Crustacean

The Amphipoda is a highly speciose order of crustaceans with a life cycle characterized by direct development and no larval stage, making them interesting models for studies on marine speciation. The family Hyperiidae Dana, 1852 is a strictly pelagic group of Amphipoda. In northern latitudes, free-swimming hyperiids belonging to the genus Themisto (Guérin, 1825) are important components of marine ecosystems in term of abundance and biomass, but little is known about their genetic relationships. We present the ﬁrst multi-locus molecular phylogenetic assessment of the Themisto in the Northern Hemisphere. We performed Bayesian and maximum likelihood reconstructions based on three nuclear loci (18S rDNA, 28S rDNA, and Histone 3) and mitochondrial cytochrome oxidase I data on eight specimens of Themisto from the Paciﬁc, Arctic, and Atlantic oceans. We also provide an updated molecular phylogeny of Hyperiidae. Based on our multi-locus phylogeny, we report the presence of cryptic species in the North Paciﬁc. Our results are discussed in the light of marine speciation.


INTRODUCTION
Many groups of marine crustaceans are notoriously difficult to identify to the species level by traditional approaches due to their enormous morphological diversity (Radulovici et al., 2009), the frequent presence of morphological homoplasy (Knowlton, 1993;Hurt et al., 2013), and because precise taxonomic descriptions are lacking for many groups (Knowlton, 1993;Radulovici, 2012).The use of cytochrome oxidase I DNA-barcoding (COI; Hebert et al., 2003) represents an efficient tool to characterize species of crustaceans and to flag cryptic species (Witt et al., 2006;Costa et al., 2007;Radulovici et al., 2009;Lörz et al., 2012).DNA-barcoding has also been instrumental in resolving local phylogenies (Costa et al., 2007).The use of a single gene in phylogeny reconstruction nevertheless implies some drawbacks.Each gene within a species has its own evolutionary history and many genes do not directly reflect the evolution of the species.Incomplete lineage sorting, gene duplication, or horizontal gene transfer can cause the difference between gene phylogeny and species phylogeny (Maddison 1997).Using data from multiple genes, ideally from both mitochondrial and nuclear genes, thus represents an invaluable tool for the construction of phylogenetic relationships and evolutionary history of species, in particular when other types of data are lacking.
In the high latitudes of both hemispheres, species belonging to the hyperiid genus Themisto (Guérin, 1825) are important in terms of abundance and biomass (Bowman 1960;Auel et al., 2002).They are mainly carnivorous, and play key roles in pelagic ecosystems of polar oceans, linking lower and higher trophic levels (Auel et al., 2002;Yamada et al., 2004;Watts & Tarling, 2012).Themisto is thought to include seven species: Themisto abyssorum (Boeck, 1871), T. australis (Stebbing, 1888), T. compressa (Göes, 1865), T. gaudichaudii (Guérin, 1825), T. japonica (Bovallius, 1887), T. libellula (Lichtenstein in Mandt, 1822), and T. pacifica (Stebbing, 1888) (Bowman, 1960;Zeidler, 2004).Some species of Themisto exhibit clear morphological and ecological differences among congeners and have distinct geographical distributions (Fig. 1; Bowman, 1960;Wing, 1976;Percy, 1993;Dalpadado, 2001;Auel et al., 2002;Yamada et al., 2004;Marion et al., 2008;Prokopowicz et al., 2009;Kraft et al., 2013), but some species are not well differentiated in terms of morphology, which complicates their identification (Yamada et al., 2004).Since the description of the genus by Guérin (1925), the taxonomy of the genus has undergone a series of changes (reviewed by Schneppenheim & Weigmann-Haas, 1986;Zeidler, 2004).Three generic names (Parathemisto Boeck, 1871; Euthemisto Bovallius, 1887, and Themisto) have been used in the classification of its various species.These revisions involved re-arranging known species among different genera.Schneppenheim & Weigmann-Haas (1986) first attempted to build a molecular phylogeny of Themisto based on allozymes in conjunction with morphological data.Their study revealed the presence of four species: T. abyssorum, T. compressa, and T. libellula in the North Atlantic, and T. gaudichaudii in Antarctic waters.No northern Pacific species of Themisto were included in their analysis, however, so the position of T. japonica and T. pacifica within this genus remained unknown.Browne et al. (2007) conducted a phylogenetic analysis of Hyperiidae Dana, 1852 using COI alone, but only two species were included, T. japonica and T. pacifica.Hurt et al. (2013) subsequently attempted to reconstruct the phylogeny of Hyperiidae using COI as well as three nuclear genes, but as in Browne et al. (2007), they only used two species.The phylogenetic relationships among the species of Themisto species have remained largely uncharacterized.
We provide the first molecular phylogeny of Themisto using COI and three nuclear genes (histone 3, 18S rDNA, 28S rDNA).We chose one mitochondrial gene (COI), one nuclear protein-coding gene (histone 3-H3) and two nuclear ribosomal genes (18S rDNA and 28S rDNA) because they have been successfully employed to investigate phylogenetic history (Giribet et al., 2001;Koenemann et al., 2010), including the phylogeny of Hyperiidae (Hurt et al., 2013).We explored the phylogenetic relationships among Themisto species in the Northern Hemisphere by sampling individuals from several locations in the Arctic Ocean and adjacent seas.To better understand the phylogenetic position of the species of Themisto within Hyperiidea, we also assessed their relationship with the representatives of Hyperia (Latreille, 1823) and Hyperoche (Bovallius, 1887) using the COI gene and multilocus data.The resulting phylogenies (gene-trees and species-tree) will help our understanding of marine speciation in this group.

Samples collection and identification
Six hyperiid individuals were collected with 500 and 750 μm mesh size horizontal or vertical nets from various locations in the Canadian Arctic Archipelago during the ArcticNet expedition in 2011 (Table 1).Specimens from the southeastern Bering Sea were collected in 2012 during the Bering-Aleutian Salmon International Survey with a 500 μm mesh bongo net (Eisner et al., 2014).Specimens from Prince William Sound in the Gulf of Alaska were collected in 2013 with a 500 μm mesh ring net.Two specimens from Japan were collected in 2012 from the Oyashio and Kuroshio region and provided by Hiroomi Myiamoto (Atmosphere and Ocean Research Institute, University of Tokyo).One specimen from Norway was collected in 2010 and provided by A. Radulovici (Université du Québec à Rimouski).All amphipods were preserved in 95% ethanol until later analyses.
The morphological identification of each specimen was performed using the taxonomic keys of Bowman & Gruner (1973), Vinogradov et al., (1996), andZeidler (2000).Some individuals were only identified to the genus level because of taxonomic uncertainty.Individuals that were not identified to the species or genus level were also included in our phylogeny and labelled as "unknown".
A 658 bp fragment of the COI gene was amplified for all species using the LCO1490/HCO2198 primers (Folmer et al., 1994) according to the protocol of Dionne et al. (2011).Fragments of circa 300 bp, 1500 bp, and 1800 bp were amplified for the H3, 18S rDNA and 28S rDNA, respectively.The reaction mix contained 1X PCR buffer , 2.2-4mM MgCl 2 (Table 2) , 0.5 mM dNTPs, 0.4 μM of each primer, 1.5 U of Taq DNA polymerase (Life Technologies, Mississauga, ON, Canada), DNA template (around 40-80 ng), and water for a final volume of 25 μl.PCR reactions for each gene were described in Table 2. PCR were performed with an initial denaturation step of 3 min at 94 °C, followed by 40 cycles of 40 s at 94 °C, 40 s at the annealing temperature and 60 s at 72 °C, and a final elongation step of 5 min at 72 °C.All PCR products were verified on a 1.5% agarose gel and direct sequenced by Génome Québec (McGill University, Montréal, Canada).All sequences have been deposited in GenBank under the accession numbers MF356305 to MF443278.All sequences were checked manually on MEGA 5 (Tamura et al., 2011).Multiple sequences alignments were performed with MUSCLE (Edgar, 2004) and were verified by eye.Gblock 0.91 (Castresana, 2000) was used due to the presence of poorly aligned regions in the 18S rDNA and 28S rDNA datasets.
Two dataset were obtained, one containing eight specimens of Themisto and two outgroups and the other containing all the hyperiid specimens from this study and other hyperiid sequences mined from Genbank (Table 1).

Saturation
Saturation in nucleic sequences can decrease the phylogenetic signal contained in a sequence and lead to inaccurate phylogeny reconstruction (Xia 2009).If saturation is detected, some alternatives can be used to minimize its effect, like excluding the third codon positions from the nucleic sequences, or converting the nucleic acid sequences into amino acid sequences because the latter are much more conserved.The presence of saturation within each gene was therefore tested for the Themisto and Hyperiidae datasets with the Xia test (Xia et al., 2003) available in DAMBE 5.3.27(Xia & Xie, 2003).This method tests if the saturation present within the nucleic acid sequences is problematic for the phylogeny reconstruction (Xia, 2009).Saturation was observed for the mtDNA COI gene at the family and at the genus levels but not for the three nuclear genes (Supplementary material Figures S1-S4), indicating a loss of the phylogenetic signal for the COI dataset.

Genetic distances
Divergence at COI among all Themisto species was estimated using the Kimura two parameters model on all Themisto sequences.All ambiguous positions were removed for each sequence pair.Evolutionary analyses were conducted in MEGA 5 (Tamura et al., 2011).

Phylogenetic analyses
Phylogenetic analyses were performed with two methods: maximum likelihood and Bayesian approaches.The maximum likelihood method evaluates several trees and parameters by selecting those that maximize the likelihood, meaning those that render the observed data the most plausible, whereas in Bayesian reconstruction, these parameters are considered as random variables with statistical distributions (Yang & Rannala, 2012).To estimate the branch support for the maximum-likelihood trees, an approximate likelihood ratio test (aLTR) like Shimodaira-Hasegawa was calculated as described in Anisimova and Gascuel (2006) and implemented in PhyML 3.0 (Guindon et al., 2010;Dereeper et al., 2008; http://phylogeny.lirmm.fr/).Bayesian inferences were performed under the appropriate evolutionary model on two runs for 5,000,000 generations until convergence was observed.Trees were sampled every 1,000 generations and the first 25% of sampled tree were discarded as burn-in.Bayesian reconstructions were performed using MrBayes 3.2.5 (Ronquist et al., 2012).The posterior probabilities (PP) were calculated with the 50% majority-rule consensus tree.Values of PP ≥ 0.95 were considered as strong support for the clade, whereas PP values between 0.85 and 0.94 were seen as intermediate support.The evolutionary models for the phylogenetic analyses were selected independently for each gene using TOPALi v2.5 (Milne et al., 2004)  To verify the possibility to concatenate all four genes, we tested the presence of congruency between the mitochondrial and nuclear genes by using the congruence among distance matrices approach (CADM) (Legendre & Lapointe, 2004;Campbell et al., 2009) available in the package ape in R (Paradis et al., 2004).The CADM approach tests the presence of incongruence among the distance matrices.The significance of the CADM test was performed with a permutation test with 1,000 permutations.The Kimura two parameters model was chosen as a distance measure and calculated in MEGA5 (Tamura et al., 2011).As no incongruence was detected between the four genes, all the sequences were concatenated and maximum-likelihood and Bayesian phylogenetic reconstruction were performed on this dataset.For the Bayesian reconstruction of concatenated datasets, separate partitions were assigned to each gene and rates were allowed to vary across partitions.
Since incomplete lineage sorting can lead to poor estimation of species trees when building a concatenated genes phylogeny (see Heled & Drummond, 2009), we reconstructed the species tree by using a Bayesian method including deep coalescence processes as described by Liu (2008) and implemented in BEST.This method assumes that genes are independent and that there is no introgression between species, and allows the estimation of time since divergence and population parameters based on unlinked genes.We used a partitioned dataset and set up the appropriate model of evolution as determined with TOPALi v2 (Milne et al., 2004).

Phylogeny of Hyperiidae
Two approaches were retained for the reconstruction of the family phylogeny.We first used the most complete dataset available to reconstruct the family phylogeny based on the COI nucleic sequences.Since saturation was detected, we performed the family phylogeny based on the amino acids sequences of COI.To do so, we set the correct codon data-frame by using the complete amino acid sequence of the COI gene from Onisimus nanseni (G.O.Sars, 1900) (Uristidae Hurley, 1963) (Genbank accession number: NC_013819).Maximum-likelihood and Bayesian phylogenies were performed on 203 amino acids.Not all species were sequenced for all four genes, leading to missing data.Several studies suggest that including taxa with missing data can improve phylogeny (Wiens, 2006).We therefore used all data available for each species to reconstruct a family phylogeny.We concatenated each sequence available for each species, resulting in a 6,795 bp dataset and after running Gblocks, the final nucleotide dataset was 3,592 bp long.The evolutionary models for the phylogenetic analyses were selected using TOPALi v2.5 (Milne et al., 2004) under the Bayesian Information Criterion (BIC), amino acids under the mitochondrial reverse model with gamma site model (BIC = 4290.16),and incomplete dataset under the GTR + I + G model (BIC = 5 6526.11).Each phylogeny was rooted using an outgroup species belonging to another family of Amphipoda, Cystisoma pellucida (Willemöes-Suhm, 1874) (Cystisomatidae Willemöes-Suhm, 1875).

Phylogenies of Themisto
The final concatenated dataset for the Themisto phylogeny was 4,040 bp long, including 611 bp of COI, 283 bp of H3, 1434 bp of 18S and 1712 bp of 28S characters.No indels were detected for the COI and the H3 alignments.In both 18S and 28S alignments, indels were observed and ranged from 1 to 14 bp and from 1 to 49 bp, respectively.The divergence estimates among the species of Themisto were around 10%, except between T. compressa and T. gaudichaudii (3.83%;Table 3).The highest genetic divergence was observed between T. abyssorum and T. gaudichaudii, and the lowest divergence was observed between T. libellula (Prince William Sound, Gulf of Alaska) and T. libellula (Bering Sea) (0.3%) and the two specimens of T. japonica (0.66%).Twelve percent of divergence was observed between T. libellula from the Arctic and those from the Bering Sea and Gulf of Alaska, suggesting the presence of cryptic species.All single gene phylogenies (Supplementary material Figures S5-S12) were congruent (W = 0.821; P = 0.0001).All genes were therefore concatenated for further analyses.
The maximum-likelihood and Bayesian trees constructed on concatenated dataset under the HKY+ G model (BIC = 23,432.13)supported a strong monophyletic clade for all species of Themisto (aLTR = 0.731; PP = 1) (Fig. 2).Three supported groups were also found, a first group contained the Arctic T. abyssorum and Atlantic T. compressa (aLTR = 0.97; PP = 0.99), a second group included the Arctic T. libellula and North Pacific Themisto libellula (Gulf of Alaska and Bering Sea) (aLTR = 0.91; PP = 1); and a third group included the Pacific species T. japonica and T. pacifica (aLTR = 0.92; PP = 0.99).Results from our species reconstruction based on multilocus dataset recovered the same pattern observed in the concatenated gene trees (Fig. 3).All species of Themisto were grouped in a strong monophyletic clade (PP = 1).A Pacific group that comprised T. japonica and T. pacifica was identified (PP = 0.99).The Atlantic group composed by T. abyssorum and T. compressa (PP = 0.91) was also recovered.It was impossible to conclude if T. libellula belonged to the Atlantic or Pacific groups (PP = 0.55).

Phylogeny of Hyperiidae
Among the 203 amino acids from the COI gene, 88 sites were variables and 61 were informative.Most of this variation occurred at the third codon position.The Bayesian inference and maximum likelihood trees reconstructed under the mitochondrial reverse model with a gamma distribution (BIC = 4,290.16)output a wellresolved tree that provided strong support for the monophyly of Themisto (PP = 0.96;aLTR = 0.931;Figs. 4,5).A polytomy was observed within Themisto based on the analyses of amino acids but not with the nucleotide phylogenies (Supplementary material Figures S2 and S3).
Both amino acids trees differed in the placement of Iulopis loveni (Bovallius, 1887) and the links among the species of Hyperia and Hyperoche.There was also incongruence in both phylogenies in the placement of two individuals morphologically identified as Hyperia galba (Montagu, 1815): one grouped with an unknown specimen (323inc1) (PP = 1; aLTR = 0.91) and the other with Hyperoche capucinus (Barnard, 1930) (PP = 0.91; aLTR = 0.76).The topology of the amino acid tree did not differ significantly from the phylogenies established with the nucleotide analyses except in the positions of some species and the unresolved phylogenetic links among species of Themisto (Supplementary material Figures S2 and S3).Neither phylogeny fully resolved the placement of Hyperiella dilatata (Stebbing, 1888).In the amino acid phylogeny, Hyperietta vosseleri (Stebbing, 1904) is grouped with Phronimopsis s pinifera (Claus, 1879) and the Lestrigonus (H.Milne Edwards, 1830) specimens (PP = 0.7; aLTR = 0.82) instead of Hyperioides sibaginis (Stebbing, 1888) as observed in the nucleotide phylogenies (Supplementary material Figures S2 and S3).Both amino acids and nucleotide phylogenies suggested that Iulopis loveni is at the base of Hyperiidae and Themisto is branching from Hyperiella dilatata and species of Hyperoche/Hyperia (Figs. 4,5).

DISCUSSION
This study provides the most detailed and up-to-date phylogeny of Themisto in the Northern Hemisphere.It differs from previous studies by including for the first time Atlantic, Arctic, and Pacific species of the genus.Our results showed the usefulness of genetic data to recover evolutionary relationships and to flag potential cryptic species.

Species of Themisto
Both concatenated gene and species trees revealed substantial differences in the genetic makeup of T. libellula, indicating the presence of a confounded species of Themisto.This lineage that inhabits the Bering Sea and Prince William Sound (northern Gulf of Alaska) showed more than 12% divergence at COI from all other species of Themisto.Themisto libellula has been traditionally known to inhabit Arctic regions (Dunbar, 1957) and to coexist in the Bering Sea with T. pacifica, which is also found near the coast of California (Wing, 1976).We report the presence of another species in these regions differing from both T. pacifica and T. libellula, suggesting the need for a taxonomic revision of this genus.The smallest divergence for COI was observed between T. compressa and T. gaudichaudii.Themisto gaudichaudii was thought to have a bipolar distribution (Bowman, 1960), but based on allozymic data, such distribution was due to the presence of two cryptic species, T. compressa in the Northern Hemisphere and T. gaudichaudii in the southern waters (Schneppenheim & Wagman-Hass, 1986).Nonetheless, the small divergence observed between these two species is indicative of a putatively more recent speciation event compared to the other species of Themisto.A small genetic divergence was also observed between Arctic and Antarctic clades of the abyssal amphipod Eurythenes gryllus (Lichtenstein in Mandt, 1822) and in three morphospecies of planktonic Foraminifera, suggesting more connections among poles than previously suspected (Darling et al., 2000;Havermans et al., 2013).
Based on the species tree, we retrieved two distinct groups, a strictly Pacific group composed by T. japonica and T. pacifica, and an Arctic-Atlantic group that comprises all other species of Themisto.
Previous studies conducted in the North Pacific have found a discontinuity among Pacific and Bering Sea populations (Addison & Hart, 2005;Coyer et al., 2010), suggesting that Alaska populations have a complex history.Periodical closures of the Bering Strait by both sea ice and the presence of the Bering land bridge during the Pleistocene glacial cycles disrupted the exchanges between the marine faunas of the Pacific and Arctic oceans (Marincovich & Gladenkov, 2001;Hu et al., 2010;Polyak et al., 2010).The isolation between these basins has probably contributed to the divergence between populations of the same species (Hardy et al., 2010).Even if these populations came back in contact at a later time, they could have accumulated enough genetic differences to avoid or reduce success of reproduction between them.This reinforcement could have therefore contributed to increase the genetic isolation among previously separated populations (Coyne & Orr, 2004).Using a molecular clock to date the time of divergence among these species would have helped to infer some hypotheses on the cause of their separation.Because saturation was detected in the mitochondrial gene and in absence of fossil for the hyperiids, however, we were not able to calibrate and date the time when these species diverged.Inferring the cause of speciation among Themisto congeners in absence of time and ecological data is therefore not yet possible.
The discovery of cryptic species along with recent work on holozooplanktonic species (Lee & Frost, 2002and Goetze, 2003, 2010 for copepods;Lörz et al., 2012 for amphipods) beg the question of how marine speciation develops.Habitat preferences or vicariant events can contribute to the differentiation of population leading to the origin of speciation processes in the marine environment (Palumbi, 1994;Norris, 2000;Norris & Hull, 2011).Ecological habitat preferences are known among species of Themisto (Bowman, 1960;Wing, 1976;Koszteyn et al., 1995;Eiane & Daase, 2002;Yamada et al., 2004), each species being associated with a range of temperatures indicating a preference for specific water masses (Koszteyn et al., 1995).It will be of particular interest to compare the specific habitat of Themisto libellula in the Bering Sea with that in the Arctic Ocean to see if the they are associated with different habitats.

Phylogeny of Hyperiidae
Our results branched Themisto with Hyperiella dilatata and Hyperoche/ Hyperia specimens.In the maximum-likelihood and Bayesian phylogenies based on COI of Hurt et al. (2013), species of Themisto either grouped with Hyperoche or at the base of Hyperiidae.In our case, most of our phylogenies based on DNA and amino acids supported the idea of Iulopis loveni at the base of Hyperiidae, and Themisto branched with Hyperiella dilatata and Hyperoche.Our phylogenies of Hyperiidae revealed that species classified in the same genus were not grouped together, as observed for Hyperiella dilatata and Hyperiella antarctica (Bovallius, 1887) as well as for Hyperietta parviceps (Bowman, 1973) and Hyperietta vosseleri.We cannot exclude erroneous morphological identification of these species.Like Browne et al. (2007) and Hurt et al. (2013), we were not able to resolve the phylogenetic relationship between the genera Hyperoche and Hyperia.In particular, two species identified as Hyperia galba were placed in two different clades.It will be useful to investigate in more details the molecular relationships within these two genera in the light of newly available taxonomic descriptions (e.g., Zeidler, 2015).Both DNA and amino acid phylogenies produced similar branching order for most species.Small incongruences resided in the branching of Hyperiella dilatata and Hyperoche martinezii (Müller, 1864).The former branched out of the Themisto clade in the DNA phylogeny but not in the amino acid one.Nucleic acid phylogenies may outperform amino acid phylogenies due to their faster evolving potential (Simmons et al., 2002).The presence of a polytomy for Themisto observed in the amino acid phylogeny but not in the  DNA-based phylogeny seems to support this idea.It should be pointed out that our family phylogeny based on amino acids was produced with a single mitochondrial gene.Several studies have demonstrated the relevance of COI when inferring phylogeny in amphipods (Lörz & Held, 2004;Browne et al., 2007).Molecular phylogenies based on a single gene are nevertheless not the same as the true species phylogenies due to specific processes involved in gene evolution that do not match evolution of the species (Maddison, 1997;Nichols, 2001).For example, incomplete lineage sorting predominates in shallow, divergent species, leading to uninformative phylogenetic histories (Moore, 1995;Degnan & Rosberg, 2009).The use of mitochondrial gene and the presence of saturation could also lead to inaccurate phylogenetic reconstructions.It has been suggested that including taxa with missing data can be beneficial for phylogeny reconstruction (Wiens, 2006).Adding individuals with missing data has helped to resolve the position of some species.Future work on nuclear genes is thus needed to improve the resolution of the clades that are presently poorly supported.

Limits of the techniques
Our single gene phylogenies are mostly congruent with the separation among the species of Themisto from the Pacific, Arctic, and Atlantic.The main difference we noted between maximum-likelihood and inferences was on the support of the nodes.Lower values of support were associated in maximum likelihood, as it is often reported in the literature (Anisimova & Gascuel, 2006).Besides, the placement of one outgroup (Hyperoche capucinus) failed in two maximum likelihood phylogenies (Histone 3 and 28S).Less phylogenetic resolution was observed with Histone 3, in particular one of the two outgroups, falling into the same clade.This gene is known to be highly conserved in eukaryotic genomes (Rooney et al., 2002) and it could be of limited help to reconstruct phylogeny among recently-diverged species.Similarly, the nucleotide sequences of 18S rDNA and 28S rDNA translated into some ambiguities (Chu et al., 2009).The large gaps observed in the sequences of these genes for different genera might have complicated their alignments and hence phylogenetic resolution.

Figure 2 .
Figure 2. Phylogeny of the genus Themisto inferred with the concatenated dataset.aLTR supports and Bayesian posterior probabilities are shown above and below each node, respectively.

Figure 4 .
Figure 4. Bayesian phylogeny of hyperiids performed with amino acids sequences of the cytochrome oxidase I gene under the mitochondrial reversible model with a gamma distribution.Numbers represent posterior probabilities.Downloaded from https://academic.oup.com/jcb/article/37/6/732/4566033 by guest on 28 July 2024

Table 1 .
List of additional species used in the study.Coordinates are expressed in decimal degrees (DD).NA, data not available.

Table 2 .
List of PCR settings used.

Table 3 .
Estimate of divergence between species of Themisto based on cytochrome oxidase I DNA sequences.Percentages of divergence among Themisto species are shown.