Evaluation of combined morphological and molecular techniques for marine nematode (Terschellingia spp.) identification
- First Online:
- Cite this article as:
- Bhadury, P., Austen, M.C., Bilton, D.T. et al. Mar Biol (2008) 154: 509. doi:10.1007/s00227-008-0945-8
- 308 Views
Marine nematodes, which play an important role in many ecosystems, include a number of apparently cosmopolitan taxa that exhibit broad biogeographic ranges even though there is no obvious dispersal phase in their lifecycle. In this study, standard taxonomic approaches to marine nematode identification in conjunction with multivariate statistical analysis of morphometric data were compared with molecular techniques. Specimens of the marine nematode Terschellingia longicaudata that had been identified by their morphological features were investigated from a range of localities (East and West Atlantic, Bahrain, Malaysia) and habitats (estuarine, intertidal, subtidal) using molecular approaches based on the amplification and sequencing of the small subunit ribosomal RNA (18S rRNA). The study revealed that the majority of the morphologically defined T. longicaudata specimens share a single 18S rRNA sequence and apparently belong to a single taxon distributed from the British Isles to Malaysia. In addition, 18S rRNA analysis also revealed two additional sequences. One of these sequences was found in both the British Isles and Mexico, the other was recorded only from British waters. Individuals collected in Bahrain and identified from their morphology as T. longicaudata had two highly divergent 18S rRNA sequences. Separate morphological and morphometric approaches to identification of specimens from the same sites that had been formalin-preserved did not support evidence of multiple genotypes revealed previously by molecular analysis. Current taxonomy based on morphological characters detected using light-microscopy may be unable to discriminate possible species complexes. Biodiversity of marine nematodes may often be underestimated due to the presence of morphologically cryptic species complexes. High-throughput techniques such as DNA barcoding would aid in species identification but may require thorough analysis of multiple nuclear and mitochondrial molecular markers.
Marine nematodes are important to the functioning of estuarine and marine ecosystems and often dominate the benthic meiofauna in terms of abundance and diversity (Austen 2004; Lambshead 2004). They play an important role in the decomposition process, aid in recycling of nutrients in marine environments (Austen 2004) and are sensitive to changes in the environment caused by pollution (e.g. Lambshead 1986; Austen and McEvoy 1997; Boyd et al. 2000).
Despite this importance, taxonomic intractability of the group has limited applications in routine biomonitoring or ecological surveys. Identification of nematodes relies on detailed morphological examination (Platt and Warwick 1988) that requires specialist taxonomic expertise and therefore this group of meiobenthic organisms remains largely ignored. With the advent of molecular technologies, new tools are being increasingly used to accelerate identification of taxonomically intractable groups such as marine nematodes. In the last few years molecular technologies such as DGGE and DNA barcoding approaches have been tested towards assessment of nematode diversity and for rapid identifications (Cook et al. 2005; De Ley et al. 2005; Bhadury et al. 2006a, b). Issues related to the phylogeny of marine nematodes have been also addressed in detail using molecular approaches (Meldal et al. 2007). DNA barcoding, based on the analysis of a segment of the genome has been proposed as a way for rapid identification of marine nematodes (Rogers and Lambshead 2004; De Ley et al. 2005; Bhadury et al. 2006b). Molecular markers such as the nuclear small subunit and large subunit ribosomal RNA (18S rRNA and 28S rRNA) have been the preferred choice in nematode barcoding studies (De Ley et al. 2005; Bhadury et al. 2006a). Most of the barcoding studies on marine nematodes have focused on British and North American waters. Yet if molecular markers are to be truly successful in barcoding studies, there is a need to evaluate them in nematode specimens collected from different biogeographic locations.
The main purpose of this study was to collect nematodes from different biogeographic locations, identify them using standard taxonomic approaches as well as morphometric analysis, and then test whether this standard taxonomy corresponds to molecular data. This is particularly important because incongruence between the two methods may result in misidentification and wrongful interpretation during molecular ecological surveys employing methods such as DNA barcoding. To evaluate the phenotypic and genotypic methods, a common and abundant marine nematode, Terschellingia longicaudata De Man, 1907 was selected. T. longicaudata, as currently understood, has an extensive geographical range (Gerlach and Riemann 1973, 1974), being reported extensively from British estuarine and coastal waters (Platt and Warwick 1983), and from the Atlantic coast of France, the Black Sea, the USA (Gulf of Mexico), China (Qingdao province), New Zealand and the Solomon Islands (Sergeeva 1991; Zhang and Ji 1994; Burgess et al. 2005).
Materials and methods
Details of the localities and habitats from where sediments were collected in this study
Subtidal muddy sand
Subtidal sandy mud
Estuarine intertidal mud
Estuarine subtidal mud
Cancún (Atlantic coast)
Ras al Barr
Sheltered bay with mangrove vegetation
Sungai Merbok estuary
Estuarine mangrove vegetation
Specimen preparation and taxonomic identification
Sediments from each location except Merbok (Malaysia) were divided into two parts. One part was fixed in molecular grade ethanol for molecular work and the other part fixed in formalin for morphometric studies. Washed meiofauna were extracted from formalin-fixed sediments, following Somerfield et al. (2005) protocol with the exception of Merbok samples. Merbok specimens collected in 1998 were initially stored in 4% buffered formalin and then mounted onto slides for studying the nematode community structure (Somerfield et al. 1998). Extraction and identification of specimens from alcohol-fixed sediment was carried out following the protocol described in Bhadury et al. (2006b).
Each specimen for molecular analysis was placed in anhydrous glycerol on a microscopic slide using standard techniques (Somerfield et al. 2005) and then identified under a compound microscope using diagnostic morphological characters by expert taxonomists. Fifteen individual specimens were identified from each of the following sites: the Tamar estuary, the Plym Estuary, the National Marine Monitoring Programme site on the Humber Estuary (NMMP) and Rame Head. Fewer individual specimens were available for morphological identification at the other sites: there were five individuals from each of North Tubli Bay and Ras al Barr in Bahrain, and from Brittany (Northern France), two individuals from Merbok and three individuals from Cancún (Mexico). In total 80 individuals were identified based on morphological characters (as detailed in Platt and Warwick 1983) that distinguish T. longicaudata from other marine nematode taxa.
Characters measured for each Terschellingia longicaudata specimen processed
Females and males
Body length; maximum body diameter; anal body diameter; tail length; oesophagus length; oesophageal bulb diameter; head diameter; amphid diameter; cephalic seta length; sub-cephalic seta length; somatic seta length; cervical seta length
Distance from vulva to head
Gubernaculum length; Spicule length
The 18S rRNA gene from all 80 specimens was targeted for amplification and sequencing. Amplification of the mitochondrial cytochrome oxidase I (COXI), 16S and cytochrome b genes were abandoned due to low amplicon yield from the specimens. The entire process starting from meiofauna extraction, identification of nematodes followed by DNA extraction and subsequent throughput steps took between 5 and 7 days.
DNA extraction, PCR amplification and sequencing of the 18S rRNA gene
DNA was extracted from nematodes following the Bhadury et al. (2006b) protocol. Two primers, MN18F (5′-CGCGAATRGCTCATTACAACAGC-3′) and Nem_18S_R (5′-GGGCGGTATCTGATCGCC-3′), were used to amplify approximately 926 bp of the 18S rRNA gene (Floyd et al. 2005; Bhadury et al. 2006a). For PCR parameters, see Bhadury et al. (2006b). PCR fragments were subsequently sequenced in both directions using the same set of primers.
DNA extraction and PCR amplification from formalin-stored Merbok (Malaysia) samples
Merbok T. longicaudata specimens were carefully removed from microscopic slides and then subjected to DNA extraction using a hot-lysis protocol described previously by Bhadury et al. (2007). Following extraction, genomic DNA from each individual was subjected to amplification and sequencing (for protocols see Bhadury et al. 2007).
Sequence quality and phylogenetic analysis
Sequences were checked for ambiguities and errors using ChromasPro and were subsequently aligned in Clustal-X using default parameters (Thompson et al. 1997; Jeanmougin et al. 1998). Authenticated marine nematode 18S rRNA sequences (926 bp length) from GenBank and EMBL databases were also included in the alignments (accession numbers in the phylogenetic tree). A neighbor-joining (NJ) tree was constructed in MEGA v3.0 (Kumar et al. 2004) using the gamma corrected Kimura two parameter distance method (Blaxter et al. 1998). The NJ tree was subsequently validated with bootstrap analysis using 1,000 replicates.
Morphometric data analysis
Multivariate analyses of the morphometric data were performed using PRIMER version 5.1 (Clarke and Warwick 2001). Lower triangular dissimilarity matrices were constructed in PRIMER using normalized Euclidean distance, without prior data transformation. Non-metric multidimensional scaling (MDS) (Clarke and Gorley 2001), an ordination technique which is robust in representing high dimensional data (indicated by acceptable stress values) was then applied to the dissimilarity data (Clarke and Warwick 2001). MDS ordinations were carried out separately using data from: all individuals with all characters; all individuals with non-sexual characters only; all females with all characters; and all males with all characters. Analysis of Similarity (ANOSIM) was applied to determine the degree and significance of differences between populations based on multiple morphometric characters. Similarity percentages (SIMPER) was used to investigate the contribution of individual morphological characters to either the significant differences (in ANOSIM) between sites (a priori testing) or between populations separated posteriori in the clustering pattern in MDS ordinations.
Successful amplification and sequencing were achieved in all the T. longicaudata specimens identified previously on key morphological characters. The majority of the sequences showed 100% identity with known T. longicaudata sequences held online at GenBank and EMBL (Accession Nos. AY854230 and AM234716). Most of the 18S rRNA sequences from the Tamar estuary, Plym estuary, Rame Head, NMMP (UK), Brittany (France), Cancún (Mexico) and all sequences from Merbok (Malaysia) were identical. For Merbok samples, only 345 bp of the 18S rRNA gene were amplifiable. All the sequences have been submitted to EMBL (Accession numbers AM941225 to AM941304).
Multivariate analyses of T. longicaudata specimens
Multi-dimensional scaling (MDS) plots for females (44 specimens) (Fig. 2c) and males (16 specimens) (Fig. 2d) from all geographical locations did not exhibit significant differences when compared with previous MDS plots.
ANOSIM of T. longicaudata specimens
Two sites, Sungai Merbok and Cancún were excluded from ANOSIM due to insufficient replicate specimens. ANOSIM analysis of males only was not carried out, again due to lack of sufficient specimens from all sites.
R values from one-way ANOSIM pairwise tests
Data analysed sites compared
All individuals, all charactersa
All individuals non-sexual charactersb
All females, all charactersc
Ras al Barr
Ras al Barr
Ras al Barr
Ras al Barr
Ras al Barr
Ras al Barr
Similarity percentage calculations were determined on all specimens (male and female) based on all morphometric characters followed by exclusion of sexual characters and then separately on males and females. Merbok was excluded from the analysis because of the limited availability of specimens. Combining male and female data and with all characters taken into consideration, the lowest and the highest percentage dissimilarity values (average squared Euclidean distances between samples) were 16 (Plym and Cancún) and 43 (Brittany and Ras al Barr) respectively. Various characters contributed towards separation of the populations but the importance of these characters varied across the sites being compared and no consistent patterns could be discerned. This was also the case when SIMPER analysis was carried out on combined male and female data but excluding sexual characters for which the lowest and highest average percentage dissimilarity values were recorded as 16 (Plym and Cancún) and 41 (Brittany and Ras al Barr) respectively and when females only were analysed where the percentage dissimilarity values between sites were again generally low, the lowest value being 16 (Cancún, Plym estuary) and the highest 41 (Cancún, North Tubli Bay).
We evaluated the 18S rRNA gene as a diagnostic marker in our study since its sequences are generally species specific and it contains both conserved (primer design) and variable (taxonomic distinction) regions (Blaxter et al. 1998; Foucher and Wilson 2002; Cook et al. 2005; Bhadury et al. 2006a). The majority of apparent T. longicaudata specimens identified using diagnostic morphological characters in a traditional taxonomic approach were consistent at the 18S rRNA level based on molecular analysis. This indicates a reasonable degree of congruency between the morphological and molecular methods. In this study more than 70% of the specimens collected from different locations shared a single 18S rRNA sequence.
As stated in the results, there were some exceptions to this general pattern; two additional sequence types were detected besides the commonest ‘T. longicaudata’ 18S rRNA sequence. The first sequence type (referred to as type A in the phylogenetic tree Fig. 1) was detected from the Tamar estuary (one specimen) and Rame Head (four specimens) in UK and Cancún (one specimen) in Mexico. Sequence type A differed by 25 base pairs from the commonest ‘T. longicaudata’ 18S rRNA sequence. In addition, seven specimens from the NMMP site had a sequence (type B) that differed by 19 base pairs from the commonest ‘T. longicaudata’ 18S rRNA sequence.
Two completely different 18S rRNA sequence types were detected in specimens collected along the Bahrain coast (Ras al Barr and North Tubli Bay) showing only 90% similarity to the commonest T. longicaudata sequence. All of the sequences from Ras al Barr were 99% identical to the Sphaerolaimus hirsutus (Sphaerolaimidae) 18S rRNA sequence and differed by 4 base pairs only (Fig. 1). The North Tubli 18S rRNA sequences were different again with highest identity to Parodontophora sp. (99% identity level), a genus in the family Axonolaimidae and a relatively distant phylogenetic lineage from that occupied by T. longicaudata, which is in the family Linhomoeidae. The presence of these aberrant sequences in specimens from Bahrain, identified as T. longicaudata, is intriguing, and could have a number of possible explanations.
One of the most plausible explanations for such sequence variation in Bahrain specimens is that they are a result of misidentification. Most nematodes are a few millimeters in length and juveniles are always difficult to identify compared to adults, especially when identifications are based on minute morphological characters. When nematodes are fixed in molecular grade ethanol for molecular analysis, their bodies can shrink rapidly making identification much more difficult. This may have confounded selection of specimens for molecular analysis based on morphology. Molecular and morphometric studies were carried out on different sets of specimens because of the effect of formalin preservation on molecular techniques and the effect of alcohol on morphological features. T. longicaudata, S. hirsutus and Parodontophora sp. are distinguished using morphological characters that are markedly different, although these taxa all belong to the same order within the phylum Nematoda. Specimens selected from the Bahrain sites for molecular analyses may have been juveniles, which had not yet developed the adult characters that allow positive identification (Platt and Warwick 1988), and could inadvertently have been assigned to the wrong genus. In contrast, taxonomic misidentification could be ruled out for specimens that showed two sequence types (A and B) since the 18S rRNA sequences indicated that they belong to the family Linhomoeidea and were identified as T. longicaudata based on key identifying characters. These two sequence types could also represent other species within the genus Terschellingia or entirely new set of divergent sequences. The most puzzling question is the observed taxonomic characters of these specimens collected from different biogeographic regions. All these specimens displayed characters that are diagnostic of T. longicaudata and were identified by experienced taxonomists, yet from the molecular perspective some of the genotypes were different from the dominant T. longicaudata 18S rRNA sequence. The morphological identification used in this study was based on key characters routinely used for identifying T. longicaudata specimens and the results of this study highlight the necessity of using ultrastructure based taxonomy in conjunction with video capture methods (see De Ley et al. 2005) for obtaining maximal morphological resolution. Other possible explanations for the observed sequence disparities in Bahrain specimens may be due to carry-over of contaminant DNA between DNA extracts or from PCR product to DNA extract or possible workflow and data analysis errors. A combination of any these factors in the molecular assembly line could have contributed to the observed molecular disparity in the specimens.
The application of morphometrics to populations collected from different geographical locations suggested that these are indistinguishable morphologically. With the exception of the Bahrain samples, multivariate analysis did not generally point to the presence of consistently distinct groups, either within the populations from different locations, among the populations from different UK locations, or among populations around the globe, despite the molecular findings. MDS ordinations showed very little difference between populations, although some specimens were outliers in the ordination plots. These specimens differed to a certain extent from the other populations based on the presence or absence of certain morphometric characters. The morphometric approach was not, apparently, sufficiently powerful to distinguish between specimens with two different sequences found in the molecular analyses (groups A and B). These specimens may indeed be morphologically indistinguishable, or alternatively not have been present in the samples selected for morphological analysis. The morphometric measurements were based on traditional characters while the ultrastructure-based characters were not measured in this study. As a result the ultrastructural characters displaying actual differences in the measurable morphology of the specimens with different sequence types (groups A and B) could have gone unnoticed. The genotypic patterns that were observed in molecular studies were not visible during phenotypic characterization and this is not surprising because the morphometric analysis was not carried out on the same set of individuals that were later tested for molecular analysis. This was because of the effect of the preservation techniques and its drawback on subsequent molecular methods. As a result specimens have sequences different from the commonest T. longicaudata sequence were not distinguishable during morphometric and subsequent statistical analyses.
As the majority of the specimens from Atlantic and Malaysia shared a common 18S rRNA sequence, T. longicaudata has a broad geographic range. In addition, these samples come from a wide range of marine habitats, including intertidal mud and shelf sediments at 70 m depth and both estuarine and fully saline coastal waters, indicating that the ecological range of this taxon also appears to be genuinely wide. The most interesting outcome of this study has been the detection of two additional sequences types (A & B) with sequence type A detected from waters of both the UK and Mexico. The two sequence types may possibly represent divergent sequences or sequences from previously undescribed species belonging to the genus Terschellingia. An obvious question is how some of these sequence types are so broadly distributed? The widespread distribution could have occurred relatively recently with the advent of intercontinental seafaring and contamination from ballast (rocks, sand and water) that has spread relatively non-motile organisms to a much greater degree than would be expected. Their existence in a range of ecological habitats with attendant environmental stressors such as changes in salinity and dessication during tidal cycles suggests a high degree of environmental tolerance in these specimens with different sequences. Mitochondrial genes such as the cytochrome oxidase I (COXI) could provide further information on gene flow patterns and cryptic level diversity within this taxon, but amplification of this gene in marine nematodes is extremely difficult and unreliable (Bhadury et al. 2006b). In addition, mitochondrial primers are generally used for barcoding studies (Hebert et al. 2003; Bucklin et al. 2007). Derycke et al. (2005, 2007) were successfully able to amplify COXI gene from two nematode species Pellioditis marina and Geomonhystera disjuncta and showed cryptic diversity within both taxa. In the present study we did not obtain reliable amplification of putative T. longicaudata specimens using the primers of Derycke et al. (2005). However, we were able to amplify COXI gene from two species of Sabatieria (unpublished) which may indicate that the primer sites were not conserved in T. longicaudata. Several issues relating to frequent failure of COXI gene amplification in marine nematodes have been detailed thoroughly by De Ley et al. (2005). In addition, the authors developed degenerate primers targeting the mitochondrial 16S and cytochrome b genes but amplification failed in T. longicaudata specimens. In future more robust mitochondrial primers need to be developed encompassing diverse marine nematode groups. The failure to amplify any of the mitochondrial genes in this study has prevented us from obtaining a better understanding of the possible presence of divergent or cryptic species complexes within the T. longicaudata specimens.
One of the important outcomes of this study has been the further evaluation of the Bhadury et al. (2007) method for extraction of DNA from archived nematode specimens. The authors were able to extract and amplify sequences of 345 bp from the partial 5′end of the 18S rRNA molecule for Merbok specimens and found that all the sequences were 100% identical to commonest T. longicaudata genotype. The Merbok sequences were also identical to the genotype A sequences (based on the 345 bp region), but differed by 6 base pairs from the genotype B sequences. When compared with the Merbok specimens, N Tubli and Ras al Barr sequences varied by 50 and 58 bp, respectively. The DNA extraction method applied in this study could be used in future to elucidate population genetic structure of archived meiofaunal specimens at the molecular level.
This study shows that careful evaluation might be needed for selection of markers for DNA barcoding studies since a highly conserved gene such as the 18S rRNA may also have some level of variation within marine nematode populations (as shown in this study). Therefore, careful consideration and evaluation involving analysis of multiple specimens from different geographic locations should be undertaken before a particular marker is selected as a standard for DNA barcoding. An approach based on more than one molecular marker, preferably nuclear and mitochondrial genes as suggested by Derycke et al. (2007) may prove to be more parsimonious in population genetics and future barcoding studies of marine nematodes.
It seems that traditional marine nematode taxonomy, based on the analysis of superficial morphological characters, may be insufficient if we are to fully understand the species-level biodiversity of this meiobenthic group. Developments in ultrastructure-based taxonomy that are used in terrestrial biodiversity (Bert et al. 2006; Giblin-Davis et al. 2006) should be explored for the identification of marine nematodes and associated diversity patterns. As discussed by Bickford et al. (2006) it is essential that cryptic species can be identified regarding analysis of biodiversity patterns, particularly with respect to conservation and habitat management. Identifications of nematodes based on morphology and probably ultra-structural features should therefore ideally be integrated with molecular approaches (DNA barcoding) during biodiversity surveys.
Punyasloke Bhadury acknowledges Plymouth Marine Laboratory (PML) for the provision of a Ph.D. Studentship. The authors would like to thank Hazel Needham, Sarah Dashfield and Andrea McEvoy for help with the taxonomic identification, and Paul Somerfield, Michaela Schratzberger and Rachel Jones for providing nematode specimens. The authors wish to thank the three anonymous reviewers for their help in improving this manuscript. This is a contribution to the PML Biodiversity and Sustainable Ecosystems project. The authors acknowledge the support by the MarBEF Network of Excellence ‘Marine Biodiversity and Ecosystem Functioning’ which is funded by the Sustainable Development, Global Change and Ecosystems Programme of the European Community’s Sixth Framework Programme (contract no. GOCE-CT-2003-505446).