades were supported via GMYC analysis; East

Australia and New Zealand (red) and west Australia (purple and black). Node labels reflect locations represented by individuals contributing to node (West

Australia, 1 = Mandurah, 2 = Woodman’s Point, 3 = Albany, 4 = Cape Le Grand,

5 = Esperance; East Australia, 1 = Wallaga Lake, 2 = Port Stephens, 3 =

Narooma).

2.4.2 Morphological analyses

Males: 17 morphological traits were recorded from 36 mature male octopods

(Table S15). 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 model was run without the interaction. A significant difference was recorded among four coasts for the multivariate model based upon 16 (and MLd as co-variate) morphological traits (Pillai Trace = 2.070, F =

2.503, df = 48,54, p = 0.001). Visualisation of the male PC biplot showed individuals from the east coast of the Australian mainland, Tasmania and New

Zealand could not be distinguished, while west Australian individuals formed a group, distinct from eastern Australian and New Zealand individuals (Fig. 4).

Individuals from west Australia were characterised as having greater SCR3 and

ALR3 (PC1) in comparison to individuals from the east Australian mainland,

Tasmania and New Zealand. No distinction based upon WD and HW among locations was recorded (PC2).

Fig. 4: PC biplot of male O. tetricus complex individuals. X axis represents PC1

(explaining 73.6% of total variation) and is driven primarily by the SCR3 and

ALR3. Y axis represents PC2 (explaining 13.7% of total variation) and is driven primarily by WD and HW.

DFA showed a significant difference among individuals from the east Australian mainland, 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 west Australia to a single group comprised solely of west

Australian individuals (Table 2). DFA assigned 83% (n = 15) of east Australian

33 individuals to the east Australian group, with 17% (n = 3) allocated to the

Tasmanian group. Furthermore, 88% (n = 7) of Tasmanian individuals to the

Tasmanian group, whilst 12% (n = 1) were assigned to the east Australian group.

All individuals from New Zealand (n = 3) were allocated into the east Australian group.

Table 2: Male Discriminant Function Analysis: Jackknifed classification matrix.

East Australia Tasmania New Zealand Western Australia % correct

East Australia 15 0 3 0 83

Tasmania 3 0 0 0 0

New Zealand 1 0 7 0 88

Western Australia 0 0 0 7 100

Ranking of CLs determined male SCR3 to contribute the most variation among groups (Table S8). 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), whilst 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-

34 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: Ten morphological traits were recorded from 25 mature female octopods (Table S16). 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 (Pillai Trace =

0.122, F = 1.989, df = 18, 28, p = 0.05). 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 (Fig. 5).

Fig. 5: Principal component biplot of female individuals of the O. tetricus complex. X axis represents PC1 (explaining 70.7% of total variation) and is driven primarily by HW and SCL3. Y axis represents PC2 (explaining 29.3% of total variation) and is driven primarily by SCR3 and ALL3.

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 3). 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.

Table 3: Female Discriminant Function Analysis: Jackknifed classification matrix.

East Australia Tasmania Western Australia % correct

East Australia 13 0 1 93

Tasmania 1 2 0 67

Western Australia 4 1 3 38

2.5 Discussion

Species level relationships: The main focus of this study was to resolve the taxonomic status of the Australasian O. tetricus complex. Molecular and morphology-based results are consistent with the hypothesis that disjunct populations of O. tetricus from Australia’s east coast (including Tasmania), and from Western Australia are distinct species. In addition, findings of this study support the hypothesis that O. gibbsi of New Zealand is synonymous with east

Australian O. tetricus (Norman & Hochberg, 2005). Therefore, we consider O. gibbsi to be a junior synonym of O. tetricus, as it will hereafter be referred to.

In the present study, interspecific variation of COI between eastern O. tetricus and west Australian 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., (2004). 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 (Strugnell et al., 2008a; Undheim et al., 2010). For example, interspecific variation was found to be 1-2% and 2-3.3% for the octopod genera Pareledone (Allcock et al., 2007) and Thaumeledone (Strugnell et al., 2008a) 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., (Undheim et al., 2010) for O. vulgaris. Low nucleotide divergence between octopod species in this, and previous, studies contrasted higher levels of divergence recorded among moths, butterflies and birds, which range from 5.8-9.1% (Moore, 1995; Hebert et al.,

2003; Hebert et al., 2004).

The result of the GMYC analysis suggests west Australian O. cf. tetricus is a distinct species from O. tetricus, and supported the synonymy of O. gibbsi with

O. tetricus. However, the support for a second cryptic Western Australian species incongruent with the phylogenetic and morphology-based results of this study, which show no such cryptic speciation. One explanation for this result is the tendency for GMYC analyses to ‘over-split’ taxa, which may result from gaps in knowledge (i.e. more species exist than is currently known) or real failures of

GMYC analyses (Talavera et al., 2013).

Talavera et al., (2013) investigated the ability of GMYC analyses 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. In the present study, no evidence for a second cryptic west Australian species was detected by phylogenetic or morphology-based analyses. Furthermore, as interspecific variability between O. cf. tetricus and O. tetricus was greater relative to the low intraspecific variability within each individual group, the discovery of a second cryptic Western Australian species is considered likely to be an artefact of ‘over-splitting’ by GMYC analysis.

Multivariate morphology-based analyses showed congruence in detecting significant differences between individuals from east Australia/New Zealand and west Australia; although female morphology was less reliable for discriminating species. Male morphology was able to successfully delimit O. tetricus and O. cf. tetricus. Sucker numbers on males’ third right arm (hectocotylus) contributed the greatest amount of variation between O. tetricus and O. cf. tetricus, with O. cf.

38 tetricus having significantly greater sucker numbers. The hectocotylus is used by males to pass sperm to females during mating. The ligula and calamus, unique to this modified arm, provide a limit to the emergence of new suckers at a relatively early stage of ontogeny (Toll, 1988). Toll (1988) investigated sucker counts on the hectocotylus (HASC) among 12 species of the sub-family

Octopodinae, and demonstrated its value in identification and delimitation of otherwise morphologically similar octopods. Sucker numbers on the hectocotylised arm were considered 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 (Toll, 1988) or environmental influence further reinforces the usefulness of male HASC in cryptic cephalopod taxonomy.

Biogeographic factors: Speciation between O. 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 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

39 western coasts, isolating populations which allowed for genetic differentiation to commence (Wilson & Allen, 1987).

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

(O'Hara & Poore, 2000; O'Loughlin et al., 2003), and gastropods (Dartnall, 1974;

Waters et al., 2005). 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 (Anderson, 1997), 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 O. tetricus populations suggests ongoing gene flow across the

Tasman Sea; a 2000 km wide marine body separating the two landmasses. Due to the benthic shallow-water habit of O. tetricus adults (Norman, 2000), connectivity between New Zealand and east Australian populations is likely attributable to trans-Tasman dispersal during the planktonic paralarval stage; although adults of the genus Octopus can raft on floating wood or drifting

40 macroalgae (Thiel & Gutow, 2005), 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 (Brasher et al., 1992; Ovenden et al., 1992; Booth & Ovenden, 2000) and morwong (cheilodactylid) fishes (Grewe et al., 1994; Burridge & Smolenski,

2003). Planktonic larval durations (PLD) for the Octopodinae appear much shorter (35-60 days; reviewed in Villanueva 1995) than those of the lobster J. edwardsii (2 years; Booth & Phillips, 1994) and cheilodactylid fishes (1 year;

Burridge, 1999). Octopus paralarvae appear to be active and often constant swimmers (Joll, 1977; Villanueva & Norman, 2008), 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 (2003). Octopod paralarvae have been observed rafting on macroalgal and other drift debris (Smale & Buchan, 1981), which may function as habitat for post-settlement juveniles until arrival at suitable shallow-water habitat.

Alternatively, paralarvae of some octopods can delay settlement in the absence of suitable habitat (Strugnell et al., 2004). These ‘super-paralarvae’ obtain larger sizes and more developed swimming capabilities, while retaining paralarval morphological characters (reviewed in Villanueva & Norman, 2008), and may facilitate trans-Tasman dispersal for O. 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 O. 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 (Ridgway, 2007).

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 (REDMAP, 2013).

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 (Ling et al., 2009).

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 O. vulgaris, supporting previous speculations that these taxa are closely related (Norman & Kubodera, 2006). 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 42 latitudes (Frakes et al., 1987); 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 (Valentine, 1984; White, 1986; Burridge & White, 2000). In addition, anti- tropical affinities between other subtropical Australasian-Japanese/Asian octopods have also 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. 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 O. vulgaris complex revealed in this study directly questions the purported cosmopolitan distribution of O. vulgaris, and supports hypotheses regarding the existence of numerous cryptic O. vulgaris-like species (Norman & Hochberg, 2005; Norman & Kubodera, 2006; Teske et al.,

2007). Norman and Kubodera (2006) previously suggested the possibility of an

Asian O. 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 O. 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

43 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 O. tetricus species complex. Molecular and morphological results support east Australian O. 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 O. vulgaris complex revealed in this study adds support to hypotheses regarding the existence of numerous cryptic O. vulgaris-like species, warranting taxonomic revision of the O. vulgaris species complex to aid in the management of this significant global marine resource.

2.6 Supplementary information

Table S1: Specimen information for individuals sequenced in the present study.

Sample Species Location Region 12S 16S COI COIII Cytb

NQ001 O.tetricus NSW Wallaga Lake * * * *

NQ002 O.tetricus NSW Wallaga Lake * * * *

NQ003 O.tetricus NSW Wallaga Lake * * * *

NQ010 O.tetricus NSW Wallaga Lake * * * *

NQ011 O.tetricus NSW Wallaga Lake * * * *

NQ015 O.tetricus NSW Narooma * * * *

NQ028 O.tetricus NSW Shoal Bay, Port Stephens * * *

NQ029 O.tetricus NSW Shoal Bay, Port Stephens * * * *

TAS4113 O.tetricus Tasmania Flinders Island * *

TAS4123 O.tetricus Tasmania Flinders Island * * *

TAS4124 O.tetricus Tasmania Flinders Island * *

TAS4126 O.tetricus Tasmania Flinders Island * *

TAS4132 O.tetricus Tasmania Flinders Island *

LML1 O. gibbsi NZ Leigh Marine Lab * * * * *

LML4 O. gibbsi NZ Leigh Marine Lab * * * * *

LML8 O. gibbsi NZ Leigh Marine Lab * * * * *

NZ1 O. gibbsi NZ Leigh Marine Lab * * * * *

ct123 O.cf. tetricus WA Woodmans Point * * * * *

ct133 O.cf. tetricus WA Town Jetty, Albany * * * * *

SWA006 O.cf. tetricus WA Lucky Bay, Cape Le Grand * * *

SWA007 O.cf. tetricus WA Lucky Bay, Cape Le Grand * * * *

SWA008 O.cf. tetricus WA Lucky Bay, Cape Le Grand * * *

SWA009 O.cf. tetricus WA Esperance * * * *

SWA010 O.cf. tetricus WA Esperance * * * *

WAM6701 O.cf. tetricus WA Mandurah * *

WAM6702 O.cf. tetricus WA Mandurah * *

WAM6703 O.cf. tetricus WA Mandurah * *

WAM6704 O.cf. tetricus WA Mandurah * * * *

WAM6705 O.cf. tetricus WA Mandurah *

WAM6706 O.cf. tetricus WA Mandurah * * * *

WAM6707 O.cf. tetricus WA Mandurah * * * *

WAM6708 O.cf. tetricus WA Mandurah * *

WAM6709 O.cf. tetricus WA Mandurah * *

WAM6710 O.cf. tetricus WA Mandurah * * * *

SAVULG01 O.vulgaris South Africa Port Elizabeth *

OVAL1 O.vulgaris Spain Perpignan * * *

OVAL2 O.vulgaris Spain Perpignan * * *

OVAL3 O.vulgaris Spain Perpignan * * *

OVAL4 O.vulgaris Spain Perpignan * * *

OVAL5 O.vulgaris Spain Perpignan * * *

Locations – NSW = New South Wales, NZ = New Zealand, WA = Western Australia 45

Table S2: Specimen information for individuals accessed via GenBank for use in the present study.

Accession # Species Location Gene(s) AJ390318 O. mimus Chile 16S AJ012128 O. mimus Chile COIII AJ250480 O. mimus Costa Rica 16S AJ390319 O. mimus Costa Rica COIII HQ846021 O. vulgaris China 16S HQ846110 O. vulgaris China COI HQ846061 O. vulgaris China 16S HQ846154 O. vulgaris China COI NC006353 O. vulgaris Japan 12S, 16S, COI, COIII, Cytb AB430546 O. vulgaris Japan COI AB573217 O. vulgaris Japan COIII AB430547 O. vulgaris Japan COI AB573219 O. vulgaris Japan COIII AB430548 O. vulgaris Japan COI AB573218 O. vulgaris Japan COIII FN424379 O. vulgaris St Paul and Amsterdam Islands COI FN424382 O. vulgaris St Paul and Amsterdam Islands COIII FN424380 O. vulgaris St Paul and Amsterdam Islands COI FN424383 O. vulgaris St Paul and Amsterdam Islands COIII FN424381 O. vulgaris St Paul and Amsterdam Islands COI FN424384 O. vulgaris St Paul and Amsterdam Islands COIII AJ628241 O. vulgaris South Africa COIII AJ628204 O. vulgaris South Africa Cytb DQ683234 O. vulgaris South Africa, Durban 16S DQ683214 O. vulgaris South Africa, Durban COI DQ683235 O. vulgaris South Africa, Durban 16S DQ683215 O. vulgaris South Africa, Durban COI DQ683236 O. vulgaris South Africa, Durban 16S DQ683216 O. vulgaris South Africa, Durban COI DQ683237 O. vulgaris South Africa, Durban 16S DQ683217 O. vulgaris South Africa, Durban COI DQ683238 O. vulgaris South Africa, Durban 16S DQ683218 O. vulgaris South Africa, Durban COI DQ683239 O. vulgaris South Africa, Durban 16S DQ683219 O. vulgaris South Africa, Durban COI DQ683247 O. vulgaris Spain, Galicia 16S DQ683221 O. vulgaris Spain, Galicia COI DQ683230 O. vulgaris South Africa, Hout Bay 16S DQ683208 O. vulgaris South Africa, Hout Bay COI DQ683228 O. vulgaris South Africa, Port Elizabeth 16S DQ683212 O. vulgaris South Africa, Port Elizabeth COI DQ683232 O. vulgaris South Africa, Struisbaai 16S DQ683210 O. vulgaris South Africa, Struisbaai COI DQ683244 O. vulgaris Africa, Senegal 16S DQ683224 O. vulgaris Africa, Senegal COI DQ683241 O. vulgaris Tristan da Chuna 16S DQ683205 O. vulgaris Tristan da Chuna COI DQ683240 O. vulgaris South Africa, Umhlanga 16S DQ683220 O. vulgaris South Africa, Umhlanga COI

Table S3: Specimen information for individuals where morphological traits were recorded during the present study.

Catalogue # Institution Species Coast Location Latitude, longitude

C126244 AM O. tetricus East Nelson Head, Port Stephens, NSW -32.716667, 152.166667

C156208 AM O. tetricus East Shelly Beach, North Manly, NSW -33.800000, 151.300000

C171669 AM O. tetricus East Parsley Bay, NSW -35.866667, 151.283333

C171685 AM O. tetricus East Woody Head, Iluka, NSW -29.358015, 153.354721

F78082 AM O. tetricus East Merewether, NSW -32.98333, 151.7833300

F78281a AM O. tetricus East Newcastle, NSW -32.916667, 151.950000

F78283b AM O. tetricus East Newcastle, NSW -32.866667, 152.016667

F160334 MV O. tetricus East Wallaga Lake, NSW -36.350000, 150.050000

F182057 MV O. tetricus East Narooma Inlet, NSW -36.218238, 150.132300

F182058 MV O. tetricus East Narooma Inlet, NSW -36.218238, 150.132300

F200317 MV O. tetricus East Wreck Bay, NSW -34.200000, 150.716667

F200318a MV O. tetricus East Wreck Bay, NSW -35.216667, 150.716667

F200318b MV O. tetricus East Wreck Bay, NSW -35.216667, 150.716667

F200318c MV O. tetricus East Wreck Bay, NSW -35.216667, 150.716667

F200319 MV O. tetricus East Long Reef, Sydney, NSW -33.733333, 151.316667

F200320 MV O. tetricus East Merimbula Harbour, NSW -36.883333, 149.916667

F200321 MV O. tetricus East Shelly Beach, North Manly, NSW -33.800000, 151.300000

F200323 MV O. tetricus East Shelly Beach, North Manly, NSW -33.800000, 151.300000

F200324 MV O. tetricus East Long Reef, Sydney, NSW -33.733333, 151.316667

F77273 MV O. tetricus East Newcastle, NSW -32.966667, 151.783333

F77274 MV O. tetricus East Tathra, NSW -36.616667, 150.066667

F78281(B) MV O. tetricus East Newcastle, NSW -32.933333, 151.950000

F78283c MV O. tetricus East Newcastle, NSW -32.866667, 152.016667

F80438 MV O. tetricus East Amity Point, North Stradbroke Island, QLD -27.398309, 153.442056

F80439 MV O. tetricus East Ned’s Beach, Lord Howe Island, NSW -31.524986, 159.060767

F80440 MV O. tetricus East North Stradbroke Island, QLD -27.668056, 153.484722

F80442a MV O. tetricus East Sydney, NSW -33.867487, 151.206990

F80442b MV O. tetricus East Sydney, NSW -33.867487, 151.206990

F80445 MV O. tetricus East Potters Point, Karnell, NSW -34.045000, 151.211667

F80446 MV O. tetricus East Potters Point, Karnell, NSW -34.040000, 151.211667

F85370a MV O. tetricus East Wreck Bay, NSW -35.200000, 150.733333

F180696 MV O. tetricus TAS Flinders Island, TAS -39.772767, 148.470439

F180697 MV O. tetricus TAS Flinders Island, TAS -39.772767, 148.470437

F180698 MV O. tetricus TAS Flinders Island, TAS -39.772767, 148.470440

F180699 MV O. tetricus TAS Flinders Island, TAS -39.772767, 148.470433

F180700 MV O. tetricus TAS Flinders Island, TAS -39.772767, 148.470435

F180701 MV O. tetricus TAS Flinders Island, TAS -39.772767, 148.470441

F180702 MV O. tetricus TAS Flinders Island, TAS -39.772767, 148.470444

F180704 MV O. tetricus TAS Flinders Island, TAS -39.772767, 148.470438

F180705 MV O. tetricus TAS Flinders Island, TAS -39.772767, 148.470443

F180706 MV O. tetricus TAS Flinders Island, TAS -39.772767, 148.470436

F180707 MV O. tetricus TAS Flinders Island, TAS -39.772767, 148.470442

310 6-83-1 AM O. cf. tetricus West Fathom Bank, off Garden Island, WA -32.242955, 115.698630

F160302 MV O. cf. tetricus West Busselton Jetty, WA -33.650000, 155.333333

F160306 MV O. cf. tetricus West Esperance boat wharf, WA -33.000000, 121.000000

F160320 MV O. cf. tetricus West Esperance tanker jetty, WA -33.868611, 121.903889

F160321 MV O. cf. tetricus West Esperance tanker jetty, WA -33.868611, 121.903889

F160325 MV O. cf. tetricus West Princess royal harbour, Albany, WA -35.075000, 117.925000

F200326a MV O. cf. tetricus West Busselton jetty, WA -33.650000, 155.333333

F200327 MV O. cf. tetricus West Lucky bay, Cape Le Grand National Park, WA -33.970413, 122.269592

F200328 MV O. cf. tetricus West Peaceful bay, 30 km East of Walpole WA -35.041944, 116.930752

F200329 MV O. cf. tetricus West Rockingham Grain Jetty, WA -32.255945, 115.751492

F200330 MV O. cf. tetricus West Rockingham Grain Jetty, WA -32.255945, 115.751492

F200331 MV O. cf. tetricus West Esperance tanker jetty, WA -33.868611, 121.903889

F200334 MV O. cf. tetricus West Town jetty, Albany, WA -35.030625, 117.886519

F80447 MV O. cf. tetricus West Woodmans Point, Perth, WA -32.125398, 115.758562

Institutions – AM = Australian Museum, Sydney, MV = Museum Victoria

Locations – NSW = New South Wales, QLD = Queensland TAS = Tasmania, WA = Western Australia

Table S4: Canonical correlation (CC) output for male octopod multivariate analysis.

CC Value

1 0.943

2 0.774

3 0.762

Table S5: Canonical loadings (CL) output for male octopod multivariate analysis.

Trait CL1 CL2 CL3

MW 0.073 0.07 -0.07

HW 0.196 0.259 -0.131

AW 0.14 0.238 -0.371

SDn 0.05 0.121 -0.304

WD 0.181 0.353 -0.152

ALL3 0.058 -0.217 0.425

ALR3 0.468 -0.274 -0.16

LSDL2 -0.055 -0.014 -0.236

LSDL3 -0.119 -0.097 -0.305

LSDR2 -0.158 0.025 -0.302

LSDR3 -0.139 -0.014 -0.382

SCL3 0.256 -0.103 0.312

SCR3 0.686 -0.033 0.435

LL 0.068 -0.287 -0.252

CL -0.017 -0.233 -0.09

TOL 0.096 -0.165 0.017

Table S6: Principal components (PC) calculated from canonical correlation and canonical loading outputs from male octopod multivariate analysis.

Catalogue number Location PC1 PC2 PC3

C126244 East Australia 1.398564154 -0.063633217 -1.82498345

C171685 East Australia 1.656582137 -0.099131475 -1.168465778

F160334 East Australia 1.574527919 0.363012939 -1.085086723

F77273 East Australia 1.685109747 -0.300965674 -0.96476396

F77274 East Australia 1.568450827 -0.501585849 -1.131766839

F78281(B) East Australia 1.351409537 0.050502997 -1.693015773

F80438 East Australia 1.710610872 -0.313058649 -1.481486248

F80439 East Australia 1.622998216 -0.385155637 -0.850786645

F80440 East Australia 1.818444188 -0.611405995 -1.468833801

F80445 East Australia 1.467906926 -0.369658284 -1.688319729

F200319 East Australia 1.790746655 -0.313842653 -1.63515882

F200324 East Australia 1.592004277 -0.221444375 -1.720480192

F200323 East Australia 1.522560376 -0.168410585 -1.272437234

F200321 East Australia 1.362814495 -0.009377961 -1.689063827

F182058 East Australia 1.912142313 -0.183517245 -2.880150654

F182057 East Australia 1.916011784 -0.192024188 -2.626587877

F200317 East Australia 1.670192661 -0.389328693 -3.064599549

F200318b East Australia 1.679143563 -0.506436436 -1.140385436

F180706 Tasmania 1.710481088 -0.548840695 -1.922252402

F180698 Tasmania 1.573629475 -0.601183599 -1.217745373

F180707 Tasmania 1.893463296 -0.91658773 -2.227068944

F180699 Tasmania 1.547096979 -0.47192172 -1.45772354

F180700 Tasmania 1.661304028 -0.506232288 -1.473664311

F180697 Tasmania 1.656125045 -0.68637114 -1.317081058

F180696 Tasmania 1.396853251 -0.53629949 -1.298398701

F180702 Tasmania 1.476770081 -0.546380117 -1.386584132

NMNZM.118421 New Zealand 2.025951471 -0.327920877 -1.827612087

NMNZM.118305 New Zealand 1.830826997 -0.206752673 -1.595982188

NMNZM.118425 New Zealand 1.505001738 -0.350295695 -1.179311946

310 6-83-1 Western Australia 2.472296823 -0.611333741 -1.313158431

F200330 Western Australia 2.385776448 -0.391990469 -1.225903028

F160306 Western Australia 2.238991724 -0.444423706 -1.501335696

F200327 Western Australia 2.244534044 -0.520498612 -1.278073611

F200329 Western Australia 2.164884583 -0.479896537 -1.694166644

F200328 Western Australia 2.795060035 -0.557242263 -1.37264284

F200326a Western Australia 2.39664783 -0.466762457 -1.718671345

Table S7: Eigenvalues and principal component (PC) variance contribution outputs from male octopod multivariate analysis.

Eigenvalue PC contribution (%)

1 8.029263844 73.6

2 1.49423831 13.7

3 1.384608781 12.7

Total 10.90811093 100

Table S8: Ranked canonical loadings (CL) from male octopod multivariate analysis; based upon contribution to principal components (PC).

Trait PC contribution Rank

SCR3 50.95 1

ALR3 38.20 2

SCL3 20.25 3

WD 18.16 4

HW 17.98 5

AW 13.57 6

SDeR2 11.97 7

SDeR3 10.42 8

SDeL3 10.09 9

TOL 9.33 10

LL 8.94 11

ALL3 7.24 12

MW 6.33 13

SDn 5.34 14

CL 4.44 15

SDeL2 4.24 16

Table S9: Canonical correlation (CC) output for female octopod multivariate analysis.

CC Value

1 0.817

2 0.674

Table S10: Canonical loadings (CL) output for female octopod multivariate analysis.

Trait CL1 CL2

MW 0.098 0.067

HW 0.553 0.086

AW 0.176 -0.006

SDn 0.177 0.158

WD 0.297 -0.33

ALL3 0.31 0.225

ALR3 0.018 -0.071

SCL3 0.493 0.037

SCR3 0.118 -0.315

Table S11: Principal components (PC) calculated from canonical correlation and canonical loading outputs from female octopod multivariate analysis.

Catalogue number Location PC1 PC2

C156208 East Australia 2.033727 -0.17161

C171669 East Australia 2.105987 -0.15068

F78082 East Australia 2.068866 -0.24301

F78281a East Australia 2.233548 -0.12304

F78283b East Australia 1.930682 -0.27489

F80442a East Australia 2.057868 -0.14028

F80442b East Australia 2.301519 -0.14622

F80446 East Australia 1.697702 -0.12759

F85370a East Australia 2.169133 -0.19117

F200320 East Australia 2.247057 -0.17498

F200319 East Australia 2.738278 -0.17541

F200318a East Australia 2.211669 -0.14806

F200318c East Australia 1.685601 -0.10294

F78283c East Australia 2.07602 -0.26773

F180704 Tasmania 2.186593 -0.12979

F180701 Tasmania 2.216311 -0.01233

F180705 Tasmania 2.93645 0.054955

F200334 Western Australia 2.967886 -0.16133

F160320 Western Australia 2.371038 -0.08675

F160321 Western Australia 2.548854 -0.23915

F160325 Western Australia 2.2995 -0.21815

F80447 Western Australia 2.241005 -0.05659

F200327 Western Australia 2.365949 -0.21258

F200331 Western Australia 2.799516 -0.14644

F160302 Western Australia 2.37516 -0.14816

Table S12: Eigenvalues and principal component (PC) variance contribution outputs from female octopod multivariate analysis.

Eigenvalue PC contribution (%)

1 2.007419303 70.7 2 0.832428114 29.3 Total 2.839847417 100

Table S13: Timing of divergence estimates (Tamura-Nei genetic distance) for

Octopus tetricus (East Australia and New Zealand) and O. cf. tetricus (Western

Australia).

O. tetricus O. cf. tetricus Divergence (million years) - + O. tetricus 0.0021 0.0336 4.4 3.2 6.9 O. cf. tetricus 0.0336 0.0018

Table S14: Timing of divergence estimates (Tamura-Nei genetic distance) for the

Australasian tetricus complex and Japanese/Chinese representatives of the

Octopus vulgaris group.

Japan/China Australasia Divergence (million years) - + Japan/China 0.0016 0.0570 7.4 5.4 11.6 Australasia 0.0570 0.0019

Table S15: Summary of male morphological measurements taken from preserved museum specimens of Octopus tetricus (east Australia and Tasmania), O. cf. tetricus (Western Australia and O. gibbsi (New Zealand).

Catalogue # Institution Location MLd MW HW AW SDn WD ALL3 ALR3 SDeL2 SDeL3 SDeR2 SDeR3 SCL3 SCR3 LL CL TOL C126244 AM East Australia 69.5 67.2 46.9 23.9 12.3 93 172 193 17.5 16.1 16.6 15.8 162 135 3.6 1.7 11.7 C171685 AM East Australia 65 67.2 46.9 23.9 12.3 93 279 249 11 11.3 11.3 10.9 220 136 2.9 1.5 9.7 F160334 MV East Australia 54.3 67.2 46.9 23.9 12.3 93 147 163 17.5 8.2 7.8 7.4 187 143 1.5 0.6 7.5 F77273 MV East Australia 80.4 49.9 44.6 19.3 9.7 66 311 266 12.1 11.8 12.1 11.5 211 158 3.1 1.2 11.3 F77274 MV East Australia 86.1 59.6 31.4 15.9 10 86 340 284 13.6 11.9 12.4 13 197 133 3.9 1.7 14 F78281b AM East Australia 63.7 64.3 45.2 24.4 12.3 91 227 198 17.5 16.1 16.6 15.8 159 128 2.7 1.1 10.7 F80438 MV East Australia 89.6 58.7 43.1 24 9.8 99 307 285 17.5 16.1 16.6 15.8 242 153 3.7 1.7 12.9 F80439 MV East Australia 69.9 55.3 37.8 16.2 9.7 63 298 242 12.7 11.4 10.9 10.7 234 154 3 1.5 10.8 F80440 MV East Australia 110 66.6 40.9 21.3 12.1 101 435 370 14.7 14.9 14.6 15 190 144 4.6 2.3 15.9 F80445 MV East Australia 80.1 53.5 41.7 23.5 11.7 73 289 287 18.5 17.5 17.3 16.4 217 121 4.1 1.3 13.1 F200319 MV East Australia 93.6 55.2 48.7 30.1 13.6 78 365 333 17.6 16.5 17 16.1 217 153 3.5 1.5 15.1 F200324 MV East Australia 86.7 59.5 44.1 25.5 12.2 88 289 307 18.5 17.8 19.6 17.2 217 138 3 1.3 14.7 F200323 MV East Australia 85.6 51.8 39 20.5 12.1 74 289 260 14.9 14.5 16.2 14.2 217 139 2.6 0.7 12.8 F200321 MV East Australia 61.7 64.3 45.2 24.4 12.3 91 211 181 17.5 16.1 16.6 15.8 201 127 3 1.5 9.4 F182058 MV East Australia 121 88.1 65.5 40 17.3 114 289 426 25.7 24.7 24.7 25.7 217 143 4.2 2.1 19.7 F182057 MV East Australia 113.7 100.5 64.3 35.4 16.5 113 289 390 22.4 22.6 24.7 21.7 217 150 4.8 2.1 16 F200317 MV East Australia 122.4 80.6 52.9 28.6 16.4 142 399 456 31.9 26.8 30.2 29.1 142 142 5.2 2.3 14.6 F200318b MV East Australia 84.4 57.1 34.1 17.4 9.6 81 345 292 14.1 16.1 14.4 13.1 220 156 3.9 1.3 15.5 F180706 MV Tasmania 113.2 82.7 58.7 22.9 14 82 470 365 18.7 19.5 23 19.7 195 136 4.5 2 18.1 F180698 MV Tasmania 101 70.9 37.5 19.5 10.9 72 525 287 17.6 16.5 17 16.1 217 140 3.4 1.8 15 F180707 MV Tasmania 134.9 71 52.5 29.2 16.3 81 580 468 21.1 25.6 23 22.3 224 146 5.7 2.8 17.6 F180699 MV Tasmania 90.8 54 34.6 17.7 11.7 81 315 301 17.6 16.5 17 16.1 203 143 3.6 1.5 15.6 F180700 MV Tasmania 110.9 67.8 37.5 22 11.1 86 408 338 17.6 16.5 17 16.1 212 139 3.8 1.7 14.6 F180697 MV Tasmania 100.8 56.2 30.2 17.2 9.9 86 403 360 17.6 16.5 17 16.1 218 146 3.4 2.1 16.7 F180696 MV Tasmania 92.5 56.8 40.7 16.8 9.9 61 413 251 17.6 16.5 17 16.1 201 132 3.7 1.6 12.3 F180702 MV Tasmania 85 58.4 34.4 16.1 9.2 68 267 268 17.6 16.5 17 16.1 231 143 3.9 1.7 13.8 NMNZM.118421 O’Shea (1999) New Zealand 135.5 97 62.5 23.9 12.3 142 532 418 17.6 16.5 21.3 21.3 217 150 4.1 2.2 14.6 NMNZM.118305 O’Shea (1999) New Zealand 121.5 55.5 52 23.9 12.3 107 393 360 17.6 16.5 22.9 17 217 162 3 1.2 14.6 NMNZM.118425 O’Shea (1999) New Zealand 89.5 59.5 37.2 23.9 12.3 62 365 192 17.6 16.5 11.5 11.6 217 142 3 1.5 14.6 310 6-83-1 AM Western Australia 90 74.6 56.1 27.2 12.9 110 557 469 17.3 15.8 16 14.9 257 201 4.5 1.6 24.3 F200330 MV Western Australia 95.8 79.1 53.9 23.7 11.2 114 420 444 16.7 15.9 15.1 14.3 281 209 3.5 1.7 14.3

F160306 MV Western Australia 91.7 76.1 54.5 26.3 14.3 115 458 421 17.3 15.8 15.4 14.9 250 177 4.2 1.9 18.7 F200327 MV Western Australia 111.7 66.6 52.2 22.9 12.6 98 458 426 17.3 15.8 15.4 14.9 283 192 4.4 1.6 13.5 F200329 MV Western Australia 114.2 74.6 56.4 30.3 13.1 96 381 384 17.9 15.6 15 15.5 230 177 4.6 2.4 14.7 F200328 MV Western Australia 163.4 83.5 65.8 34.3 14.1 107 559 544 17.3 15.8 15.4 14.9 291 218 3.9 1.7 26.8 F200326a MV Western Australia 127 68 54 25.7 12.4 127 365 520 17.6 16.5 17 16.1 217 207 5 2.1 11.9

Missing data (shown in bold) has been replaced by the global mean of the respective trait. Institutions; AM = Australian Museum, Sydney, MV = Museum Victoria.

Table S16: Summary of female morphological measurements taken from preserved museum specimens of Octopus tetricus (east

Australia and Tasmania), O. cf. tetricus (Western Australia and O. gibbsi (New Zealand).

Catalogue # Institution Location MLd MW HW AW SDn WD ALL3 ALR3 SCL3 SCR3

C156208 AM East Australia 67.6 65.4 44.6 20.7 12.2 84 212 232 197 194

C171669 AM East Australia 62.1 65.4 44.6 20.7 12.2 84 240 208 218 199

F78082 AM East Australia 117.8 65.4 31.6 19.5 12 96 315 335 218 229

F78281a AM East Australia 70.3 65.4 44.6 20.7 12.2 84 323 209 235 222

F78283b AM East Australia 53.2 65.4 44.6 20.7 12.2 84 158 327 163 222

F80442a MV East Australia 96 64.9 34.7 16.8 9.5 69 358 356 247 234

F80442b MV East Australia 126.3 74.8 38.1 16.9 11.8 88 452 418 238 254

F80446 MV East Australia 60.3 45.2 34.4 14.1 8.5 59 230 216 201 183

F85370a MV East Australia 100 65.4 44.6 20.7 12.2 84 287 400 216 221

F200320 MV East Australia 120.2 65.4 44.6 20.7 12.2 84 323 435 235 224

F200319 MV East Australia 104.5 80.1 52.2 28.8 15.2 100 437 409 265 290

F200318a MV East Australia 104.8 70.1 43.1 21 14.1 86 311 356 214 222

F200318c MV East Australia 84.4 47.7 31.5 13.4 9 52 259 243 204 198

F78283c AM East Australia 52.6 65.4 44.6 20.7 12.2 84 152 327 235 222

F180704 MV Tasmania 102.2 64.9 42.8 20 12.5 74 323 369 235 230

F180701 MV Tasmania 99.7 65.8 43.8 19.5 12.1 72 418 367 216 196

F180705 MV Tasmania 135.4 88.7 57.5 25.7 17.2 99 524 327 302 222

NMNZM.90320 O’Shea (1999) New Zealand 48 35.7 30 20.7 6 45 141 313 206 194

NMNZM.131569 O’Shea (1999) New Zealand 48.9 28 27 20.7 5.9 40 283 141 235 145

NMNZM.118426 O’Shea (1999) New Zealand 137 90.5 55 20.7 16.8 112 424 485 235 243

F200334 MV Western Australia 114.1 97 61.5 31.4 15 139 474 327 235 222

F160320 MV Western Australia 84.2 59.2 44.6 20 20 75 304 327 264 264

F160321 MV Western Australia 90.9 64.6 53.9 19.5 11.2 101 396 438 264 264

F160325 MV Western Australia 97.3 59.3 48 17.5 9.7 82 350 355 257 266

F80447 MV Western Australia 87.4 69.3 45.2 19.6 11.5 79 356 322 241 186

F200327 MV Western Australia 88.3 64.3 51.9 20.4 12 82 240 277 290 253

F200331 MV Western Australia 119.6 69 57.4 25.7 13.2 119 457 327 276 222

F160302 MV Western Australia 71.4 68.9 51.8 22 13.2 97 302 305 236 200

Missing data (shown in bold) has been replaced by the global mean of the respective trait. Institutions; AM = Australian Museum, Sydney, MV = Museum Victoria.

3. Morphological assessment of the Octopus vulgaris species complex evaluated in light of molecular-based phylogenetic inferences.

This chapter is published as:

Amor, M. D., Norman, M. D., Roura, A., Leite, T. S., Lu, C. C., Reid, A. L.,

Hochberg, F. G., Perales-Raya, C., Gleadall, I. G., Zheng, X. D., Silvey, C., &

Strugnell, J. M. (2016). Morphological assessment of the Octopus vulgaris species complex evaluated in light of molecular-based phylogenetic inferences.

Zoologica Scripta. doi:10.1111/zsc.12207/abstract

3.1 Abstract

Cryptic species are common in the ocean, particularly among marine invertebrates such as octopuses. Delineating cryptic species is particularly problematic in octopus taxonomy where the plasticity recorded among taxonomic characters often results in low resolution at the species level. This study investigated the morphological relationships among seven phylogenetic clades (identified using cytochrome c oxidase subunit I) of the broadly distributed Octopus vulgaris species complex and close relatives. Morphological analyses in the present study were successful in delimiting O. sinensis,

Brazilian O. vulgaris and O. vulgaris sensu stricto, which was congruent with the molecular findings of this study. Analyses based on male morphology were successful in distinguishing 14 of 15 total pairwise comparisons, and proved to be a more reliable indicator of species-level relationships in comparison to female morphology. The majority of characters with the greatest discriminatory power were male sexual traits. Significant morphological differences were also recorded among sampling localities of conspecifics, with phenotype showing correlation with local environmental data. The findings of this study support the hypothesis that multiple O. vulgaris-like species are currently being incorrectly treated under a single species name, O. vulgaris. Octopuses being exported globally under the name O. vulgaris are of extremely high fisheries market value and profile. Our findings have potentially significant implications for the naming and conservation of commercially harvested members of this species complex throughout their ranges.

3.2 Introduction

The marine environment has traditionally been thought of as a large continuous system with relatively few barriers to dispersal. Organisms with an effective dispersal capability may therefore have the potential to maintain global genetic homogeneity (Waples, 1987). However, dispersal distances of pelagic larvae are influenced by several physiological and biological factors (Hohenlohe, 2004) and are often unknown (Knowlton, 1993). Several examples exist where

59 organisms once thought to be cosmopolitan in distribution, are now understood to represent morphologically similar yet genetically distinct cryptic species with relatively restricted distributions (Knowlton, 1993; Klautau et al., 1999; Bickford et al., 2007). Cryptic species are common among marine invertebrates

(Knowlton, 1993), many of which lack identifiable delineating morphological traits (Klautau et al., 1999). This results in cryptic taxa being ‘lumped’ into single morphospecies, despite being genetically distinguishable. Cryptic diversity is often missed due to an inability to recognise distinguishing morphological traits, distortion of specimens through preservation, and/or an inability to quantify the chemical recognition/communication systems that delineate species.

One marine group where cryptic species are common are the cephalopods, including squids and octopuses (Norman et al., 2014a, 2014b). Taxonomy

(Norman & Hochberg, 2005; Norman et al., 2014b) and phylogenetic relationships (Carlini et al., 2001; Guzik et al., 2005; Strugnell et al., 2008a,

2008b; Kaneko et al., 2011; Acosta-Jofré et al., 2012; Strugnell et al., 2013) within the benthic octopuses has received greater attention in recent years, with a number of cryptic species being identified (Pickford & McConnaughey, 1949;

Söller et al., 2000; Allcock, 2005; Leite et al., 2008; Allcock et al., 2011; Amor et al., 2014; Reid & Wilson, 2015). The difficulties in identifying octopuses and understanding their evolutionary relationships are well illustrated by the current uncertainty and confusion surrounding the phylogeny and taxonomy of genus

Octopus Cuvier, 1797 (type genus of the family Octopodidae d'Orbigny, 1839).

Octopus has long been considered a ‘catch all’ genus (e.g., Nesis, 1998), with few morphological characters available for distinguishing among closely related

60 taxa. More recently, the genus Octopus was characterised by a muscular mantle and arms, saccular mantle with a wide opening, two rows of suckers on each arm, hectocotylised third right arm, terminal organ with diverticulum, functional ink sac, well-developed anal flaps, absence of water pouches on the oral surface of webs and a benthic adult life history (Norman & Sweeney, 1997;

Sweeney & Roper, 1998).

Species-level taxonomy of octopuses has been hindered due to morphological plasticity (Robson, 1929; Pickford, 1945; Voight, 1994; O'Shea, 1999) since their characteristic soft body has few hard structures (Bookstein et al., 1985) and is subject to distortion upon preservation (Pickford, 1964; Burgess, 1966;

Voight, 2001). This means that using morphological characters to distinguish closely related species is particularly difficult (e.g., Norman & Kubodera, 2006); however, recent morphology-based studies suggest that benthic octopuses can be delineated based on discrete phenotypic differences (Gleadall et al., 2010;

Gleadall, 2013; Amor et al., 2014; Gleadall, 2016). Recent taxonomic revisions

(O'Shea, 1999; Norman et al., 2014a) and molecular-based phylogenetic studies (Guzik et al., 2005; Kaneko et al., 2011; Acosta-Jofré et al., 2012; Lü et al., 2013) have confirmed that the genus Octopus is polyphyletic, containing a large assemblage of species groups comprising a number of different genera.

The species group most similar in morphology and behaviour to the type species of the genus (Octopus vulgaris Cuvier, 1797) has been identified as the

‘O. vulgaris species group,’ based on general similarities in overall size, mantle shape, arm length and skin sculpture (Robson, 1929). Species in this group are now considered to comprise the genus Octopus sensu stricto (O'Shea, 1999).

Historically, O. vulgaris was considered to be a cosmopolitan species. First reported from the Mediterranean Sea and eastern North Atlantic, O. vulgaris has been reported from the sub-tropical waters of Australasia, Europe, Africa,

Asia and the Americas. However, recent analyses (Söller et al., 2000; Leite et al., 2008; Amor et al., 2014, 2015; Gleadall, 2016) suggest that populations previously treated as O. vulgaris comprise a complex of morphologically similar but genetically distinct vulgaris-like species (the ‘O. vulgaris species complex’).

Octopus vulgaris sensu stricto (s. s.) occurs in the Mediterranean and eastern

North Atlantic. Other members of this species complex include several species

‘Types’ which have been recognised based on geographic isolation and lack of plausible gene flow mechanisms (Norman et al., 2014a; Fig. 1). Type I occurs in the Caribbean and Gulf of Mexico; Type II in the western South Atlantic along the coast of Brazil; and Type III occurs in the eastern South Atlantic and the

Indian Ocean, along the coast of South Africa. Type IV occurs in subtropical to temperate eastern Asia. Octopus jollyorum Reid and Wilson, 2015 was described based on a phylogenetic analysis that included ‘O. vulgaris’ individuals sampled throughout its known range and the discovery of a member of the group in the Kermadec Islands. Octopus jollyorum was used to recognise a clade that included specimens from Japan and Taiwan, as no type specimen existed for a potential available name, O. sinensis. Subsequently, Gleadall

(2016) designated a neotype for O. sinensis. While not stated in Gleadall

(2016), we now recognise O. jollyorum as a junior synonym of O. sinensis and the latter name is used for representatives of this clade in the current study.

Recent molecular-based analyses support the hypothesis that O. vulgaris s. s.,

O. sinensis and O. vulgaris Type II represent distinct species within the O. vulgaris species complex (Amor et al., 2015). However, the only recent morphological comparisons undertaken to investigate the taxonomic relationships among members of the O. vulgaris species complex are those between O. vulgaris s. s. and O. insularis Leite & Haimovici, 2008 (in Leite et al., 2008) and O. sinensis (Gleadall, 2016). This study employs the first ever global scale sampling strategy to investigate morphological variation and determine the validity of morphological based identifications among members and close relatives of the O. vulgaris species complex. We combine analyses conducted using conventional morphological traits and a more extensive data set. Phylogenetic analyses based on the mitochondrial ‘barcode of life’ gene cytochrome c oxidase subunit I (COI) are also used to provide insights into taxonomic resolution among taxa currently being treated as O. vulgaris.

3.3 Methods

3.3.1 Sampling

Whole specimens and tissue samples of O. vulgaris species group individuals were sourced from museums, university collections and fish markets from the

Atlantic, Indian and Pacific oceans and the Mediterranean Sea (Fig. 1, Table 1).

Tissue samples were stored in 70-90% ethanol. After tissue samples were

63 taken, whole specimens were fixed in 10% formalin and later preserved in 70% ethanol following methods outlined in (Roper & Sweeney, 1983).

Fig. 1: Sampling localities (triangles) for whole animals/tissue samples of members of the Octopus vulgaris species group and close relatives.

Distributions of O. vulgaris sensu stricto and species ‘Types’ are shaded in dark grey (Norman et al. 2014a). Distributions of non-O. vulgaris species are represented by dashed lines. Externally sourced data (Banyuls-sur-Mer,

France; Table 1) are represented by a circle.

Table 1: Sample data for Octopus species analysed in the present study. Sample type refers to the type of data used: whole = whole animals, tissue = tissue samples or data = existing data from the literature.

Species/Type Location Institution Sample Type Reference O. vulgaris s. s. Banyuls-sur-Mer, France Santa Barbara Museum of Natural History Data

O. vulgaris s. s. Galicia, Spain Consejo Superior de Investigaciones Científicas (CSIC), Vigo Whole/Tissue

O. vulgaris s. s. Mauritania Instituto Español de Oceanografía (IEO), Tenerife Whole/Tissue

O. sinensis China Fisheries College, Ocean University of China, Qingdao Whole/Tissue Reid and Wilson (2015)

O. sinensis Keelung / Da si, Taiwan National Taiwan Ocean University, Keelung Whole/Tissue Reid and Wilson (2015)

O. sinensis Kermadec Islands, New Zealand Australian Museum, Sydney Whole/Tissue Reid and Wilson (2015)

O. sinensis Kyushu / Sendai, Japan Tohoku University, Sendai Whole/Tissue

Type II (Brazil) Pontal do Paraná, Brazil Universidade Federal do Paraná (UFPR) Whole

O. insularis Rio Grande do Norte/Brazil Universidade Federal do Rio Grande do Norte (UFRN) Whole

O. insularis Saint Peter and Saint Paul Archipelago, Brazil Universidade Federal do Rio Grande do Norte (UFRN) Whole

O. insularis Trindade Island, Brazil Universidade Federal do Rio Grande do Norte (UFRN) Whole

O. mimus Tocapilla / Pisagua, Chile Consejo Superior de Investigaciones Científicas (CSIC), Vigo Data Guerra et al., (1999)

O. tetricus New South Wales, Australia Museum Victoria Whole/Tissue

O. tetricus Tasmania, Australia Museum Victoria Tissue

O. cf. tetricus Western Australia, Australia Fisheries and Marine Research Laboratories, Western Australia Whole/Tissue Museum Victoria

3.3.2 Molecular analyses

Sequencing: Genomic DNA was extracted from mantle or arm tissue samples of

1-2 mm3 (avoiding skin where possible) using a QIAGEN DNeasy Blood & Tissue

Kit according to the manufacturer’s instructions. Partial COI sequences were amplified via PCR using the universal primers LCO1490 and HCO2198 (Folmer et al., 1994). PCR solutions (25 µL) were composed of 0.5 µL forward primer (10

µM), 0.5 µL reverse primer (10 µM), 12.5 µL MyTaq Red Mix (Bioline), 9.5 µL

H2O and 2 µL DNA (5-10 ng total concentration). PCR cycle conditions were as follows: a single initial denaturing step (two minutes at 95°C); 35 cycles of denaturing (30 seconds at 95°C); annealing (30 seconds at 48°C); and extension

(30 seconds at 72°C); and a single final extension step (five minutes at 72°C).

PCR products were sequenced by Macrogen Inc (Seoul, Korea). COI sequences generated in this study were deposited in GenBank under accession numbers

KU525758-KU525769. Additional sequences from previously published work were obtained from GenBank (Table S1). Octopus cyanea was selected as the outgroup to root the phylogenetic tree (Amor et al., 2015), as it is the closest known relative of the ingroup (Acosta-Jofré et al., 2012). Multiple sequence alignment of the 482 base pair partial COI fragments was performed using the

‘Muscle Alignment’ feature (Edgar, 2004) within Geneious v7.1.3 (created by

Biomatters; available from http://www.geneious.com/).

Molecular-based phylogenetic analyses: jModelTest v0.1.1 (Posada, 2008) was used to carry out statistical selection of best-fit models of nucleotide substitution of the COI alignment. The appropriate model (GTR+G) was chosen based on

‘goodness of fit’ via the Akaike Information Criterion (AIC; Akaike, 1974).

Maximum likelihood (ML) topologies were constructed using RAxML v8.0.19

(Stamatakis, 2014). Strength of support for internal nodes of ML construction was measured using 1,000 rapid bootstrap replicates. Bayesian inference (BI) marginal posterior probabilities were calculated using MrBayes v3.2 (Ronquist &

Huelsenbeck, 2003). Model parameter values were treated as unknown and were estimated. Random starting trees were used and the analysis was run for fifteen million generations, sampling the Markov chain every 1,000 generations.

An average standard deviation of split frequencies of <0.01 was used as a guide to ensure the two independent analyses had converged. The program Tracer v1.3 (Rambaut et al., 2014) was then used to ensure Markov chains had reached stationarity, and to determine the correct ‘burn-in’ for the analysis.

3.3.3 Morphological analyses

Standard morphological characters were measured using digital callipers following Roper and Voss (1983) and Norman and Sweeney (1997): dorsal mantle length (MLd), ventral mantle length (MLv), mantle width (MW), head width

(HW), funnel length (FL), free funnel length (FFL), gill length (GL), enlarged sucker diameter (SDe), non-enlarged sucker diameter (SDn), specialisations at the tip of the males hectocotylised (third right) arm (ligula length, LL; calamus length, CL), terminal organ length (TOL) and arm width (AW) were all recorded to the nearest 0.1 mm. Web depth (WD) was measured from the beak opening to the mid-point of the web sector; and the length of the arms on the left (ALL1-4) and right (ALR1-4) side from the beak opening to the arm tip, were measured to the nearest 1 mm using stretch-resistant cord. The number of suckers on the left 67 third arm (SCL) and the right third arm (SCR; which for males is the sucker count of the hectocotylised arm, HASC) were counted with the aid of a dissecting microscope. Arm lengths and sucker counts were excluded where damage to an arm was perceived to inhibit growth, suckers appeared damaged and no scars/remnants were visible, or arm regeneration was evident (Tables S2 and

S3). All missing data due to these exclusions were replaced with the ‘local’ mean of that trait across the geographic location as missing data was not permitted in analyses.

Morphological datasets were recorded only for mature males and females. To account for differences attributed to variation in overall size, and to allow for investigation of size free trait variation, all morphometric and meristic traits (with the exception of SC, FFL, LL and DL) were transformed to indices, dividing each trait by the dorsal mantle length (a proxy for body size) of the respective specimen. The remaining indices were obtained as follows: sucker counts of each arm were divided by the respective arm length, FFL was divided by FL, LL was divided by CL, and DL was divided by TOL. Morphological relationships were investigated using the complete set of traits recorded during the present study (25 traits for males; 20 traits for females; Tables S2 and S3, respectively).

For comparison with published data, a reduced number of traits was also analysed independently (12 traits for males; 8 traits for females; see traits marked with '*' in Tables S2 and S3, respectively). The reduced set of traits were

MLd, MW, HW, FL, FFL, WD, ALL3/R3, SDn, SCL3/R3 (HASC, males only), LL

(males only) and CL (males only). Analyses of reduced and complete trait data sets were performed on males and females separately to enable the inclusion of male-specific reproductive characters in morphological analyses.

Morphological indices of both males and females were mean scale transformed

(Berner, 2011), and normalised using the ‘normalise variables’ function in

PRIMER E+ v6 and PERMANOVA+ (Anderson et al., 2008) to allow for comparisons of traits despite differing scales of measurement. All morphological analyses were performed using PRIMER E+ v6 (Clarke & Gorley, 2006) and

PERMANOVA+ (Anderson et al., 2008). Collinearity and redundancy of morphological traits was investigated via Principal Component Analysis (PCA) vector plots, Draftsman plots and Spearman correlation matrices as detailed in the user manual (Anderson et al., 2008). Highly correlated variables (R2 ≥ 85%) were considered redundant. The effect of within-clade multivariate dispersion

(i.e. the significance of within-clade variation contributing to between-clade differences) was investigated via permutational distance-based tests for homogeneity of multivariate dispersions (PERMDISP). Differences in morphological traits among sampled individuals were analysed via permutational multivariate ANOVA (PERMANOVA). A resemblance matrix based on Euclidean distance was calculated. To visualise the relationships among locations, PCA was performed using the COI-based phylogenetic clade as an independent factor to group individuals into taxonomically informative entities. Variable contributions to variation were investigated via Similarity Percentages (SIMPER) analysis (Clarke, 1993). In order to evaluate the discriminative power of the morphological traits used, estimates of group assignment were performed using

Canonical Analysis of Principal Components (CAP).

3.3.4 Comparative analyses

Environmental data were incorporated to estimate correlations between morphological variation and each environmental predictor variable. Mean annual

(1900-1997) sea surface temperature (SST), sea bottom temperature (SBT) and salinity were obtained from NOAA (2014). A distance based linear model

(Anderson et al., 2008) was used to perform a marginal test on each environmental variable to determine the overall morphological variation explained. To quantify the variability in the morphological resemblance matrix that was explained by environmental variables, a step-wise sequential test was performed using the AIC to select the model of best fit.

3.4 Results

3.4.1 Phylogenetic relationships

Topologies resulting from molecular-based ML and BI analyses showed a highly supported monophyletic clade containing O. insularis, O. mimus Gould, 1852, O. bimaculoides Pickford and McConnaughey, 1949, and O. maya Voss and Solís

Ramírez, 1966 (bootstrap value [BS] = 95, posterior probability [PP] = 1; Fig. 2).

This clade was sister taxon to (1) a clade containing O. hummelincki Adam,

1936, and (2) a clade containing the O. vulgaris species complex, O. tetricus

Gould, 1852, and O. cf. tetricus of Australasia (BS = 64, PP = 0.66). All members of the O. vulgaris species complex formed a highly supported monophyletic clade which also included O. tetricus and O. cf. tetricus (BS = 95, PP = 1; O. vulgaris group). The O. vulgaris species complex formed three distinct monophyletic clades, which corresponded to three of the O. vulgaris ‘Types’ 70 described in Norman et al., (2014a): Clade 9, O. sinensis (Asia and Kermadec Is;

BS = 75, PP = 1); Clade 10, O. vulgaris Type II (southern Brazil: BS = 69, PP =

0.83); and Clade 11, O. vulgaris s. s. and O. vulgaris Type III (South Africa: BS =

88, PP = 1), which also included a single individual from southern Brazil.

Fig. 2: Bayesian topology depicting the relationships among members of the

Octopus vulgaris species group and close relatives. Analyses are based on partial sequence of the mitochondrial COI gene, showing Bayesian Inference

Posterior Probabilities above and Maximum Likelihood bootstrap values below major nodes. Outgroup is O. cyanea. Node labels represent geographic localities of each haplotype. Clade number is also shown (C1-11). Octopus vulgaris

‘Types’ refer to; Mediterranean/NE Atlantic (O. vulgaris s. s.), southern Brazil

(Type II) and South Africa (Type III) (Norman et al. 2014a). Haplotype characters in parentheses correspond to individuals in Table S1.

3.4.2 Morphological relationships

Comparison of complete and reduced trait datasets: PERMANOVA comparisons and assignment of individuals to their a priori molecular-based phylogenetic clade via CAP were more successful using male and female complete trait datasets (Tables 2-3 and S8-S11). Analyses based on the reduced trait datasets are presented in online supplementary data associated with this manuscript.

Analyses based on the complete trait data sets are presented below.

Analyses of male specimens: Male arm lengths (L2, L3, L4 and R2) displayed

≥85% correlation with each other. Arm length data was most complete for arm

L3, therefore ALL3 was retained while the remaining correlated arm lengths were considered redundant and excluded from analyses. Within-clade variation had no significant impact on among clade analyses (p = >0.05). A significant difference was recorded among the six molecular-based phylogenetic clades investigated

(Pseudo-F = 5.2805, df = 5, p = 0.001). Pairwise comparisons among these six phylogenetic clades showed 14/15 (93%) significant differences (Table 2). All members of the O. vulgaris species complex were distinguished based on morphological analyses (p = <0.02). Octopus vulgaris s. s. and O. sinensis were distinguished primarily by differences in GL and ALR4. Octopus vulgaris s. s. was distinguished from O. vulgaris Type II primarily by SDe. Octopus sinensis was distinguished from O. vulgaris Type II by significantly longer gills (GL).

Octopus insularis specimens were found to be morphologically distinct from all other taxa in the O. vulgaris species complex (p = <0.002). The greatest sources of variation between O. vulgaris s. s. and O. insularis were attributed to differences in ALR3 and HASC. Octopus vulgaris Type II and O. insularis were

72 primarily distinguished by DL and HASC. Octopus sinensis and O. insularis were distinguished by variations in GL and TOL. Octopus tetricus and O. cf. tetricus differed significantly from each other (p = 0.012), particularly through differences in SCL3 and DL. No morphological differences were found between Octopus vulgaris s. s. and O. cf. tetricus (p = 0.095).

Table 2: Pairwise comparisons of male Octopus vulgaris species group and O. insularis individuals based on 25 morphological traits. Lower left diagonal represents PERMANOVA results with significant differences (p = <0.05) highlighted in bold. Upper right diagonal represents results of SIMPER analyses showing traits that contribute most to variation between groups. SIMPER results are also shown in bold if corresponding PERMANOVA showed a significant difference. Far right column represents the percentage of individuals assigned to their a priori group via Canonical Analysis of Principal Components (CAP) analysis (see Table S4 for full CAP analysis table).

O. vulgaris s. s. O. sinensis Type II (Brazil) O. insularis O. tetricus O. cf. tetricus Correct (%)

O. vulgaris s. s. - GLL/ALR4 SDe SCR3/ALR3 SCL3/SDn SCL3/DL 84.2 O. sinensis 0.003 - GLL/GLR TOL/GLL SDn SCL3/TOL 75.0 Type II (Brazil) 0.011 0.001 - DL/SCR3 SCL3/AW SCL3/DL 81.8 O. insularis 0.002 0.001 0.001 - SDe ALR3 66.7 O. tetricus 0.009 0.001 0.001 0.001 - SCL3/DL 80.0 O. cf. tetricus 0.095 0.001 0.01 0.001 0.012 - 100

Visualisation of the principal component biplot for males (Fig. 3a) showed O. vulgaris s. s. and O. vulgaris Type II males to have greater levels of morphological variability in comparison to other taxa, which was demonstrated by their occupation of highly positive and highly negative PC1 and PC2 spaces.

Octopus vulgaris s. s., O. sinensis and Brazilian O. vulgaris Type II showed the least discrimination, although O. vulgaris s. s. and O. vulgaris Type II individuals had relatively longer arms than O. sinensis. Octopus vulgaris Type II individuals had relatively fewer suckers on the third arm pair than O. vulgaris s. s. and O. sinensis. Octopus tetricus, O. cf. tetricus and O. insularis demonstrated negative

PC2 loadings attributed to high sucker numbers. Octopus tetricus and O. insularis showed the least overlap with other taxa included in the analysis but O. cf. tetricus overlapped with all members of the O. vulgaris species complex.

Fig. 3: Principal component biplot of male (a) and female (b) Octopus vulgaris species group and O. insularis individuals grouped by COI-based phylogenetic clade. Analysis is based upon 25 and 20 morphological traits, respectively.

Octopus vulgaris Type II refers to individuals from southern Brazil.

Of the 68 male individuals analysed, 54 (79%) were correctly assigned to their a priori group via CAP (Table 2). For O. vulgaris s. s., 16 individuals (84%) were correctly classified: the remainder were misclassified as O. sinensis (n = 3).

Twelve O. sinensis individuals (75%) were correctly assigned to their a priori

74 group, with the remaining individuals being misclassified as O. vulgaris s. s. (n =

1), Brazilian Type II (n = 1), O. insularis (n = 1) or O. cf. tetricus (n = 1). Nine O. vulgaris Type II individuals (82%) were correctly classified while the remaining individuals were misclassified as O. vulgaris s. s. (n = 1) and O. insularis (n = 1).

Eight O. insularis individuals were correctly assigned (67%), with the remaining individuals misclassified as O. vulgaris s. s. (n = 1), O. tetricus (n = 1) or O. cf. tetricus (n = 2). Four O. tetricus individuals (80%) were correctly assigned, with the remaining individual being misclassified as O. sinensis. All O. cf. tetricus individuals (n = 5) were correctly assigned to their respective a priori group.

Analysis of female specimens: Significant within-clade variation was recorded for

O. vulgaris s. s. and O. insularis females (p = 0.03). The main-effects model showed significant morphological differences among the six molecular-based phylogenetic clades of female individuals (Pseudo-F = 3.8184, df = 5, p = 0.001).

Pairwise comparisons showed that 10/15 (67%) comparisons had significant morphology-based differences (Table 3). All members of the O. vulgaris species complex (O. vulgaris s. s., O. sinensis and O. vulgaris Type II) were successfully distinguished based on multivariate morphological analyses (p = ≤0.01). Octopus vulgaris s. s. and O. sinensis were distinguished primarily by arm length (L3) and sucker diameter. Arm width was the primary source of variation between O. vulgaris s. s. and O. vulgaris Type II. Octopus sinensis and O. vulgaris Type II were found to differ in gill length and arm width. All members of the O. vulgaris species complex were distinguished from O. insularis (p = ≤0.003). Variation between O. vulgaris s. s. and O. insularis was primarily attributed to differences in the number of suckers on the third arm pair, which was also the greatest source of variation between O. vulgaris Type II and O. insularis. Octopus sinensis and O. insularis were best delineated by the variation in sucker numbers

75 on the third left arm. Octopus tetricus and O. cf. tetricus were unable to be distinguished based on morphology (p = 0.3).

Table 3: Pairwise comparisons of female Octopus vulgaris species group and O. insularis individuals based on 20 morphological traits. Lower left diagonal represents PERMANOVA results with significant differences (p = <0.05) highlighted in bold. Upper right diagonal represents results of SIMPER analyses showing traits that contribute most to variation between groups. SIMPER results are also shown in bold if corresponding PERMANOVA showed a significant difference. Asterisks represent pairwise comparisons effected by significant within clade variation. Far right column represents the percentage of individuals assigned to their a priori group via CAP analysis (see Table S6 for full CAP analysis table).

O. vulgaris s. s. O. sinensis Type II (Brazil) O. insularis O. tetricus O. cf. tetricus Correct (%)

O. vulgaris s. s. - ALL3/SDn AW SCR/L3* SCR3/HW SCL/R3 76.2 O. sinensis 0.004 - GLL/AW SCL3 SCR3/FL HW 50 Type II (Brazil) 0.01 0.001 - SCR/L3 SCR3/AW AW 71.4 O. insularis 0.001* 0.001 0.003 - SCL3/FL ALL1/3 100 O. tetricus 0.053 0.119 0.039 0.004 - HW 75 O. cf. tetricus 0.181 0.05 0.041 0.012 0.114 - 50

Visualisation of the principal component biplot for females (Fig. 3b) showed that

O. vulgaris s. s. and O. sinensis had the most morphological variability, with highly positive and negative PC1 and PC2 loadings. Octopus vulgaris Type II was characterised by positive PC2 loadings (low SCL/R3). Octopus insularis

76 individuals formed a distinct group characterised by positive PC1 and negative

PC2 loadings (low arm lengths and high sucker counts).

Overall, 41 of the 62 analysed female individuals (66%) were correctly assigned via CAP (Table 3). Sixteen O. vulgaris s. s. individuals (76%) were correctly classified, while four individuals were misclassified as O. sinensis and a single individual as O. tetricus. Ten O. sinensis individuals (50%) were correctly assigned to their a priori group, with the remaining individuals being misclassified as O. vulgaris s. s. (n = 5), O. insularis (n = 1), O. tetricus (n = 2) and O. cf. tetricus (n = 2). Five O. vulgaris Type II individuals (71%) were correctly assigned, with a single individual misclassified as O. vulgaris s. s., O. sinensis and O. tetricus. All O. insularis individuals (n = 6) were correctly assigned, while

75% of O. tetricus and 50% of O. cf. tetricus individuals were assigned correctly.

Reduced trait analyses of male O. vulgaris s. s.: Significant differences were recorded among Galician, Mediterranean and Mauritanian males (p = 0.001), with the pairwise multivariate model showing a significant difference among the three localities (p = ≤0.004; Table 4).

Table 4: Pairwise comparisons of male Octopus vulgaris sensu stricto individuals based on 12 morphological traits. Lower left diagonal represents PERMANOVA results with significant differences (p = <0.05) highlighted in bold. Upper right diagonal represents results of SIMPER analyses showing traits that contribute most to variation between groups. SIMPER results are also shown in bold if corresponding PERMANOVA showed a significant difference. Far right column represents the percentage of individuals assigned to their a priori group via CAP analysis (see Table S5 for full CAP analysis table).

Galicia Mediterranean Mauritania Correct (%)

Galicia - ALR3/SCL3 ALR3 80 Mediterranean p=0.004 - FFL/LL 88.9 Mauritania p=0.001 p=0.003 - 100

A PC biplot (Fig. 4a) showed that each sampling locality for O. vulgaris s. s. males could be distinguished, although there was some overlap. Individuals from the Mediterranean were found to have greater sucker numbers (L3, R3) in comparison to Galician and Mauritanian (eastern North Atlantic) individuals.

Galician and Mauritanian individuals were able to be distinguished along PC1, as

Galician males had longer arms (L3, R3).

Fig. 4: Principal component biplot 27 Octopus vulgaris sensu stricto males (a) and females (b) grouped by locality. Analysis is based on 12 and 8 morphological traits, respectively. Octopus vulgaris Type II refers to individuals from southern Brazil.

Based on the CAP, 24 of the 27 O. vulgaris s. s. males (89%) were correctly assigned (Table 4). All individuals from Mauritania (n = 8) were successfully assigned to their correct sampling locality: eight of the nine Mediterranean individuals (89%) were correctly assigned, with a single individual being misclassified as Galician; and eight of the ten Galician individuals (80%) were correctly assigned, with the remaining two individuals misclassified as

Mauritanian.

Variation attributable to environmental data explained 31.4% of the variation in male morphology (R2 = 0.31354). Investigating each trait independently via marginal tests, SST explained 21.3% (p = 0.001) and SBT 21.2 % (p = 0.001) of the variation. Sequential tests revealed that SST accounted for 21.3% of the variation seen in the morphological data (p = 0.002). Once SST was accounted

79 for, SBT explained a further 10% of the variation (p = 0.002). Latitude, longitude and depth did not explain any further variation, although each was found to explain a significant amount of the variation in morphology when analysed independently (p = 0.001, p = 0.005 and p = 0.023, respectively),

Reduced trait analyses of female O. vulgaris s. s.: A significant difference was recorded among Galician, Mediterranean and Mauritanian females (p = 0.001), with the pairwise multivariate model showing a significant difference among the three localities (p = ≤0.002; Table 5).

Table 5: Pairwise comparisons of female Octopus vulgaris sensu stricto individuals based on eight morphological traits. Lower left diagonal represents

PERMANOVA results with significant differences (p = <0.05) highlighted in bold.

Upper right diagonal represents results of SIMPER analyses showing traits that contribute most to variation between groups. SIMPER results are also shown in bold if corresponding PERMANOVA showed a significant difference. Far right column represents the percentage of individuals assigned to their a priori group via CAP analysis (see Table S7 for full CAP analysis table).

Galicia Mauritania Mediterranean Correct (%)

Galicia - SCL3/HW FFL/FL 90 Mauritania p=0.001 - FFL/SCL3 100 Mediterranean p=0.002 p=0.001 - 100

A PC biplot (Fig. 4b) distinguished O. vulgaris s. s. females by locality.

Individuals from the eastern North Atlantic (Galicia and Mauritania) were more similar to each other than they were to Mediterranean females, which have

80 longer funnels (FL). Individuals from the eastern North Atlantic differed, with

Galician males possessing more suckers (SCL/SCR) and a larger head (HW).

Of 27 female O. vulgaris s. s. individuals, 26 (96%) were correctly assigned to their a priori group (Table 5). Individuals from Mauritania and the Mediterranean

(France) were all assigned with 100% accuracy, and nine of the ten Galician individuals were assigned correctly (90%), with the remaining individual misclassified as Mediterranean.

Of the overall variation in female morphology, 33.9% was correlated with variation in environmental data (R2 = 0.33854). Investigating each trait independently via marginal tests showed that latitude explained 20.8% (p =

0.001) and SST 18.8% (p = 0.001) of the variation. In sequential tests, latitude accounted for 20.8% of the morphological variation (p = 0.001). With latitude accounted for, SST explained a further 13% of the variation (p = 0.002); and once both latitude and SST were accounted for, SBT, longitude and depth explained no further variation (although a significant amount of variation in morphology was explained when these parameters were analysed independently: p = 0.002, p = 0.001 and p = 0.001, respectively).

3.5 Discussion

Molecular-based phylogenetic analyses of O. vulgaris species group individuals in the present study support the presence globally of five distinct clades, which were used as a discriminant factor in morphological analyses. Multivariate morphological analyses using conventional morphological traits were successful

81 in distinguishing the majority of these clades and support the hypothesis of greater species-level diversity within the O. vulgaris species complex (O. vulgaris s. s., O. vulgaris Type II and O. sinensis). Although each of these species was successfully distinguished, further distinctions were detected among the sampling localities of O. vulgaris s. s., suggesting a requirement of broad sampling across the known distribution to ensure robust future morphological analyses of this group.

Previous molecular-based phylogenetic analyses using five mitochondrial genes placed Chinese and Japanese O. vulgaris into a well-supported monophyletic clade, distinct from all other members of the O. vulgaris species complex (Amor et al., 2014). Reid and Wilson (2015) considered mitochondrial-based differences to warrant the distinction of Kermadec Island individuals from O. vulgaris s. s., establishing the name O. jollyorum for this clade, which also encompassed Asian Type IV O. vulgaris individuals. The recent designation of a neotype for O. sinensis, effectively renames the clade member-taxa and places

O. jollyorum in synonymy with O. sinensis. We formally synonymise the two species here. The latter species was redescribed by Gleadall (2016) and can be distinguished from O. vulgaris with the former species having shorter arms and fewer suckers. Although, individuals from Asia and the Kermadec Islands are currently understood to comprise a single species, the substantial geographic distance between these two geographic regions warrants further investigation into their species-level diversity.

Vidal et al., (2010) compared the morphology and chromatophore patterns of O. vulgaris paralarvae from the eastern North Atlantic (Galicia, Spain; O. vulgaris s. s.) and the western South Atlantic (southern Brazil; O. vulgaris Type II), noting

82 considerable differences in chromatophore numbers. These differences support the hypothesis that O. vulgaris Type II is distinct from O. vulgaris s. s. The present study reports differences in adult morphology and places individuals from southern Brazil into a monophyletic clade, distinct from O. vulgaris s. s. and

O. sinensis. We therefore recognise O. vulgaris Type II as a distinct species within the O. vulgaris species complex.

Superficial morphological similarity among species in the O. vulgaris species complex had resulted in the assumption that O. vulgaris is a single cosmopolitan species. Despite estimates of 3-15 million years divergence between

Australasian/Asian taxa (Amor et al., 2014) and 19-41 million years divergence between O. insularis and other members of the O. vulgaris species group (Amor et al., 2015), principal component plots show that the morphology of these taxa is relatively conservative. The distinct molecular-based clades within the O. vulgaris species complex have allopatric distributions, therefore the selective pressures to adapt their phenotype due to interspecific competition may be reduced. Differentiation in morphological traits is often most extreme where closely related species occur in sympatry (Brown & Wilson, 1956), which is thought to limit resource overlap and interspecific competition and allow otherwise directly competing taxa to co-exist. Such ‘ecological character displacement’ appears to be a common strategy among closely related taxa and has been documented in a number of plant, reptile, mammal, bird, fish and snail taxa (Dayan & Simberloff, 2005). One exception within the O. vulgaris species group is the parapatric distribution of O. vulgaris Type II (sub-tropical southern

Brazil) and O. insularis (mid-Atlantic islands and tropical northern Brazil).

Although these two taxa are relatively distantly related, they are very similar in

83 morphology, which may represent a unique opportunity to investigate the extent of this phenomenon within the O. vulgaris group.

The sexual traits of male individuals were found to be important characters for morphology-based species discrimination in the O. vulgaris species complex, confirming the utility of male sexual traits in cephalopod systematics. Similar findings associated with other animal groups also show that sexual traits are more variable than non-sexual traits (Pomiankowski & Moller, 1995), and are often the only reliable delimiting characters among species (Arnqvist, 1998).

Amor et al., (2014) used 17 morphological characters (five of which were sexual traits) to distinguish O. tetricus (from New Zealand and the eastern coast of

Australia) and O. cf. tetricus (west Australia). HASC was found to be the primary source of variation between the two species, with significantly greater values for

O. cf. tetricus. However, a study of the genus Pareledone found that morphological traits (including three sexual traits) were unsuccessful in resolving species-level relationships, however clear genus level resolution was achieved

(Allcock et al., 2008).

The utility of HASC has previously been demonstrated in species-level resolution of octopuses (Toll, 1988). Among 12 species, Toll (1988) reported HASC values to be relatively fixed among conspecifics. In contrast, the present study found

HASC values for O. vulgaris s. s. differed significantly among sampling localities.

Individuals from the Mediterranean (France) and the eastern North Atlantic

(Spain) had overlapping but significantly differing HASC values (144-168 and

156-183, respectively). Mauritanian specimens were found to have significantly lower HASC values (114-150) than those for both France and Spain. The significant differences in HASC values reported within O. vulgaris s. s. are

84 considered to represent population-level differences. Alternatively, since specimens from Mauritania display minimal overlap in this character compared with those from France and Spain, this may indicate the presence of further species-level diversity within O. vulgaris s. s. Such a wide range in HASC values within O. vulgaris s. s. therefore suggests the need for caution in basing species within this group on discrimination between HASC values. Voight (2012) questioned the validity of using HASC as a species delimiting trait, also citing wide variation in HASC as a potential problem for species-level inferences, concluding that variation in sucker numbers of ≤15% between potential species should be interpreted with caution. Although, variation in HASC values among

Australasian members of the O. vulgaris species group showed western

Australian O. cf. tetricus have ~40% greater sucker numbers than those for eastern Australian O. tetricus, which was determined to reflect species-level differences (Amor et al., 2014).

The discriminatory power of female based morphological analyses was weaker than that for males. In the complete and reduced trait datasets, more morphological traits were available for males (male-specific reproductive characters) and these traits were found to be important in distinguishing among the molecular based phylogenetic clades. In contrast, the female traits found to have the greatest delimiting power among species were non-sexual. Sexual traits, particularly the hectocotylus, are also important distinguishing taxonomic characters for many cephalopods (Bello, 1995; Brakoniecki, 1996; Norman & Lu,

1997; O'Dor & Lipinski, 1998; von Byern & Klepal, 2010). In comparison to body size and shape traits (which are likely to be less phenotypically and genetically variable between species), sexual traits are often exaggerated and diverse among close relatives (Pomiankowski & Moller, 1995), making them ideal

85 candidates for distinguishing among species. While sexual traits were the primary source of morphological variation in the present study, non-sexual traits for both male and female morphology were successful in distinguishing among sampling localities of O. vulgaris s. s. (Galicia, France and Mauritania).

The need for greater taxonomic resolution within the family Octopodidae is particularly important in light of the growing global exploitation of octopuses as a commercial fisheries resource (Norman & Finn, 2014). Global production of octopuses exceeds 350,000 tonnes with a total export value of US$1.07 billion, surpassing many valuable finfish fisheries (FAO, 2012). A major limitation of the global catch statistics reported by the FAO is the poor state of octopus taxonomy, with only five (O. vulgaris, O. maya, Eledone cirrhosa, Eledone moschata and Enteroctopus dofleini) of the estimated 100 species of commercially harvested octopuses listed in global statistics (Norman & Finn,

2014). As the majority of octopus fisheries world-wide are in decline (Norman &

Finn, 2014), this low taxonomic resolution highlights the requirement for more accurate species identification in order to develop more sustainable octopod fisheries practices. Octopuses being exported globally under the name O. vulgaris are of extremely high market value and profile (Norman et al., 2014a), particularly in north-western Africa, the largest single-species octopus fishery in the world (FAO, 2012). Aquaculture and captive growing of wild caught juveniles are receiving increasing funding, particularly in China (Norman et al., 2014a).

Differences among geographical areas in hatchling features and paralarvae viability (Iglesias et al., 2007, 2014) may also be linked to taxonomic differences.

The findings presented here support the hypothesis that multiple O. vulgaris-like species are currently being incorrectly treated under a single species name. Our findings therefore have significant implications for the naming, marketing, value,

86 documentation and potentially conservation of commercially harvested members of this species complex throughout their ranges.

3.6 Supplementary information

Fig. S1: Principal Component biplot of a) male and b) female Octopus vulgaris species group individuals and close relatives, grouped by COI based phylogenetic clade. Analysis is based upon 12 and 8 morphological traits respectively. X and Y axes represent PC1 and PC2 respectively. Octopus vulgaris Type II refers to individuals from southern Brazil.

Table S1: List of COI sequences accessed from GenBank and corresponding species and locality information. Clade and haplotype information correspond to

Fig. 2 of the present study. Asterisks reflect individuals requiring updated identifications in GenBank.

Accession number Species Location Clade Haplotype AB191280 O. cyanea Japan O AB430534 O. cyanea Japan O AB430535 O. cyanea Japan O KC894940 O. salutii Mediterranean Sea 1 KC894941 O. salutii Mediterranean Sea 1 AF377967 O. bimaculoides Santa Barbara, USA 2 KF774309 O. bimaculoides Mexico 2 GU355923 O. mimus Chile/Peru 3 3b KP056550 O. mimus Chile 3 3b GU355924 O. mimus Chile/Peru 3 3a GU355925 O. mimus Callao, Peru 3 3a GU355926 O. mimus Chile/Peru 3 3a KP056551 O. mimus Chile 3 3a GU362545 O. maya Mexico 4 HQ214117 O. maya Mexico 4 KP056552 O. insularis St Helena 5 5a KP056553 O. insularis Ascension Island 5 5a KP056554 O. insularis Ascension Island 5 5a KP056555 O. insularis Ascension Island 5 5a KF844025 O. insularis Salvador, Brazil 5 KF844022 O. insularis Barra Grande, Brazil 5 KF844023 O. insularis Barra Grande, Brazil 5 KF844024 O. insularis Barra Grande, Brazil 5

KF844044 O. hummelincki Ceará, Brazil 6 KJ605246 O. tetricus Australia: New South Wales 7 7a KJ605247 O. tetricus Australia: Wallaga Lake 7 7a KJ605249 O. tetricus Australia: Wallaga Lake 7 7a KJ605250 O. tetricus Australia: Wallaga Lake 7 7a KJ605251 O. tetricus Australia: Narooma 7 7a KJ605252 O. tetricus Australia: Port Stephens 7 7a KJ605253 O. tetricus Australia: Port Stephens 7 7a KJ605255 O. tetricus Australia: Flinders Island 7 7a KJ605256 O. tetricus Australia: Flinders Island 7 7a KJ605248 O. tetricus Australia: Wallaga Lake 7 7b KJ605254 O. tetricus Australia: Flinders Island 7 7b KJ605257 O. tetricus Australia: Flinders Island 7 7b KJ605258 O. tetricus New Zealand: Leigh 7 7b KJ605259 O. tetricus New Zealand: Leigh 7 7b KJ605260 O. tetricus New Zealand: Leigh 7 7b KJ605261 O. tetricus New Zealand: Leigh 7 7b KJ605264 O. cf. tetricus Australia: Cape Le Grand 8 8a KJ605263 O. cf. tetricus Australia: Albany 8 8a KJ605269 O. cf. tetricus Australia: Mandurah 8 8a KJ605262 O. cf. tetricus Australia: Woodman Point 8 8a KJ605273 O. cf. tetricus Australia: Mandurah 8 8a KJ605271 O. cf. tetricus Australia: Mandurah 8 8a KJ605267 O. cf. tetricus Australia: Esperance 8 8a KJ605266 O. cf. tetricus Australia: Cape Le Grand 8 8a KJ605265 O. cf. tetricus Australia: Cape Le Grand 8 8a KJ605268 O. cf. tetricus Australia: Esperance 8 8a KJ605275 O. cf. tetricus Australia: Mandurah 8 8a KJ605278 O. cf. tetricus Australia: Mandurah 8 8a KJ605272 O. cf. tetricus Australia: Mandurah 8 8a 88

KJ605274 O. cf. tetricus Australia: Mandurah 8 8a KJ605276 O. cf. tetricus Australia: Mandurah 8 8a KJ605277 O. cf. tetricus Australia: Mandurah 8 KJ605270 O. cf. tetricus Australia: Mandurah 8 HQ846110* O. sinensis China 9 9a KU525758* O. sinensis Taiwan 9 9a KU525759* O. sinensis Japan 9 9a NC006353* O. sinensis Japan 9 9a AB430546* O. sinensis Japan: Hyougo, Akashi, Futami 9 9b AB430547* O. sinensis Japan: Kanagawa, Misaki 9 9b AB430548* O. sinensis Japan: East China Sea 9 9b HQ846154* O. sinensis China 9

KU525760* O. sinensis Kermadec Island 9 KF844041 O. vulgaris Brazil: Santa Catarina 10 10a KF844039 O. vulgaris Brazil: Paraná 10 10a KF844038 O. vulgaris Brazil: Paraná 10 10a KF844037 O. vulgaris Brazil: São Paulo 10 10a KF844036 O. vulgaris Brazil: São Paulo 10 10a KF844035 O. vulgaris Brazil: São Paulo 10 10a KF844034 O. vulgaris Brazil:Rio de Janeiro 10 10a KF844033 O. vulgaris Brazil:Rio de Janeiro 10 10a KF844032 O. vulgaris Brazil:Rio de Janeiro 10 10a KF844029 O. vulgaris Brazil: Pará 10 10a KF844028 O. vulgaris Brazil: Pará 10 10a KF844026 O. vulgaris Brazil: Amapá 10 10a KF844040 O. vulgaris Brazil: Santa Catarina 10 KF844031 O. vulgaris Brazil: Bahia 10 KF844030 O. vulgaris Brazil: Pará 10 DQ683227 O. vulgaris Spain: Mediterranean 11 11a KJ605279 O. vulgaris South Africa: Port Elizabeth 11 11a FN424379 O. vulgaris St Paul and Amsterdam Islands 11 11a FN424381 O. vulgaris St Paul and Amsterdam Islands 11 11a FN424380 O. vulgaris St Paul and Amsterdam Islands 11 11a KU525761 O. vulgaris Morocco: Cabo Blanco 11 11a DQ683215 O. vulgaris South Africa: Durban 11 11a DQ683217 O. vulgaris South Africa: Durban 11 11a DQ683218 O. vulgaris South Africa: Durban 11 11a DQ683219 O. vulgaris South Africa: Durban 11 11a DQ683208 O. vulgaris South Africa: Hout Bay 11 11a DQ683212 O. vulgaris South Africa: Port Elizabeth 11 11a DQ683210 O. vulgaris South Africa: Struisbaai 11 11a DQ683205 O. vulgaris Tristan da Cunha 11 11a DQ683220 O. vulgaris South Africa: Umhlanga 11 11a DQ683221 O. vulgaris Spain: Galicia 11 11b KJ605280 O. vulgaris France: Perpignan 11 11b KJ605282 O. vulgaris France: Perpignan 11 11b KJ605284 O. vulgaris France: Perpignan 11 11b KF844043 O. vulgaris Portugal 11 11b KF844042 O. vulgaris Portugal 11 11b KJ605281 O. vulgaris France: Perpignan 11 11b KJ605283 O. vulgaris France: Perpignan 11 11b KU525762 O. vulgaris Spain: Tenerife 11 11b KU525763 O. vulgaris Spain: Galicia 11 11b KU525764 O. vulgaris Spain: Tenerife 11 11b KU525765 O. vulgaris Spain: Galicia 11 11b KU525766 O. vulgaris Morocco: Cabo Blanco 11 11b KU525767 O. vulgaris Chile: Juan Fernandez Island 11 11c KU525768 O. vulgaris Chile: Juan Fernandez Island 11 11c DQ683214 O. vulgaris South Africa: Durban 11 DQ683216 O. vulgaris South Africa: Durban 11 KU525769 O. vulgaris Morocco: Cabo Blanco 11 KF489451 O. vulgaris India: Kerala 11 KF844027 O. vulgaris Brazil: Pará 11

Table S2: Raw male morphological data for Octopus vulgaris species group individuals and close relatives.

Species/ Location Catalogue# MLd MLv MW HW FL FFL WD ALL1 ALL2 ALL3 ALL4 ALR1 ALR2 ALR3 ALR4 AW SDn SDe SCL3 SCR3 GLL GLR LL CL TOL DL Type (institution) * * * * * * * * * * * * * Octopus cf. Western F200326a 127 84 68 54 45.9 34.1 127 * 594 * 480 508 593 520 482 25.7 12.4 15 * 207 41.97 35.5 5 2.1 11.9 * tetricus Australia (MV) Octopus cf. Western F200327 111.7 77 66.6 52.2 40.6 21.1 98 395 411 458 * 392 467 426 * 22.9 12.6 17.3 283 192 32.3 31.5 4.4 1.6 13.5 * tetricus Australia (MV) Octopus cf. Western F200328 163.4 104.3 83.5 65.8 44.2 29 107 431 494 559 455 480 437 544 535 34.3 15.4 17.3 291 218 47.3 46.6 3.9 1.7 26.8 9.8 tetricus Australia (MV) Octopus cf. Western F200329 114.2 80.8 74.6 56.4 40.7 20.3 118 388 442 381 420 357 448 384 407 30.3 13.1 17.9 230 177 * * 4.6 2.4 14.7 5.9 tetricus Australia (MV) Octopus cf. Western F200330 95.8 79.6 79.1 53.9 38.1 29.1 114 407 466 420 446 432 497 444 436 23.7 12.2 16.7 281 209 * * 3.5 1.7 14.3 5.5 tetricus Australia (MV) Fujian Octopus Province, OcfvC10 153 83.1 92.3 48.5 50.86 36.46 103 370 * * 294 339 * * * 21.39 12.71 24.91 * 126.6 38.25 36.2 * * * * sinensis China (OUC) Fujian Octopus Province, OcfvC11 120.6 87.14 78 46.7 40.6 29.99 86 * 480 * * 315 * 398 274 17.9 11.6 20.7 * 130 45.1 46 2.94 1.31 26.7 6.55 sinensis China (OUC) Zhejiang Octopus Province, OcfvC12 112.8 61.7 58.1 39.9 35.4 14.2 81 337 436 417 347 337 482 374 333 18.2 9.4 16.3 199 137 33.2 33.2 3.47 1.63 21.97 7.96 sinensis China (OUC) Fujian Octopus Province, OcfvC6 107.66 87.3 78.8 41.2 42.7 33.8 91 306 407 385 316 344 426 367 371 21.5 13.7 20.4 160 121 40.37 42.05 3.64 1.19 28.04 12.69 sinensis China (OUC) Zhejiang Octopus Province, OcfvC8 102.1 58.66 59.81 32.22 36.89 19.9 70 328 444 385 363 298 395 294 305 20.38 10.1 15.57 173 123 37.85 41.05 2.5 1.16 20.04 5.03 sinensis China (OUC) Fujian Octopus Province, OcfvC9 165 89.8 87.7 46.3 50.8 35.8 120 390 463 520 439 360 530 429 400 17.52 13.4 18.2 208 122 49.48 58.62 3.71 1.22 28.01 8.25 sinensis China (OUC) Octopus Kyushu, IGG307 88.1 70.5 64.9 44 27.5 18.3 105 320 385 326 320 290 368 271 313 20 8.9 14.5 229 135 29.9 29.8 3.5 2.03 18.5 6.8 sinensis Japan (MV) Octopus Kyushu, IGG309 125 80 90 50 45 35 93 360 400 395 370 325 396 310 355 23.8 11.8 15.1 * 130 41.5 43.6 3.92 1.73 23.98 7.38 sinensis Japan (MV) Octopus Yilan County, OcfvT10 107.8 73.4 69.1 36.9 38.8 30.2 82 300 338 391 * 334 * 286 306 13.7 8.3 12.8 188 112 42.4 40.4 3.51 1.45 20.2 4.2 sinensis Taiwan (MV) Octopus Yilan County, OcfvT11 155.5 115.4 89.5 55.5 61.4 34.7 138 459 473 647 * 510 * 533 512 25.1 13.5 19.3 222 146 59.3 56 4.58 1.32 33 9.4 sinensis Taiwan (MV) Octopus Yilan County, OcfvT12 86.8 68.8 61.5 39.1 39.9 24.7 75 * 325 255 280 * 325 285 275 13.9 7.2 13.8 * 127 35.7 35.1 2.39 1 14 5.35 sinensis Taiwan (MV) Octopus Yilan County, OcfvT13 125 90 95 55 45 35 100 390 480 460 385 350 475 380 390 17.3 8.8 15 211 113 49.7 47.5 3.89 0.63 28.77 8.67 sinensis Taiwan (MV) Octopus Yilan County, OcfvT14 103.8 79.4 71.2 40.7 48.4 36.7 85 323 380 455 315 297 418 297 355 16.6 8.5 14.1 193 117 42.4 44.8 2.94 1.38 19.7 6.5 sinensis Taiwan (MV) Octopus Yilan County, OcfvT15 89 64.1 57.7 35.4 27.4 25.8 90 275 375 325 315 * 325 290 270 15.7 8.6 11.4 * 131 29 * 1.68 0.7 * * sinensis Taiwan (MV) Octopus Yilan County, OcfvT16 93.6 65.1 65.2 39 36.4 20.9 76 * 328 * 272 276 255 288 240 15.4 7.6 10.9 * 111 36.8 36.7 2.75 1.62 16.8 7.1 sinensis Taiwan (MV) Octopus Yilan County, OcfvT9 111.9 84.8 76.9 42.2 46 24 110 374 391 395 375 322 433 350 365 18 10.1 15.3 179 127 42.7 44.6 3.9 1.4 27.7 6.8 sinensis Taiwan (MV) Fernando de Octopus Noronha, OiB10 108 93.2 80 46.6 44.4 * 73 * 363.5 * * 300 251 250 310 14.7 * 12.2 * * * * 3 1.6 13 5.81 insularis Brazil (UFRN) Fernando de Octopus Noronha, OiB12 100 82.2 68.2 46.8 40 * 92 * 363.5 280 * * * 256 248 15.6 * 11.1 * 122 * * * 1.75 15.06 5.97 insularis Brazil (UFRN) Fernando de Octopus Noronha, OiB14 105 68.1 67.9 54.6 30.4 * 87 308 363.5 244 * 270 343 203 255 16.6 * 10.6 * * * * 3.8 1.7 15 6 insularis Brazil (UFRN) 90

St Peter St Paul 94 70 64.7 43.4 36.6 * 79 226 363.5 * * 271 290 222 260 15 * 10.8 * * * * 2.04 1.28 13.2 5 Octopus Archipelago, OiB17 insularis Brazil (UFRN) St Peter St Paul 78 52.6 60.8 35.5 33 * 65 * 260 265 251 222 * 220 260 11.6 * 11 * * * * 3 2.2 12.8 4 Octopus Archipelago, OiB21 insularis Brazil (UFRN) St Peter St Paul 98 70.7 69.6 48.3 * * 99 * 363.5 * 306 290 285 266 346 15.5 * 10.5 * * * * 3.3 2.5 14.1 5.7 Octopus Archipelago, OiB22 insularis Brazil (UFRN) St Peter St Paul 114 76.4 67.8 42.25 38.1 * 104 * 363.5 * 312.6 344 * 265 * 14.8 * 12 * 142 * * 3.75 2.1 17.45 5.6 Octopus Archipelago, OiB9 insularis Brazil (UFRN) St Peter St Paul MOA-DOL- 124.83 70.5 78.92 51.22 37.42 24.5 108 352 467 431 * * 408 400 412 27.27 11.39 * 199 134 29.77 * 2.64 1.09 * * Octopus Archipelago, 66 insularis Brazil (UFRN) MOA-DOL- Octopus Pernambuco, 67 105.27 73.67 69.91 57.28 32.59 23.18 97 323 * 277 * 291 * 237 298 26.29 12.98 13.63 196 117 25.12 24.68 1.9 0.34 13.9 5.87 insularis Brazil (UFRN) Rio de MOA-DOL- Octopus Janeiro, 88 90.91 66.73 64.97 46.1 34.71 23.93 88 257 * 368 305 308 307 274 321 20.1 9.11 * 224 136 27.93 25.06 2.87 1.34 12.01 4.45 insularis Brazil (UFRN) Rio Grande MOA-DOL- Octopus do Norte, 65 142.04 82.75 98.16 55.06 48.57 34.63 150 418 * * * 373 450 366 414 27.34 12.36 13.06 * 125 37.18 37.73 3.36 1.09 * * insularis Brazil (UFRN) Octopus Northern 49524 108.54 64.87 91.42 56.34 40.88 26.3 111 * * 376 344 351 * 205 410 23.86 10.94 17.35 199 * * 28.4 2.29 1.26 * * insularis Brazil (UFRN) New South Octopus Wales, F182058 121 88.4 88.1 65.5 35 33 114 404 * 289 384 * 494 426 * 40 17.3 25.7 * 143 41.8 36.5 4.2 2.1 19.7 * tetricus Australia (MV) New South Octopus Wales, F200319 93.6 71.8 55.2 48.7 34 24.4 78 289 355 * 324 * 358 333 358 30.1 13.6 17.6 * 126 32.9 27.9 3.5 1.5 15.1 3.4 tetricus Australia (MV) New South Octopus Wales, F200323 85.6 69.1 51.8 39 26.5 25.9 74 259 322 289 298 205 320 260 * 20.5 12.1 16.2 * 139 35.4 34 2.6 0.7 12.8 5.4 tetricus Australia (MV) New South Octopus Wales, F200324 86.7 58.2 59.5 44.1 31.2 20.3 88 240 287 289 308 * 316 307 * 25.5 12.2 19.6 * 138 29.5 22.8 3 1.3 14.7 4.5 tetricus Australia (MV) New South Octopus Wales, F182057 113.7 84.9 100.5 64.3 51.5 24 113 397 * 289 358 * 475 390 * 35.4 16.5 24.7 * 150 35 35 4.8 2.1 16 * tetricus Australia (MV) Octopus Mauritania, OvAf12 130 87 96 55 * * 125 405 490 450 425 330 470 335 460 30 10 19 216 150 37 42 * 1.56 18 * vulgaris Africa (MV) Octopus Mauritania, OvAf13 94.4 61.5 99.5 45.2 41.1 26.2 110 327 422 410 375 321 405 354 370 19.8 10 18.9 200 138 40.7 40.1 2.4 1 18.3 6.7 vulgaris Africa (MV) Octopus Mauritania, OvAf14 133.1 80.4 77.6 51 44.6 29.9 119 363 504 518 401 351 441 349 412 22.7 10.9 22.5 230 133 48.6 51 4.22 1.41 20.18 6.51 vulgaris Africa (MV) Octopus Mauritania, OvAf15 130 100 83 43 41.5 26 110 400 510 485 355 410 490 325 367 50.5 10 25 212 122 43 46 4.6 1.62 22.5 6.5 vulgaris Africa (MV) Octopus Mauritania, OvAf16 119 82.8 92.4 52.6 44.4 33.1 112 355 412 453 396 328 404 367 364 23.1 9.8 20.3 198 136 36.8 41.9 3.46 1.01 25.21 6.12 vulgaris Africa (MV) Octopus Mauritania, OvAf17 135 80 110 60 48 35 100 385 400 380 295 360 440 325 350 26.3 9.6 20.8 164 131 42.3 33.5 4.09 1.88 18.6 4.12 vulgaris Africa (MV) Octopus Mauritania, OvAf18 140 95 110 55 55 35 100 335 390 420 330 * 420 320 300 22.13 10.5 23.1 * * 43.4 53.2 4.45 1.94 22.2 7.5 vulgaris Africa (MV) Octopus Mauritania, OvAf19 121.2 65.4 63 44.9 41.9 32.2 105 379 489 483 322 354 396 376 361 20.5 9.4 19 215 131 46 32.1 4.78 1.88 22.4 6.5 vulgaris Africa (MV)

Octopus Mauritania, OvAf20 125 82.8 86.4 54.8 44.5 30.2 92 314 367 332 318 * 433.3 283 281 23.6 10.13 19.4 202 * 45.8 47.5 4.11 1.74 27.9 6.61 vulgaris Africa (MV) Type II Rio Grande OvB11 113 74.4 84.4 39.15 43.3 * 102 500 590 550 458 450 505 400 435 12.2 * 12.4 208 126 * * 42 20 180 55 (Brazil) do Sul, Brazil (UFRN) Type II Rio Grande OvB14 160 90.5 97.3 44.4 56.2 * 172 580 683 710 620 615 720 580 630 10 * 14.4 209 128 * * 58 28 360 90 (Brazil) do Sul, Brazil (UFRN) Type II Rio Grande OvB15 142.45 98 102.8 53 61.8 * 115 500 580 576 520 482 560 502 550 12.8 * 14.4 195 121 * * 60 20 280 70 (Brazil) do Sul, Brazil (UFRN) Itajaí, Santa Type II Catarina, OvB16 102 74.4 83.5 52.1 44 * 115 * 595 560 * 515 610 385 420 9.6 * 10 226 125 * * 73 36 132 58 (Brazil) Brazil (UFRN) Type II Rio Grande OvB17 125 100 92.5 50 * * 120 460 540 * 500 425 560 378 475 20 * 14 * 117 * * 55 25 176 50 (Brazil) do Sul, Brazil (UFRN) Type II Rio Grande OvB18 150 95.6 107.3 48.8 57 * 110 410 570 530 450 450 500 405 440 15.5 * 14.4 * * * * 48 23 258 54 (Brazil) do Sul, Brazil (UFRN) Type II Rio Grande OvB19 110 78.8 83.4 44.4 46.6 * 85 * * 430 330 410 420 262 330 14.4 * 12 * * * * 50 20 124 47 (Brazil) do Sul, Brazil (UFRN) Type II Paraná, OVBPP2 106.41 70.15 62.01 49.56 34.55 27.08 74 287 388 422 378 335 409 311 359 25.06 12.34 16.24 232 135 23.15 23.9 3.61 2.49 14.7 3.81 (Brazil) Brazil (UFPR) Santa Type II Catarina, OvB20 107 75 90.5 53.5 * * 90 302 405 420 370 345 410 315 392 * * * 192 128 * * 61 35 191 43 (Brazil) Brazil (UFRN) Type II OVBPP3 101.27 69.62 65.2 48.93 36.73 20.98 78 368 * 468 417 324 * 354 * 24.93 15.8 15.8 237 146 25.1 29.76 3.79 1.78 15.13 3.72 (Brazil) Bala-Marina (UFPR) Santa Type II Catarina, OVBFL2 100.24 74.61 76.28 55.46 49.41 30.43 91 * 628 593 * 527 643 442 493 30.67 11.38 20.55 225 115 38.71 37.17 6.04 3.06 19.28 5.5 (Brazil) Brazil (UFPR) Octopus Cies Island, OvG11 118 80 75 57 47 32 106 480 650 * 560 434.6 610 460 555 29.8 12.5 29.6 * 173 31.3 29.2 5.24 2.47 22.4 6.6 vulgaris Spain (MV) Octopus Cies Island, OvG12 119.6 82 80.5 61.1 39.6 33.3 115 455 * * 496 380 573 573 490 26.3 13.4 21.7 * 165 35.8 35.1 4.6 1.93 19.06 7 vulgaris Spain (MV) Octopus Cies Island, OvG13 132 100 98 73 24.8 27.1 140 394 530 500 465 405 420 * 485 27 10.8 22 * 183 38.14 41 5.36 2.38 19.27 6.48 vulgaris Spain (MV) Octopus Cies Island, OvG14 83.2 75.5 63.1 51.5 43.7 27.9 122 440 * 545 490 435 565 440 475 25.3 11.4 21.3 298 175 * * 4.7 2.36 17.7 6.04 vulgaris Spain (MV) Octopus Cies Island, OvG15 105 79.1 87.3 58.9 53.3 25.5 127 417 655 599 * 485 674 524 563 28.7 14.4 26.1 * 174 35.9 33.4 4.57 2.33 24 9.23 vulgaris Spain (MV) Octopus Cies Island, OvG16 112.4 73.1 82.3 58.8 42.8 32.5 114 432 601 590 508 456 539 455 519 25.9 13.7 21.9 237 169 32.9 28.8 3.02 2.04 22.73 7.78 vulgaris Spain (MV) Octopus Cies Island, OvG17 134 90.6 91.1 58.4 41.6 30.6 110 520 606 630 489 424 618 464 501 27.5 12.7 23.8 236 156 37.2 48.9 5.99 2.91 23.11 7.35 vulgaris Spain (MV) Octopus Cies Island, OvG18 103 79.2 82.6 55 39.8 26.8 89 387 510 487 405 398 507 455 427 24.5 12.1 22.3 245 165 29.8 31.7 3.69 2.15 14.5 7.65 vulgaris Spain (MV) Octopus Cies Island, OvG19 104.1 80.2 78.3 60.1 47.4 28.7 116 494 670 659 554 494 574 497 562 29.7 11.5 24.2 260 181 46.7 38.7 4.58 2.18 19.6 6.6 vulgaris Spain (MV) Octopus Cies Island, OvG20 134.3 85.5 93.8 65 43 16.4 130 430 * 570 470 434.6 587 430 789 27.1 13 26.7 265 171 38.4 40.3 4.06 1.64 20.75 5.35 vulgaris Spain (MV) Banyules- Octopus seu-Mer, (MNHN)- 210 - 150 90 78 40 200 - - 875 - - - 675 - - 20 - 264 155 - - 9.5 5 - - vulgaris France 1992 Banyules- Octopus seu-Mer, (MNHN)- 105 - 63 58 43 21 75 - - 350 - - - 307 - - 10 - 250 168 - - 4 1.5 - - vulgaris France 3464 Banyules- Octopus seu-Mer, OvF12 108 - 88 55 51 20 104 - - 390 - - - 371 - - 12.5 - 219 154 - - 5 2 - - vulgaris France (SBMNH) Banyules- Octopus seu-Mer, OvF13 108 - 80 54 45 25 90 - - 418 - - - 390 - - 12 - 240 162 - - 6 2.5 - - vulgaris France (USNM) Banyules- Octopus seu-Mer, OvF14 95 - 71 50 45 25 94 - - 397 - - - 350 - - 12 - 238 226 - - 4.5 2 - - vulgaris France (USNM) 92

Banyules- Octopus seu-Mer, LA89-11 134 - 115 75 55 28 108 - - 626 - - - 497 - - 17 - 286 158 - - 6 2.5 - - vulgaris France (MZUF) Banyules- Octopus seu-Mer, LA89-15 83 - 66 43 35 14 68 - - 390 - - - 280 - - 9 - 219 146 - - 3.5 1.5 - - vulgaris France (MZUF) Banyules- Octopus seu-Mer, (MNHN)- 142 - 85 61 50 29 110 - - 585 - - - 460 - - 12 - 180 144 - - 7 2 - - vulgaris France 1991 Octopus OvSp1 119.8 - 92 64.8 52.1 22.9 100 - - * - - - 366 - - 18.31 - * 144 - - 5.17 2.05 - - vulgaris Rosas, Spain (MV) Pisagua and MNCN- Octopus Tocopilla, 11V 130 - 74 51 37 22 122 - - 482 - - - 393 - - * - 248 149 - - 6 2 - - mimus Chile (CSIC) Pisagua and MNCN- Octopus Tocopilla, 12V 143 - 82 52 38 17 102 - - * - - - 412 - - * - * 129 - - 6 2 - - mimus Chile (CSIC) * = missing data, - = data not available from source, institutions; AM = Australia Museum, CSIC = Consejo Superior de Investigaciones Científicas (Spain), MNHN = Muséum National d’Histoire Naturelle (France), MV = Museum Victoria (Australia), MZUF = Museo Zoologico “La Specola”, Università di Firenze (Italy), OUC = Ocean University of China, SBMNH = Santa Barbara Museum of Natural History (California), UFPR = Federal University of Paraná (Brazil), UFRG = Federal University of Rio Grande (Brazil), UFRN = Federal University of Rio Grande do Norte (Brazil), USNM = National Museum of Natural History, Smithsonian Institution (Washington).

Table S3: Raw female morphological data for Octopus vulgaris species group individuals and close relatives.

Catalogue# MLd* MLv MW* HW* FL* FFL* WD* ALL1 ALL2 ALL3* ALL4 ALR1 ALR2 ALR3 ALR4 AW SDn* SCL3* SCR3 GLL GLR User ID Species/Type Location (Institution) OcftA3 F160325 Octopus cf. tetricus Western Australia (MV) 97.3 68.7 59.3 48 33.5 26.8 89 291 355 350 336 270 330 355 305 17.5 9.7 257 266 * * OcftA4 F200327 Octopus cf. tetricus Western Australia (MV) 88.3 64.1 64.3 51.9 30 16.4 89 295 390 * 369 * * 277 371 20.4 12 290 253 21.6 19.5 OcftA5 F200331 Octopus cf. tetricus Western Australia (MV) 119.6 77.8 69 57.4 43.2 24.2 119 419 438 457 * 367 439 * * 25.7 13.2 276 * 37.7 44.9 OcftA6 F200334 Octopus cf. tetricus Western Australia (MV) 114.1 99.1 97 61.5 47.5 23.9 139 407 * 474 449 431 * 490 * 31.4 15 235 198 44.3 37.8 OcfvC2 Fujian Province, OUC15 Octopus sinensis China (OUC) 114.22 75.78 79.24 39.76 42.47 24.22 101 330 372 381 357 312 435 449 321 16.69 13.03 * * * * OcfvC3 Zhejiang Province, OUC16 Octopus sinensis China (OUC) 124 59.95 69.59 41.99 39.5 23.25 89 233 * * * 251 * 331 320 20.02 * * 188 41.82 38.42 OcfvC5 Fujian Province, OUC2 Octopus sinensis China (OUC) 123.3 81.3 83.3 42.6 46.9 42.1 94 375 474 455 380 365 465 457 387 18 9.2 227 205 46.8 43.9 IGG304 OcfvJ1 Octopus sinensis Kyushu, Japan (MV) 89.9 54.9 52 30.3 39.5 28.6 100 305 385 365 325 330 410 414 275 17.9 9.9 212 271 30.1 21.2 IGG305 OcfvJ2 Octopus sinensis Kyushu, Japan (MV) 111 75 75 35 40.7 30.1 95 310 365 380 317 280 340 340 340 16.8 10.9 227 230 34.7 32.8 IGG306 OcfvJ3 Octopus sinensis Kyushu, Japan (MV) 82.5 64.7 57.2 45.1 31.1 25.5 83 273 355 335 310 257 270 327 314 18.9 11.3 254 250 28.8 27 IGG308 OcfvJ4 Octopus sinensis Kyushu, Japan (MV) 100.6 59 80 42.7 34.2 27.1 89 405 449 452 398 405 432 414 360 16.9 13.1 220 231 42.4 38.8 IGG313 OcfvJ5 Octopus sinensis Kyushu, Japan (MV) 94 62.3 61.1 41.8 26 15.4 75 295 355 356 331 278 374 367 325 20.1 9.2 * * 35.1 27.9 OcfvK3 North Mayer Is. c.975975 Octopus sinensis Kermadec Is (AM) 109.18 72.44 91.94 55.83 45.21 28.1 126 418 * 496 469 * * * 456 34.48 19.8 272 * 31.83 35.3 OcfvK4 Roal Is. Kermadec AIMMA119968 Octopus tetricus Is. (AM) 98.69 74.84 76.81 49.95 45.81 28.23 94 335 429 375 285 318 * 401 268 20.58 11.19 302 308 24.8 22.44 OcfvT1 OcfvT1 Octopus sinensis Yilan County, Taiwan (MV) 117.9 78.8 38.5 42.2 42.9 23.1 109 309 * 376 321 314 402 417 * 20 9.9 189 200 37.95 40.6 OcfvT18 OcfvT18 Octopus sinensis Keelung, Taiwan (MV) 97.5 66.7 79.8 51.7 43.5 30.3 99 * 360 366 * 279 358 377 337 19.6 8.6 208 * 39.3 43.6 OcfvT19 OcfvT19 Octopus sinensis Keelung, Taiwan (MV) 129.6 96.7 91.4 51.9 49.9 33.4 101 325 423 417 392 372 414 434 379 24.6 11.4 215 228 51.9 51.9 OcfvT2 OcfvT2 Octopus sinensis Yilan County, Taiwan (MV) 132.7 95.5 91.8 52.6 52.6 27.3 94 391 460 480 * 364 400 393 341 19.3 10.3 163 * 62.6 49.2 OcfvT3 OcfvT3 Octopus sinensis Yilan County, Taiwan (MV) 108.2 81.5 63.4 36.7 35.3 29.4 121 247 429 343 358 339 * 385 353 17.6 8.7 193 194 37.1 41.1 OcfvT4 OcfvT4 Octopus sinensis Yilan County, Taiwan (MV) 106.2 78.2 69.3 40 48.8 24.9 78 316 * 371 335 319 404 368 315 16.6 8.8 196 214 * * OcfvT5 OcfvT5 Octopus sinensis Yilan County, Taiwan (MV) 116.1 80.9 83.5 49.3 51 33.4 95 368 * 462 340 320 381 489 393 20.5 10 191 216 47.4 48 OcfvT6 OcfvT6 Octopus sinensis Yilan County, Taiwan (MV) 133.1 97.9 73.2 39.9 49.8 22.8 99 395 * 454 396 * 445 468 376 20.8 9 243 238 51.8 59.6 OcfvT7 OcfvT7 Octopus sinensis Yilan County, Taiwan (MV) 106.5 68.7 76 43.2 46 31.3 86 297 392 371 312 346 371 383 307 19.3 7.4 202 208 40.7 42.5 OcfvT8 OcfvT8 Octopus sinensis Yilan County, Taiwan (MV) 135.4 95.2 83.5 45.8 52.3 32.2 123 432 535 503 * * 534 * 450 24.2 11.8 228 * 47.3 49.3 OiB1 OiB1 St Peter St Paul Octopus insularis Archipelago, Brazil (UFRN) 113.01 95.5 91.62 54.76 42.31 19.85 110 * 381 * * * * * * 29.21 14.4 221 * 24.66 28.36 Fernando de OiB11 OiB11 Octopus insularis Noronha, Brazil (UFRN) 95 64.5 70 44.5 31 * 75 * 242 278 * 271 * 280 274 14.2 * * 229 * * Fernando de OiB15 OiB15 Octopus insularis Noronha, Brazil (UFRN) 104 75 75.5 47.7 34.5 * 73 260 * 271 304 290 * 302 * 10 * * * * * 94

Fernando de OiB16 OiB16 Octopus insularis Noronha, Brazil (UFRN) 95 64 75 44.6 36.3 * 95 230 280 * 260 256 274 290 290 11 * * * * * St Peter St Paul OiB20 OiB20 Octopus insularis Archipelago, Brazil (UFRN) 120 91 76 50.5 41.6 * 94 * 320 * * * * * * * * * * * * OiB7 Rio Grande do MOA-DOL-42 Octopus insularis Norte, Brazil (UFRN) 109.1 61.46 92.83 48.71 45.67 27.63 119 * 423 * 433 409 450 * 462 28.02 11.79 * * * 43.07 OtA3 New South Wales, F200318a Octopus tetricus Australia (MV) 104.5 83.2 80.1 52.2 35.9 32.1 100 336 397 437 397 333 387 409 364 28.8 15.2 265 290 38.7 37.9 OtA4 New South Wales, F200318c Octopus tetricus Australia (MV) 120.2 89.4 65.4 44.6 42 36.4 105 379 500 447 425 352 412 435 427 25.4 12.2 235 224 48.7 37.3 OtA5 New South Wales, F200319 Octopus tetricus Australia (MV) 117.8 82.7 65.4 31.6 35.9 23.8 96 286 342 315 308 288 369 335 * 19.5 12 218 229 41.3 44.5 OtA7 New South Wales, F200320 Octopus tetricus Australia (MV) 126.3 94.6 74.8 38.1 34.2 26 106 338 449 452 449 360 449 418 400 17.9 11.8 238 254 33.6 38.6 OvAf1 OvAf1 Octopus vulgaris Mauritania (MV) 114.14 60.14 97.6 45.3 42.8 27.7 122 310 357 434 355 * 456 394 347 20.9 10.8 186 169 40.9 44.4 OvAf10 OvAf10 Octopus vulgaris Mauritania (MV) 139.3 82.1 110.2 53.9 52.3 32 122 * 413 671 * 443 * 551 * 25.8 11.2 * 212 53.2 55.4 OvAf11 OvAf11 Octopus vulgaris Mauritania (MV) 137.9 76 87.1 50.8 54 36.1 112 381 490 468 432 361 531 466 434 25.5 11.8 227 225 56.6 52.9 OvAf2 OvAf2 Octopus vulgaris Mauritania (MV) 114.1 60.1 97.6 45.3 42.8 27.7 122 310 357 434 355 * 456 394 347 20.9 10.8 183 * 40.9 44.4 OvAf3 OvAf3 Octopus vulgaris Mauritania (MV) 115.5 84 104.8 52.4 51.9 36.8 128 373 483 495 418 405 472 452 331 29.3 12 201 185 46.7 44.4 OvAf4 OvAf4 Octopus vulgaris Mauritania (MV) 110.5 75.4 94.9 42.9 40.9 39.4 105 339 417 471 382 341 416 432 351 23.1 9.8 208 220 45 47.7 OvAf5 OvAf5 Octopus vulgaris Mauritania (MV) 141.6 89.3 89.8 54.7 51.3 36.1 121 377 461 410 * 381 460 395 383 25.2 11.8 205 190 59.5 51.4 OvAf6 OvAf6 Octopus vulgaris Mauritania (MV) 103.7 76.1 81.4 49 33.8 31.4 109 366 495 431 363 326 399 429 336 23.7 11.1 195 192 53.3 50 OvAf7 OvAf7 Octopus vulgaris Mauritania (MV) 126.3 78.2 75.8 43.8 50.8 31.9 109 444 579 534 440 370 * 456 435 18.6 9.7 230 205 48.4 49.1 OvAf8 OvAf8 Octopus vulgaris Mauritania (MV) 136.2 83.1 96.2 53.2 42.4 28.2 115 344 335 440 386 405 474 452 * 222.8 10.7 190 172 56.7 53.4 OvAf9 OvAf9 Octopus vulgaris Mauritania (MV) 133.3 83 93.8 56.7 54.7 36.8 112 352 459 485 375 357 454 445 403 27.6 11.2 182 170 55.5 52.1 OvB1 OvB1 Santa Catarina, Type II (Brazil) Brazil (UFPR) 105.7 82.19 74.33 53.02 40.76 29.19 73 268 358 362 296 281 333 376 318 22.32 13.79 216 229 29.43 29.65 Rio Grande do Sul, OvB10 OvB10 Type II (Brazil) Brazil (UFRN) 140 82.4 108 48.2 52.8 * 131 555 600 760 600 595 700 670 570 8.5 * * 221 * * Rio Grande do Sul, OvB12 OvB12 Type II (Brazil) Brazil (UFRN) 124 84.15 81.3 43.5 41 * 79 * * 450 360 258 430 447 388 7.2 * * 215 * * Rio Grande do Sul, OvB13 OvB13 Type II (Brazil) Brazil (UFRN) 172 111.5 112.2 50.5 51.6 * 115 * 595 * 530 497 584 585 475 10.6 * * * * * Santa Catarina, OvB5 OvB5 Type II (Brazil) Brazil (UFRG) 98.12 74.93 72.99 46.59 39.32 24.01 100 * 460 469 474 500 546 550 495 16.33 11.88 212 229 32.14 34.52 Rio Grande do Sul, OvB8 OvB8 Type II (Brazil) Brazil (UFRN) 127 87 72 40 43 * 105 450 * 470 460 * 540 450 450 8 * * * * * Rio Grande do Sul, OvB9 OvB9 Type II (Brazil) Brazil (UFRN) 142 90.7 91.55 42.4 50.85 * 112 550 620 630 575 530 650 630 543 9 * 220 221 * * OvG1 OvG1 Octopus vulgaris Cies Island, Spain (MV) 123.5 90.9 76.1 60.7 48.7 27.4 127 495 * 600 541 495 586 589 520 26.3 17.1 291 289 42.4 35 OvG10 OvG10 Octopus vulgaris Cies Island, Spain (MV) 115.3 87.2 75.3 51 41.5 23.2 119 427 558 536 422 401 544 534 435 25.9 18.2 258 260 33.6 31.8 OvG2 OvG2 Octopus vulgaris Cies Island, Spain (MV) 101.1 72.2 74.4 50.5 40.8 24.9 102 431 * 457 468 * 471 504 444 23.8 14.6 212 248 32.1 30.2 OvG3 OvG3 Octopus vulgaris Cies Island, Spain (MV) 130.8 86.3 88.7 65 43.4 36.1 105 430 556 497 458 475 498 575 562 274.6 25.6 275 271 28.2 26.6

OvG4 OvG4 Octopus vulgaris Cies Island, Spain (MV) 115.2 75.7 83.9 56.6 43.8 32.8 105 442 534 524 524 472 569 610 550 25.9 15.7 255 286 * * OvG5 OvG5 Octopus vulgaris Cies Island, Spain (MV) 139.8 94.3 90.5 63.1 53.2 26.7 119 399 540 554 495 457 592 614 500 31.3 18.5 252 263 31.95 33 OvG6 OvG6 Octopus vulgaris Cies Island, Spain (MV) 109.3 77.7 92.7 67.3 44.12 33.7 110 420 551 550 469 * * 565 498 29.3 18.3 245 237 26.1 31.7 OvG7 OvG7 Octopus vulgaris Cies Island, Spain (MV) 134.4 95.2 76.6 61.2 44 23 104 467 * * 521 * 541 560 501 28.2 18.1 * 264 28 27 OvG8 OvG8 Octopus vulgaris Cies Island, Spain (MV) 111 87 55.6 72.6 33.9 30.5 120 * 49.5 555 48 * 473 * * 23.4 18.8 288 * 23.8 27.2 OvG9 OvG9 Octopus vulgaris Cies Island, Spain (MV) 121.4 92.6 64.1 58.9 40.4 26.3 100 422 * 570 496 430 510 561 464 29.7 18.7 270 261 42.5 38.1 Pisagua and 180 91 44 50 32 105 593 21 292 Om3 Octopus mimus Tocopilla, Chile (MNCN)9V ------Pisagua and 170 98 54 43 28 112 510 20 297 Om4 Octopus mimus Tocopilla, Chile (MNCN)10V ------Pisagua and 165 85 48 50 30 82 560 19 292 Om5 Octopus mimus Tocopilla, Chile (MNCN)8V ------Pisagua and 120 75 39 38 25 110 660 23 272 Om6 Octopus mimus Tocopilla, Chile (MNCN)6V ------Banyules-seu-Mer, 134 100 65 63 30 122 584 20 266 OvF1 Octopus vulgaris France (MNHN)-1993 ------Banyules-seu-Mer, 140 90 56 50 24 88 408 19 226 OvF2 Octopus vulgaris France (MNHN)-1994 ------Banyules-seu-Mer, 112 78 46 44 23 75 331 17 238 OvF3 Octopus vulgaris France (MNHN)3465 ------Banyules-seu-Mer, OvF4 115 88 64 46 21 98 515 16 268 OvF4 Octopus vulgaris France (SBMNH) ------Banyules-seu-Mer, OvF5 90 73 45 44 19 77 353 18 230 OvF5 Octopus vulgaris France (USNM) ------Banyules-seu-Mer, OVf6 84 62 48 42 14 80 351 18 253 OvF6 Octopus vulgaris France (USNM) ------* = missing data, - = data not available from source, institutions; AM = Australia Museum, CSIC = Consejo Superior de Investigaciones Científicas (Spain), MNHN = Muséum National d’Histoire Naturelle (France), MV = Museum Victoria (Australia), MZUF = Museo Zoologico “La Specola”, Università di Firenze (Italy), OUC = Ocean University of China, SBMNH = Santa Barbara Museum of Natural History (California), UFPR = Federal University of Paraná (Brazil), UFRG = Federal University of Rio Grande (Brazil), UFRN = Federal University of Rio Grande do Norte (Brazil), USNM = National Museum of Natural History, Smithsonian Institution (Washington).

Table S4: Canonical Analysis of Principal Components (CAP), based on 25 morphological traits of male Octopus vulgaris species group and O. insularis individuals. Values represent the number of individuals classified into a respective group based on their morphology.

Original group O. vulgaris s. s. O. sinensis Brazil (Type II) O. insularis O. tetricus O. cf. tetricus Correct (%)

O. vulgaris s. s. 16 3 84.2

O. sinensis 1 12 1 1 1 75.0

Brazil (Type II) 1 9 1 81.8

O. insularis 1 8 1 2 66.7

O. tetricus 1 4 80.0

O. cf. tetricus 5 100

Table S5: Canonical Analysis of Principal Components (CAP) results, based on

12 morphological traits among 27 male Octopus vulgaris sensu stricto individuals.

Original group Galicia Mediterranean Mauritania Correct (%)

Galicia 8 2 80

Mediterranean 1 8 88.9

Mauritania 8 100

Table S6: Canonical Analysis of Principal Components (CAP), based on 20 morphological traits of female Octopus vulgaris species group and O. insularis individuals. Values represent the number of individuals classified into a respective group based on their morphology.

Original group O. vulgaris s. s. O. sinensis Brazil (Type II) O. insularis O. tetricus O. cf. tetricus Correct (%)

O. vulgaris s. s. 16 4 1 76.2

O. sinensis 5 10 1 2 2 50

Brazil (Type II) 1 1 5 1 71.4

O. insularis 6 100

O. tetricus 1 3 75

O. cf. tetricus 1 1 2 50

Table S7: Canonical Analysis of Principal Components (CAP) results, based on ten morphological traits among 27 female Octopus vulgaris sensu stricto individuals.

Original group Galicia Mauritania Mediterranean Correct (%)

Galicia 9 1 90

Mauritania 11 100 Mediterranean 6 100

Table S8: Pairwise comparisons of male members of the Octopus vulgaris species group and close relatives, based on 12 morphological traits. Lower left diagonal represents PERMANOVA results with significant differences (p=<0.05) highlighted in bold. Upper right diagonal represents results of SIMPER analyses showing traits that contribute most to variation between groups. SIMPER results are also shown in bold if corresponding PERMANOVA showed a significant difference.

O. vulgaris s. s. O. sinensis Type II (Brazil) O. insularis O. tetricus O. cf. tetricus O. mimus

O. vulgaris s. s. - FL CL LL/SCR3 SCR3 SCR3/HW FL/FFL

O. sinensis 0.005* - CL LL/SCR3 SCL3SCR3 SCR3/ALL3 MW/FFL

Type II (Brazil) 0.002 0.001 - SCR3/CL SCR3/CL SCR3/CL FFL/CL

O. insularis 0.001* 0.001 0.001 - SCL3SDn SDn FFL/LL

O. tetricus 0.002 0.001 0.001 0.002 - SCR3/WD SCL3/FL

O. cf. tetricus 0.235 0.024 0.005 0.006 0.05 - FFL/ALL3

O. mimus 0.107* 0.179* 0.047 0.009 0.04 0.243 -

Table S9: Canonical Analysis of Principal Components (CAP), based on 12 morphological traits of male Octopus vulgaris species group individuals. 59 out of

79 (70.9%) individuals were correctly assigned to their respective groups.

Original group O. vulgaris s. s. O. sinensis Brazil (Type II) O. insularis O. tetricus O. cf. tetricus O. mimus Correct (%)

O. vulgaris s. s. 19 4 2 3 67.9

O. sinensis 14 1 1 87.5

Brazil (Type II) 1 8 2 72.7

O. insularis 2 8 1 1 66.7

O. tetricus 1 4 80.0

O. cf. tetricus 1 2 40.0 2 O. mimus 1 1 50

Table S10: Pairwise comparisons of female Octopus vulgaris species group

individuals and close relatives, based on 10 morphological traits. Lower left

diagonal represents PERMANOVA results with significant differences (p=<0.05)

highlighted in bold. Upper right diagonal represents results of SIMPER analyses

showing traits that contribute most to variation between groups. SIMPER results

are also shown in bold if corresponding PERMANOVA showed a significant

difference.

O. vulgaris s. s. O. sinensis Type II (Brazil) O. insularis O. tetricus O. cf. tetricus O. mimus

O. vulgaris s. s. - ALL3/SDn WD/HW ALL3/SCL3 MW/FFL HW SCL3

O. sinensis 0.012 - WD SCL3 FL/FFL HW SCL3/FL

Type II (Brazil) 0.059 0.176 - SCL3 FFL HW SCL3

O. insularis 0.001 0.001 0.001 - ALL3/MW ALL3 HW/SCL3

O. tetricus 0.006 0.049 0.056 0.005 - HW SCL3/WD

O. cf. tetricus 0.465 0.034 0.029 0.004 0.087 - HW/WD

O. mimus 0.001 0.001 0.044 0.003 0.21 0.063 -

Table S11: Canonical Analysis of Principal Components (CAP), based on 10

morphological traits of female Octopus vulgaris species group individuals and

close relatives. 34 out of 72 (47.2%) individuals were correctly assigned to their

respective groups.

Original group O. vulgaris s. s. O. sinensis Brazil (Type II) O. insularis O. tetricus O. cf. tetricus O. mimus Correct (%)

O. vulgaris s. s. 10 10 2 2 2 1 37

O. sinensis 1 8 4 3 4 40

Type II (Brazil) 2 4 1 57.1

O. insularis 5 1 83.3

O. tetricus 1 1 2 50

O. cf. tetricus 1 75 3 O. mimus 1 1 2 50

100

4. Genome-wide sequencing uncovers cryptic diversity and mito- nuclear discordance in the Octopus vulgaris species complex.

4.1 Abstract

Cryptic speciation in the marine environment is a more common occurrence than previously thought. Morphological similarity among closely related species can lead to reduced estimates of diversity, which can have significant implications for the management and conservation of taxa. The present study investigated the species-level relationships among the O. vulgaris species group. 447 genome wide loci were obtained via double digest RADseq. Phylogenetic analyses, genome wide concordance and species tree estimation support the distinction of six species within the O. vulgaris species group; O. vulgaris s. s. (Mediterranean and north-east Atlantic), O. vulgaris Type II (southern Brazil) and O. vulgaris

Type III (South Africa), O. tetricus (east Australia), O. cf. tetricus (west Australia) and O. sinensis (Asia). Members of the O. vulgaris group comprise the most valuable octopod fishery in the world. The improved species-level resolution provided by this study improves our understanding of species distributions within this group. These findings have significant implications for the naming and description of species, which may inform the appropriate management of the group via improved accuracy of global fisheries catch statistics.

101

4.2 Introduction

It is now well established that many marine organisms once proposed to be cosmopolitan are in fact cryptic species with relatively restricted geographic ranges. Underestimated species diversity often results from an inability to identify distinguishing morphological differences, which do not always correspond with genetic change and/or speciation (Bickford et al., 2007). This results in cryptic taxa being ‘lumped’ into single morphospecies, despite the potential for being genetically distinguished. Recent studies employing molecular-based techniques have detected cryptic species across a wide range of invertebrates. Examples are particularly common among marine invertebrates (Knowlton, 1993), spanning several phyla including Porifera (Solé-Cava et al., 1991), Cnidaria (Dawson &

Jacobs, 2001), Echinodermata (Sponer & Roy, 2002), Bryozoa (Gomez et al.,

2007), Rotifera (Suatoni et al., 2006) and Mollusca (Hunt et al., 2010; Norman et al., 2014a). Within molluscs cryptic species are commonly reported in cephalopods, especially squids and octopuses (Söller et al., 2000; Allcock, 2005;

Leite et al., 2008; Allcock et al., 2011; Amor et al., 2014; Norman et al., 2014a,

2014b; Reid & Wilson, 2015).

Species-level taxonomy within the family Octopodidae d'Orbigny, 1839 has been hindered due to morphological plasticity of octopuses (Voight, 1994). The soft bodies on which standardised external measurements are based (Robson, 1929;

Roper & Voss, 1983) are subject to distortion upon preservation (Burgess, 1966;

Voight, 2001) and octopods have few hard structures (Bookstein et al., 1985) to base morphological comparisons on. One Octopus species group likely contains a suite of species similar in morphology and behaviour to the type species of the

102 genus, Octopus vulgaris Cuvier, 1797. This group has been named the ‘Octopus vulgaris species group’ based on general similarities in overall size, mantle shape, arm length and skin sculpture (Robson, 1929). More recently, this group has been considered the only valid species group belonging in the Octopus genus in the strict sense, designated by the term Octopus sensu stricto (s. s.;

O'Shea, 1999).

Octopuses are highly valued as a fisheries resource world-wide, with global production exceeding 350,000 tonnes and a total export value of US$1.07 billion

(FAO, 2012), which exceeds many valuable finfish fisheries. A major limitation of the global catch statistics reported by the FAO is the poor state of octopus taxonomy, as of the estimated 100 species of octopus harvested only four (O. vulgaris, O. maya, Eledone cirrhosa and E. moschata) are listed in global statistics (Norman & Finn, 2014). Octopus vulgaris is a highly valued fisheries target species throughout its range, particularly in north-west Africa, the largest single-species octopus fishery in the world (FAO, 2012). However, the majority of octopus fisheries world-wide are in decline (Norman & Finn, 2014), which highlights the requirement for improved taxonomy to allow more accurate species-level identification and inform more sustainable management strategies.

Historically O. vulgaris was considered a cosmopolitan species. First reported from the Mediterranean Sea and north-east Atlantic, O. vulgaris has been reported from Australasia, Europe, Africa, Asia and the Americas. However, recent analyses based on mitochondrial (mt) DNA (Söller et al., 2000; Leite et al., 2008; Acosta-Jofré et al., 2012; Amor et al., 2014, 2015, 2016) suggest populations previously treated as O. vulgaris comprise a complex of morphologically similar but genetically distinct O. vulgaris-like species (the

103

‘Octopus vulgaris species complex’). Within the O. vulgaris species complex, several hypothesised species ‘Types’ have been described (Norman et al.,

2014a). Octopus vulgaris s. s. occurs in the Mediterranean and north-east

Atlantic (Fig. 1). Type I occurs in the Caribbean and Gulf of Mexico. Type II occurs in the south-west Atlantic along the coast of Brazil. Type III occurs in the

South Atlantic and Indian Oceans, along the coast of South Africa. Octopus sinensis d’Orbigny, 1841 occurs in the north-west Pacific Ocean, from Japan to

China and Taiwan, as has also been recorded in the south-west Pacific Ocean, around the Kermadec Islands (Reid & Wilson, 2015).

Previous attempts at phylogenetic resolution of Octopus vulgaris have incorporated a limited number of genetic markers derived from mtDNA (Söller et al., 2000; Oosthuizen et al., 2004; Warnke et al., 2004; Guerra et al., 2010; Sales et al., 2013; Reid & Wilson, 2015). The use of a single marker (DNA barcoding) is often used to aid in the discovery of new species or assign individuals with unknown taxonomy to an existing species (Hebert et al., 2003, 2004). However,

DNA barcoding is potentially confused with attempts to resolve phylogenetic relationships, of which it has limited capacity (Moritz & Cicero, 2004). Previous studies have shown DNA barcoding of well-studied groups to be relatively accurate for species-level inference (Hebert et al., 2003), however investigations of groups with incomplete sampling (Meyer & Paulay, 2005) and high intraspecific variability (Meier et al., 2006) performed relatively poorly.

Previously, mtDNA had been used to determine the distinct species status of O. insularis (Leite et al., 2008), O. cf. tetricus and O. tetricus (Amor et al., 2014), with Amor et al., (2014) also showing significant morphology-based differences.

The finding that O. insularis, O. tetricus and O. cf. tetricus are distinct species

104 was supported by a recent morphological investigation into the O. vulgaris group

(Amor et al., 2016). Although mtDNA supported the species-level resolution observed for these three taxa in the present study, several inconsistencies, such as differing levels of phylogenetic resolution, clade support and alternate topologies were also evident.

The shortcomings of using mtDNA for species level inferences are well noted

(Edwards et al., 2005; Edwards & Bensch, 2009). Rapidly evolving mtDNA is subject to homoplasy, whereby a single base position is the site of repeat mutations, potentially masking evolutionary signal at deeper timescales. The haploid and maternally inherited nature of mtDNA also means its effective population size is one quarter that of nuclear DNA, therefore it is more likely to reveal reduced diversity in species that have undergone recent population contractions/bottlenecks. Finally, mtDNA evolves as a single linked locus, therefore the use of multiple mtDNA genes for species tree inference can be problematic due to the random nature of lineage sorting during speciation.

The advancement of sequencing technology from traditional Sanger sequencing to next-generation sequencing (NGS) has allowed for the relatively cheap production of extremely large volumes of data (Metzker, 2010). NGS has enabled rapid genotyping of thousands of markers from virtually any genome of interest (Stapley et al., 2010), even with little or no previous available genetic information (Davey et al., 2011). This is of particular interest for studying non- model organisms such as the O. vulgaris species complex where no reference genome is available. This study aims to utilise genome-wide data to investigate the phylogenetic and species-level relationships among members of the O.

105 vulgaris species group. Furthermore, we will contrast our findings with those from previous mtDNA-based studies.

4.3 Methods

4.3.1 Sampling

Tissue samples of individuals belonging to the O. vulgaris species group and close relative, O. insularis were obtained from 16 localities around the world (Fig.

1, Table 1), and were stored in ~90% ethanol at -80°C until processing.

Fig. 1: Sampling locations (triangles) of Octopus vulgaris species group and O. insularis individuals included in the present study. Distributions of O. vulgaris sensu stricto and species ‘Types’ are shaded as per Norman et al., (2014a); purple: O. vulgaris s. s., blue: Type I (Caribbean/ Gulf of Mexico), green: Type II 106

(Brazil) and orange: Type III (South Africa). Distributions of non-O. vulgaris species are shaded in yellow: O. sinensis, dark blue: O. cf. tetricus, light blue: O. tetricus and red: O. insularis.

Table 1: Sampling localities of Octopus vulgaris species group, and O. insularis individuals included in the present study.

Species Location n Latitude Longitude

O. vulgaris s. s. Perpignan, France 3 42.48352 3.13145 O. vulgaris s. s. Galicia, Spain 2 42.22719 -8.89403 O. vulgaris s. s. Tenerife, Spain 2 28.29156 -16.62913 O. vulgaris Type II Rio de Janeiro, Brazil 3 -23.15000 -44.23330 O. vulgaris Type III Hamburg, South Africa 3 -33.29446 27.48093 O. sinensis Zhejiang Province, China 2 30.71616 121.36597 O. sinensis Zhejiang Province, China 1 26.17578 119.63425 O. sinensis Yilan County, Taiwan 3 24.94063 121.90000 O. sinensis Hato-No-Kama, Japan 3 32.60504 130.41029 O. tetricus Merimbula, Australia 3 -36.89175 149.91045 O. tetricus Flinders Is, Australia 1 -39.98364 148.05269 O. tetricus Leigh, New Zealand 1 -36.29165 174.80965 O. cf. tetricus Mandurah, Australia 3 -32.12540 115.75856 O. insularis Natal, Brazil 3 -3.85381 -32.42379 O. insularis Ascension Island 2 -7.94672 -14.35592 O. insularis Georgetown, St Helena 1 -15.96501 -5.70892

4.3.2 Library preparation and sequencing

Genomic DNA was extracted from mantle or arm tissue (~1 mm3) using a

QIAGEN DNeasy Blood and Tissue Kit according to the manufacturer’s instructions, except for the final elution which was repeated twice in a single aliquot of 57°C ‘low TE buffer’ to increase DNA yield. Where possible, skin was

107 trimmed from tissue as a noticeable decrease in PCR efficiency was observed when skin was included in the reaction. We therefore recommend the removal of skin from octopus samples during NGS library preparation protocols to reduce the amount of PCR inhibitors and contaminants. A double digest RAD library was prepared using a modified version of the Peterson et al., (2012) protocol (written by SRD and MDA - available at http://dx.doi.org/10.13140/RG.2.1.1693.8488).

Samples were randomly allocated to a position in a 96-well, round bottom PCR plate to avoid preparation biases.

Genomic DNA was digested for 18 hours in 30 µL reactions composed of 1 µL

EcoRI-HF and 2 µL ClaI restriction enzymes (New England Biolabs), 3 µL

CutSmart buffer (New England Biolabs) 20 µL DNA solution (50 ng concentration) and 4 µL H2O. Barcodes/adapters were ligated to digested DNA fragments in 40 µL reactions composed of 30 µL DNA digestion solution, 1 µL T4

DNA ligase, 4 µL T4 DNA ligase buffer (New England Biolabs), 1 µL H2O, 2 µL common anti-sense adapter (2 µM) and 2 µL sense adapter (2 µM) which was unique to each sample. Ligation solutions were incubated at 16-18°C for 1 hour then 2 µL EDTA (0.5 M) was added to stop the reaction. Non-ligated adapters were removed using a 0.7x (DNA solution volume) AMPure XP (Agencourt) magnetic bead purification. The purified DNA pellets were suspended in 12 µL

40°C H2O.

20 µL PCR reactions were performed using 0.5 µM of each indexed sense and anti-sense primer and (10 µM), 10 µL KAPA HiFi Real Time PCR master mix

(KAPA Biosystems) and 9 µL size selected DNA solutions. PCR cycle conditions included a single initial denaturing step (98°C for 2 minutes) and 18 cycles of denaturing (98°C for 15 seconds), annealing (60°C for 30 seconds) and

108 extension (72°C for 30 seconds). Amplified DNA solutions were purified using

AMPure XP beads/PEG 6000 solution (0.7x DNA solution volume), quantified using a Qubit® 2.0 Fluorometer (Invitrogen) and pooled in eqimolar concentrations (15.4 ng). The pooled library was ran on a 1.5% agarose gel and fragments between 300-350 base pairs were excised and purified using a

Wizard® SV Gel and PCR Clean-Up System (Promega). The size selected library was diluted to 12 pM and sequencing was performed using a 600 cycle

(paired-end) v3 MiSeq Reagent Kit on an Illumina MiSeq with 10% PhiX spiked into the run.

4.3.3 Quality filtering and bioinformatics pipeline

Raw paired-end reads were merged using PEAR v0.9.4 (Zhang et al., 2014).

Merged and unmerged (read one only) reads were demultiplexed into individual sample read-sets based on their corresponding ligated inline barcode and indexed adapter using the ‘process_radtags’ feature of STACKS v1.2.7 (Catchen et al., 2013). This step was also used to trim all reads to 250 base pairs (bp) as a noticeable decrease in read quality was recorded after 250 bp, after which the remaining low quality reads were discarded (based on phred 30 quality score).

The remaining reads were then filtered to exclude microbial contamination using

Kraken v0.10.4 (Wood & Salzberg, 2014) and unclassified, non-microbial reads were retained for phylogenetic analyses. Unmerged read-two data were not included in analyses as these reads did not meet the assumption that all reads are unlinked.

109

4.3.4 Mapping to mitochondrial genome

Demultiplexed filtered reads for each sample were mapped using Bowtie v2.2.2

(Langmead & Salzberg, 2012) to the Japanese O. vulgaris mitochondrial genome (Genbank accession NC006353). BAM files were generated using

Samtools v0.1.19 (Li, 2011). Mapped reads were imported into CLC genomics

Workbench v7.0.4 (https://www.qiagenbioinformatics.com). A depth of five reads per individual surrounding restriction sites was required before consensus sequences were obtained manually. Multiple sequence alignment was performed on consensus sequences using the Muscle algorithm (Edgar, 2004). jModelTest v0.1.1 (Posada, 2008) was used to carry out statistical selection of best-fit models of nucleotide substitution of the alignment. The most appropriate substitution model (GTR+G) was selected based on ‘goodness of fit’ via the

Akaike Information Criterion (AIC; Akaike, 1974).

Maximum Likelihood (ML) analyses were performed using RAxML v8.0.19

(Stamatakis, 2014). Strength of support for internal nodes of ML construction was measured using 1,000 rapid bootstrap (BS) replicates. Bayesian Inference

(BI) marginal posterior probabilities (PP) were calculated using MrBayes v3.2.5

(Ronquist & Huelsenbeck, 2003). Model parameter values were treated as unknown and were estimated. Random starting trees were used and analysis was run for 15 million generation sampling the Markov chain every 1,000 generations. After removing the initial 10% of samples, split frequencies within

MrBayes as well as the program Tracer v1.5 (Rambaut et al., 2014) were used to ensure independent Markov chains had converged and reached stationarity. 110

4.3.5 De novo assembly

De novo assembly of unclassified RAD loci was performed using PyRAD v3.0.4

(Eaton, 2014). Further sequence quality filtering was performed to convert base calls with a score of <30 into N’s, whilst excluding reads with >5 N’s. A minimum of five reads per individual was required for clustering of putative loci, and those with fewer than five reads were excluded. Reads were clustered at 90% similarity within each individual. Putative loci containing more than two alleles were excluded as potential paralogs. Loci were clustered among samples at 90% similarity threshold. Any locus containing one or more heterozygous sites across more than three samples were excluded. In cases where individuals were missing a given locus, gaps were replaced with N’s in the multiple sequence alignment output.

Three datasets were generated and analysed. The first was composed of all samples (n = 34; O. vulgaris species complex, O. tetricus, O. cf. tetricus and O. insularis). A minimum of 26 individuals (76%) were required for each loci to be included in the final data-matrix. The second data set included only the two individuals with the highest number of reads per taxon (O. insularis included three individuals to include North Brazil and Ascension). A minimum of 12/15 individuals (80%) was required per locus for inclusion in the final data-matrix.

The final data set included only individuals belonging to the O. vulgaris species complex, O. tetricus and O. cf. tetricus (n=28). Of the 28 individuals, a minimum of 21 individuals per locus (75%) was required for inclusion in the final data- matrix. ML phylogenies were constructed using the GTR+G model RaxML 111 v8.0.19 (Stamatakis, 2014). Strength of support for internal nodes of ML construction was measured using 1000 rapid BS replicates.

4.3.6 Genome-wide concordance

Support for sub-optimal topologies was investigated via the partitioned RAD phylogenetic analysis approach (Hipp et al., 2014) using the RADami package

(Hipp, 2014) in R v3.1.1 (R Team, 2015). To visualise the number of loci supporting the optimal tree generated from the O. vulgaris species complex data- matrix, relative to neighbouring sub-optimal trees, a candidate pool of 250 trees was generated via nearest-neighbour interchange (NNI) for comparison with the

RAxML optimal tree (i.e. ‘best tree’). A set of unique trees for each locus was then generated by pruning the 251 trees to only those tips present in each locus.

Site likelihoods for each locus-tree were calculated in RAxML v8.0.19

(Stamatakis, 2014) under the GTR+G model using the original RADseq data- matrix. The likelihood of each tree was then plotted against the number of loci favouring and disfavouring each tree.

4.3.7 Phylogenetic hypothesis testing

To assess confidence of the present studies optimal topology relative to previously published mitochondrial based alternate topologies, the Approximately

Unbiased (AU) test (Shimodaira, 2002) was performed using the software

112 package CONSEL (Shimodaira & Hasegawa, 2001). Three topologies were investigated; the optimal topology generated by analysis of the O. vulgaris species complex RADseq data-matrix, an alternate hypothesis whereby O. sinensis was sister taxon to a clade containing O. tetricus and O. cf. tetricus

(mtDNA; Amor et al., 2014), and a second alternate hypothesis whereby O. vulgaris Type III and O. vulgaris s. s. form a monophyletic clade (mtDNA; Guerra et al., 2010; Amor et al., 2015). Constrained topologies were constructed using the present studies concatenated RADseq data-matrix. Site based likelihoods for each alternate tree were estimated using RAxML v8.0.19 (Stamatakis, 2014) under the GTR+G. Significance values for the AU test and BS/PP values were calculated for each topology.

4.3.8 Species tree estimation

Coalescent based species tree estimation on the Octopus vulgaris species complex data-matrix was conducted following a quartet inference approach

(Chifman & Kubatko, 2014) using the SVDQuartet command in PAUP v4.0a146 for Unix/Linux (Swofford, 2003). The ML based analysis was performed using

100,000 randomly generated quartets and 1,000 BS replicates.

4.4 Results

113

4.4.1 Data assembly

Sequencing resulted in 17.25 million paired end reads that contained a maximum of a single error per 1,000 bases (Phred quality score of 30). Of the quality paired end reads, 15.49 million (89.8%) overlapped and were merged into ‘single end data’. Merged and non-merged (read one only) reads were demultiplexed according to their barcodes, which resulted in 13.8 and 1.7 million reads being retained respectively (90.2% of raw reads). An average of 9% of total reads were classified to the bacterial database and were discarded, therefore 91% of the demultiplexed reads were retained for analysis with an average of 307,316 reads per individual (standard deviation (SD) = 130,692). The 28 O. vulgaris species group individuals (excluding O. insularis) had an average of 336,892 reads per individual (SD = 111,675).

4.4.2 Mapping to mitochondrial genome

A partial ND2 sequence alignment was obtained after mapping all demultiplexed reads to the mitochondrial genome of O. vulgaris (Accession: NC_006353.1).

Sequence data was obtained for O. vulgaris s. s., O. vulgaris Type II, O. vulgaris

Type III, O. sinensis and O. tetricus which all had a single EcoRI cut site within their mt genome. All O. insularis (Brazil, Ascension Island and St Helena) and O. cf. tetricus individuals were missing data for this region, which may be the result of mutation(s) within the five base pair EcoRI cut site. ML analysis of O. vulgaris species group individuals placed the five taxa into four distinct monophyletic 114 clades (BS = >72; Fig. S4) which is consistent with a previous COI based phylogeny of the species group (Amor et al., 2015). Each taxon was placed into a distinct clade with the exception of O. vulgaris s. s. and O. vulgaris Type III which composed a single monophyletic clade.

4.4.3 Phylogenetic inference

Maximum likelihood analysis of the O. vulgaris species complex and O. insularis individuals using the 165 RAD loci supported seven phylogenetic clades (Fig. 2).

Octopus insularis formed a highly supported monophyletic clade distinct form the

O. vulgaris species complex (BS = 100). A large branch length separated O. insularis from members of the O. vulgaris species complex. All O. vulgaris species complex individuals were represented in at least 71% of the 165 RAD loci, however O. insularis individuals were represented in 50% or fewer loci.

Octopus vulgaris Type II (Brazil) was found to be the sister taxon to the remaining O. vulgaris species complex members and was therefore selected to root the phylogeny for subsequent analyses including only O. vulgaris species complex individuals.

115

Fig. 2: Maximum likelihood phylogeny depicting the relationships among members of the Octopus vulgaris species complex, O. insularis, O. tetricus and

O. cf. tetricus. Analysis was conducted using 165 concatenated RAD loci (40,938 base pairs) using the GTR+G evolutionary model in RAxML v8.0.19 (Stamatakis,

2014). ML likelihood values are displayed below major nodes. Tip labels represent sampling localities.

Maximum likelihood analysis of the Octopus vulgaris species group individuals

(excluding O. insularis) based on 447 RAD loci recovered six highly supported monophyletic clades (Fig. 3). Five of the six phylogenetic clades corresponded to

116 the species previously identified based on differences in morphology (Amor et al., 2016). These included O. vulgaris Type II (BS = 100), O. vulgaris s. s. (BS =

100), O. sinensis (BS = 100), O. tetricus (BS = 100) and O. cf. tetricus (BS =

100). Furthermore, individuals from South Africa (O. vulgaris Type III) also formed a distinct monophyletic clade (BS = 100).

117

Fig. 3: Maximum likelihood phylogeny depicting the relationships among members of the Octopus vulgaris species group. Analysis was conducted using

447 concatenated double digest RADseq loci (110,822 base pairs) using the

GTR+G evolutionary model in RAxML v8.0.19 (Stamatakis, 2014). Maximum likelihood bootstrap values are displayed below major nodes. Tip labels correspond to sampling localities in Table 1.

4.4.4 Genome-wide concordance and hypothesis testing

Partitioned RAD analysis (Hipp et al., 2014) was performed using an NNI approach to investigate the support for sub-optimal trees. The RAxML best tree was shown to be the most reliable representation of the RADseq data-matrix, as it was favoured by the greatest number of RAD loci, and disfavoured by fewer than 10 loci. The phylogeny presented in the present study (Fig. 3) contrasted previous mtDNA-based topologies. Topological comparisons conducted via AU test showed that the RAxML best tree was the highest ranked topology (PP = 1;

Table 1). The tree constraining the monophyly of O. vulgaris Type III and O. vulgaris s. s. (based on mtDNA analyses) was ranked second. This tree was not significantly worse than the topology presented in this study, although the probability supporting it was substantially lower (PP = 7E-07). Enforcing the monophyly of O. vulgaris Type III and O. vulgaris s. s. resulted in a sister taxon relationship of two distinct clades, and did not result in a single clade

‘conspecific’ relationship obtained via mtDNA-based analyses. The final constrained topology forced O. sinensis to be the sister taxon to O. tetricus and

O. cf. tetricus, as reported by mtDNA-based analyses of Amor et al., (2014). This 118

tree was ranked third and was significantly worse than the topology obtained in

the present study (P = 0.024, PP = 7E-43).

Table 2: Approximate Unbiased (AU) test results comparing the present studies

phylogenetic topology (Fig. 3) with contrasting results from previous studies. The

Bayesian Posterior probability (PP) calculated under the Bayesian information

criterion (BIC) is also shown. Topologies are rejected with a p-value of <0.05.

Tree rank Constraint p-value PP Reference 1 No constraint 0.671 1 2 Forced monophyly of O. vulgaris s. s. and O. vulgaris Type III 0.402 7.00E-07 Amor et al., (2015) 3 O. sinensis as sister taxon to O. tetricus and O. cf. tetricus 0.024 7.00E-43 Amor et al., (2014)

4.4.5 Species tree estimation

Species tree estimation supported the presence of six species within the

Octopus vulgaris species group (Fig. 4). Each species was highly supported (BS

= >99.1), with O. vulgaris s. s., O. vulgaris Type II, O. vulgaris Type III, O. cf.

tetricus, O. tetricus and O. sinensis supported as distinct species.

119

Fig. 4: Species tree estimation supporting six species within the Octopus vulgaris species group. Analysis was based on 447 concatenated RAD loci (110,822 base pairs) using the SVDQuartets algorithm (Chifman & Kubatko, 2014) implemented in PAUP v4.0a146 for Unix/Linux (Swofford, 2003).

4.5 Discussion

Genome-wide RAD loci were analysed to investigate the phylogenetic and species-level relationships within the highest value fisheries target species of octopus in the world. This study provides evidence for the existence of three cryptic species within the O. vulgaris species complex, which has previously been treated as a single taxonomic unit; O. vulgaris s. s. (Mediterranean and north-east Atlantic), O. vulgaris Type II (southern Brazil) and O. vulgaris Type III

(South Africa). I also confirm that O. insularis is distinct from the O. vulgaris species group, as previously suggested (Leite et al., 2008). In concordance with previous studies, phylogenetic relationships and species tree estimation in this study support the distinct species status of O. tetricus, O. cf. tetricus and O.

120 sinensis, and confirm that New Zealand’s O. gibbsi is a junior synonym of east

Australia’s O. tetricus (Amor et al. 2014).

Several issues have been associated with the use of a single genetic locus

(particularly mtDNA) for inferring phylogenetic relationships, specifically that individual trees based on a single gene (or linked genome such as the mitochondria) do not necessarily reflect species trees (Page & Charleston,

1997). The use of a single genetic marker (DNA barcoding) is often applied to aid in the discovery of new species or assign individuals with unknown taxonomy to an existing species (Hebert et al., 2003, 2004). However, DNA barcoding is commonly confused with attempts to resolve phylogenetic relationships, for which it has limited capacity (Moritz & Cicero, 2004). The accuracy of DNA barcoding for identifying species has been shown in well-studied groups (Hebert et al., 2003), however investigations of groups with incomplete sampling (Meyer

& Paulay, 2005) and high intraspecific variability (Meier et al., 2006) performed relatively poorly.

The shortcomings of relying solely on mtDNA-based analyses for species-level inferences are well noted (Moritz & Cicero, 2004; Edwards et al., 2005; Edwards

& Bensch, 2009). Homoplasy (where a single base position is the site of repeat mutations), post-hybridisation introgression and transfer of mtDNA paralogs to the nucleus are known to potentially ‘mask’ evolutionary signal of mtDNA.

Furthermore, the haploid and maternally inherited nature of mtDNA means its effective population size is one quarter that of nuclear DNA, therefore it is more likely to reveal reduced diversity in species that have undergone recent population contractions/bottlenecks. Finally, mtDNA evolves as a single linked

121 locus, therefore the use of multiple mtDNA genes for species tree inference can be problematic due to the random nature of lineage sorting during speciation.

The present study showed nuDNA-based analyses provided greater resolution than mtDNA, with clear species-level differentiation evident between Octopus vulgaris s. s. and O. vulgaris Type III. Where clear biogeographic inconsistencies are present between mtDNA and nuDNA (as seen in the O. vulgaris group), lineage sorting is often ruled out as a cause of discordance (Toews & Brelsford,

2012). In circumstances where nuDNA shows greater levels of geographic structure, isolation followed by secondary contact and hybridisation is hypothesised to be the most common driver of mito-nuclear discordance.

As mtDNA based analyses still display geographic structure among other members of the O. vulgaris group, a potential explanation for the lack of phylogenetic signal between O. vulgaris s. s. and South Africa may be a result of hybridisation following secondary contact. Teske et al., (2007) investigated the geographic structure of O. vulgaris Type III and identified two distinct mtDNA lineages. The most common lineage was reported to be abundant along the entire South African coastline, whilst the rarer lineage was reported in two individuals from the south-eastern extreme of the known distribution (Durban).

There may therefore be a selective advantage for the foreign mtDNA throughout

South Africa as it appears to be at or near fixation. However, the rate and mode of gene flow between these taxa is unclear, especially considering the relatively restricted geographical ranges of O. vulgaris group taxa. Therefore, further investigation is required to validate this hypothesis.

The relationships among O. sinensis, O. tetricus and O. cf. tetricus identified in the present study contrasted those of previous mtDNA based analyses. Analyses

122 of Amor et al., (2014) showed O. sinensis (treated as Asian O. vulgaris) to be sister taxon to a clade composed of O. tetricus and O. cf. tetricus. The present study rejected this topology, finding the relationship whereby O. cf. tetricus was sister taxon to a clade containing O. tetricus and O. sinensis to be significantly more probable. As the relationships depicted among these taxa in the present study are based on genome-wide evidence, we consider them to be a more reliable representation of the true relationships among O. sinensis, O. tetricus and O. cf. tetricus.

The phylogenetic analysis in the present study also showed overall improvements in support values throughout the tree. In particular, previous studies based on analyses of mtDNA resulted in poor support for the clade containing O. vulgaris Type II individuals (Sales et al., 2013; Amor et al., 2015).

Datasets composed of multiple unlinked nuDNA loci have a far greater ability to resolve phylogenetic relationships in comparison to mtDNA alone (Edwards et al., 2005; Edwards & Bensch, 2009). The present study highlights the advantages of including multiple loci from throughout the genome when investigating species-level relationships in the O. vulgaris group. Future mtDNA- based analyses should incorporate a morphological component or aim to include nuDNA loci to improve the reliability of species-level inferences, or all of the above.

Rather than being able to maintain large-scale genetic homogeneity, each species identified in the present study is relatively restricted in terms of their geographic distribution, despite the dispersal capacity of the planktonic paralarval stage recently described for O. vulgaris s. s. (Roura et al., 2016, in press). Recent studies investigating phylogenetic diversity within the O. vulgaris

123 species group suggest gene flow can be maintained at a distance of 2,000-3,000 kilometers (Amor et al., 2014, 2015). The present study confirms these findings, however two cases where a single species is hypothesised to occur in allopatric northern and southern hemisphere populations could not be verified.

Based on analyses of mtDNA, O. sinensis has the widest distribution of any taxa within the O. vulgaris group. A previous analysis of mtDNA suggests individuals from Asia and the Kermadec Islands are conspecifics (Reid & Wilson, 2015), despite allopatric distributions in the North and South Pacific Ocean. Preliminary mtDNA-based analyses also suggest that individuals from the Caribbean/Gulf of

Mexico (Type I) and South Brazil (Type II) are conspecifics (de Lima et al.,

2017). The above examples suggest these taxa are able to maintain gene flow despite connectivity requiring trans-tropical dispersal. However, as mtDNA is known to underestimate species diversity, and individuals from the

Caribbean/Gulf of Mexico and Kermadec Islands could not be included in the present study, these preliminary findings require validation.

The present study successfully resolved the relationships within the O. vulgaris species complex, however including O. insularis resulted in a substantial decrease in the number of loci obtained from de novo assembly. To obtain a highly supported phylogeny, only two individuals with the highest number of reads per species were included in a subsequent assembly. However, there was substantial non-random missing data in the resulting data-matrix driven by the comparatively large genetic distance between O. insularis and the ingroup.

Sequencing a greater number of reads per individual may be beneficial to future studies investigating diversity within the genus Octopus.

124

Octopuses being exported globally under the name O. vulgaris are of extremely high market value and profile (Norman et al., 2014a). Aquaculture and captive growing of wild caught juveniles are receiving increasing profile and funding, particularly in China. The findings presented here along with broader research by the authors have significant implications for the naming, marketing, value, documentation and potentially conservation of commercially harvested members of this species complex throughout their ranges. The present study increases our understanding of the species diversity and geographic boundaries of species within the O. vulgaris complex and their close relatives. This information has important implications for the appropriate management of this highly valued fisheries resource.

125

4.6 Supplementary information

Table S1: Number of raw reads passing Q30 quality filter (<0.1 error per 100

bases), merged and non-merged reads and retained reads post demultiplexing.

Merging Demultiplexing Merged reads Non-merged reads Retained merged reads Retained non-merged reads Raw reads (percent of raw reads) (Percent of raw reads) (percent of merged reads) (percent of non-merged reads) Index 01 4,459,932 4,012,707 (89.97%) 446,752 (10.02%) 3,514,922 (87.59%) 436,392 (97.68%)

Index 02 3,244,750 2,926,059 (90.18%) 318,343 (9.81%) 2,622,626 (89.63%) 312,830 (98.27%)

Index 03 2,508,086 2,238,066 (89.23%) 269,795 (10.76%) 2,009,842 (89.80%) 265,842 (98.53%)

Index 04 2,660,240 2,380,111 (89.47%) 279,875 (10.52%) 2,138,222 (89.84%) 273,131 (97.59%)

Index 05 1,677,924 1,515,524 (90.32%) 162,226 (9.67%) 1,381,054 (91.13%) 159,104 (98.08%)

Index 06 2,702,199 2,413,451 (89.31%) 288,521 (10.68%) 2,166,441 (89.77%) 283,122 (98.13%)

Total reads 17,253,131 15,485,918 (89.76%) 1,765,512 (10.23%) 13,833,107 (89.33%) 1,730,421 (98.01%)

126

Table S2: Number of merged and non-merged (read one only) reads classified

(discarded) and unclassified (retained) to the bacterial database using Kraken

v0.10.4

Merged reads Un-merged reads (read one only) Index Barcode Demultiplexed Classified Unclassified Demultiplexed Classified Unclassified Total unclassified IDX01 RAD03 617,403 42,932 574,471 77,471 6,239 71,232 645,703 IDX01 RAD04 483,337 30,125 453,212 71,139 4,364 66,775 519,987 IDX01 RAD05 360,064 25,201 334,862 50,787 3,090 47,697 382,559 IDX01 RAD06 259,101 16,962 242,139 34,696 2,121 32,575 274,714 IDX01 RAD08 295,215 19,473 275,742 48,754 2,978 45,776 321,518 IDX01 RAD12 492,858 30,176 462,682 56,252 2,794 53,458 516,140 IDX01 RAD26 647,202 358,045 289,157 60,038 27,989 32,049 321,206 IDX01 RAD48 353,802 23,686 330,116 36,628 2,230 34,398 364,514 IDX02 RAD03 353,261 23,875 329,386 17,729 1,056 16,673 346,059 IDX02 RAD04 385,820 25,956 359,864 54,518 3,609 50,909 410,773 IDX02 RAD05 458,592 30,649 427,943 67,502 4,163 63,339 491,282 IDX02 RAD06 442,511 27,824 414,687 45,772 2,719 43,053 457,740 IDX02 RAD12 169,352 12,277 157,075 24,924 1,538 23,386 180,461 IDX02 RAD26 443,536 28,740 414,796 58,396 4,356 54,040 468,836 IDX02 RAD48 364,856 25,928 338,928 37,726 2,535 35,191 374,119 IDX03 RAD03 239,004 19,229 219,775 15,023 1,053 13,970 233,745 IDX03 RAD04 261,326 17,982 243,344 54,681 3,463 51,218 294,562 IDX03 RAD05 325,228 26,015 299,213 40,310 2,997 37,313 336,526 IDX03 RAD06 294,347 20,490 273,857 34,821 2,283 32,538 306,395 IDX03 RAD08 284,789 20,736 264,053 31,037 1,890 29,147 293,200 IDX03 RAD12 350,946 25,859 325,087 43,420 2,887 40,533 365,620 IDX03 RAD26 215,410 15,109 200,301 34,487 2,681 31,806 232,107 IDX03 RAD48 35,718 2,535 33,183 11,676 879 10,797 43,980 IDX04 RAD03 356,888 24,626 332,262 24,315 1,542 22,773 355,035 IDX04 RAD04 103,593 7,284 96,309 16,613 1,022 15,591 111,900 IDX04 RAD05 128,018 8,968 119,050 25,918 1,707 24,211 143,261 IDX04 RAD06 123,694 8,495 115,199 16,150 1,066 15,084 130,283 IDX04 RAD08 426,350 22,660 403,690 74,357 3,730 70,627 474,317 IDX04 RAD12 321,915 21,183 300,732 34,330 2,063 32,267 332,999 IDX04 RAD26 308,587 20,740 287,847 31,325 2,859 28,466 316,313 IDX04 RAD48 366,378 28,922 337,456 49,835 3,831 46,004 383,460 IDX05 RAD03 252,865 16,931 235,934 34,344 2,643 31,701 267,635 IDX05 RAD04 383,967 26,524 357,443 37,213 2,493 34,720 392,163 IDX05 RAD05 114,451 8,194 106,257 15,973 995 14,978 121,235 IDX05 RAD06 179,966 11,848 168,118 25,119 1,713 23,406 191,524 IDX05 RAD12 210,825 14,308 196,517 16,960 1,114 15,846 212,363 IDX05 RAD26 180,079 13,160 166,919 19,793 1,716 18,077 184,996 IDX05 RAD48 56,583 4,331 52,252 6,872 535 6,337 58,589 IDX06 RAD03 259,843 17,262 242,581 30,809 1,956 28,853 271,434 IDX06 RAD04 159,240 11,345 147,895 25,325 1,690 23,635 171,530 IDX06 RAD05 478,452 33,582 444,870 48,632 3,173 45,459 490,329 IDX06 RAD06 155,218 12,153 143,065 21,224 1,809 19,415 162,480 IDX06 RAD08 287,018 19,981 267,037 45,383 3,161 42,222 309,259 IDX06 RAD12 255,848 17,507 238,341 40,708 2,550 38,158 276,499 IDX06 RAD26 237,551 13,047 224,504 35,820 3,159 32,661 257,165 IDX06 RAD48 330,010 22,639 307,371 34,832 2,189 32,643 340,014 Total 13,811,017 1,255,494 12,555,522 1,719,637 138,630 1,581,007 14,136,529 Mean 300,240 27,293 272,946 37,383 3,014 34,370 307,316 Standard deviation 135,630 50,533 117,076 17,279 3,928 15,890 130,692

127

Fig. S1: Meta data analysis of RAD loci with a clustering similarity of 90%, minimum depth of 5 and a required 12/15 (80%) Octopus vulgaris species complex and O. insularis individuals per locus. Each taxa is represented by the two individuals with the greatest number of reads (Octopus insularis is represented by three individuals to include Brazil and Ascension sampling localities). X and Y axis phylogeny shows unrooted relationship via RAXML ‘best tree’ (individual labels are shown on opposite right side). Black circles represent the proportion of shared loci an individual has with another individual (scale from

50-100% - lower right of image). Diagonal red circles represent the proportion of overall loci each individual is represented in. Upper black bar represents the proportion of loci each individual shares will all other individuals. Octopus

128 insularis (far left three individuals on the X axis) are separated by a large branch length and display greater levels of missing data compared to all other taxa.

Fig. S2: Meta data analysis of RAD loci with a clustering similarity of 90%, minimum depth of 5 and a required 21/28 (75%) Octopus vulgaris species complex individuals per locus. X and Y axis phylogeny shows unrooted relationship via RAXML ‘best tree’ (individual labels are shown on opposite right side). Black circles represent the proportion of shared loci an individual has with another individual (scale from 50-100% - lower right of image). Diagonal red circles represent the proportion of overall loci each individual is represented in.

129

Upper black bar represents the proportion of loci each individual shares will all other individuals.

Fig. S3: Number of loci favouring (left) and disfavouring (right) each tree, with the

RAxML optimal tree highlighted in red. Points above/below the regression line have greater/fewer loci support than expected respectively. Points outside the dotted prediction intervals represent tree(s) that are more strongly supported than expected given their likelihood values.

130

Fig. S4: Phylogenetic analysis of individuals belonging to the O. vulgaris species- group. Sequence data (partial ND2 gene) was obtained via double digest

RADseq. Maximum Likelihood Bootstrap values are displayed below each node.

131

5. Reconstructing the biogeographic history of speciation within the

Octopus vulgaris species group.

5.1 Abstract

Improving our knowledge of the historical processes that drove speciation will help increase our understanding of the present day distribution of lineages. The present study aimed to investigate the processes leading to speciation among six newly recognised species within the O. vulgaris species group using 447 genome-wide loci and 18 morphological traits. Molecular-based divergence time estimation inferred that species within the O. vulgaris group evolved from their most recent common ancestor over the last 3-8 million years, depending of the molecular rate used. Model testing was implemented to determine the most likely scenario of evolution among extant species via molecular-based ancestral area reconstruction. The most favoured scenario estimated the ancestor of the group to have been a widespread species maintaining population connectivity around the world. Isolation events, rather than founder event- based speciation, was determined to have driven diversification within the O. vulgaris group. This finding was supported by low levels of morphological disparity throughout the history of the group, which was in contrast to expectations under scenarios of increased ecological opportunity, such as the colonisation of newly available habitat/niches. The modern-day dispersive potential of each species’ paralarvae appears to be relatively restricted, with past connectivity among continents appearing to depend on favourable global temperatures and current systems.

132

These findings suggest a limited ability for extant members of the O. vulgaris group to establish or maintain gene flow between hemispheres, which may help inform country-specific management practices of currently overharvested populations.

5.2 Introduction

Understanding the biogeographic history of closely related species is an important step in increasing our knowledge of the present day distributions of lineages (Ree et al., 2005) and the processes that drove the emergence of new taxa (Crisp et al., 2011). Well resolved taxonomic relationships combined with present day distributional data enables the inference of historical processes/events that led to diversification (Andersson, 1996). This discipline, known as ancestral area reconstruction (AAR) can be combined with estimates of timing of divergence to place diversification in context with past geological events that potentially shaped the present day distribution of extant taxa (Crisp et al., 2011; Ronquist & Sanmartín, 2011).

The availability of previously uninhabited environments presents an increase in ecological opportunity, which has been associated with responses in population density, habitat use and trait variation (Schluter, 1996; Yoder et al., 2010). The rapid diversification of lineages during adaptation to newly available niches

(adaptive radiation) suggests that morphological disparity is linked to ecological opportunity (Mahler et al., 2010). High levels of morphological disparity through time are therefore expected under scenarios where evolution was the result of

133 colonising previously uninhabited environments (e.g. founder events). Therefore, analyses inferring historical morphological disparity are potentially useful for complimenting molecular data-based AARs that aim to determine if speciation via founder events played a significant role in the diversification of lineages of interest.

Although multiple distinct AAR models are available to researchers, until recently comparisons of these models was not possible due to contrasting methods of implementation, such as parsimony or maximum likelihood (Ronquist &

Sanmartín, 2011; Matzke, 2014). One potential problem with having to choose a model to implement is that each available model differs in its assumptions

(Ronquist & Sanmartín, 2011). Realising this limitation, Matzke (2014) developed a method to implement three of the most commonly used AAR models (and potentially more unnamed models) under ML, therefore allowing direct model testing. Since the study of Matzke (2014), studies implementing explicit model testing have revealed that the model itself can have a great effect on interpretations (Zhang et al., 2016). For example, the ability to perform model selection has since highlighted the significant role of founder-events in explaining the modern-day assemblages of oceanic island chains (Matos-Maraví et al.,

2014; Tänzler et al., 2014; Voelker et al., 2014; Shaw et al., 2015; Berger et al.,

2016; Zhang et al., 2016).

Accurate inference of biogeographic histories relies on well resolved taxonomic relationships of modern day taxa within the group of interest (Andersson, 1996).

One potential hindrance to attempting ancestral reconstructions within the marine environment is the known prevalence of cryptic species (Knowlton,

1993). Several examples exist where organisms once thought to be widespread

134 are now understood to represent morphologically similar yet genetically distinct cryptic species with relatively restricted distributions (Knowlton, 1993; Klautau et al., 1999; Bickford et al., 2007). This is particularly common among cephalopods, including squids and octopuses (Norman et al., 2014a; Norman et al., 2014b).

Within this group, cryptic speciation is well illustrated by the previous uncertainty and confusion surrounding the species-level relationships of Octopus vulgaris

Cuvier, 1979 (Mangold, 1998).

First reported from the Mediterranean Sea and eastern North Atlantic, O. vulgaris is distributed within sub-tropical waters around the world. The morphological and phylogenetic analyses presented in this thesis (chapters two, three and four) have identified strong biogeographic structuring of within ‘O. vulgaris’, and have identified that this ‘species’ in fact represents a complex of morphologically similar, yet genetically distinct species. Investigating the processes that underpinned historical diversification within this group may improve our understanding of how they coped with past fluctuations in climate and habitat availability (Mace et al., 2003). The recently resolved relationships within the O. vulgaris group enable the inference and dating of historical divergence events, which may provide insight into how modern day distributions arose.

The present study aims to test (1) the significance of a dispersive life history in the diversification of the O. vulgaris group by investigating the occurrence/frequency of founder event based speciation. Founder event-based speciation results from a long distance colonisation event that leads to the colonisation of a previously uninhabited region that is immediately genetically isolated. In direct contrast to hypothesis one, this study aims to test whether (2) the O. vulgaris group had a global ancestral distribution, in which case vicariance

135 events/isolation would have played a role in diversification. Analyses of morphological disparity through time will also be conducted to determine if historical morphological diversification supports scenarios of founder event- based speciation.

All members of the O. vulgaris group are distributed throughout the sub-tropical

Atlantic, Indian and Pacific Oceans, in the northern and southern hemispheres

(Fig. 1). To determine whether (3) the anti-tropical distribution is the result of a single acute event (such as a rapid rise in temperature) or a more gradual process the present study aims to investigate the timing of major anti-tropical cladogenesis events and their correlation with historical temperatures. Finally, disjunct populations from the Mediterranean and South Africa are often considered to represent a single species (Voss & Day, 1962; Smale & Buchan,

1981; Guerra et al., 2010), despite a lack of plausible gene flow mechanisms

(Norman et al., 2014a). Human based dispersal (via ship hulls or ballast water) has been suggested to facilitate connectivity between otherwise isolated populations (e.g. South Africa; Teske et al., 2007). The present study therefore also investigates whether (4) the timing of divergence between O. vulgaris Type

III (South Africa) and other members of the species group is consistent with human based dispersal.

5.3 Methods

5.3.1 Sampling and data collection

136

The present study took advantage of the morphological and molecular data presented in chapters three and four, respectively. The molecular data set consisted of the sequence alignment of 447 loci, which was composed from sequence data of 28 individuals from 13 sampling localities (Fig.1, Table 1). The methods used to obtain these data are outlined in chapter 4.2.5 ‘De novo assembly’. The morphological dataset utilised in the present study was composed of 17 traits recommended for use in Octopus systematics (Roper &

Voss, 1983) in common between mature male (n = 56) and female (n = 56) individuals (Tables S2 and S3; Chapter three); dorsal mantle length (MLd), ventral mantle length (MLv), mantle width (MW), head width (HW), funnel length

(FL), free funnel length (FFL), web depth (WD), the length of the arms on the left

(ALL1 and 4) and right (ALR1, 2 and 4) side of the body, arm width (AW), non- enlarged sucker diameter (SDn), the number of suckers on the left third arm

(SCL) and the right third arm (SCR) and gill length (GL). Details for how these measurements were collected can be found in chapter 3.3.3 ‘Morphological analyses’.

To account for differences attributed to variation in overall size, and to allow for investigation of size free trait variation, all morphometric and meristic traits (with the exception of SC and FFL) were transformed to indices, dividing each trait by the MLd (a proxy for body size) of the respective specimen. The remaining indices were obtained as follows: sucker counts of each arm were divided by the respective arm length and FFL was divided by FL. Morphological indices of both males and females were mean scale transformed (Berner, 2011), and normalised using the ‘normalise variables’ function in PRIMER E+ v6 and

PERMANOVA+ (Anderson et al., 2008) to allow for comparisons of traits despite differing scales of measurement.

137

Fig. 1: Sampling localities (triangles) of members of the Octopus vulgaris species group. Distributions of O. vulgaris sensu stricto and species ‘Types’ are shaded as per Norman et al., (2014a); purple: O. vulgaris s. s., blue: Type I (Caribbean/

Gulf of Mexico), green: Type II (Brazil), orange: Type III (South Africa) and yellow: Type IV (Asia). Distributions of non-vulgaris species are shaded in red:

O. insularis, dark blue (O. cf. tetricus) and light blue (O. tetricus).

Table 1: Sampling localities of Octopus vulgaris species group individuals included in the present study.

Species Location n Latitude Longitude

138

5.3.2 Molecular-based analyses

Combined phylogenetic analyses and divergence time estimation: The consensus sequence for each species was obtained and aligned using Geneious v7.1.7 (Kearse et al., 2012). Statistical selection of best-fit models of nucleotide substitution were undertaken using jModelTest v0.1.1 (Posada, 2008). The appropriate model (GTR+G) was chosen based on ‘goodness of fit’ via the

Akaike Information Criterion (AIC; Akaike, 1974). Ultrametric topologies and divergence times of the most recent common ancestor (MRCA) were co- estimated using Bayesian Inference (BI) implemented within BEAST v2.3.1 and its associated software (Bouckaert et al., 2014). All analyses were conducted using a log-normal relaxed clock (Drummond et al., 2006) under the Yule- speciation process (Aldous, 2001) with a normal prior probability distribution (Ho

& Phillips, 2009). Topologies were estimated using random starting trees, gamma site models and log-normal relaxed clocks and were run for 750 million generations, sampling the Markov chain every 5,000. Three independent analyses were carried out and the resulting log and tree files were combined.

The program Tracer v1.3 (Rambaut et al., 2014) was used to ensure Markov chains had reached stationarity, effective sampling size (ESS) was adequate

(>200) and to determine the correct ‘burn-in’ for the analysis. The ‘maximum clade credibility tree’ was then obtained.

Estimates of mutation rates for Octopus or cephalopods are not available due to their limited fossil record. Bivalve molluscs (class Bivalvia) have the most complete fossil record of any mollusc (>90%; Foote & Sepkoski, 1999; Forey et al., 2004) and are the closest relative of the study taxa with available estimated

139 molecular rates of evolution. Therefore, estimates of absolute divergence times in the present study were reliant on the use of substitution rates of the distantly related molluscan bivalve family Arcidae Lamarck, 1809, which was also the case for a recent study investigating the historical population expansion of the giant squid genus Architeuthis (Winkelmann et al., 2013). The present study used three mutation rates, each calculated using the nuclear marker histone 3

(Marko, 2002); (1) 0.14-0.2%, (2) 0.09-0.1% and (3) 0.02-0.03% per million years. Each rate was calculated using fossil calibration points including the split between (1) the subfamilies Noetiinae and Striarcinae, (2) Anadarinae and the clade containing subgenera Fugleria and Cucullaearca and (3) Anadara s.s. and the sub-genus Grandiarca. Mean values at the mid-point of each rate range were calculated and the upper and lower range were set using a normal prior probability distribution. Analyses were run for 750 million generations, sampling every 5,000 generations with the aim of achieving effective sample size (ESS) values of >200 for the estimated ages of internal nodes.

Molecular-based AAR: Sampling locality data were associated with the tips of the molecular-based BI topology. Three models of AAR for the internal nodes of the phylogenetic topology were utilised; (1) Dispersal Extinction Cladogenesis (DEC;

Ree, 2005; Ree & Smith, 2008), (2) a model based on Dispersal/Vicariance

Analysis (DIVA; Ronquist, 1997) and (3) a range evolution model similar to

Bayesian Area Analysis (BAYAREA; Landis et al., 2013). The appropriate AAR model was chosen based on ‘goodness of fit’ via the AIC (Akaike, 1974).

Model assumptions: The three models utilised in the present study each have unique assumptions regarding the processes allowed during cladogenesis events (Table 2). Each model allows expansion (dispersal) and contraction (local extinction) and for species to have sympatric distributions within a single area. 140

Implementation in BioGeoBears (Matzke, 2013) also allowed for the process founder event-based speciation to be modelled in each of the three models. The models differed in whether they allowed/disallowed (1) sympatry of a single taxon to occur in multiple areas, (2) sympatry in a subset of total inhabited areas

(3) isolation of one area from another and (4) isolation of multiple areas from two or more areas.

Table 2: Processes allowed by three commonly used ancestral area reconstruction models. All models originally excluded the possibility to model founder event based speciation (marked by asterisk), however this parameter has been added via the implementation of each model in BioGeoBears (Matzke,

2013). Models utilised are Dispersal Extinction Cladogenesis (DEC; Ree, 2005;

Ree & Smith, 2008), a model based on Dispersal Vicariance Analysis (DIVA;

Ronquist, 1997) and a Range Evolution model similar to Bayesian Area Analysis

(BAYAREA; Landis et al., 2013).

Model Assumptions DEC DIVA BAYAREA

Range expansion Yes Yes Yes

Range contraction Yes Yes Yes

Sympatry within a single area Yes Yes Yes

Widespread sympatry occurring in >1 range No No Yes

Sympatry in a sub-set of the total areas inhabited Yes No No

Vicariance leading to isolation of a single area from multiple areas Yes Yes No

Vicariance leading to isolation of >1 area from multiple areas No Yes No

Dispersal leading to a founder event in a previously uninhabited area No* No* No*

Hypothesised modes of speciation: Based on the phylogenetic topology presented in chapter four, four hypotheses regarding the potential evolutionary history of the O. vulgaris species group were tested. (1) The MRCA of this group

141 originated in a single area and speciation occurred via founder event dispersal, the free parameter +J was added within BioGeoBears. This model allowed for a taxon to ‘jump’ to a previously uninhabited area and was tested using all three

AAR models. (2) In contrast to hypothesis one, the likelihood that vicariance led to the isolation of a taxon was tested. In this model a single taxon was allowed to be present in up to six areas simultaneously, with a subsequent loss of connectivity leading to the isolation of a single area (DEC and DIVA).

The likelihood that (3) a single event in time was responsible for either the isolation of all modern day taxa in the northern hemisphere from those in the southern hemisphere was tested (DIVA). This required all taxa with present day distributions in the northern hemisphere to be isolated from those in the southern hemisphere during a single event. Finally, (4) human-based dispersal (e.g. dispersal via ship hulls or ballast water) was tested using multiple lines of evidence. First, a single species must have a modern day distribution in two discrete areas, or a divergence event must have been estimated to occur within a time frame consistent with human-mediated dispersal. Second, a dispersal event had to occur whereby a taxon colonised a previously uninhabited area while remaining extant in the original area (DEC, DIVA and BAYAREA).

5.3.1 Morphology-based disparity through time

Principal component analysis (PCA), performed using PRIMER E+ v6 (Anderson et al., 2008), was used to transform morphology-based indices into axes of morphological variation (principal components; PC). Mean values of the PC explaining the greatest proportion of morphological variation (PC1) for each 142 individual were calculated for each clade within the O. vulgaris species group.

PCA also enabled investigation of each traits contribution to morphological variability among individuals. To investigate the levels of morphological variation through time, disparity through time (DTT) analysis was performed using the package Geiger v2.0.6 (Harmon et al., 2009) in R v3.2.3 (R Team, 2015). 1000 simulations of morphological trait evolution were conducted under Brownian motion.

5.4 Results

5.4.1 Divergence time estimation

Divergence time estimation: The O. vulgaris group was estimated to have evolved within the past eight million years (Fig. 2), throughout the late Miocene,

Pliocene or Pleistocene depending on the molecular rate used. Two of the three molecular rates used to calibrate analyses resulted in estimates that the group evolved within the last 3 Ma (Fig. 2; green and red topologies). Analysis based on the remaining molecular rate estimated a relatively older divergence from the

MRCA (Fig. 2; black topology).

143

Fig. 2: Divergence time estimation within the Octopus vulgaris group. Analyses are performed using three mutation rates that were all calculated for the molluscan family Arcidae using the nuclear marker histone 3 (Marko, 2002);

(green topology) 0.14-0.2%, (red topology) 0.09-0.1% and (black topology) 0.02-

0.03% divergence per million years. Node bars represent 95% highest posterior densities for divergence estimates. Relative temperature is based off documented cooling of the climate from 10 to 6 Ma and the re-establishment of the major Antarctic ice sheet (Zachos et al., 2001), increasing temperature from

6-3.2 Ma (Morley & Dworetzky, 1991; Haywood & Valdes, 2004) and subsequent

144 cooling from 3.2 Ma (Sosdian & Rosenthal, 2009) which also marked the origin of the northern hemisphere ice sheets (Zachos et al., 2001).

5.4.2 Ancestral area reconstruction

Hypothesised modes of speciation: The most favoured ancestral distribution model was the DIVA-like model which allowed for a taxon to be present in the maximum available number of areas (Table 3). The DIVA-like model also potentially allowed for range expansion (dispersal) and contraction (local extinction), sympatry within a single area or vicariance leading to isolation of a taxon. This model showed that sequential events isolated individual populations and were responsible for the speciation of a once widespread ancestor (Fig. 3).

Models that were less favoured included those that allowed for founder event- based speciation and/or restricting the number of areas an ancestor could occupy. Therefore, the hypothesis that the MRCA of O. vulgaris was present in a single location and speciation resulted from founder event type dispersal into new areas was not supported.

Estimates based on three molecular rates of evolution suggest the isolation and speciation of extant taxa took place within the last 3-8 Ma (Fig. 2). Two separate instances of taxa being isolated into northern and southern hemispheres occurred; between O. tetricus and O. sinensis as well as O. vulgaris s. s. and O. vulgaris Type III. The three obtained divergence time estimates found that these separate isolation events occurred between one and five million years apart (Fig.

2). The hypothesis where a single acute event in time resulted in the anti-tropical distribution of all O. vulgaris group taxa therefore was not supported. Finally, the 145 most recent estimate for the origin of O. vulgaris Type III was approximately 1.5

Ma, therefore the hypothesis that human mediated dispersal maintained connectivity between this taxon and another member of the O. vulgaris group was also not supported.

Table 2: Molecular-based ancestral state reconstruction models based on algorithms from three software packages; Dispersal Extinction Cladogensis

(Ree, 2005; Ree & Smith, 2008), Dispersal Vicariance Analysis (Ronquist, 1997) and Bayesian Area Analysis (Landis et al., 2013). Models where founder events may lead to speciation are indicated by ‘+J’.

Maximum areas allowed AIC

Model 2 3 4 5 6 6

DEC -30.75926 -24.58628 -20.25876 -20.25876 -13.22304 32.45

DEC+J -21.33982 -19.15190 -16.90956 -16.90956 -13.22470 34.45

DIVA -25.65969 -20.74872 -16.98403 -14.82469 -12.65230 31.30

DIVA+J -18.03833 -18.18017 -15.72719 -14.06433 -12.65277 33.31

BAYAREA -46.94521 -47.12643 -47.12789 -47.12789 -47.12789 44.14

BAYAREA+J -20.98466 -20.98467 -20.98466 -20.98466 -20.98466 47.97

146

Fig. 3: Ancestral area reconstruction of six species within the Octopus vulgaris group, based on 447 genome-wide loci. Phylogenetic representation of ancestral distribution is based on the most likely DIVA-like reconstruction, showing that the

O. vulgaris group had a widespread historical ancestry. Each taxon was allowed

147 to occur in up to six areas at a single time. Node colours represent the potential area each species/ancestor was estimated to have occupied.

5.4.1 Morphological disparity through time

PCA revealed that the majority of morphological variability was represented by

PC1 (51.5%; Table 3), whilst PC2 only represented 15.7%. The contribution of each morphological trait to the overall variability captured by PC1 was relatively even (between 3.4-7.2%).

Table 3: Principal component loadings for PC1, which explains the majority

(51.5%) of morphological variability among 17 (minus dorsal mantle length) morphological indices. Analysis shows a relatively even contribution of all the standard morphological traits utilised in Octopus systematics to PC1, showing that PC1 is a good representation of overall body morphology.

Morphological variable PC1 loading (51.5%) PC1 contribution (%)

Head width 0.293 7.23

Sucker diameter 0.278 6.86

Arm length (L3) 0.275 6.79

Arm length (R3) 0.272 6.72

Web depth 0.271 6.69

Arm length (L1) 0.271 6.69

Arm length (R1) 0.27 6.67

Arm length (R4) 0.27 6.67

Arm width 0.267 6.59

Mantle length (ventral) 0.238 5.88

Mantle width 0.233 5.75

Sucker number (L3) 0.218 5.38

Sucker number (R3) 0.21 5.19

Funnel length 0.206 5.09

Free funnel length 0.197 4.86

Gill length (left) 0.143 3.53

Gill length (right) 0.138 3.41 148

Morphological DTT analyses showed that observed values of phenotypic variation did not differ significantly from simulations under Brownian motion (Fig.

4). With increasing ecological opportunity, such as during the colonisation of newly available niches, a corresponding increase in morphological disparity is hypothesised (Roughgarden, 1972; Stanley, 1973; Nosil & Reimchen, 2005).

The lack of historical morphological variation within the O. vulgaris group suggests that dispersal into previously uninhabited areas was not a common occurrence, which adds further support to the hypothesis of a widespread ancestral distribution of the group.

Fig. 4: Observed historic morphological disparity (relative time until present) among phylogenetic clades during the evolution of the Octopus vulgaris group

(solid line). Based on principal component loadings (PC1) of 18 male and female traits. Grey shading represents 95% confidence interval of 1000 simulations of morphological disparity under Brownian motion, whilst dashed line represents the median simulated value.

149

5.5 Discussion

The present study combined analyses of genome-wide molecular data and morphological traits to investigate the ancestral origins and historical divergence events within the O. vulgaris group. Morphological disparity through time and

AAR analyses suggest the ancestor of the O. vulgaris group was distributed throughout the temperate and tropical oceans of the world. Our data suggests that isolation of individual regions/localities, rather than dispersal-based founder events was responsible for the restriction of gene flow and subsequent speciation among present day taxa.

Dispersal-based colonisation events have been found to facilitate speciation in several marine organisms with dispersive planktonic paralarvae (Litsios et al.,

2014; Owens, 2015; Thacker, 2015). In the present study, however, AAR suggested that regional isolation events led to speciation of a widespread taxon, disfavouring scenarios where dispersal underpinned speciation. In a laboratory- based setting, the maximum duration of the dispersive paralarval phase of O. vulgaris was recorded at 60 days (Villanueva & Norman, 2008). Close relatives,

O. tetricus and O. insularis have both been shown to maintain gene flow across

2000-3000 km of open water (Amor et al., 2014, 2015). Present day species distributions within the O. vulgaris group suggest that Octopus are generally unable to establish or maintain gene flow between continents, or where distances are greater than ~3000 km. The findings of the present study suggest that this was not always the case, and that historical climate/geographic conditions allowed for large scale gene flow among widespread populations.

150

During the evolution of the O. vulgaris group, the position of the world’s continents were relatively similar to today (Zachos et al., 2001), however several changes to the global climate have been documented. The two most recent divergence time estimates obtained in this study suggested that modern day O. vulgaris group taxa evolved from their MRCA during the last 3 Ma. Over the last

3 Ma the climate changed from relatively warm conditions with a reduced equatorial-polar temperature gradient to a relatively cool climate with an increase in Antarctic ice volume and the establishment of the Arctic ice sheet (Zachos et al., 2001; Sosdian & Rosenthal, 2009). Present day O. vulgaris group taxa have sub-tropical to temperate distributions in both hemispheres, and were estimated to have previously inhabited equatorial waters. These divergence time estimates suggest modern anti-tropical distributions originated during a period of global cooling, rather than the poleward range shift of each species being the result of increasingly warm and less suitable equatorial temperatures. A potential explanation for this scenario is the documented decrease in strength of many major ocean currents (Haywood & Valdes, 2004). Therefore, previously connected populations may have been unable to maintain gene flow due to a decreased transequatorial dispersal potential.

The present studies earliest estimate of divergence time suggested the O. vulgaris group evolved during the last 8 Ma. From 10 to 6 Ma ocean temperatures cooled and the major Antarctic ice sheet was re-established

(Zachos et al., 2001). Temperatures then increased between 2-3°C at mid- latitudes and 5-10°C in higher latitudes from 6-3.2 Ma (Morley & Dworetzky,

1991; Haywood & Valdes, 2004). This may have resulted in unsuitably warm temperatures at low and mid-latitudes, potentially driving a poleward range shift which may have led to the isolation of northern and southern hemisphere

151 populations. Furthermore, the poleward shift in distribution may have meant that paralarvae were no longer able to take advantage of strong equatorial currents for transportation between continents. Oceanic gyres in the Indian, southern

Atlantic and northern Atlantic oceans potentially provide a means of transport among continents, and are strong barriers to dispersal of plankton between hemispheres (Goetze et al., 2015). However, paralarvae of extant O. vulgaris group taxa appear to only be able to maintain gene flow across 2000-3000 km areas of ocean and are unable to maintain gene flow among major continents.

Morphological disparity is expected to increase significantly with corresponding ecological opportunity, which arises during the colonisation of newly available niches (Roughgarden, 1972; Stanley, 1973; Nosil & Reimchen, 2005). Significant levels of historical morphological disparity were not inferred within the O. vulgaris group. This supports the AAR findings of a widespread ancestor, suggesting that dispersal and colonisation of previously uninhabited niches did not play a role in the evolution of the O. vulgaris group. Previous investigations into morphological disparity have been able to take advantage of known within-clade phenotypic differences, such as distinct feeding mechanics (Arbour & López-Fernández,

2013) or those identified to be significant by PCA, such as fin morphology

(Astudillo-Clavijo et al., 2015). The present study, however, showed that >50% of overall variation was explained by a single PC, and that all 18 traits contributed relatively evenly to this axis. Species within the O. vulgaris group are generalist predators that prey upon crustaceans, fishes, molluscs and polychaetes

(Mangold, 1983). The selective pressure to adapt different phenotypes among allopatric species may therefore remain relatively low, since there appears to be little requirement for unique predatory specialisations among modern day taxa.

152

Morphological differentiation is often greatest where closely related species occur in sympatry (Brown & Wilson, 1956). This is thought to limit interactions between taxa that share a common resource and allow them to co-exist (Abrams

& Cortez, 2015). ‘Ecological character displacement’ appears to be a common strategy among closely related taxa and has been documented in a number of plant, reptile, mammal, bird, fish and snail taxa (Dayan & Simberloff, 2005).

Although exceptions are known among octopuses (e.g. O. bimaculoides and O. bimaculatus), enhanced levels of morphological variability in sympatric species are seen among members of the Pareledone charcoti complex (Allcock, 2005).

All eight species within this group have sympatric distributions, with six of these species commonly occurring at overlapping depths (between 100-450 m).

Closely related P. charcoti group taxa display distinct differences in papillae number, size, patterns and proportion of mantle coverage. Papillae of cephalopods have known to be associated with camouflage and signalling/communication behaviours that are linked to the behavioural ecology of a species (Allen et al., 2014).

Histones are among the most conserved proteins known, displaying high levels of conservation across widely divergent eukaryotes (Waterborg, 2011).

Evolutionary rates obtained from histone 3 sequence data of bivalve molluscs may therefore provide a good representation of the expected mutation rates of this protein in cephalopods (although the dataset used in the present study did not include histone 3 as it was based on non-coding nuDNA). Variation in mutation rate within a genome is primarily associated with differing levels of GC- content (Wolfe et al., 1989), recombination and gene density (Lercher & Hurst,

2002), and there is little evidence for significant differences between coding and non-coding regions (Wolfe et al., 1989). Mutation rates, however, are known to

153 be time-dependent and exponentially decline with increasing time into the past

(Ho et al., 2011). The use of evolutionary rates calibrated at different nodes throughout a phylogeny may therefore bias divergence time estimations. Two of the molecular rates used in the present study were calculated at generic level nodes, while one was calibrated from a split within a cryptic species complex

(Fig. 1: C1/C2 and C3 respectively; in Marko, 2002). The latter calibration point is therefore considered to provide the most appropriate molecular rates of evolution for the present study. Therefore, the most recent estimate of divergence times (based on the most recent calibration point) are potentially the most reliable.

A limitation of the present study was the absence of O. vulgaris Type I from the

Caribbean and Gulf of Mexico, which may influence AAR models if this taxon is found to be a distinct species. Mitochondrial DNA-based evidence suggests this taxon is conspecific with O. vulgaris Type II from southern Brazil (de Lima et al.,

2017). If this relationship is confirmed, it would be a rare example of trans- tropical gene flow within the O. vulgaris group. Octopus sinensis is currently understood to have trans-tropical populations (Reid & Wilson, 2015; Amor et al.,

2016; Gleadall, 2016), however further investigation is required to confirm this.

Furthermore, no morphological or nuclear DNA based evidence currently supports this finding. Analyses based on mtDNA consistently place individuals from South Africa into a single monophyletic clade with those from the

Mediterranean and eastern North Atlantic, suggesting the two are conspecific

(Guerra et al., 2010). Conversely, analyses based on a far greater number genome-wide nuclear loci prove these are two distinct species (chapter four).

Obtaining genome-wide nuclear DNA for O. vulgaris Type I will help to obtain a complete view of the evolutionary history within the O. vulgaris species complex.

154

5.6 Supplementary information

Table S1: Principal component Analysis results for members of the Octopus vulgaris species group included in this study.

Species/Type Location Institution Catalogue # PC1 PC2 Octopus cf. tetricus Western Australia MV F200326a -0.15957 -0.79591 Octopus cf. tetricus Western Australia MV F200327 -0.72394 -0.71749 Octopus cf. tetricus Western Australia MV F200328 2.55220 0.16222 Octopus cf. tetricus Western Australia MV F200329 -0.84146 -1.58100 Octopus cf. tetricus Western Australia MV F200330 -4.80580 -1.98690 Type IV (Asia) Fujian Province, China OUC OcfvC10 3.88720 2.48110 Type IV (Asia) Fujian Province, China OUC OcfvC11 1.52150 1.08050 Type IV (Asia) Zhejiang Province, China OUC OcfvC12 2.31570 1.88130 Type IV (Asia) Fujian Province, China OUC OcfvC6 -0.82554 0.52594 Type IV (Asia) Zhejiang Province, China OUC OcfvC8 1.74670 1.34260 Type IV (Asia) Fujian Province, China OUC OcfvC9 4.51720 2.13100 Type IV (Asia) Kyushu, Japan MV IGG307 -1.50040 -1.86230 Type IV (Asia) Kyushu, Japan MV IGG309 2.12060 -0.32057 Type IV (Asia) Yilan County, Taiwan MV OcfvT10 2.20500 1.21470 Type IV (Asia) Yilan County, Taiwan MV OcfvT11 0.60771 2.75420 Type IV (Asia) Yilan County, Taiwan MV OcfvT12 -0.80999 -1.18730 Type IV (Asia) Yilan County, Taiwan MV OcfvT13 1.08260 1.65940 Type IV (Asia) Yilan County, Taiwan MV OcfvT14 0.13473 0.81662 Type IV (Asia) Yilan County, Taiwan MV OcfvT15 0.29260 -0.25545 Type IV (Asia) Yilan County, Taiwan MV OcfvT16 0.28521 0.58357 Type IV (Asia) Yilan County, Taiwan MV OcfvT9 0.10350 1.70390 Octopus insularis Fernando de Noronha, Brazil UFRGN OiB10 1.78220 -2.11980 Octopus insularis Fernando de Noronha, Brazil UFRGN OiB12 1.03300 -2.38510 Octopus insularis Fernando de Noronha, Brazil UFRGN OiB14 3.01240 -3.87020 Octopus insularis St Peter St Paul Archipelago, Brazil UFRGN OiB17 1.62150 -2.68330 Octopus insularis St Peter St Paul Archipelago, Brazil UFRGN OiB21 -0.54751 -2.87150 Octopus insularis St Peter St Paul Archipelago, Brazil UFRGN OiB22 1.14410 -2.29290 Octopus insularis St Peter St Paul Archipelago, Brazil UFRGN OiB9 2.82710 -1.13230 Octopus insularis St Peter St Paul Archipelago, Brazil UFRGN MOA-DOL-66 2.53680 0.35216 Octopus insularis Pernambuco, Brazil UFRGN MOA-DOL-67 1.80670 -2.85730 Octopus insularis Rio de Janeiro, Brazil UFRGN MOA-DOL-88 -0.03280 -1.49600 Octopus insularis Rio Grande do Norte, Brazil UFRGN MOA-DOL-65 2.84880 -0.71355 Octopus insularis Northern Brazil UFRG 49524 0.62037 -2.27980 Octopus tetricus New South Wales, Australia MV F182058 -0.32874 -3.11520 Octopus tetricus New South Wales, Australia MV F200319 -0.58466 -2.23060 Octopus tetricus New South Wales, Australia MV F200323 -0.30472 -3.55100 Octopus tetricus New South Wales, Australia MV F200324 -0.92527 -2.35060

155

Octopus tetricus New South Wales, Australia MV F182057 -1.49810 -2.59920 Octopus vulgaris Mauritania, Africa MV OvAf12 1.18280 -0.35584 Octopus vulgaris Mauritania, Africa MV OvAf13 -2.90470 -0.03508 Octopus vulgaris Mauritania, Africa MV OvAf14 1.80770 0.83391 Octopus vulgaris Mauritania, Africa MV OvAf15 0.66738 0.37338 Octopus vulgaris Mauritania, Africa MV OvAf16 0.57378 1.00760 Octopus vulgaris Mauritania, Africa MV OvAf17 2.53080 0.23516 Octopus vulgaris Mauritania, Africa MV OvAf18 2.60670 0.00881 Octopus vulgaris Mauritania, Africa MV OvAf19 1.77870 1.76060 Octopus vulgaris Mauritania, Africa MV OvAf20 2.41410 -0.49234 Type II (Brazil) Rio Grande do Sul, Brazil UFRGN OvB11 -0.68861 2.57090 Type II (Brazil) Rio Grande do Sul, Brazil UFRGN OvB14 1.14920 5.07170 Type II (Brazil) Rio Grande do Sul, Brazil UFRGN OvB15 0.77292 3.93170 Type II (Brazil) Itajaí, Santa Catarina, Brazil UFRGN OvB16 -3.33550 1.71780 Type II (Brazil) Rio Grande do Sul, Brazil UFRGN OvB17 -0.20628 1.53680 Type II (Brazil) Rio Grande do Sul, Brazil UFRGN OvB18 3.04840 2.63660 Type II (Brazil) Rio Grande do Sul, Brazil UFRGN OvB19 0.63653 -0.00431 Type II (Brazil) Paraná, Brazil UFPR OVBPP2 1.37030 -0.81485 Type II (Brazil) Santa Catarina, Brazil UFRGN OvB20 -0.37932 0.61080 Type II (Brazil) Bala-Marina UFPR OVBPP3 -1.13460 -0.03604 Type II (Brazil) Santa Catarina, Brazil UFRG OVBFL2 -5.54530 2.26360 Octopus vulgaris Cies Island, Spain MV OvG11 -2.48550 1.18640 Octopus vulgaris Cies Island, Spain MV OvG12 -2.01630 0.73069 Octopus vulgaris Cies Island, Spain MV OvG13 -0.20487 -1.25200 Octopus vulgaris Cies Island, Spain MV OvG14 -9.52460 0.14689 Octopus vulgaris Cies Island, Spain MV OvG15 -6.46480 1.28390 Octopus vulgaris Cies Island, Spain MV OvG16 -2.97180 0.65910 Octopus vulgaris Cies Island, Spain MV OvG17 -0.63880 1.41890 Octopus vulgaris Cies Island, Spain MV OvG18 -3.13900 -0.70453 Octopus vulgaris Cies Island, Spain MV OvG19 -6.56910 1.25580 Octopus vulgaris Cies Island, Spain MV OvG20 -1.04700 1.01970

156

6. Discussion

6.1 Thesis overview

The combination of morphological and molecular-based analyses conducted throughout this thesis provide conclusive evidence for the existence of six species within the O. vulgaris group. Analyses of morphology, mitochondrial

DNA and genome-wide nuclear loci were conducted to investigate the relationships of individuals from across the global range of the O. vulgaris group.

Chapter two combined analyses of mtDNA and morphology to support the distinction of east Australian from west Australian O. cf. tetricus, as well as confirm O. gibbsi as a junior synonym of O. tetricus. Placed in the broader context of the O. vulgaris group, phylogenetic analyses also highlighted the distinction of Asian individuals from all other localities of the O. vulgaris group.

Chapter three verified the distinction of O. vulgaris s. s., O. sinensis, O. vulgaris

Type II, O. tetricus and O. cf. tetricus based on comprehensive morphological analyses. These morphological analyses were also successful in distinguishing the O. vulgaris group from their distant relatives, O. insularis and O. mimus.

Chapter four presented phylogenetic inferences and species tree estimation based on 447 genome-wide loci. This chapter supported the findings of chapter three and provided evidence for the distinction of a sixth species within the O. vulgaris group; O. vulgaris Type III (South Africa). In chapter five, ancestral area reconstruction and analysis of morphological disparity through time suggested the most recent common ancestor of the O. vulgaris group was a widespread species that existed between 3-8 Ma. Speciation among extant taxa was likely to have been driven by historical events. Fluctuations in climate and current 157 strength throughout the late Miocene, Pliocene and early Pleistocene likely shaped the distribution of modern day taxa.

6.2 Implications

The findings presented within this thesis have significant implications for the naming and description, management and conservation of members of the O. vulgaris group throughout their ranges. New species descriptions are required for

O. tetricus (east Australia) and O. vulgaris s. s. (Mediterranean and eastern north

Atlantic) as type material is non-extant for both species (Table 1). Furthermore,

O. cf. tetricus (west Australia), O. vulgaris Type II (southern Brazil) and Type III

(South Africa) all require formal taxonomic descriptions. Past fisheries reports have only listed catch statistics for five (O. vulgaris, O. maya, Eledone cirrhosa,

Eledone moschata and Enteroctopus dofleini) of the estimated 100 species of commercially harvested octopuses (Norman & Finn 2014). The amended species-level resolution provided by this thesis improves our understanding of the geographical distributions of this species group. Catch statistics based on these findings will more accurately reflect the stock health of each individual species, enhancing their value as a tool for informing management and conservation decisions.

158

Table 1: The current state of taxonomy for Octopus Cuvier, 1797 species

investigated within this thesis.

Taxa Distribution Status Comments on taxonomy O. gibbsi O’Shea, 1999 Northern New Zealand Synonym of tetricus O. jollyorum Reid and Wilson, 2015 Kermadec Is. Synonym of Further molecular-based work required to confirm synonymy with sinensis sinensis d’Orbigny, 1841 O. tetricus Gould, 1852 Southeast Australia, Valid Requires re-description and neotype designation northern New Zealand O. cf. tetricus Southwest Australia Valid Requires description, naming and type material designation

O. sinensis d’Orbigny, 1841 Japan, China, Taiwan Valid Description and type material available (Gleadall, 2016)

O. vulgaris Cuvier, 1797 Mediterranean, eastern Valid Requires re-description and neotype designation North Atlantic O. vulgaris Type I* Caribbean, Gulf of Mexico Unresolved Possible synonym: O. vulgaris Type II. Possible available names: americanus Baker in Denys de Montfort, 1802, carolinensis Verrill 1884, tayrona Guerrero-Kommritz and Camelo-Guerin 2015

O. vulgaris Type II Southern Brazil Valid Possible synonym: O. vulgaris Type I (requires confirmation prior to description) O. vulgaris Type III South Africa Valid Requires description, naming and type material designation. Possible available name: argus Krauss 1848

O. vulgaris (India)* Kerala, India Unresolved Requires morphological and molecular-based identification

* Denotes O. vulgaris group taxon not included within this thesis.

A large number of pelagic, benthopelagic and demersal cephalopods have

shown an increase in abundance over the past decade (Doubleday et al., 2016).

However, benthic octopus populations are expected to perform less favourably

under projected climate change scenarios (Andre et al., 2009; Andre et al.,

2010). As octopuses are becoming increasingly important as a commercial

fisheries resource (Norman & Finn, 2014), they are also expected to experience

increasing exploitation (Xavier et al., 2015), which may exacerbate climate-

induced effects (Harley et al., 2006). The collective studies within this thesis will

enhance future fine-scale demographic studies by providing a realistic

understanding of gene flow among local coastlines and neighbouring

islands/continents. Regional investigations into current stock health are required

to ensure appropriate population numbers and levels of genetic diversity, which

are essential for the sustainability of economically important fisheries stocks.

159

6.3 Limitations

Sampling O. vulgaris individuals from all known localities was not possible during the completion of this thesis. Individuals from the Caribbean and Gulf of Mexico

(O. vulgaris Type I) were not able to be obtained, despite significant efforts.

Therefore, the relationship of O. vulgaris Type I within the O. vulgaris species complex remains the subject of future research. Evidence based on mitochondrial DNA (de Lima et al., 2017) suggests individuals from the

Caribbean and Gulf of Mexico are conspecific with O. vulgaris Type II (southern

Brazil). Confirmation of this preliminary finding should precede the description of the Brazilian taxon, given that the name O. americanus Baker in Denys de

Montfort, 1802 is likely synonymous with O. vulgaris Type I.

The molecular-based component of this thesis would have benefited from the availability of a reference genome for O. vulgaris. Although museum and university collections were a valuable resource for tissue samples, the age and condition of some samples led to poor DNA quality. NGS approaches, including the RADseq method implemented in chapter four, rely on high quality input DNA.

As such, key samples, including individuals from India, were excluded from chapter four. In the absence of a reference genome, optimal de novo assembly relies on high quality sequence reads. Low quality reads obtained as a result of poor template DNA quality, therefore, substantially impede this process. This limitation would potentially be overcome via the ability to map reads to an appropriate reference sequence, which may allow the recovery of several potentially phylogenetically informative reads from otherwise unusable samples.

160

Another limitation to this study is that species divergence time estimates were based on mutation rates of a distantly related bivalve mollusc family (Marko,

2002). This was because the O. vulgaris group lack a comprehensive fossil record and molecular rates of evolution. Previous studies investigating divergence times of cephalopods have also used molecular rates for this bivalve family (Winkelmann et al., 2013). However, as mutation rates across the tree of life do not follow a strict molecular clock, rates from a more closely related organism would result in greater confidence in divergence time estimation.

6.4 Future research

Under current CO2 emission scenarios, global surface temperatures are projected to increase 0.3-6.4°C (Solomon et al., 2007). This increase in global temperature (and associated SST) will undoubtedly influence biodiversity

(Walther et al., 2002). Altered ocean currents are expected to influence larval transport and impact marine population dynamics (Harley et al., 2006). Many marine species distributions are expected to shift (latitude and/or depth) as a direct result of increasing temperature (Walther et al., 2002; Perry et al., 2005).

This is particularly likely for pelagic species and those with planktonic larvae, such as O. vulgaris (Fields et al., 1993). Furthermore, an increase in the frequency and intensity of invasive species introductions is expected, which may have negative consequences for native species (Bellard et al., 2013) as immigrants are more likely to be adapted to the increased local temperature

(Fields et al., 1993).

161

Several marine species have shifted their distribution poleward in response to global warming (Walther et al., 2002; Perry et al., 2005). Within the O. vulgaris group, O. tetricus was recently reported to have undergone a southward range extension into the waters off Flinders Island, Tasmania (Ramos et al., 2014).

Ramos et al., (2014) observed individuals from the newly inhabited range to have relatively fast growth rates, small body sizes and a reduced generation time, which were suggested to be favourable attributes for the success of invasive species. Members of the O. vulgaris group may, therefore, be able to successfully adapt to projected increased temperatures by altering their distributions poleward. This may have significant implications for local fisheries

(Madin et al., 2012), as fisheries may decline in the original range and new fisheries become available poleward. Furthermore, cephalopods, including octopuses, are known to be voracious generalist predators (Clarke, 1996), which may negatively impact communities in the range extension areas (Sorte et al.,

2010). Octopus vulgaris Type III is the only species within this group that is known to be distributed at the southern limit of a landmass. Alteration to this species’ depth distribution may be a viable response to increasing temperatures, as this has been previously observed in several marine species (e.g. Perry et al.,

2005; Caputi et al., 2009).

Chapter two speculated that the allopatric distributions of O. tetricus and O. cf. tetricus may have resulted from the cooling of the Southern Ocean, which potentially forced populations northwards along the western and eastern coasts of Australia. The projected increase in SST resulting from global warming may provide an opportunity for secondary contact between these two species along the southern coast of Australia. The potential for hybridisation among members of the O. vulgaris group is unknown, however chapter four speculated that the

162 shared mtDNA genotype between O. vulgaris s. s. and O. vulgaris Type III

(South Africa) may be the result of introgression during hybridisation.

Furthermore, examples of hybridisation have been recorded in other cephalopods, for example, the southern calamari squid Sepioteuthis australis is suspected of representing two cryptic species that hybridise when they come into contact (Triantafillos & Adams, 2001). Hybridisation may serve to promote increased genetic diversity and therefore adaptive capacity (Sherwin & Moritz,

2000), however, if hybrid progeny are reproductively sterile, wasted mating efforts would result in a decreased reproductive success.

Sequence data from an ‘unverified COI-like gene’ is available for a relative of the

O. vulgaris group from the coast of Kerala, India (GenBank accession

KF489451; Appukuttan & Vijayamma, unpublished). Analysis of mtDNA places this individual from tropical Neendakara, India (12°N) within the O. vulgaris species complex (in a monophyletic clade with O. vulgaris s. s. and Type III; chapter two). Chapter five estimated that the O. vulgaris group’s MRCA post- dated the closure of the Tethys. This taxa is therefore potentially (1) a native species most closely related to other members of the O. vulgaris group from the

Indian Ocean (O. vulgaris Type III and O. cf. tetricus) or (2) an introduced relative potentially the result of human mediated dispersal (shipping). Further sampling from the coast of India is required to confirm the status of this newly discovered relative of the O. vulgaris group.

Significant morphological variation was recorded among sampling localities of O. vulgaris s. s. (Mediterranean, Galicia and Mauritania). HASC values are a commonly used character for species level delimitation of octopuses, and this trait differed significantly between Mauritanian and Mediterranean/Galician

163 individuals, with zero overlap between these two groups. Genome-wide sequence data was included for individuals from the Mediterranean and Galicia in chapter four, however individuals from Mauritania were unable to be included due to issues with tissue quality. This result presents an opportunity to investigate levels of intraspecific morphological variation in this group and may reveal further cryptic species-level diversity within O. vulgaris s. s.

A major benefit of using RADseq data for species-level investigations is the volume of data obtained enabling a genome-wide perspective of phylogenetic relationships. However, RADseq is limited by relatively short read lengths that often result in individual loci that are unable to resolve phylogenetic relationships independently. Therefore individual loci are unlikely to be useful for delimiting species, and as a result obtaining informative loci that can be sequenced in a more traditional way (such as Sanger sequencing) is unlikely. Previous work has shown that data obtained via RADseq has the potential to successfully resolve phylogenetic relationships among nodes ranging in divergence times from 5-360 million years (Cariou et al., 2013; Gonen et al., 2015). The sequence data presented in chapter five may therefore be further utilised as a reference catalog for future RADseq analyses among members of the genus Octopus, and potentially deeper phylogenetic relationships.

Octopus sinensis (also treated within as O. vulgaris Type IV) is currently described as occurring in the waters surrounding China, Taiwan, Japan and the

Kermadec Islands. If the currently recognised distribution is accurate O. sinensis will be the first confirmed member of the O. vulgaris group with a distribution spanning northern and southern hemispheres. Limited specimens are available from the Kermadecs at present, hampering the ability to make reliable

164 morphology-based distinction of this locality from Asia. The present study was unable to obtain sequence data of sufficient quality to include individuals from the Kermadec Islands in phylogenomic analyses. As genome-wide evidence presented within this thesis contrasts with mtDNA based phylogenies, further investigation into individuals from Asia and the Kermadecs may prove useful in verifying their currently understood taxonomic relationships.

6.5 Summary

In this thesis, I combined analyses of mtDNA, genome wide nuclear loci and morphological traits to investigate species-level relationships, historical timing of divergence and the past processes that led to speciation among members of the

O. vulgaris group. The collective studies within this thesis comprise the most extensive sampling effort and accumulation of data for investigating global species within this group. I used molecular and morphological evidence to successfully delimit six species within the O. vulgaris group. Ancestral reconstruction suggested that the ancestor of the group was a wide-spread species, able to maintain gene flow among populations in the Atlantic, Indian and

East Pacific Oceans. I estimated that within the last 8 million years, isolation of populations resulted from fluctuations in climate and drove speciation of modern day taxa. These results provide a better understanding of the species-level relationships within the O. vulgaris group. The improved understanding of species distributions will enable more accurate reporting of fisheries catch statistics, which is beneficial for the appropriate management and conservation of species’ that comprise the world’s most valuable octopod fishery. 165

7. References

Abrams, P. A., & Cortez, M. H. (2015). Is competition needed for ecological

character displacement? Does displacement decrease competition?

Evolution, 69, 3039-3053.

Acosta-Jofré, M. S., Sahade, R., Laudien, J., & Chiappero, M. B. (2012). A

contribution to the understanding of phylogenetic relationships among

species of the genus Octopus (Octopodidae: Cephalopoda). Scientia

Marina, 76, 311-318.

Akaike, H. (1974). A new look at the statistical model identification. IEEE

Transactions on Automatic Control, 19, 716-723.

Aldous, D. J. (2001). Stochastic models and descriptive statistics for

phylogenetic trees, from Yule to today. Statistical Science, 23-34.

Allcock, A. L. (2005). On the confusion surrounding Pareledone charcoti (Joubin,

1905) (Cephalopoda: Octopodidae): endemic radiation in the Southern

Ocean. Zoological Journal of the Linnean Society, 143, 75-108.

Allcock, A. L., Barratt, I., Eleaume, M., Linse, K., Norman, M. D., Smith, P. J.,

Steinke, D., Stevens, D. W., & Strugnell, J. M. (2011). Cryptic speciation

and the circumpolarity debate: a case study on endemic Southern Ocean

octopuses using the COI barcode of life. Deep Sea Research Part II:

Topical Studies in Oceanography, 58, 242-249.

Allcock, A. L., Strugnell, J. M., & Johnson, M. (2008). How useful are the

recommended counts and indices in the systematics of the Octopodidae

(Mollusca: Cephalopoda). Biological Journal of the Linnean Society, 95,

205-218.

166

Allcock, A. L., Strugnell, J. M., Prodohl, P., Piatkowski, U., & Vecchione, M.

(2007). A new species of Pareledone (Cephalopoda : Octopodidae) from

Antarctic Peninsula Waters. Polar Biology, 30, 883-893.

Allen, J. J., Bell, G. R., Kuzirian, A. M., Velankar, S. S., & Hanlon, R. T. (2014).

Comparative morphology of changeable skin papillae in octopus and

cuttlefish. Journal of Morphology, 275, 371-390.

Amor, M. D., Doyle, S. R., Norman, M. D., Hall, N. E., Robinson, A., Leite, T. S.,

Roura, A., & Strugnell, J. M. (in prep). Genome-wide sequencing

uncovers cryptic diversity and mito-nuclear discordance in the Octopus

vulgaris species complex.

Amor, M. D., Laptikhovsky, V., Norman, M. D., & Strugnell, J. M. (2015). Genetic

evidence extends the known distribution of Octopus insularis to the mid-

Atlantic islands Ascension and St Helena. Journal of the Marine Biological

Association of the United Kingdom, 1-6.

Amor, M. D., Norman, M. D., Cameron, H. E., & Strugnell, J. M. (2014). Allopatric

speciation within a cryptic species complex of Australasian octopuses.

Plos One, 9, e98982. doi:98910.91371/journal.pone.0098982.

Amor, M. D., Norman, M. D., Roura, A., Leite, T. S., Lu, C. C., Reid, A. L.,

Hochberg, F. G., Perales-Raya, C., Gleadall, I. G., Zheng, X. D., Silvey,

C., & Strugnell, J. M. (2016). Morphological assessment of the Octopus

vulgaris species complex evaluated in light of molecular-based

phylogenetic inferences. Zoologica Scripta.

Anderson, M. J., Gorley, R. N., & Clarke, K. R. (2008). PERMANOVA+ for

PRIMER: guide to software and statistical methods. PRIMER-E Ltd.

Anderson, T. J. (1997). Habitat selection and shelter use by Octopus tetricus.

Marine Ecology Progress Series, 150, 137-148.

167

Andersson, L. (1996). An ontological dilemma: epistemology and methodology of

historical biogeography. Journal of Biogeography, 23, 269-277.

Andre, J., Grist, E. P. M., Semmens, J. M., Pecl, G. T., & Segawa, S. (2009).

Effects of temperature on energetics and the growth pattern of benthic

octopuses. Marine Ecology-Progress Series, 374, 167-179.

Andre, J., Haddon, M., & Pecl, G. T. (2010). Modelling climate-change-induced

nonlinear thresholds in cephalopod population dynamics. Global Change

Biology, 16, 2866-2875.

Appukuttan, N. B., & Vijayamma, S. (unpublished). Taxonomy and Diversity of

Cephalopods (Phylum: Mollusca) of Kerala coast.

Arbour, J. H., & López-Fernández, H. (2013). Ecological variation in South

American geophagine cichlids arose during an early burst of adaptive

morphological and functional evolution. Proceedings of the Royal Society

of London B: Biological Sciences, 280, 20130849.

Arnqvist, G. (1998). Comparative evidence for the evolution of genitalia by

sexual selection. Nature, 393, 784-786.

Astudillo-Clavijo, V., Arbour, J. H., & López-Fernández, H. (2015). Selection

towards different adaptive optima drove the early diversification of

locomotor phenotypes in the radiation of Neotropical geophagine cichlids.

BMC Evolutionary Biology, 15, 1.

Avise, J. C., Nelson, W. S., & Sugita, H. (1994). A Speciational History of "Living

Fossils": Molecular Evolutionary Patterns in Horseshoe Crabs. Evolution,

48, 1986-2001.

Baird, N. A., Etter, P. D., Atwood, T. S., Currey, M. C., Shiver, A. L., Lewis, Z. A.,

Selker, E. U., Cresko, W. A., & Johnson, E. A. (2008). Rapid SNP

168

Discovery and Genetic Mapping Using Sequenced RAD Markers. Plos

One, 3.

Bellard, C., Thuiller, W., Leroy, B., Genovesi, P., Bakkenes, M., & Courchamp, F.

(2013). Will climate change promote future invasions? Global Change

Biology, 19, 3740-3748.

Bello, G. (1995). A key for the identification of the Mediterranean sepiolids

(Mollusca: Cephalopoda). Bulletin de l'Institut Oceanographique, Monaco,

16 (Special), 41-55.

Berger, B. A., Kriebel, R., Spalink, D., & Sytsma, K. J. (2016). Divergence times,

historical biogeography, and shifts in speciation rates of Myrtales.

Molecular Phylogenetics and Evolution, 95, 116-136.

Berner, D. (2011). Size correction in biology: how reliable are approaches based

on (common) Principal Component Analysis? Oecologia, 166, 961-971.

Bickford, D., Lohman, D. J., Sodhi, N. S., Ng, P. K. L., Meier, R., Winker, K.,

Ingram, K. K., & Das, I. (2007). Cryptic species as a window on diversity

and conservation. Trends in Ecology & Evolution, 22, 148-155.

Boal, J. (2006). Social recognition: a top down view of cephalopod behaviour.

Vie et Milieu, 56, 69-80.

Bookstein, F. L., Chernoff, B. C., Elder, R. L., Humphries, J. M., Smith, G. R., &

Strauss, R. E. (1985). Morphometrics in evolutionary biology. Academy of

Natural Sciences of Philadelphia, Special Publication 15.

Booth, J. D., & Ovenden, J. R. (2000). Distribution of Jasus spp. (Decapoda:

Palinuridae) phyllosomas in southern waters: implications for larval

recruitment. Marine Ecology Progress Series, 200, 241-255.

Booth, J. D., & Phillips, B. F. (1994). Early life history of spiny lobster.

Crustaceana, 66, 271-294.

169

Bouckaert, R., Heled, J., Kühnert, D., Vaughan, T., Wu, C.-H., Xie, D., Suchard,

M. A., Rambaut, A., & Drummond, A. J. (2014). BEAST 2: A Software

Platform for Bayesian Evolutionary Analysis. PLoS Comput Biol, 10,

e1003537.

Boyle, P. (1983). Ventilation rate and arousal in the octopus. Journal of

Experimental Marine Biology and Ecology, 69, 129-136.

Brakoniecki, T. F. (1996). A revision of the genus Pickfordiateuthis Voss, 1953

(Cephalopoda; Myopsida). Bulletin of Marine Science, 58, 9-28.

Brasher, D. J., Ovenden, J. R., Booth, J. D., & White, R. W. G. (1992). Genetic

subdivision of Australian and New Zealand populations of Jasus verreauxi

(Decapoda, Palinuridae) - Preliminary evidence from the mitochondrial

genome. New Zealand Journal of Marine and Freshwater Research, 26,

53-58.

Brown, W. L., & Wilson, E. O. (1956). Character Displacement. Systematic

Zoology, 5, 49-64.

Bryant, D., Bouckaert, R., Felsenstein, J., Rosenberg, N. A., & RoyChoudhury,

A. (2012). Inferring species trees directly from biallelic genetic markers:

bypassing gene trees in a full coalescent analysis. Molecular Biology and

Evolution, 29, 1917-1932.

Burbrink, F. T., & Pyron, R. A. (2010). How does ecological opportunity influence

rates of speciation, extinction, and morphological diversification in New

World ratsnakes (tribe Lampropeltini)? Evolution, 64, 934-943.

Burgess, L. A. (1966). A Study of the morphology and biology of Octopus

hummelincki Adam, 1936 (Mollusca: Cephalopoda). Bulletin of Marine

Science, 16, 762-813.

170

Burridge, C. P. (1999). Molecular Phylogeny of Nemadactylus and Acantholatris

(Perciformes: Cirrhitoidea: Cheilodactylidae), with Implications for

Taxonomy and Biogeography. Molecular Phylogenetics and Evolution, 13,

93-109.

Burridge, C. P., & Smolenski, A. J. (2003). Lack of genetic divergence found with

microsatellite DNA markers in the tarakihi Nemadactylus macropterus.

New Zealand Journal of Marine and Freshwater Research, 37, 223-230.

Burridge, C. P., & White, R. W. G. (2000). Molecular phylogeny of the antitropical

subgenus Goniistius (Perciformes : Cheilodactylidae : Cheilodactylus):

evidence for multiple transequatorial divergences and non-monophyly.

Biological Journal of the Linnean Society, 70, 435-458.

Caputi, N., Melville-Smith, R., de Lestang, S., Pearce, A., & Feng, M. (2009). The

effect of climate change on the western rock lobster (Panulirus cygnus)

fishery of Western Australia. Canadian Journal of Fisheries and Aquatic

Sciences, 67, 85-96.

Cariou, M., Duret, L., & Charlat, S. (2013). Is RAD‐seq suitable for phylogenetic

inference? An in silico assessment and optimization. Ecology and

Evolution, 3, 846-852.

Carlini, D. B., Young, R. E., & Vecchione, M. (2001). A molecular phylogeny of

the Octopoda (Mollusca: Cephalopoda) evaluated in light of morphological

evidence. Molecular Phylogenetics and Evolution, 21, 388-397.

Catchen, J., Hohenlohe, P. A., Bassham, S., Amores, A., & Cresko, W. A.

(2013). Stacks: an analysis tool set for population genomics. Molecular

Ecology, 22, 3124-3140.

Chifman, J., & Kubatko, L. (2014). Quartet Inference from SNP Data Under the

Coalescent Model. Bioinformatics, 30, 3317-3324.

171

Chiswell, S. M., Wilkin, J., Booth, J. D., & Stanton, B. (2003). Trans-Tasman Sea

larval transport: Is Australia a source for New Zealand rock lobsters?

Marine Ecology Progress Series, 247, 173-182.

Chou, J., Gupta, A., Yaduvanshi, S., Davidson, R., Nute, M., Mirarab, S., &

Warnow, T. (2015). A comparative study of SVDquartets and other

coalescent-based species tree estimation methods. BMC Genomics, 16,

S2.

Clarke, K. R. (1993). Non-parametric multivariate analyses of changes in

community structure. Australian Journal of Ecology, 18, 117-143.

Clarke, K. R., & Gorley, R. N. (2006). PRIMER v6: User Manual/Tutorial.

PRIMER-E, : Plymouth.

Clarke, M. R. (1986). A Handbook for the Identification of Cephalopod Beaks.

Oxford: Clarendon.

Clarke, M. R. (1996). The role of cephalopods in the world's oceans: General

conclusion and the future. Philosophical Transactions: Biological

Sciences, 351, 1105-1112.

Crisp, M. D., Trewick, S. A., & Cook, L. G. (2011). Hypothesis testing in

biogeography. Trends in Ecology & Evolution, 26, 66-72.

Dartnall, A. (1974). Littoral biogeography: Biogeography and ecology in

Tasmania (ed. by W.D. Williams). The Hague: Dr. Junk.

Davey, J. W., Hohenlohe, P. A., Etter, P. D., Boone, J. Q., Catchen, J. M., &

Blaxter, M. L. (2011). Genome-wide genetic marker discovery and

genotyping using next-generation sequencing. Nature Reviews Genetics,

12, 499-510.

Dawson, M. N., & Jacobs, D. K. (2001). Molecular evidence for cryptic species of

Aurelia aurita (Cnidaria, Scyphozoa). Biological Bulletin, 200, 92-96.

172

Dayan, T., & Simberloff, D. (2005). Ecological and community-wide character

displacement: the next generation. Ecology Letters, 8, 875-894. de Lima, F. D., Berbel-Filho, W. M., Leite, T. S., Rosa, C. & Lima, S. M. Q.

(2017). Occurrence of Octopus insularis Leite & Haimovici, 2008 in the

Tropical Northwestern Atlantic and implications of species

misidentification in the management of octopus fisheries. Marine

Biodiversity, 1-12.

Degnan, J. H., & Rosenberg, N. A. (2009). Gene tree discordance, phylogenetic

inference and the multispecies coalescent. Trends in Ecology & Evolution,

24, 332-340.

Doubleday, Z. A., Prowse, T. A., Arkhipkin, A., Pierce, G. J., Semmens, J., Steer,

M., Leporati, S. C., Lourenço, S., Quetglas, A., & Sauer, W. (2016).

Global proliferation of cephalopods. Current Biology, 26, R406-R407.

Drummond, A., Suchard, M., Xie, D., & Rambaut, A. (2012). Bayesian

phylogenetics with BEAUti and the BEAST 1.7. Molecular Biology and

Evolution, 29, 1969-1973.

Drummond, A. J., Ho, S. Y., Phillips, M. J., & Rambaut, A. (2006). Relaxed

phylogenetics and dating with confidence. PLoS Biol, 4, e88.

Eaton, D. A. R. (2014). PyRAD: assembly of de novo RADseq loci for

phylogenetic analyses. Bioinformatics, 30, 1844-1849.

Edgar, G. (2000). Australian Marine Life: The Plants and Animals of Temperate

Waters (Revised Edition). Sydney, Australia: Reed New Holland.

Edgar, R. C. (2004). MUSCLE: multiple sequence alignment with high accuracy

and high throughput. Nucleic Acids Research, 32, 1792-1797.

173

Edwards, S., & Bensch, S. (2009). Looking forwards or looking backwards in

avian phylogeography? A comment on. Molecular Ecology, 18, 2930-

2933.

Edwards, S. V., Kingan, S. B., Calkins, J. D., Balakrishnan, C. N., Jennings, W.

B., Swanson, W. J., & Sorenson, M. D. (2005). Speciation in birds: genes,

geography, and sexual selection. Proceedings of the National Academy of

Sciences, 102, 6550-6557.

Author 2009. Splits: Species’ Limits by Threshold Statistics. R package version

1.0. URL: http://R-Forge.R-project.org/projects/splits/.

FAO (2011). Global Production Statistics 2007:

http://www.fao.org/figis/servlet/TabSelector

FAO (2012). Fisheries and Acquaculture Statistics 2010. FAO Yearbook. Food

and Agriculture Organization of the United Nations. Rome.

Fields, P., Graham, J., Rosenblatt, R., & Somero, G. (1993). Effects of expected

global climate change on marine faunas. Trends in Ecology & Evolution,

8, 361-367.

Folmer, O., Black, M., Hoeh, W., Lutz, R., & Vrijenhoek, R. (1994). DNA primers

for amplification of mitochondrial cytochrome c oxidase subunit from

diverse metazoan invertebrates. Molecular Marine Biology and

Biotechnology, 3, 294-299.

Foote, M., & Sepkoski, J. J. (1999). Absolute measures of the completeness of

the fossil record. Nature, 398, 415-417.

Forey, P. L., Fortey, R. A., Kenrick, P., & Smith, A. B. (2004). Taxonomy and

fossils: a critical appraisal. Philosophical Transactions of the Royal

Society of London B: Biological Sciences, 359, 639-653.

174

Frakes, I., McGowran, B., & Bowler, J. (1987). Evolution of Australian

environments. pp 1-16 in: Dyne, G.R. & Walton, D.W. (eds.) Fauna of

Australia: Volume 1A: General articles. Canberra: Australian Government

Publishing Service.

Gleadall, I. G. (2013). A molecular sequence proxy for Muusoctopus januarii and

calibration of recent divergence among a group of mesobenthic

octopuses. Journal of Experimental Marine Biology and Ecology, 447,

106-122.

Gleadall, I. G. (2016). Octopus sinensis d’Orbigny, 1841 (Cephalopoda:

Octopodidae): Valid species name for the commercially valuable east

Asian common octopus. Species Diversity, 21, 31-42.

Gleadall, I. G., Guerrero-Kommritz, J., Hochberg Jr, F. G., & Laptikhovsky, V. V.

(2010). The inkless octopuses (Cephalopoda: Octopodidae) of the

southwest Atlantic. Zoological Science, 27, 528-553.

Goetze, E., Andrews, K. R., Peijnenburg, K. T., Portner, E., & Norton, E. L.

(2015). Temporal stability of genetic structure in a mesopelagic copepod.

Plos One, 10, e0136087.

Gomez, A., Wright, P. J., Lunt, D. H., Cancino, J. M., Carvalho, G. R., & Hughes,

R. N. (2007). Mating trials validate the use of DNA barcoding to reveal

cryptic speciation of a marine bryozoan taxon. Proceedings of the Royal

Society B-Biological Sciences, 274, 199-207.

Gonen, S., Bishop, S. C., & Houston, R. D. (2015). Exploring the utility of cross-

laboratory RAD-sequencing datasets for phylogenetic analysis. BMC

research notes, 8, 1.

175

Graziadei, P. (1964). Electron microscopy of some primary receptors in the

sucker ofOctopus vulgaris. Zeitschrift für Zellforschung und

mikroskopische Anatomie, 64, 510-522.

Grewe, P. M., Smolenski, A. J., & Ward, R. D. (1994). Mitochondrial DNA

diversity in jackass morwong (Nemadactylus macropterus: Teleostei) from

Australian and New Zealand waters. Canadian Journal of Fisheries and

Aquatic Sciences, 51, 1101-1109.

Guerra, A., Cortez, T., & Rocha, F. (1999). Redescripción del pulpo de los

Changos, Octopus mimus Gould, 1852, del litoral chileno-peruano

(Mollusca, Cephalopoda). Iberus, 17, 37-57.

Guerra, A., Roura, A., Gonzalez, A. F., Pascual, S., Cherel, Y., & Perez-Losada,

M. (2010). Morphological and genetic evidence that Octopus vulgaris

Cuvier, 1797 inhabits Amsterdam and Saint Paul Islands (Southern Indian

Ocean). ICES Journal of Marine Science, 67, 1401-1407.

Guindon, S., Dufayard, J. F., Lefort, V., Anisimova, M., Hordijk, W., & Gascuel,

O. (2010). New algorithms and methods to estimate Maximum-Likelihood

phylogenies: Assessing the performance of PhyML 3.0. Systematic

Biology, 59, 307-321.

Guzik, M. T., Norman, M. D., & Crozier, R. H. (2005). Molecular phylogeny of the

benthic shallow-water octopuses (Cephalopoda: Octopodinae). Molecular

Phylogenetics and Evolution, 37, 235-248.

Hague, T., Florini, M., & Andrews, P. L. (2013). Preliminary in vitro functional

evidence for reflex responses to noxious stimuli in the arms of Octopus

vulgaris. Journal of Experimental Marine Biology and Ecology, 447, 100-

105.

176

Harley, C. D. G., Hughes, A. R., Hultgren, K. M., Miner, B. G., Sorte, C. J. B.,

Thornber, C. S., Rodriguez, L. F., Tomanek, L., & Williams, S. L. (2006).

The impacts of climate change in coastal marine systems. Ecology

Letters, 9, 228-241.

Harmon, L., Weir, J., Brock, C., Glor, R., Challenger, W., & Hunt, G. (2009).

geiger: Analysis of evolutionary diversification. R package version, 1.

Haywood, A. M., & Valdes, P. J. (2004). Modelling Pliocene warmth: contribution

of atmosphere, oceans and cryosphere. Earth and Planetary Science

Letters, 218, 363-377.

Hebert, P. D. N., Cywinska, A., Ball, S. L., & DeWaard, J. R. (2003). Biological

identifications through DNA barcodes. Proceedings of the Royal Society

of London Series B-Biological Sciences, 270, 313-321.

Hebert, P. D. N., Stoeckle, M. Y., Zemlak, T. S., & Francis, C. M. (2004).

Identification of birds through DNA barcodes. PLoS Biol, 2, e312.

Heled, J., & Drummond, A. J. (2010). Bayesian Inference of Species Trees from

Multilocus Data. Molecular Biology and Evolution, 27, 570-580.

Henry, C. S. (1994). Singing and cryptic speciation insects. Trends in Ecology &

Evolution, 9, 388-392.

Hipp, A. L. (2014). RADami: R Package for Phylogenetic Analysis of RADseq

Data. R package version 1.0-3. url: http://CRAN.R-

project.org/package=RADami.

Hipp, A. L., Eaton, D. A. R., Cavender-Bares, J., Fitzek, E., Nipper, R., & Manos,

P. S. (2014). A Framework Phylogeny of the American Oak Clade Based

on Sequenced RAD Data. Plos One, 9, e93975.

177

Ho, S. Y. W., Lanfear, R., Bromham, L., Phillips, M. J., Soubrier, J., Rodrigo, A.

G., & Cooper, A. (2011). Time-dependent rates of molecular evolution.

Molecular Ecology, 20, 3087-3101.

Ho, S. Y. W., & Phillips, M. J. (2009). Accounting for calibration uncertainty in

phylogenetic estimation of evolutionary divergence times. Systematic

Biology, 58, 367-380.

Hohenlohe, P. A. (2004). Limits to gene flow in marine animals with planktonic

larvae: models of Littorina species around Point Conception, California.

Biological Journal of the Linnean Society, 82, 169-187.

Hunt, B., Strugnell, J., Bednarsek, N., Linse, K., Nelson, R. J., Pakhomov, E.,

Seibel, B., Steinke, D., & Würzberg, L. (2010). Poles apart: the “bipolar”

pteropod species Limacina helicina is genetically distinct between the

Arctic and Antarctic oceans. Plos One, 5, e9835.

Iglesias, J., Pazos, G., Fernández, J., Sánchez, F., Otero, J., Domingues, P.,

Lago, M., & Linares, F. (2014). The effects of using crab zoeae (Maja

brachydactyla) on growth and biochemical composition of Octopus

vulgaris (Cuvier 1797) paralarvae. Aquaculture International, 22, 1041-

1051.

Iglesias, J., Sánchez, F. J., Bersano, J., Carrasco, J. F., Dhont, J., Fuentes, L.,

Linares, F., Muñoz, J. L., Okumura, S., & Roo, J. (2007). Rearing of

Octopus vulgaris paralarvae: present status, bottlenecks and trends.

Aquaculture, 266, 1-15.

Iribarne, O. O. (1991). Life history and distribution of the small south-western

Atlantic octopus, Octopus tehuelchus. Journal of Zoology, 223, 549-565.

Joll, L. M. (1976). Mating, egg-laying and hatching of Octopus tetricus (Mollusca,

Cephalopoda) in laboratory. Marine Biology, 36, 327-333.

178

Joll, L. M. (1977). Growth and food intake of Octopus tetricus (Mollusca,

Cephalopoda in aquaria. Australian Journal of Marine and Freshwater

Research, 28, 45-56.

Joll, L. M. (1978). Observations on embryonic development of Octopus tetricus

(Mollusca, Cephalopoda). Australian Journal of Marine and Freshwater

Research, 29, 19-30.

Joll, L. M. 1983. Octopus tetricus. In P. R. Boyle (Ed) Cephalopod Life Cycles

pp. 325-334). New York: Academic Press.

Kaneko, N., Kubodera, T., & Iguchi, A. (2011). Taxonomic study of shallow-water

octopuses (Cephalopoda: Octopodidae) in Japan and adjacent waters

using mitochondrial genes with perspectives on octopus DNA barcoding.

Malacologia, 54, 97-108.

Kearse, M., Moir, R., Wilson, A., Stones-Havas, S., Cheung, M., Sturrock, S.,

Buxton, S., Cooper, A., Markowitz, S., & Duran, C. (2012). Geneious

Basic: an integrated and extendable desktop software platform for the

organization and analysis of sequence data. Bioinformatics, 28, 1647-

1649.

Klautau, M., Russo, C. A. M., Lazoski, C., Boury-Esnault, N., Thorpe, J. P., &

Solé-Cava, A. M. (1999). Does cosmopolitanism result from

overconservative systematics? A case study using the marine sponge

Chondrilla nucula. Evolution, 53, 1414-1422.

Knowlton, N. (1993). Sibling species in the sea. Annual Review of Ecology and

Systematics, 24, 189-216.

Knowlton, N. (2000). Molecular genetic analyses of species boundaries in the

sea. Hydrobiologia, 420, 73-90.

179

Kröger, B., Vinther, J., & Fuchs, D. (2011). Cephalopod origin and evolution: a

congruent picture emerging from fossils, development and molecules.

Bioessays, 33, 602-613.

Laboratories, D. (2001). Extraction of DNA from tissue: High salt method.

Version 1.0. School of Biological Sciences, University of Liverpool, L69

7ZD. UK. Available:

http://www,genomics.liv.ac.uk/animal/research/isolation.pdf. Accessed 25

January 2011.

Landis, M. J., Matzke, N. J., Moore, B. R., & Huelsenbeck, J. P. (2013). Bayesian

analysis of biogeography when the number of areas is large. Systematic

Biology, 62, 789-804.

Langmead, B., & Salzberg, S. L. (2012). Fast gapped-read alignment with Bowtie

2. Nature Methods, 9, 357-359.

Larget, B. R., Kotha, S. K., Dewey, C. N., & Ané, C. (2010). BUCKy: gene

tree/species tree reconciliation with Bayesian concordance analysis.

Bioinformatics, 26, 2910-2911.

Larkin, M. A., Blackshields, G., Brown, N. P., Chenna, R., McGettigan, P. A.,

McWilliam, H., Valentin, F., Wallace, I. M., Wilm, A., Lopez, R.,

Thompson, J. D., Gibson, T. J., & Higgins, D. G. (2007). Clustal W and

Clustal X Version 2.0. Bioinformatics, 23, 2947-2948.

Leite, T. S., Haimovici, M., Molina, W., & Warnke, K. (2008). Morphological and

genetic description of Octopus insularis, a new cryptic species in the

Octopus vulgaris complex (Cephalopoda : Octopodidae) from the tropical

South-Western Atlantic. Journal of Molluscan Studies, 74, 63-74.

180

Lemmon, E. M., & Lemmon, A. R. (2013). High-throughput genomic data in

systematics and phylogenetics. Annual Review of Ecology, Evolution, and

Systematics, 44, 99-121.

Lercher, M. J., & Hurst, L. D. (2002). Human SNP variability and mutation rate

are higher in regions of high recombination. Trends in Genetics, 18, 337-

340.

Li, H. (2011). A statistical framework for SNP calling, mutation discovery,

association mapping and population genetical parameter estimation from

sequencing data. Bioinformatics, 27, 2987-2993.

Ling, S. D., Johnson, C. R., Frusher, S. D., & Ridgway, K. R. (2009). Overfishing

reduces resilience of kelp beds to climate-driven catastrophic phase shift.

Proceedings of the National Academy of Sciences.

Litsios, G., Pearman, P. B., Lanterbecq, D., Tolou, N., & Salamin, N. (2014). The

radiation of the clownfishes has two geographical replicates. Journal of

Biogeography, 41, 2140-2149.

Liu, L. (2008). BEST: Bayesian estimation of species trees under the coalescent

model. Bioinformatics, 24, 2542-2543.

Liu, L., & Pearl, D. K. (2007). Species trees from gene trees: reconstructing

Bayesian posterior distributions of a species phylogeny using estimated

gene tree distributions. Systematic Biology, 56, 504-514.

Lu, C., Boucher-Rodoni, R., & Tillier, A. (1995). Catalogue of types of recent

Cephalopoda in the Muséum national d'Histoire naturelle (France).

Bulletin du Muséum national d'histoire naturelle. Section A, Zoologie,

biologie et écologie animales, 17, 307-343.

181

Lu, C., & Ickeringill, R. (2002). Cephalopod beak identification and biomass

estimation techniques: tools for dietary studies of southern Australian

finfishes: Museum Victoria.

Lü, Z., Cui, W., Liu, L., Li, H., & Wu, C. (2013). Phylogenetic relationships among

Octopodidae species in coastal waters of China inferred from two

mitochondrial DNA gene sequences. Genet Mol Res, 12, 3755-3765.

Mace, G. M., Gittleman, J. L., & Purvis, A. (2003). Preserving the tree of life.

Science, 300, 1707-1709.

Madin, E. M., Ban, N. C., Doubleday, Z. A., Holmes, T. H., Pecl, G. T., & Smith,

F. (2012). Socio-economic and management implications of range-shifting

species in marine systems. Global Environmental Change, 22, 137-146.

Mahler, D. L., Revell, L. J., Glor, R. E., & Losos, J. B. (2010). Ecological

opportunity and the rate of morphological evolution in the diversification of

Greater Antillean anoles. Evolution, 64, 2731-2745.

Mangold, K. 1983. Octopus vulgaris. In P. R. Boyle (Ed) Cephalopod Life Cycles

pp. 335-364). London: New York Academic Press.

Mangold, K. (1998). The Octopodinae from the Eastern Atlantic Ocean and the

Mediterranean Sea. Washington, D. C.: Smithsonian Institution Press.

Marko, P. B. (2002). Fossil calibration of molecular clocks and the divergence

times of geminate species pairs separated by the Isthmus of Panama.

Molecular Biology and Evolution, 19, 2005-2021.

Matos-Maraví, P., Águila, R. N., Peña, C., Miller, J. Y., Sourakov, A., &

Wahlberg, N. (2014). Causes of endemic radiation in the Caribbean:

evidence from the historical biogeography and diversification of the

butterfly genus Calisto (Nymphalidae: Satyrinae: Satyrini). BMC

Evolutionary Biology, 14, 1.

182

Matzke, N. J. (2013). Probabilistic historical biogeography: new models for

founder-event speciation, imperfect detection, and fossils allow improved

accuracy and model-testing. Frontiers of Biogeography, 5.

Matzke, N. J. (2014). Model selection in historical biogeography reveals that

founder-event speciation is a crucial process in island clades. Systematic

Biology, syu056.

Mayr, E. (1963). Animal Species and Evolution. Cambridge: Harvard University

Press.

Meier, R., Shiyang, K., Vaidya, G., & Ng, P. K. (2006). DNA barcoding and

taxonomy in Diptera: a tale of high intraspecific variability and low

identification success. Systematic Biology, 55, 715-728.

Metzker, M. L. (2010). Applications of Next-Generation Sequencing technologies

- the next generation. Nature Reviews Genetics, 11, 31-46.

Meyer, C. P., & Paulay, G. (2005). DNA barcoding: error rates based on

comprehensive sampling. PLoS Biol, 3, e422.

Mirarab, S., Bayzid, M. S., Boussau, B., & Warnow, T. (2014). Statistical binning

enables an accurate coalescent-based estimation of the avian tree.

Science, 346, 1250463.

Mirarab, S., Bayzid, M. S., & Warnow, T. (2016). Evaluating summary methods

for multilocus species tree estimation in the presence of incomplete

lineage sorting. Systematic Biology, 65, 366-380.

Monaghan, M. T., Wild, R., Elliot, M., Fujisawa, T., Balke, M., Inward, D. J. G.,

Lees, D. C., Ranaivosolo, R., Eggleton, P., Barraclough, T. G., & Vogler,

A. P. (2009). Accelerated species inventory on Madagascar using

coalescent-based models of species delineation. Systematic Biology, 58,

298-311.

183

Moore, W. S. (1995). Inferring phylogenies from mtDNA variation - mitochondrial-

gene trees versus nuclear-gene trees. Evolution, 49, 718-726.

Moritz, C., & Cicero, C. (2004). DNA Barcoding: Promise and pitfalls. PLoS Biol,

2, e354.

Morley, J. J., & Dworetzky, B. A. (1991). Evolving Pliocene-Pleistocene climate:

A North Pacific perspective. Quaternary Science Reviews, 10, 225-237.

Narum, S. R., Buerkle, C. A., Davey, J. W., Miller, M. R., & Hohenlohe, P. A.

(2013). Genotyping-by-sequencing in ecological and conservation

genomics. Molecular Ecology, 22, 2841-2847.

Nesis, K. N. (1998). Biodiversity and systematics in cephalopods: unresolved

problems require an integrated approach. South African Journal of Marine

Science-Suid-Afrikaanse Tydskrif Vir Seewetenskap, 20, 165-173.

NOAA (2014). http://www.esrl.noaa.gov/psd/data/gridded/data.nodc.woa98.html.

Accessed 15 September 2015.

Norman, M., & Kubodera, T. (2006). Taxonomy and biogeography of an

Australian subtropical octopus with Japanese affinities. Proceedings of the

7th and 8th symposia on collection building and natural history studies in

Asia and the Pacific Rim, 1, 171-189.

Norman, M. D. (2000). Cephalopods: A World Guide. Hackenheim, Germany:

ConchBooks.

Norman, M. D., & Finn, J. K. 2014. World octopod fisheries. In P. Jereb, C. F. E.

Roper, M. D. Norman & J. K. Finn (Eds) Cephalopods of the World. An

Annotated and Illustrated Catalogue of Cephalopod Species Known to

Date. Octopods and Vampire Squids pp. 9-21). Rome: Food and

Agriculture Organization of the United Nations.

184

Norman, M. D., Finn, J. K., & Hochberg, F. G. 2014a. Family Octopodidae. In P.

Jereb, C. F. E. Roper, M. D. Norman & J. K. Finn (Eds) Cephalopods of

the world. An annotated and illustrated catalogue of cephalopod species

known to date. Octopods and Vampire Squids. FAO Species Catalogue

for Fishery Purposes. No. 4. pp. 36-215). Rome: Food and Agriculture

Organization of the United Nations.

Norman, M. D., & Hochberg, F. G. (2005). The current state of octopus

taxonomy. Proceedings of the International Workshop and Symposium of

Cephalopod International Advisory Council, Phuket, 2003. Phuket Marine

Biological Center Special Publication, 66, 127-154.

Norman, M. D., Hochberg, F. G., & Finn, J. K. 2014b. Octopods and vampire

squids. In P. Jereb, C. F. E. Roper, M. D. Norman & J. K. Finn (Eds)

Cephalopods of the World. An Annotated and Illustrated Catalogue of

Cephalopod Species Known to Date. Volume 3 Octopods and Vampire

Squids pp. 33-270). Rome: Food and Agriculture Organization of the

United Nations.

Norman, M. D., & Lu, C. C. (1997). Redescription of the southern dumpling squid

Euprymna tasmanica and a revision of the genus Euprymna

(Cephalopoda: Sepiolidae). Journal of the Marine Biological Association of

the United Kingdom, 77, 1109-1137.

Norman, M. D., & Sweeney, M. J. (1997). The shallow-water octopuses

(Cephalopoda: Octopodidae) of the Philippines. Invertebrate Taxonomy,

11, 89-140.

Nosil, P., & Reimchen, T. (2005). Ecological opportunity and levels of

morphological variance within freshwater stickleback populations.

Biological Journal of the Linnean Society, 86, 297-308.

185

Nottage, J. D., West, R. J., Montgomery, S. S., & Graham, K. (2007).

Cephalopod diversity in commercial fisheries landings of New South

Wales, Australia. Reviews in Fish Biology and Fisheries, 17, 271-281.

O'Dor, R., & Lipinski, M. (1998). The genus Illex (Cephalopoda;

Ommastrephidae): characteristics, distribution and fisheries. FAO fisheries

technical paper, 1-12.

O'Hara, T. D., & Poore, G. C. B. (2000). Patterns of distribution of southern

Australian marine echinoderms and decapods. Journal of Biogeography,

27, 1321-1335.

O'Loughlin, P., Waters, J. M., & Roy, M. S. (2003). A molecular and

morphological review of the asterinid, Patiriella gunnii (Gray)

(Echinodermata: Asteroidea). Memoirs of Museum Victoria, 60, 181-195.

O'Shea, S. (1999). The marine fauna of New Zealand: Octopoda (Mollusca:

Cephalopoda): NIWA Biodiversity Memoir.

Ogden, R. S., Allcock, A., Wats, P., & Thorpe, J. (1998). The role of beak shape

in octopodid taxonomy. South African Journal of Marine Science, 20, 29-

36.

Oosthuizen, A., Jiwaji, M., & Shaw, P. (2004). Genetic analysis of the Octopus

vulgaris population on the coast of South Africa. South African Journal of

Science, 100, 603-607.

Ovenden, J. R., Brasher, D. J., & White, R. W. G. (1992). Mitochondrial DNA

analyses of the red rock lobster Jasus edwardsii supports an apparent

absence of population subdivision throughout Australasia. Marine Biology,

112, 319-326.

Owens, H. L. (2015). Evolution of codfishes (Teleostei: Gadinae) in geographical

and ecological space: evidence that physiological limits drove

186

diversification of subarctic fishes. Journal of Biogeography, 42, 1091-

1102.

Page, R. D., & Charleston, M. A. (1997). From gene to organismal phylogeny:

reconciled trees and the gene tree/species tree problem. Molecular

Phylogenetics and Evolution, 7, 231-240.

Perry, A. L., Low, P. J., Ellis, J. R., & Reynolds, J. D. (2005). Climate change and

distribution shifts in marine fishes. Science, 308, 1912-1915.

Peterson, B. K., Weber, J. N., Kay, E. H., Fisher, H. S., & Hoekstra, H. E. (2012).

Double Digest RADseq: An inexpensive method for De novo SNP

discovery and genotyping in model and non-model species. Plos One, 7.

Pfenninger, M., & Schwenk, K. (2007). Cryptic animal species are

homogeneously distributed among taxa and biogeographical regions.

BMC Evolutionary Biology, 7, 121.

Pickford, G. C. (1945). Le Poulpe Americain: A Study of the littoral Octopoda of

the Western Atlantic. Transactions of the Connecticut Academy of Arts

and Sciences, 36, 701-811.

Pickford, G. E. (1964). Octopus dofleini (Wülker), the giant octopus of the North

Pacific. Bulletin of the Bingham Oceanographic Collection in the Peabody

Museum of Natural History, Yale University, 19, 1-70.

Pickford, G. E., & McConnaughey, B. H. (1949). The Octopus bimaculatus

problem: a study in sibling species. Bulletin of the Bingham

Oceanographic Collection, 12, 1-66.

Pomiankowski, A., & Moller, A. (1995). A resolution of the lek paradox.

Proceedings of the Royal Society of London B: Biological Sciences, 260,

21-29.

187

Pons, J., Barraclough, T. G., Gomez-Zurita, J., Cardoso, A., Duran, D. P., Hazell,

S., Kamoun, S., Sumlin, W. D., & Vogler, A. P. (2006). Sequence-based

species delimitation for the DNA taxonomy of undescribed insects.

Systematic Biology, 55, 595-609.

Posada, D. (2008). jModelTest: Phylogenetic model averaging. Molecular

Biology and Evolution, 25, 1253-1256.

Quinn, G., & Keough, M. (2002a). Experimental Design and Analysis for

Biologists. New York: Cambridge University Press.

Quinn, G., & Keough, M. (2002b). Introduction to Multivariate Analyses. In:

Experimental Design and Analysis for Biologists. New York: Cambridge

University Press.

Rambaut, A., A., S. M., Xie, D., & Drummond, A. J. (2014). Tracer v1.6, Available

from http://beast.bio.ed.ac.uk/Tracer.

Rambaut, A., & Drummond, A. J. (2003). Tracer 1.3 Oxford University,

http://tree.bio.ed.ac.uk/software/tracer.

Ramos, J. E., Pecl, G. T., Moltschaniwskyj, N. A., Strugnell, J. M., León, R. I., &

Semmens, J. M. (2014). Body size, growth and life span: implications for

the polewards range shift of Octopus tetricus in south-eastern Australia.

Plos One, 9, e103480.

REDMAP (2011, 28/02). http://www.redmap.org.au/species/view/267/gloomy-

octopus/

REDMAP (2013, 26/02). http://www.redmap.org.au/region/tas/sightings/latest/

Ree, R. H. (2005). Detecting the historical signature of key innovations using

stochastic models of character evolution and cladogenesis. Evolution, 59,

257-265.

188

Ree, R. H., & Hipp, A. L. (2015). Inferring phylogenetic history from restriction

site associated DNA (RADseq).

Ree, R. H., Moore, B. R., Webb, C. O., Donoghue, M. J., & Crandall, K. (2005). A

likelihood framework for inferring the evolution of geographic range on

phylogenetic trees. Evolution, 59, 2299-2311.

Ree, R. H., & Smith, S. A. (2008). Maximum Likelihood inference of geographic

range evolution by dispersal, local extinction, and cladogenesis.

Systematic Biology, 57, 4-14.

Reid, A. L., & Wilson, N. G. (2015). Octopuses of the Kermadec Islands:

Discovery and description of a new member of the Octopus ‘vulgaris’

complex (O. jollyorum, sp. nov.) and the first description of a male

Callistoctopus kermadecensis (Berry, 1914). Bulletin of the Auckland

Museum, 20, 349-368.

Ridgway, K. R. (2007). Long-term trend and decadal variability of the southward

penetration of the East Australian Current. Geophysical Research Letters,

34, L13613.

Robson, G. C. (1929). A Monograph of the Recent Cephalopoda, Part I:

Octopodinae. London: British Museum (Natural History).

Roch, S., & Steel, M. (2015). Likelihood-based tree reconstruction on a

concatenation of aligned sequence data sets can be statistically

inconsistent. Theoretical Population Biology, 100, 56-62.

Roch, S., & Warnow, T. (2015). On the robustness to gene tree estimation error

(or lack thereof) of coalescent-based species tree methods. Systematic

Biology, syv016.

Ronquist, F. (1997). Dispersal-vicariance analysis: a new approach to the

quantification of historical biogeography. Systematic Biology, 46, 195-203.

189

Ronquist, F., & Huelsenbeck, J. P. (2003). MrBayes 3: Bayesian phylogenetic

inference under mixed models. Bioinformatics, 19, 1572-1574.

Ronquist, F., & Sanmartín, I. (2011). Phylogenetic methods in biogeography.

Annual Review of Ecology, Evolution, and Systematics, 42, 441.

Roper, C., & Voss, G. (1983). Guidelines for taxonomic description of

cephalopod species. Memoirs of the National Museum of Victoria, 44, 13-

27.

Roper, C. F. E., & Sweeney, M. J. (1983). Techniques for fixation, preservation,

and curation of cephalopods. Memoirs of the National Museum of Victoria,

44, 29-47.

Roper, C. F. E., Sweeney, M. J., & Nauen, C. E. (1984). FAO Species

Catalogue. Cephalopods of the world: An annotated and illustrated

catalogue of species of interest to fisheries. Vol 3 (125). FAO Fisheries

Synopsis. pp. 1-277.

Roughgarden, J. (1972). Evolution of niche width. American Naturalist, 683-718.

Roura, Á., Antón Álvarez‐Salgado, X., González, Á. F., Gregori, M., Rosón, G.,

Otero, J., & Guerra, Á. (2016). Life strategies of cephalopod paralarvae in

a coastal upwelling system (NW Iberian Peninsula): insights from

zooplankton community and spatio‐temporal analyses. Fisheries

Oceanography, 25, 241-258.

Roura, A., Doyle, S., Hall, N., Guerra, A., González, A. F., & Strugnell, J. M. (in

press). Wild oceanic Octopus vulgaris paralarvae provide answers for

aquaculture. Journal of Shellfish Research.

Roy, S. N., & Bargmann, R. E. (1958). Tests of multiple independence and the

associated confidence bounds. The Annals of Mathematical Statistics,

491-503.

190

R Team (2015). R: A language and environment for statistical computing. R

Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-

project.org/.

Sales, J. B. D. L., Rego, P. S. D., Hilsdorf, A. W. S., Moreira, A. A., Haimovici,

M., Tomás, A. R., Batista, B. B., Marinho, R. A., Markaida, U., Schneider,

H., & Sampaio, I. (2013). Phylogeographical features of Octopus vulgaris

and Octopus insularis in the south-eastern Atlantic based on the analysis

of mitochondrial markers. Journal of Shellfish Research, 32, 325-339.

Schluter, D. (1996). Ecological causes of adaptive radiation. American Naturalist,

S40-S64.

Schönrogge, K., Barr, B., Wardlaw, J. C., Napper, E., Gardner, M. G., Breen, J.,

Elmes, G. W., & Thomas, J. A. (2002). When rare species become

endangered: cryptic speciation in myrmecophilous hoverflies. Biological

Journal of the Linnean Society, 75, 291-300.

Shaw, A. J., Shaw, B., Johnson, M. G., Devos, N., Stenøien, H. K., Flatberg, K.

I., & Carter, B. E. (2015). Phylogenetic structure and biogeography of the

Pacific Rim clade of Sphagnum subgen. Subsecunda: haploid and

allodiploid taxa. Biological Journal of the Linnean Society, 116, 295-311.

Sherwin, W., & Moritz, C. (2000). Managing and monitoring genetic erosion. In:

Genetics, demography and viability of framented populations A. G. Young

& G. M. Clarke (Eds). Cambridge, UK: Cambridge University Press.

Shimodaira, H. (2002). An Approximately Unbiased Test of Phylogenetic Tree

Selection. Systematic Biology, 51, 492-508.

Shimodaira, H., & Hasegawa, M. (2001). CONSEL: for assessing the confidence

of phylogenetic tree selection. Bioinformatics, 17, 1246-1247.

191

Simon, C., Paabo, S., Kocher, T. D., & Wilson, A. C. (1990). Evolution of

mitochondrial ribosomal-RNA in insects as shown by the polymerase

chain-reaction. Molecular Evolution, 122, 235-244.

Smale, M., Clarke, M., Klages, N., & Roeleveld, M. (1993). Octopod beak

identification - resolution at a regional level (Cephalopoda, Octopoda:

Southern Africa). South African Journal of Marine Science, 13, 269-293.

Smale, M. J., & Buchan, P. R. (1981). Biology of Octopus vulgaris off the east

coast of South Africa. Marine Biology, 65, 1-12.

Solé-Cava, A., Klautau, M., Boury-Esnault, N., Borojecic, R., & Thorpe, J. (1991).

Genetic evidence for cryptic speciation in allopatric populations of two

cosmopolitan species of the calcareous sponge genus Clathrina. Marine

Biology, 111, 381-386.

Söller, R., Warnke, K., Saint-Paul, U., & Blohm, D. (2000). Sequence divergence

of mitochondrial DNA indicates cryptic biodiversity in Octopus vulgaris and

supports the taxonomic distinctiveness of Octopus mimus (Cephalopoda :

Octopodidae). Marine Biology, 136, 29-35.

Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K., Tignor, M.,

& Miller, H. 2007. Contribution of working group I to the fourth assessment

report of the intergovernmental panel on climate change, 2007:

Cambridge University Press, Cambridge.

Sorte, C. J., Williams, S. L., & Carlton, J. T. (2010). Marine range shifts and

species introductions: comparative spread rates and community impacts.

Global Ecology and Biogeography, 19, 303-316.

Sosdian, S., & Rosenthal, Y. (2009). Deep-sea temperature and ice volume

changes across the Pliocene-Pleistocene climate transitions. Science,

325, 306-310.

192

Sponer, R., & Roy, M. S. (2002). Phylogeographic analysis of the brooding brittle

star Amphipholis squamata (Echinodermata) along the coast of New

Zealand reveals high cryptic genetic variation and cryptic dispersal

potential. Evolution, 56, 1954-1967.

Stamatakis, A. (2014). RAxML version 8: A tool for phylogenetic analysis and

post-analysis of large phylogenies. Bioinformatics, 30, 1312-1313.

Stanley, S. M. (1973). Effects of competition on rates of evolution, with special

reference to bivalve mollusks and mammals. Systematic Biology, 22, 486-

506.

Stapley, J., Reger, J., Feulner, P. G. D., Smadja, C., Galindo, J., Ekblom, R.,

Bennison, C., Ball, A. D., Beckerman, A. P., & Slate, J. (2010). Adaptation

genomics: the next generation. Trends in Ecology & Evolution, 25, 705-

712.

Strugnell, J., Norman, M., Drummond, A., & Cooper, A. (2004). Neotenous

origins for pelagic octopuses. Current Biology, 14, R300-R301.

Strugnell, J., Norman, M., Vecchione, M., Guzik, M., & Allcock, A. L. (2013). The

ink sac clouds octopod evolutionary history. Hydrobiologia, 725, 215-235.

Strugnell, J. M., Collins, M. A., & Allcock, A. L. (2008a). Molecular evolutionary

relationships of the octopodid genus Thaumeledone (Cephalopoda :

Octopodidae) from the Southern Ocean. Antarctic Science, 20, 245-251.

Strugnell, J. M., Jackson, J., Drummond, A. J., & Cooper, A. (2006). Divergence

time estimates for major cephalopod groups: evidence from multiple

genes. Cladistics, 22, 89-96.

Strugnell, J. M., Rogers, A. D., Prodöhl, P. A., Collins, M. A., & Allcock, A. L.

(2008b). The thermohaline expressway: the Southern Ocean as a centre

of origin for deep‐sea octopuses. Cladistics, 24, 853-860.

193

Strugnell, J. M., Watts, P. C., Smith, P. J., & Allcock, A. L. (2012). Persistent

genetic signatures of historic climatic events in an Antarctic octopus.

Molecular Ecology, 21, 2775-2787.

Suatoni, E., Vicario, S., Rice, S., Snell, T., & Caccone, A. (2006). An analysis of

species boundaries and biogeographic patterns in a cryptic species

complex: the rotifer—Brachionus plicatilis. Molecular Phylogenetics and

Evolution, 41, 86-98.

Swofford, D. L. (2003). PAUP*. Phylogenetic Analysis Using Parsimony (*and

Other Methods). Version 4. Sinauer Associates, Sunderland,

Massachusetts.

Systat Software, I. (2009). Systat, Version 13 www.systat.com.

Talavera, G., Dincă, V., & Vila, R. (2013). Factors affecting species delimitations

with the GMYC model: insights from a butterfly survey. Methods in

Ecology and Evolution, n/a-n/a.

Tamura, K., & Nei, M. (1993). Estimation of the number of nucleotide

substitutions in the control region of mitochondrial DNA in humans and

chimpanzees. Molecular Biology and Evolution, 512-526.

Tamura, K., Peterson, D., Peterson, N., Stecher, G., Nei, M., & Kumar, S. (2011).

MEGA5: Molecular Evolutionary Genetics Analysis using Maximum

Likelihood, Evolutionary Distance, and Maximum Parsimony methods.

Molecular Biology and Evolution, 28.

Tänzler, R., Toussaint, E. F., Suhardjono, Y. R., Balke, M., & Riedel, A. (2014).

Multiple transgressions of Wallace's Line explain diversity of flightless

Trigonopterus weevils on Bali. Proceedings of the Royal Society of

London B: Biological Sciences, 281, 20132528.

194

Teske, P., Oosthuizen, A., Papadopoulos, I., & Barker, N. (2007).

Phylogeographic structure of Octopus vulgaris in South Africa revisited:

identification of a second lineage near Durban harbour. Marine Biology,

151, 2119-2122.

Thacker, C. E. (2015). Biogeography of goby lineages (Gobiiformes: Gobioidei):

origin, invasions and extinction throughout the Cenozoic. Journal of

Biogeography, 42, 1615-1625.

Thiel, M., & Gutow, L. (2005). The ecology of rafting in the marine environment.

II. The rafting organisms and community. Oceanography And Marine

Biology An Annual Review, 43, 279-418.

Toews, D. P. L., & Brelsford, A. (2012). The biogeography of mitochondrial and

nuclear discordance in animals. Molecular Ecology, 21, 3907-3930.

Toll, R. (1988). The use of arm sucker number in Octopodid systematics

(Cephalopoda: Octopoda). American Malacological Bulletin, 6, 207-211.

Triantafillos, L., & Adams, M. (2001). Allozyme analysis reveals a complex

population structure in the southern calamary Sepioteuthis australis from

Australia and New Zealand. Marine Ecology-Progress Series, 212, 193-

209.

Tricarico, E., Amodio, P., Ponte, G., & Fiorito, G. 2014. Cognition and recognition

in the cephalopod mollusc Octopus vulgaris: Coordinating interaction with

environment and conspecifics. Biocommunication of Animals pp. 337-

349): Springer.

Trontelj, P., & Fišer, C. (2009). Cryptic species diversity should not be trivialised.

Systematics and biodiversity, 7, 1-3.

195

Undheim, E. A. B., Norman, J. A., Thoen, H. H., & Fry, B. G. (2010). Genetic

identification of Southern Ocean octopod samples using mtCOI. Comptes

Rendus Biologies, 333, 395-404.

Valentine, J. W. (1984). Neogene marine climate trends - Implications for

biogeography and evolution of the shallow-sea biota. Geology, 12, 647-

650.

Vidal, É. A. G., Fuentes, L., & da Silva, L. B. (2010). Defining Octopus vulgaris

populations: A comparative study of the morphology and chromatophore

pattern of paralarvae from Northeastern and Southwestern Atlantic.

Fisheries Research, 106, 199-208.

Villanueva, R. (1995). Experimental rearing and growth of planktonic Octopus

vulgaris from hatching to settlement. Canadian Journal of Fisheries and

Aquatic Sciences, 52, 2639-2650.

Villanueva, R., & Norman, M. D. (2008). Biology of the Planktonic Stages of

Benthic Octopuses. Florida: Taylor & Francis.

Voelker, G., Peñalba, J. V., Huntley, J. W., & Bowie, R. C. (2014). Diversification

in an Afro-Asian songbird clade (Erythropygia–Copsychus) reveals

founder-event speciation via trans-oceanic dispersals and a southern to

northern colonization pattern in Africa. Molecular Phylogenetics and

Evolution, 73, 97-105.

Voight, J. R. (1991a). Enlarged suckers as an indicator of male maturity in

octopus. Bulletin of Marine Science, 49, 98-106.

Voight, J. R. (1991b). Morphological variation in octopod specimens -

reassessing the assumption of preservation-induced deformation.

Malacologia, 33, 241-253.

196

Voight, J. R. (1994). Morphological variation in shallow-water octopuses

(Mollusca: Cephalopoda). Journal of Zoology, 232, 491-504.

Voight, J. R. (1995). Sexual dimorphism and niche divergence in a mid-water

octopod (Cephalopoda: Bolitaenidae). Biological Bulletin, 189, 113-119.

Voight, J. R. (2001). Morphological deformation in preserved specimens of the

deep-sea octopus Graneledone. Journal of Molluscan Studies, 67, 95-

102.

Voight, J. R. (2012). Meristic variation in males of the hydrothermal vent octopus,

Muusoctopus hydrothermalis (Cephalopoda: Octopodidae). Journal of the

Marine Biological Association of the United Kingdom, 92, 361-366. von Byern, J., & Klepal, W. (2010). Re-evaluation of taxonomic characters of

Idiosepius (Cephalopoda, Mollusca). Malacologia, 52, 43-65.

Voss, G. L. 1977. Present status and new trends in cephalopod systematics.

Symposia of the Zoological Society of London pp. 49-60).

Voss, G. L., & Day, J. H. (1962). South African cephalopods. Transactions of the

Royal Society of South Africa, 36, 245-272.

Walther, G.-R., Post, E., Convey, P., Menzel, A., Parmesan, C., Beebee, T. J. C.,

Fromentin, J.-M., Hoegh-Guldberg, O., & Bairlein, F. (2002). Ecological

responses to recent climate change. Nature, 416, 389-395.

Waples, R. S. (1987). A multispecies approach to the analysis of gene flow in

marine shore fishes. Evolution, 41, 385-400.

Warnke, K., Söller, R., Blohm, D., & Saint-Paul, U. (2004). A new look at

geographic and phylogenetic relationships within the species group

surrounding Octopus vulgaris (Mollusca, Cephalopoda): indications of

very wide distribution from mitochondrial DNA sequences. Journal of

Zoological Systematics and Evolutionary Research, 42, 306-312.

197

Waterborg, J. H. (2011). Evolution of histone H3: emergence of variants and

conservation of post-translational modification sites. Biochemistry and Cell

Biology, 90, 79-95.

Waters, J. M., King, T. M., O'Loughlin, P. M., & Spencer, H. G. (2005).

Phylogeographical disjunction in abundant high-dispersal littoral

gastropods. Molecular Ecology, 14, 2789-2802.

White, B. N. (1986). The isthmian link, antitropicality and american biogeography

- distributional history of the Atherinopsinae (Pisces, Atherinidae).

Systematic Zoology, 35, 176-194.

Wilson, B., & Allen, G. (1987). Major components and distribution of marine

fauna. pp 43-68 in: Dyne, G.R. & Walton, D.W. (eds.) Fauna of Australia:

Volume 1A: General articles. Canberra: Australian Government Publishing

Service.

Winkelmann, I., Campos, P. F., Strugnell, J., Cherel, Y., Smith, P. J., Kubodera,

T., Allcock, L., Kampmann, M. L., Schroeder, H., Guerra, A., Norman, M.,

Finn, J., Ingrao, D., Clarke, M., & Gilbert, M. T. P. (2013). Mitochondrial

genome diversity and population structure of the giant squid Architeuthis:

genetics sheds new light on one of the most enigmatic marine species.

Proceedings of the Royal Society B-Biological Sciences, 280.

Wolfe, K. H., Sharp, P. M., & Li, W.-H. (1989). Mutation rates differ among

regions of the mammalian genome.

Wood, D. E., & Salzberg, S. L. (2014). Kraken: ultrafast metagenomic sequence

classification using exact alignments. Genome Biology, 15, R46-R46.

Xavier, J., Clarke, M. R., Magalhães, M. C., Stowasser, G., Blanco, C., & Cherel,

Y. (2007). Current status of using beaks to identify cephalopods: III

198

international workshop and training course on cephalopod beaks, Faial

Island, Azores, April 2007.

Xavier, J. C., Allcock, A. L., Cherel, Y., Lipinski, M. R., Pierce, G. J., Rodhouse,

P. G., Rosa, R., Shea, E. K., Strugnell, J. M., & Vidal, E. A. (2015). Future

challenges in cephalopod research. Journal of the Marine Biological

Association of the United Kingdom, 95, 999-1015.

Yoder, J., Clancey, E., Des Roches, S., Eastman, J., Gentry, L., Godsoe, W.,

Hagey, T., Jochimsen, D., Oswald, B., & Robertson, J. (2010). Ecological

opportunity and the origin of adaptive radiations. Journal of Evolutionary

Biology, 23, 1581-1596.

Zachos, J., Pagani, M., Sloan, L., Thomas, E., & Billups, K. (2001). Trends,

rhythms, and aberrations in global climate 65 Ma to present. Science,

292, 686-693.

Zhang, G., Basharat, U., Matzke, N., & Franz, N. (2016). Model selection in

statistical historical biogeography of Neotropical insects - the

Exophthalmus genus complex (Curculionidae: Entiminae). bioRxiv,

053611.

Zhang, J., Kobert, K., Flouri, T., & Stamatakis, A. (2014). PEAR: a fast and

accurate Illumina Paired-End reAd mergeR. Bioinformatics, 30, 614-620.

Zimmermann, T., Mirarab, S., & Warnow, T. (2014). BBCA: Improving the

scalability of* BEAST using random binning. BMC Genomics, 15, S11.

Zink, R. M., & Barrowclough, G. F. (2008). Mitochondrial DNA under siege in

avian phylogeography. Molecular Ecology, 17, 2107-2121.

199