The phylogeny of the annelid genus Ophryotrocha (Dorvilleidae)

Abstract

Ophyotrocha is easy to keep in the laboratory and has therefore been used in several studies of evolution and speciation. The phylogenetic relationships within the group are, however, still not clear and morphological and molecular data are contradictory. Here we attempt to shed light on the phylogeny by adding an additional gene (cytochrome c oxidase subunit I) to the previous analyses of the group. However, the results are still incongruent with the results from the morphological data. We also include a species of the genus Iphitime, and conclude that this species falls within the Ophryotrocha clade. The implications are discussed.

Keywords:

• Dorvilleidae
• Lophotrochozoa
• model organism
• molecular phylogeny

Published in collaboration with the University of Bergen and the Institute of Marine Research, Norway, and the Marine Biological Laboratory, University of Copenhagen, Denmark

Introduction

The genus Ophryotrocha was first described by Clarparède & Mencznikow in 1869 from material from a seawater aquarium in Naples. The first species, named Ophryotrocha puerilis (Clarparède & Mencznikow, 1869) was later split into two subspecies, O. puerilis puerilis from the Mediterranean and O. puerilis siberti from the Atlantic (Bacci & La Greca 1953). Since the description of the genus, more than 40 additional species of Ophryotrocha from various habitats have been described.

Many Ophryotrocha species are found in shallow warm-water habitats and often in polluted areas. Accordingly, the effects of pollution (Åkesson & Ehrenström 1984; Brooks et al. 2003; Conlan et al. 2004) and changes in temperature and salinity (Åkesson & Costlow 1978) have been investigated. Ophryotrocha have also been found in temperate regions (Oug 1978, 1994), Arctic and Antarctic regions (Fournier & Conlan 1994; Varella Petti et al. 2006), deep-sea regions (Hilbig & Blake 1991) and in association with hydrothermal vents (Blake 1985; Dreyer et al. 2005). The speciation has also been studied (e.g. Åkesson & Paxton 2005).

Many Ophryotrocha species are easy to maintain in laboratory culture, which has greatly facilitated the possibility of conducting scientific investigations. As the species are morphologically similar, jaw morphology, reproductive mode and egg morphology are widely accepted as some of the most important key characters for distinguishing between species. These structures have undergone numerous investigations (jaw morphology: Bonnier 1893, Paxton 2004, 2005; reproductive mode and egg morphology: Åkesson 1973, Ockelmann & Åkesson 1990, Sella et al. 1997, Cassai & Prevedelli 1999, Sella & Lorenzi 2000, 2003, Schleicherová et al. 2005, Prevedelli et al. 2006, Lorenzi et al. 2006). Based on reproductive mode and jaw morphology, Ophryotrocha have been divided into three informal groups:

1. The labronica group is gonochoristic. When reaching adulthood, the p-type (primitive) jaws are replaced by k-type (kompliziert) jaws.

2. The hartmanni group is hermaphroditic and has a soft, irregularly shaped egg mass. In addition their p-type jaws become replaced by k-type jaws when the individuals become adults.

3. Species belonging to the gracilis group are also hermaphrodites. The gracilis group is recognized by a fusiform egg mass surrounded by a harder cover. The p-type jaws are never replaced.

The three species groups of Ophryotrocha have also been identified in a phylogenetic study by Pleijel & Eide (1996) based on morphology. Several life table (Åkesson 1982) and life cycle (Åkesson 1976a, b; Paavo et al. 2000; Prevedelli & Simonini 2003) studies have been conducted.

The division of Ophryotrocha into three species groups has not been supported by genetic investigations. A division between gonochoristic and hermaphroditic species was found by Dahlgren et al. (2001), who utilized the mitochondrial 16S gene to study the relationship between 18 Ophryotrocha species. Their results confirmed the monophyly of the labronica group, but the hartmanni and gracilis groups were shown to be paraphyletic.

The placement of Ophryotrocha within Annelida is debated. Ophryotrocha was originally placed within the Dorvilleidae. This placement has been challenged by Orensanz (1990), who proposed the establishment of the family Iphitimidae, including Ophryotrocha and Iphitime. This view was not supported by Eibye-Jacobsen & Kristensen (1994), who found Iphitimidae not to be monophyletic and a subtaxon of Dorvilleidae sensu Orensanz (1990), but still favoured a close relationship of Ophryotrocha and Iphitime. Caryological (Robotti et al. 1991; Sella et al. 1995) and genetic (Dahlgren et al. 2001) methods have been used to investigate the relationships among ophryotrochan species, as well as their relationships to other genera. These studies have indicated close relationships among species within Ophryotrocha and between Ophryotrocha and other genera placed in Dorvilleidae.

In order to shed light on the incongruent results from the molecular and morphological phylogenetic analyses (Pleijel & Eide 1996; Dahlgren et al. 2001), the phylogenetic relationships among some species of Ophryotrocha were addressed using the mitochondrial gene cytochrome c oxidase subunit I (COI).

Two different data sets were used, one set using only COI data and one set using combined COI and 16S data. In the first analysis, a sequence from Iphitime paguri Fage & Legendre, 1934 was also included in order to be able to address the question of the relationship between Iphitime Marenzeller, 1902 and Ophryotrocha. Several outgroup taxa were chosen for this investigation. Dorvillea albomaculata was used as this was one of the outgroup species used by Dahlgren et al. (2001). Alvinella pompejana Desbruyères & Laubier, 1980 was used to aid the aligning, and was therefore also included in the analysis.

Material and methods

In total, 20 specimens belonging to 14 different Ophryotrocha species were included in the phylogenetic analyses. Additionally, one specimen of Iphitime paguri and one specimen of Dorvillea albomaculata fixed in 98% alcohol were included. A list of included species is provided in Table I. All samples were extracted by use of the Qiagen DNeasy Tissue kit following the animal tissue protocol. All species are deposited in the Bergen Museum, Natural History Collections.

Table I. Summary of samples included in the survey. The catalogue numbers are vouchers deposited in the Bergen Museum, Natural History Collections of Bergen.

Sample	Sample locality	ZMBN catalogue no.
O. puerilis	Malaga, Spain	77025
O. japonica A	LaSpecia, Italy	77030
O. japonica B	Ravenna, Italy	77021
O. japonica C	Taranto, Italy	77022
O. permanni	Tampa, USA	77034
O. labronica	Costa del Sol, Spain	77039
O. notoglandulata	Tokyo, Japan	77051
O. gracilis	Sylt, Germany	77053
O. robusta	Malaga, Spain	77054
O. alborana	Marbella, Spain	77056
O. diadema	Long Beach, USA	77057
O. hartmanni	Malaga, Spain	77060
O. rubra	Ceuta, Spain	77919
O. sp. Qingdao A	Qingdao, China	77042
O. sp. Qingdao B	Qingdao, China	77041
O. sp. Hurghada	Eiliat-Hurghada, Egypt	77064
O. sp. Sanya	Sanya, China	77061

All amplification reactions were carried out in a final volume of 50 µl. The primer pairs LCO 1490 (forward: 5′-GGT CAA CAA ATC ATA AAG ATA TTG G-′3) and HCO 2198 (reverse: 5′-TAA ACT TCA GGG TGA CCA AAA AAT CA-′3) (Folmer et al. 1994) were used. The protocol followed included 1 µl of each primer, 5 µl 10×buffer, 4 µl dNTPs, 0.25 µl Takara (polymerase) and 35.75 µl ddH₂O. In addition, 4 µl template was used. Polymerase chain reactions (PCR) were carried out in 0.2 ml PCR tubes (Eppendorf®) in a thermal cycler (Applied Biosystems GeneAmp® PCR System 2700 or MJRESEARCH PTC-200 Pelitier Thermal Cycler) both under split temperature cycle conditions: denaturation temperature 94^°C for 1.5 min; five cycles of 94^°C for 30 s, 45^°C for 1.5 min, 72^°C for 1 min; 34 cycles of 94^°C for 30 s, 51^°C for 1.5 min, 72^°C for 1 min; finally, 72^°C for 7 min and 4^°C indefinitely. The PCR products were loaded on to a 1% agarose gel stained with ethidium bromide. For visualization, an ultraviolet transilluminator and Grab-IT v2.59 computer software were used. The QIAquick PCR Purification Kit was used to clean the PCR products, according to the manufacturer's recommendations, using 30 µl of the elution buffer. Purified PCR products were loaded on to a 1% agarose gel for verification of the PCR product.

Sequencing of the PCR product was performed using the following protocol: Big-dye, 1.0 µl; sequencing buffer, 1.5 µl; primer 1, 0.5 µl; primer 2, 0.5 µl; ddH₂O, 6.0 µl. A volume of 9 µl mix and 1 µl template per sample was used. The temperature cycle was initially set to 95^°C for 5 min, followed by 25 cycles of 96^°C for 10 s, 50^°C for 5 min and 60^°C for 4 s. Finally, the temperature was set to 4^°C indefinitely. Ten microlitres ddH₂O was added before the sequences were delivered to the Sars Centre at the University of Bergen where the sequencing was performed using an ABI 3700 DNA Sequencer (Applied Biosystems). Cleaning the sequenced product was also performed by the Sars Centre, using Millipore Montage seq cleanup and a BIOMEK 2000 robot (Beckmann). Sequences were assembled and aligned using the Seqman package in Lasergene 6.1 (DNASTAR, Inc.). Gap penalties were set to 45 and gap extension penalties were set to 30 to ease the aligning. Alvinella pompejana (GeneBank accession no. AY645988) was downloaded from the NCBI GeneBank and used to ease the aligning. ClustalX (Thompson et al. 1997) and BioEdit (Hall 1999) were used for further editing and trimming the alignments.

Phylogenetic analysis was applied to two different data matrices. Dataset one (COI analysis) comprised only sequences obtained in this study. The second dataset (combined analysis) comprised 16S sequences from Dahlgren et al. (2001) and some of the COI sequences from this study. GeneBank accession numbers are provided in Table II. Alignments of COI and 16S were carried out separately and were joined afterwards for the phylogenetic analysis.

Table II. Taxon and authors for the species included in the analysis. GeneBank accession numbers for cytochrome c oxidase subunit I (COI) and the 16S sequences included in the combined analysis are also given.

Taxon	Author	Accession no. COI	Accession no. 16S
A. pompejana	Desbruyères & Laubier 1980	AY645988
I. paguri	Fage & Legendre 1934	EF464549
D. albomaculata	Åkesson & Rice 1992	EF464550	AF380115
O. alborana	nomen nudum	EF464540	AF321420
O. diadema	Åkesson 1976b	EF464534	AF321422
O. gracilis	Huth 1933	EF464545	AF321423
O. hartmanni	Huth 1933	EF464546	AF321424
O. japonica	nomen nudum	EF464541	AF321433
O. labronica	La Greca & Bacci 1962	EF464536	AF321425
O. notoglandulata	Pfannenstiel 1972	EF464542	AF321427
O. permanni	nomen nudum	EF464535	AF321428
O. puerilis	Claparède & Metschnikow 1869	EF464544	AF321429
O. robusta	nomen nudum	EF464547	AF321430
O. rubra	nomen nudum	EF464543
O. sp. Sanya	not formally described	EF464537	AF321435
O. sp. Qingdao	not formally described	EF464539	AF321434
O. sp. Hurghada	not formally described	EF464538	AF321432

Phylogenetic analyses were conducted in the program PAUP 4.0 beta 10 win (Swofford 2002). The following analyses were applied to both datasets. A simple parsimony analysis was conducted by a heuristic search. Modeltest 3.06 (Posada & Crandall 1998) was applied to find the best model for maximum likelihood. Furthermore, MrModeltest 2.2 (Nylander 2004) was applied to find the best model for Bayesian analysis. The results from this analysis were used in MrBayes 3 (Huelsenbeck & Ronquist 2003) for Bayesian inference.

Each COI sequence contained 628 base pairs. Both a Modeltest search and an iterative procedure for model optimization were conducted. The iterative procedure was based on a most-parsimonious tree as the starting tree. A GTR + I+G model was obtained as the best fitting model. The best fitting model obtained from the Modeltest search (hierarchical likelihood ratio tests) was used as a model when iteratively searching for optimized parameter values. The optimized log likelihood value was 6783.70707. Base frequencies (A = 0.2957, C = 0.1840, G = 0.1680 and T = 0.3523) and substitution rates ([A–C] = 1.1518, [A–G] = 7.5098, [A–T] = 2.4235, [C–G] = 3.4645 and [C–T] = 9.6478) were estimated. The proportion of invariable sites (I) was 0.2015 and the gamma distribution shape parameter was 0.4966.

A bootstrap analysis was conducted for COI with a neighbour-joining tree as the starting tree and by branch swapping with neighbour interchanges. Computing time allowed for only 100 bootstrap replicate searches.

A GTR + G+I model was also suggested by MrModeltest 2.2. MrBayes 3 was employed for Bayesian inference. A MCMC search was performed with 1,000,000 generations, and the sampling frequency set to 100. Burnin was set to 250 based on the equilibrium of log likelihood values.

For the combined analysis, the same procedure as used in the COI analysis was applied. Modeltest 3.06 and an iterative maximum likelihood search were used to find the model that best fit the sequence matrix. The best fitted model resulting from hierarchical likelihood ratio tests was the GTR + G+I model. The optimized log likelihood value was estimated to be 8734.7275 in the iterative searches for the best fitting model. The base frequencies were estimated to be A = 0.2964, C = 0.1744, G = 0.1834 and T = 0.3457 and the resulting rate matrix values were [A–C] = 1.1817, [A–G] = 5.3021, [A–T] = 2.3637, [C–G] = 1.5677 and [C–T] = 4.8543. The gamma distribution shape parameter was estimated to be 1.0508 and the proportion of invariable sites was estimated to be 0.2520.

MrModeltest 2.2 was used to find the model that best fitted the data for inferring phylogeny with the program MrBayes. The result was a GTR + G+I model. Phylogenetic analyses were conducted in MrBayes 3.0 with MCMC searches and with 1,000,000 generations, and the sample frequency set to 100. Burnin was set to 250 based on the equilibrium of log likelihood values.

Alignments have been submitted to treebasee, submission number SN3407-15085.

Results

COI parsimony

The sequence matrix was made up of a total of 20 different specimens from 17 different species (Table II). The parsimony search resulted in four equally parsimonious trees, 1760 steps long, in which 202 characters were constant. Of the resulting variable characters, 59 were parsimony uninformative, whereas 334 were parsimony informative. The consistency index was 0.4630 and the retention index was 0.5539. The consensus (Figure 1) of these trees was used as the starting tree in the maximum likelihood analyses.

Figure 1.  Consensus of the eight most-parsimonious trees resulting from the analysis of cytochrome c oxidase subunit I data.

COI maximum likelihood

The maximum likelihood analysis was conducted through a heuristic search. The first maximum likelihood search was based on the consensus tree from the parsimony analysis. One tree was obtained by stepwise addition and branch-swapping algorithm set to tree-bisection-reconnection. The analysis resulted in three distinct clades (Figure 2). Clade A consists of O. robusta, O. rubra, O. sp. Hurghada, O. labronica, O. sp. Sanya, O. notoglandulata and O. japonica, which all are all consistent with the so-called labronia group. Clade B consists of O. puerilis, Iphitime paguri, O. hartmanni and O. gracilis. Finally, clade C is made up of O. alborana, the two O. sp. Qingdao, and O. diadema.

Figure 2.  The single best tree from the maximum likelihood analysis of the cytochrome c oxidase subunit I data. Branch support values given in italics indicate bootstrap values, whereas those in bold are Bayesian values.

The bootstrap consensus tree included several collapsed branches, but most of the branch support values were fairly high (Figure 2). An exception was the support value of the branch leading to clade A and B. Also interesting is the branch support value of the clade O. gracilis and O. hartmanni, which shows low bootstrap support. Because the branch of O. puerilis and I. paguri is collapsed, the branches leading to clade B and to I. paguri, O. gracilis and O. hartmanni are lacking bootstrap support values.

COI Bayesian analysis

The Bayesian analysis resulted in a phylogenetic consensus tree identical to the maximum likelihood tree (not shown). The support values resulting from the Bayesian analysis are given in bold in Figure 2.

Combined analysis – parsimony

The data matrix for the combined analysis was constructed from 13 taxa and 1061 characters. In the combined analysis, the sequence of O. japonica A was chosen to represent the O. japonica species, whereas O. sp. Qingdao A was chosen to represent the O. sp. Qingdao species. A single most-parsimonious tree with length 1907 resulted when stepwise addition and branch-swapping algorithm set to tree-bisection-reconnection were applied (Figure 3). Of the characters estimated, 408 were constant. The number of variable characters which were parsimony uninformative was 138, whereas 515 of the variable characters were parsimony informative. The resulting tree had a consistency index of 0.5868 and a retention index of 0.5247. This tree was used as the starting tree for the iterative search of a maximum likelihood tree.

Figure 3.  The single most-parsimonious tree resulting from the combined analysis including only the ophryotrochans.

Combined analysis – maximum likelihood and bootstrap support

Through a stepwise addition and branch-swapping algorithm set to tree-bisection-reconnection, one maximum likelihood tree was obtained (Figure 4). The resulting phylogenetic tree showed no division of species into the clades indicated by morphology. Not all of the distinct clades that were obtained in the COI analysis were found in the combined analysis. The same species that were placed closely in the phylogenetic tree resulting from the COI analysis were also grouped closely in the combined analysis.

Figure 4.  The single best likelihood tree from the combined analysis only including species from the genus Ophryotrocha. The branch values indicate bootstrap support. The line on the bottom left corner of the figure shows the number of substitutions.

Combined analysis – Bayesian analysis

The Bayesian analysis resulted in a tree similar to the maximum likelihood tree. The resulting tree showed high posterior probability on branches. An exception was the branch leading to O. sp. Hurghada, O. japonica and O. notoglandulata, with a value of 0.68. The result of the analysis is shown in Figure 5.

Figure 5.  Bayesian consensus tree of the combined analysis. The values indicate posterior probabilities.

Discussion

Morphological congruence

The COI analysis and the combined analysis resulted in similar trees. The species were placed into two or three groups. The tree with two groups indicated one gonochoristic group: O. rubra, O. robusta, O. sp. Hurghada, O. labronica, O. sp. Sanya, O. notoglandulata, O. japonica A–C; and one hermaphroditic group: O. puerilis, O. hartmanni, O. gracilis, O. alborana, O. diadema, and O. sp. Qingdao A and B. The tree that resulted in three groups had an additional subdivision of the hermaphroditic species. One of the hermaphroditic groups contained O. puerilis, O. harmanni, O. gracilis and Iphitime paguri, whereas the other group incorporated O. sp. Qingdao A and B, O. diadema and O. alborana. By excluding O. sp. Qingdao A and B and O. alborana, which comprise most of the species in one of the two hermaphroditic groups, O. diadema was forced to regroup. As a consequence, one of the hermaphroditic clades was excluded.

Bootstrap and Bayesian analyses indicated good support for the branch leading to O. hartmanni and O. gracilis. Former analyses based on morphological data are incongruent with the grouping of O. hartmanni and O. gracilis resulting from this analysis. Ophryotrocha hartmanni and species similar to it posses k-type jaws and have an irregularly shaped egg mass. Species similar to O. gracilis possess p-type jaws and a fusiform-shaped egg mass.

Molecular investigations conducted earlier have also resulted in the placement of O. hartmanni and O. gracilis as sister species (Dahlgren et al. 2001) (Figure 6). Dahlgren et al. did not have any suggestions to explain this pattern. The constrained analysis also resulted in a significant difference between the tree obtained by Dahlgren et al. and the results obtained here. Clearly, more studies are needed to finally clarify the phylogenetic relationships within Ophryotrocha.

Figure 6.  Phylogenetic tree resulting from the maximum likelihood analysis conducted by Dahlgren et al. (2001). The filled circles indicate nodes congruent with the investigation of Pleijel & Eide (1996).

The placement of Iphitime paguri

The phylogenetic analysis placed I. paguri within the Ophryotrocha clade. The relationship between these groups has previously been argued for by Orensanz (1990). Not all agree on treating the genera as closely related. Eibye-Jacobsen & Kristensen (1994) argued that ophryotrochans and iphitimids should be placed into different families, Dorvilleidae and Iphitimidae, respectively. Høisæter & Samuelsen (2006) have investigated iphitimids and believe that the conclusions of Orensanz (1990) may deserve more credit. Investigations conducted by Paxton (2004) based on jaw morphology have also justified the conclusions of Orensanz.

The placement of I. paguri in this study indicates that this species could be placed within Ophryotrocha. Because the relationship between I. paguri and the type species I. doederleinii Marenzeller, 1902, that is the type species by monotypy, has not been investigated, we cannot at this stage conclude anything about the relative positions of Ophryotrocha and Iphitime. New investigations are necessary to resolve the iphitimid–ophryotroch relationship.

Editorial responsibility:Kenneth Halanych

Acknowledgements

Jon Kongsrud, Bergen Museum, Fredrik Pleijel, Göteborg University, and Eyvind Oug NIVA are thanked for help with obtaining references. Kenneth Meland and Endre Willassen, University of Bergen are thanked for critically reading and commenting on the manuscript. Critical comments from the subject editor and two anonymous referees greatly improved the manuscript. The study was funded by grants from the University of Bergen and from the Meltzer Foundation.

Notes

Published in collaboration with the University of Bergen and the Institute of Marine Research, Norway, and the Marine Biological Laboratory, University of Copenhagen, Denmark