Evolutionary history of Chaetognatha inferred from molecular and morphological data: a case study for body plan simplification
© Gasmi et al.; licensee BioMed Central Ltd. 2014
Received: 1 July 2014
Accepted: 30 October 2014
Published: 21 November 2014
Chaetognatha are a phylum of marine carnivorous animals which includes more than 130 extant species. The internal systematics of this group have been intensively debated since it was discovered in the 18th century. While they can be traced back to the earlier Cambrian, they are an extraordinarily homogeneous phylum at the morphological level - a fascinating characteristic that puzzled many a scientist who has tried to clarify their taxonomy. Recent studies which have attempted to reconstruct a phylogeny using molecular data have relied on single gene analyses and a somewhat restricted taxon sampling. Here, we present the first large scale phylogenetic study of Chaetognatha based on a combined analysis of nearly the complete ribosomal RNA (rRNA) genes. We use this analysis to infer the evolution of some morphological characters. This work includes 36 extant species, mainly obtained from Tara Oceans Expedition 2009/2012, that represent 16 genera and 6 of the 9 extant families.
Cladistic and phenetic analysis of morphological characters, geometric morphometrics and molecular small subunit (SSU rRNA) and large subunit (LSU rRNA) ribosomal genes phylogenies provided new insights into the relationships and the evolutionary history of Chaetognatha. We propose the following clade structure for the phylum: (((Sagittidae, Krohnittidae), Spadellidae), (Eukrohniidae, Heterokrohniidae)), with the Pterosagittidae included in the Sagittidae. The clade (Sagittidae, Krohnittidae) constitutes the monophyletic order of Aphragmophora. Molecular analyses showed that the Phragmophora are paraphyletic. The Ctenodontina/Flabellodontina and Syngonata/Chorismogonata hypotheses are invalidated on the basis of both morphological and molecular data. This new phylogeny also includes resurrected and modified genera within Sagittidae.
The distribution of some morphological characters traditionally used in systematics and for species diagnosis suggests that the diversity in Chaetognatha was produced through a process of mosaic evolution. Moreover, chaetognaths have mostly evolved by simplification of their body plan and their history shows numerous convergent events of losses and reversions. The main morphological novelty observed is the acquisition of a second pair of lateral fins in Sagittidae, which represents an adaptation to the holoplanktonic niche.
Chaetognaths are small predators of major importance in the marine ecosystem ,. They are abundant in every sea worldwide and can be traced back to the Cambrian radiation . Most of them are planktonic but a few are benthic. Chaetognaths are particularly renowned for their peculiar morphological and developmental features. These characters, as well as the affinities of the group within the metazoans, have been extensively debated by zoologists since the discovery of the phylum in the 18th century . So far the most recent phylogenetic analyses have also proved problematic for inferring their sister-group relationships within metazoans, which makes their positioning one of the most difficult issues in animal phylogeny . Numerous alternative phylogenetic hypotheses have been proposed over a long history of debate (for review ,). However, a recent hypothesis has emerged, based on morphological , and phylogenomic analyses -, where chaetognaths have been considered an early diverging member of Protostomia. The circumoral brain and the intraepithelial ventral cords have been recognized to be two of the key apomorphies of Protostomia ,. The nervous system in Chaetognatha is characterized by such a typical arrangement. However, even though the Chaetognatha partly share the Protostomia ground pattern, Perez et al.  concluded that their derived genome and morphology do not include any convincing synapomorphy that would suggest a sister-group relationship to another metazoan taxon.
Thus, based on a consensus between Tokioka, Casanova and Bieris hypotheses, the extant Chaetognatha are represented by three orders (Biphragmophora, Monophragmophora, Aphragmophora) and nine families (Heterokrohniidae, Eukrohniidae, Pterokrohniidae, Spadellidae, Krohnittellidae, Krohnittidae, Pterosagittidae, Sagittidae, Bathybelosidae).
The first molecular study of chaetognaths systematics was conducted with a short portion of the large subunit ribosomal RNA 28S (LSU rRNA) gene . These authors concluded that the LSU rRNA gene is duplicated in Chaetognatha, the division into Aphragmophora and Phragmophora is supported and several genera of the Sagittidae family described by Tokioka  and Bieri  are recovered. Papillon et al.  carried out a more extensive molecular study based on 26 sequences of the small subunit ribosomal RNA 18S (SSU rRNA) isolated from members of six extant families; they concluded that (1) similarly to LSU rRNA, a duplication of SSU rRNA gene occurred, suggesting that the whole ribosomal cluster is duplicated, (2) Tokiokas suborders Ctenodontina and Flabellodontina are not validated, (3) Casanovas hypotheses Syngonata/Chorismogonata and Monophragmophora/Biphragmophora are rejected, (4) the families Krohnittidae and Pterosagittidae are not supported, (5) three monophyletic groups are identified: Sagittidae/Krohnittidae, Spadellidae/Pterosagittidae and Eukrohniidae/Heterokrohniidae, (6) the order Aphragmophora without Pterosagitta draco is monophyletic. Since then, no molecular study has been made to further explore the systematics of this phylum. Finally, a recent barcoding analysis was highly successful at discriminating between the described species . It notably revealed little geographical structure and showed that Eukrohnia bathypelagica and Eukrohnia hamata are probably young sister-species.
Thus, even after one century of heavy debates, it has not been possible to establish a stable and reliable hypothesis on the evolutionary history of Chaetognatha. In the present work, we have conducted an extensive molecular analysis based on LSU and SSU rRNA duplicated genes. We combined, for the first time, the molecular results with a morphological classification and geometric morphometrics. In the light of our results, we present a revised phylogeny and discuss the morphology-based character systems that have traditionally been used to classify this enigmatic phylum.
Alignments and erroneous sequences
First, we identified erroneous sequences by constructing test trees from LSU and SSU rRNA genes and by then identifying which sequences from public databases came out in suspicious positions when compared with new sequences obtain in the present study. This approach revealed three chaetognath sequences that are erroneous due to contaminant or bad species diagnosis. These sequences that must be excluded from any future phylogenetic analyses were:
The SSU rRNA of Krohnitta pacifica (class I DQ351879 and class II DQ351891) from  which most likely is a Sagittidae contaminant close to Parasagitta setosa when compared with three new sequences belonging to Krohnitta subtilis.
The SSU rRNA of Sagitta sp (class I AY922316) from  which belongs to the Eukrohniidae family, close to Eukrohnia hamata and Eukrohnia bathypelagica.
We also characterised new sequences for three different specimens of Pterosagitta draco that were used instead of the erroneous SSU rRNA sequences (class I DQ351885 and class II DQ351898) from .
After excluding these contaminants, we set up five alignments available upon request (dataset 1, LSU rRNA paralogous genes; dataset 2, SSU rRNA paralogous genes; dataset 3, SSU rRNA paralogy class I; dataset 4, SSU rRNA paralogy class II; dataset 5, concatenated alignment of SSU and LSU rRNA genes).
Molecular phylogenetic analyses
Whatever the dataset considered, the Bayesian and approximate likelihood ratio (aLRT) trees obtained were almost identical to the maximum likelihood (ML) bootstrapped tree. Thus their statistical values were reported on the ML topology. However, in the case of the concatenated analysis, because only the aLRT and the Bayesian reconstructions showed a fully congruent topology, statistical values obtained from the three methods were reported on the Bayesian topology. The validity of the Krohnittidae and Heterokrohniidae families will not be discussed because both are only represented here by a single species.
Small subunit rRNA. Three data sets of SSU rRNA sequences were analysed. First, phylogenetic reconstructions were conducted with dataset 2 comprising all available SSU rRNA sequences from both paralogy classes (dataset 2?=?SSU rRNA Class I and II, 138 sequences from 33 species: 80 class I and 58 class II sequences). One of the recurrent problems when analysing the molecular data in Chaetognatha is the lack of relevant outgroups among bilaterians. So, this first analysis offered to root one paralogy class on the other, highlighting the branching order of five represented families with good statistical values and revealing the following sequence from the last common ancestor to the most derived family (Additional file 1): Eukrohniidae?+?Heterokrohniidae, Spadellidae, Krohnittidae and Sagittidae, the latter including Pterosagittidae. Similarly to the LSU rRNA analysis, Phragmophora (Heterokrohniidae?+?Eukrohniidae?+?Spadellidae) were paraphyletic while Aphragmophora (Pterosagittidae?+?Sagittidae?+?Krohnittidae) received high support in most methods applied (class I 80/1/0.98, class II 19/1/0.85).
Concatenated analysis. As it is done for large coding genes that contain considerable phylogenetic signal, rRNA genes were concatenated to increase the accuracy of the phylogenetic reconstructions (Dataset 5?=?SSU rRNA Class I, SSU rRNA Class II and LSU rRNA Class I concatenated). The total length of the concatenation based on the two paralogous SSU rRNA class I and class II (1670 and 1087 pb respectively) and LSU rRNA class I (366 pb) was 3123bp. We conserved as many taxa as possible in the concatenated approach, e.g., all species showing at least two genes and two specimens. Eukrohnia bathypelagica and Paraspadella gotoi displayed SSU rRNA sequences only for class I and Eukrohnia fowleri only for class II. The model of evolution was estimated on the full concatenated data set without any partitioning of the data. The selected model was the GTR?+???+?I with a lnL of ?15473.70 with the dataset comprising 55 sequences and lnL of ?13443.12 with the dataset comprising 53 sequences.
Cladistic and phenetic analysis by morphology
We conducted an alternative phenetic approach to include 9 quantitative characters and to estimate the overall similarities between taxa. The phenetic analysis of morphological variations was rooted on midpoint (Figure6B). As in the cladistic analysis, the phenetic clustering resulted in a good estimation of the phylogenetic relationships at the order rank with the monophyly of Phragmophora (approximately unbiased test value = 75) and Aphragmophora (78) and for assigning a species to the correct family. The phenetic approach was more congruent to molecular trees than the cladistics. It yielded the sister-group relationships between X. sorbei and Eukrohniidae (75), congruent relationships within Spadellidae with basal P. gotoi, Spadella monophyletic and sister group relationships between S. ledoyeri and S. valsalinae. P. draco was also found nested in a group comprising S. bipunctata and two Parasagitta species (87). Apart from the case of Heterokrohniidae and Krohnittidae, each having a single representative in the analysis, the remaining families, Spadellidae and Eukrohniidae, were found to be natural groups with high support (98 and 82 respectively). Sagittidae were not recovered not only because of the inclusion of P. draco but also that of Krohnitta subtilis as sister to Serratosagitta species. However, the low support for this grouping (63) raises questions over this mongrel assemblage. All Sagittidae genera except Parasagitta were monophyletic and most nodes picturing their relationships received good support. Similar to the molecular and morphological cladistic analyses, the morphological phenogram yielded a close relationship between Ferosagitta and Aidanosagitta. However, the three main Sagittidae lineages highlighted in the molecular trees were not recovered.
A phenogram was constructed on the basis of body shape similarities, using the Riemannian shape distance ?, computed on all species pairs (see Additional file 3 for comparison between different families). The purpose was not to give more insights about chaetognaths relationships but rather to test whether the two pairs of lateral fins observed in Sagittidae come from the division of a single fin or from the neoformation of the anterior pair. The phenogram of body shape similarities was more congruent with the molecular data when it was constructed using the anterior end of the posterior lateral fin as homologous to the anterior end of the unique lateral fin of species having only one pair (PH1; agglomerative coefficient = 0.84; p = 0.0098***, Additional file 4A) than when it had been constructed with the anterior end of the anterior lateral fin as homologous to the anterior end of the unique lateral fin of species having only one pair (PH2; agglomerative coefficient = 0.89; p = 0.4537, Additional file 4B).
A reassessment of chaetognaths relationships
The species studied in the present report belong to the Aphragmophora (Sagittidae, Krohnittidae and Pterosagittidae) and Phragmophora which have been divided in Biphragmophora (Heterokrohniidae) and Monophragmophora (Spadellidae and Eukrohniidae) . We were therefore able to discuss the evolutionary history of six of the nine traditional families of the phylum (according to the views of Casanova  and Bieri ). The monophyly of the Krohnittidae cannot be debated because only one of the three known species, Krohnitta subtilis, is included in the present analysis. The same applies to Heterokrohniidae represented here only by Xenokrohnia sorbei. The Eukrohniidae and Spadellidae families have been confirmed, but not the Sagittidae and Pterosagittidae (a family comprising only one species, Pterosagitta draco). Indeed, the Sagittidae sensu stricto is a paraphyletic assemblage from which P. draco derives.
We observed poorly resolved basal nodes in the LSU rRNA trees and to a lesser extent in the SSU rRNA trees as well as a lack of relationships accuracy within Sagittidae. However, when molecular analyses were based on the concatenation of the two paralogous SSU rRNA class I and class II and LSU rRNA class I the resolution for deep branching nodes and the accuracy of relationships within Sagittidae were improved (see Figure5). This indicates that when the amount of molecular data increases the phylogenetic signal does too. Thus, the low resolution of the single rRNA gene tree reconstructions is not due to a recent acceleration of diversification within Chaetognatha, as previously proposed ,, but rather to the short length of the aligned DNA segments.
Morphology and molecules produced some mixed phylogenetic results. This incongruence can be resolved by allowing several convergent losses and/or reversions of morphological characters. The lack of a character cannot be rigorously coded in cladistic due to the lack of primary homology hypothesis especially in a clade with few fossils and inappropriate outgroup. One consequence is that the amount of convergent losses can be underestimated. In such a situation, a phenetic approach should give a better estimation of the evolutionary history. In Chaetognatha, quantitative and qualitative morphological variations based on the degree of overall similarities have been more congruent with molecular topologies than cladistic did, especially concerning the relationships within Spadellidae and the position of X. sorbei as sister to Eukrohniidae. The position of K. subtilis, the only Krohnittidae relative, is also unstable according to the method used: either at the basis of all Sagittidae (cladistics on morphological data and molecular data) or sister group to Pseudosagitta within Sagittidae (phenetics on morphological data). Such an unstable position could be explained by an independent evolution of Krohnittidae during a long period from an early stage of Aphragmophora . Finally, most inconsistencies observed between molecular and morphological approaches concern the Sagittidae relationships, which shows that there is no linear relationship between the degree of morphological divergence and the time of divergence within this family.
Phragmophora Aphragmophora split and the Ctenodontina/Flabellodontina hypothesis
The clade Sagittidae?+?P. draco as sister group to Krohnittidae is highly supported by our molecular and morphological cladistic analyses and revives the Aphragmophora, a clade invalidated by Papillon et al.. However the conclusion these authors made was mainly based on the positioning of P. draco within Spadellidae, which was in agreement with morphometry and body appearance . However, the present molecular results demonstrate that the previous P. draco sequence was likely to be a contaminant. Moreover, our morphological analyses do not support the Dallot and Ibanez  conclusion, highlighting the convergence of some morphological traits between Spadellidae and P. draco. It is interesting to note that on the basis of posterior lateral fins restricted to the tail observed in Demisagitta demipenna (firstly described as Aidanosagitta demipenna), Bieri  pointed out a possible relationships between P. draco and some species belonging to Sagittidae. He noted that if D. demipenna were to lose the anterior fin and develop a pair of large floating bristles, the species would be included into the Pterosagitta genus. However, according to the first description  and a subsequent revision , even the position of the posterior lateral fins of D. demipenna is unique in Sagittidae and similar to the one-fin species P. draco, other characters conform to those of Aidanosagitta. Finally, and more importantly, the inclusion of P. draco within Sagittidae is corroborated by a recent report on the organisation of the chaetognath nervous system . Indeed, this study showed that the RFamidergic pattern of P. draco is similar to that of several Sagittidae species when compared to several Spadellidae species.
Traditionally, authors who proposed internal systematics in Chaetognatha ,,,,- identified two major groups; mainly on the basis of the presence or absence of transverse muscles - the phragms. Throughout this debate on chaetognath evolutionary trends, most authors agreed to consider the presence of phragms as a plesiomorphic state , but with slightly different hypotheses. Spadellidae were believed to have given rise to the Eukrohniidae and Heterokrohniidae according to Tokioka  whereas Casanova  considered the Heterokrohniidae as the chaetognaths that retain the highest number of plesiomorphic character states. Only Salvini-Plawen  suggested a radically different scenario which contradicted the primitiveness of phragms and identified Pterosagittidae as the sister group to all remaining families. Our results favour the ancestrality of Phragmophora and do not support Salvini-Plawens hypothesis since P. draco appears to be a highly specialised and homoplasic member of Sagittidae - as shown by various features such as the loss of the anterior lateral fins, the position of posterior lateral fins, the type of corona ciliata and a high trunk/tail length ratio. So far, the exact functional significance of phragms is unknown but their presence is correlated with a benthic lifestyle ,,,. Indeed, the creeping and predatory activity on the sea bed requires more complicated movement than the pelagic niche does. One exception concerns the pelagic Eukrohniidae which exhibit phragms in the trunk, however these are vestigial structures believed to be functionless .
According to the rooted topology obtained on the basis of both paralogy classes of SSU rRNA genes (see Additional file 1), our analyses uphold the monophyly of the Aphragmophora but contradict that of the Phragmophora. The Phragmophora appear paraphyletic - a typical situation when a clade is defined on the basis of a plesiomorphic character state. However, the morphological cladistic approach shows that the Aphragmophora are only defined by the lack of phragm or the scarce development of glandular structures on the body surface, which leaves us in an unsatisfactory situation. Based on the current knowledge of gross morphology, histology, cytology and neuroarchitecture of chaetognaths, it is simply impossible to describe any noteworthy apomorphic feature of Aphragmophora. Finally, the Aphragmophora have been divided into two suborders ,: Flabellodontina (Krohnittidae) and Ctenodontina (Sagittidae?+?Pterosagittidae). Our results show that this supplementary subdivision is unnecessary and undermines the hypothesis based on the structure of the cephalic armature that Ctenodontina could be closer to Phragmophora than to Krohnittidae .
Validity of Biphragmophora and Monophragmophora
Bieris nomenclature and Sagittidae relationships
Despite the complexity of the distribution of some morphological characters, which poses problems when assessing the relationships within Sagittidae, most of the new genera proposed by Bieri  were supported by the molecular trees. Our results unambiguously confirm the monophyly of Serratosagitta, Pseudosagitta, Flaccisagitta and Ferosagitta, and to a lesser extent the validity of large and heterogeneous assemblages such as Aidanosagitta and Parasagitta. The relationships between species belonging to Parasagitta cannot be resolved on the basis of morphological analyses and received low support in the molecular trees. The morphology of this genus remains one of the most heterogeneous on the basis of several diagnosis characters which are prone to homoplasy (Figure7): the structure and position of seminal vesicles, the presence/absence of intestinal diverticula, the presence/absence of intestinal vacuolated cells, the presence/absence of rayless zone in lateral fins and the structure of the corona ciliata if considering the inclusion of Mesosagitta minima. The status of Solidosagitta is still pending because only one species has been studied using SSU rRNA paralogous genes. However, our analyses based on morphological data and LSU rRNA sequences include two species and favour the validity of this latter genus.
Our molecular results divide the Sagittidae family into three major lineages (Figure7): Serratosagitta?+?Solidosagitta, Sagitta?+?Pterosagitta?+?Parasagitta (including M. minima) and Flaccisagitta?+?Aidanosagitta?+?Ferosagitta?+?Mesosagitta decipiens. Species with vacuolated intestine (character #10) are distributed in these three lineages. This supports the opinion of Dallot  who considered the vacuolated species plesiomorphic on the basis of their general morphology and the structure of their seminal vesicles. This also strongly suggests that the ability to develop large intestinal vacuolated cells has been lost in numerous extant Sagittidae species. The grouping of M. decipiens, Aidanosagitta, Ferosagitta and Flaccisagitta receives high support in molecular phylogenies. Morphological analyses only support close relationships between Ferosagitta and Aidanosagitta. Moreover, some morphological characters are congruent with the association between Aidanosagitta and M. decipiens: the corona ciliata begins below eye level (type B ) and intestinal diverticula are present in Mesosagitta (Decipisagitta), Aidanosagitta and Ferosagitta. However, these characters isolate Flaccisagitta from the rest of the group: the corona ciliata is short and confined to the head, starting just behind the brain and stretching to the neck (type D ) and intestinal diverticula are absent. Kinship between Sagitta bipunctata and Parasagitta species are highly supported by rRNA data and has been previously proposed by several authors. According to Tokioka, , these chaetognaths display a similar extended corona ciliata (type C). Moreover, Furnestin  and Dallot  also suggested such affinities on the basis of the structure and position of lateral fins and number of teeth and hooks. An important incongruence between molecular and morphological analyses is the sister-group relationships between Flaccisagitta and Pseudosagitta, a result yielded by morphology but invalidated by all molecular trees. This group has previously been proposed by Tokioka but not all authors agree to bring these species in a same clade. Several authors ,- suggested that lyra-gazellae-maxima was undeniably a coherent group gathered in Pseudosagitta while Flaccisagitta hexaptera constituted the sister species to F. enflata. Finally, the morphological similarities between Flaccisagitta and Pseudosagitta could be linked either to a specialised form highly adapted to the oceanic plankton (thin primary muscles, flaccid body, not wholly rayed lateral fins with gelatinous masses) or should be considered as plesiomorphic states among Sagittidae (corona ciliata type D and seminal vesicles type C).
When taken together, these remarks emphasize the need to better define the morphological and anatomical boundaries between the traditional genera of Sagittidae by re-evaluating ancestral states and homologies of important traditional diagnostic characters at the histo- and cytological levels (for instance the nervous and muscular systems, the corona ciliata, the seminal vesicles and the fins).
Taxonomic notes on the genera Parasagitta/Occulosagitta and Mesosagitta/Decipisagitta
In his attempt to improve the Sagittidae systematic, Bieri  also noticed the heterogeneity of several Sagittidae genera and modified his own classification by creating six new genera. Two species included in our analyses are concerned by these modifications: Parasagitta megalophthalma and Mesosagitta decipiens which were respectively renamed by Bieri as Occulosagitta megalophthalma and Decipisagitta decipiens. First, Mesosagitta as a natural group is contradicted. Indeed, in our molecular analyses, Mesosagitta minima always branches without any ambiguity within the Parasagitta genus while M. decipiens exhibits close relationships with Flaccisagitta, Aidanosagitta and Ferosagitta. Second P. megalophthalma always shows a close relationship with Parasagitta elegans whatever the molecular tree considered. Thus we propose (i) to invalidate the Occulosagitta genus, (ii) to rename Mesosagitta minima as Parasagitta minima and (iii) to gather the remaining Mesosagitta species (M. decipiens, M. neodecipiens, M. sibogae) into the new genus Decipisagitta.
Phylogenetic consensus from molecules and implication for chaetognaths evolutionary trends
Paleontological evidences have demonstrated the existence of chaetognaths not only in the middle Cambrian Burgess Shale biota  but also in the earlier Cambrian Chengjiang biota  with morphological features almost identical to extant species. The discovery of new deep species  leads to the conclusion that Heterokrohniidae (Biphragmophora) is the family that presents the highest number of plesiomorphic characters (i.e., the most primitive group sensu Casanova). However, our analysis raises the question of whether the lack of phragms in the tail of Spadellidae and Eukrohniidae is derived and due to convergence or represents a homologous and plesiomorphic character. The answer has far reaching consequences for our understanding of evolutionary pathways in chaetognaths. For instance, since phragms are considered important for creeping forms but are unnecessary for species that proceed by movements in the water column, this question is related to whether the stem chaetognaths were hyperbenthic or holoplanktonic species. The molecular and morphological phylogenetic results we obtained rather suggest that the Eukrohniidae and Spadellidae (Monophragmophora in Casanovas hypothesis) exhibit the most primitive state, a scenario that needs only two evolutionary steps (one acquisition of phragms in the tail of Heterokrohniidae and one loss in the trunk of Sagittidae) against three steps when considering two parallel losses in the tail of Eukrohniidae and Spadellidae respectively, followed by one loss in the trunk of Sagittidae. However, we shall see that several arguments are in favour of the latter scenario. First, morphology and body ratios of specimens found in Chengjiang biota suggest that the chaetognaths from lower Cambrian were planktonic with ecological preferences for hyperbenthic niches close to the sea bottom . Among extant genera, those showing the closest ecological features are the hyperbenthic Heterokrohniidae. Interestingly, their ecology is still observed in one species of Eukrohniidae, Eukrohnia calliops,. Second, phragms have been identified in trunk and tail of specimens from Cambrian Burgess Shale biota . Finally, the stable environment of deep oceanic waters has likely delayed the body plan evolution and could explain the conservation of some ancestral morphology in the extant deep benthoplanktonic Heterokrohniidae. All summed up, it is reasonable to assume that an arrangement of transverse muscles in trunk and tail should be regarded as the most primitive state in chaetognaths. In hypothesizing such a complex Heterokrohniidae-like ancestor, one must postulate the loss of many structures and a body plan simplification during the evolutionary history of Chaetognatha (Figure7).
In our scenario, an important split yielded two clades with different ecological niches, the strictly benthic Spadellidae and the holoplanktonic Aphragmophora lineages (Figure7). This hypothesis contradicts the ancestrality of Spadellidae . According to the comparative studies of the muscles in Chaetognatha, Spadellidae are highly derived and underwent important modifications of their muscular apparatus . The structure of their primary muscles which lack B fibres is characteristic of benthic species and is derived from AB fibres typology. Combined with our results, there is therefore strong evidence that the Spadellidae ancestor was planktonic, partly linked to the sea bed, and adapted secondarily to a strict benthic lifestyle. The highly specialized status of this family is shown by its high number of synapomorphies (characters #2, #5, #6 and #8; Additional file 2).
The Aphragmophora diversification does not display any morphological novelty and highlights a new case of body plan simplification with the loss of phragms in the trunk (Figure7). The last Aphragmophora ancestor divided into two lineages, giving rise to the current Krohnittidae and Sagittidae families. The Krohnittidae family retained some ancestral traits such as one pair of lateral fins on the trunk and tail, with the anterior end and posterior rear end at equal distance from the caudal septum (character #2) but also developed numerous autapomorphic features. They exhibit abruptly curved hooks (character #11) and anterior teeth arranged in a fan shape (character #22) and they lack posterior teeth (character #20). The peculiarity found in teeth and hooks of Krohnitta species reveals a high level of specialisation  and points to an independent evolution of Krohnittidae that started during the early stages of Aphragmophora cladogenesis as demonstrated by our results. Interestingly, the arrangement and shape of the anterior teeth of Sagitta nairi, a Sagittidae recently described  are similar to those of the genus Krohnitta suggesting a possible convergence between these unrelated species. The second Aphragmophora family, the Sagittidae, is defined by two pairs of lateral fins. Caecosagitta and Pseudosagitta can be recognized as two early off-shoots of Sagittidae. Dallot and Ibanez  have already proposed the isolation of Pseudosagitta lyra and suggested the possibility that its membership to the Sagittidae is dubious. Our results do not support such exclusion.
An important question considering body plan variation in Chaetognatha is concerned with fin evolution. More precisely, did the anterior and posterior lateral fins of Sagittidae originate from the division of the single large fin observed in Heterokrohniidae and Eukrohniidae? Casanova and Moreau  noted some similarities between the posterior lateral fin of species belonging to Pseudosagitta genus and the unique lateral fin of those belonging to Eukrohniidae, the posterior extremities of which are only slightly apart from the bodywall. The posterior and anterior lateral fins of Pseudosagitta are connected by a tegumentary bridge which reinforces the idea that the two lateral fins would have formed after the incomplete division of a unique large one. However, our morphometric analysis shows that the anterior lateral fin of two-fin species does not result from the division of a unique lateral fin but from a neo-formation after a backward movement of the anterior end of the unique fin. In other words, the posterior lateral fin of Sagittidae is homologous to the unique lateral fin of the other families. This evolutionary step constitutes a rare case of increase in body plan complexity at the anatomical level in Chaetognatha. While the possession of two pairs of lateral fins was recognized as a good synapomorphy for Sagittidae after Tokiokas classification, the inclusion of P. draco which exhibits only one posterior pair raises doubts about its validity. This means that the loss of the anterior fin did not constitute an evolutionary dead end for Sagittidae making the presence of one pair of fin a homoplasy.
As previously mentioned ,, the distribution of some morphological characters in Chaetognatha cannot be related to the phylogeny nor to the ecology suggesting a differential evolution of separate chaetognath organs (mosaic evolution). The prevalence of mosaic evolution can be demonstrated through the examination of character associations in extant chaetognaths (Figure7). Different traits and outcomes are favoured by natural selection in different species and these evolutionary pathways might be responsible for the non congruence between some of the cladistic and phenetic analyses because when mosaic selection occurs the primitive or derived nature of character states cannot be deduced on the basis of their correlation with other character states which are believed to be primitive or derived . Moreover, a common trend correlated with mosaic evolution is the prominence of homoplasy powered by common selective pressures as shown in many plants  and animals -.
Such a combination of mosaic evolution and lack of fossil records can lead to persistent problems in interpreting relationships through morphological cladistic analysis .
Ecologically-induced convergence in holoplanktonic chaetognaths
It is likely that all Aphragmophora and Eukrohniidae lineages are morphologically highly similar because of their holoplanktonic lifestyle in the pelagos , (Figure7). For instance, in spite of their separate evolutionary history, Sagittidae, Krohnittidae and Eukrohniidae share features such as the common tendency towards a body with large surface to volume ratio, the trunk elongation, the reduction of epidermal glandular structures and the reduction/loss of phragms. Because these clades do not constitute a natural grouping, it is obviously a case of convergent evolution. Functional similarity is also present at the cytological level since all these holoplanktonic species exhibit the highest proportion of B fibres in their primary muscles and a heterosarcomeric organisation of their secondary muscles . Among Sagittidae, the morphology of P. draco is more puzzling. There are numerous homoplasic reversions in this species since it exhibits a high trunk/tail length ratio and one small pair of lateral fins being restricted to the tail, a typical set of Spadellidae morphological features. This epipelagic species compensates the decrease of its surface/volume body ratio by developing a foamy epidermal collarette around most of the body combined with floating bristles.
Buoyancy ability represents an interesting area of investigation in holoplanktonic chaetognaths -. To decrease their specific gravity, Eukrohniidae show a hyper development of the unique pair of lateral fins and vestigial phragms, which are likely to be not functional and are confined to the most anterior part of the trunk . A similar trend is observed in Krohnittidae with the development of large lateral fins and phragms that have totally disappeared. The Sagittidae adapted in a different way by creating a new pair of lateral fins. This acquisition constitutes an important and unique event to adapt to holoplanktonic lifestyle in Chaetognatha and could contribute to explain the successful current biodiversity of Sagittidae. It has been recently shown that the diversification of the Euthecosomata, which are important holoplanktonic molluscs, occurred in the context of an important turn-over in the marine planktonic community due to severe environmental changes that started from the Late Palaeocene . Moreover, these morphological innovations correlated with climatic changes and species turn-over were largely shaped by shell buoyancy adaptation. One could postulate that a broad and recent diversification of Sagittidae occurred in the same evolutionary background. However, the paucity of fossils for these soft bodied invertebrates did not allow an efficient calibration of the divergence times of Chaetognatha lineages.
Molecular analyses have highlighted the homoplasy of several traditional characters and the influence of lifestyle on morphology, particularly for chaetognaths that adapted to a pelagic environment. We also propose that Chaetognatha evolved mostly through simplification of a pre-existing body plan, rather than through an increase in complexity. This constitutes a shift of paradigm in the traditional understanding of the groups evolution and prompts a re-appraisal of previous hypotheses concerning the morphological characters polarity and history. For example, it is reasonable to think that the loss of phragms and teeth could have occurred independently in different branches during chaetognath evolution. If the anterior and posterior rows of teeth are considered as homologous structures, the loss of one of these rows may be described as an event of parallel evolution. Another important traditional diagnostic character, the trunk/tail length ratio, can also be considered subject to homoplasy by reversion due to Bauplan limits (Pterosagitta draco versus Spadellidae) and convergence caused by an evolution in similar ecosystems (Sagittidae/Krohnittidae versus Eukrohniidae). Even the main Sagittidae synapomorphy represented by two pairs of lateral fins is homoplasic by reversion. Because of these numerous losses and homoplasic events, traditional morpho-anatomical traits may prove unhelpful with deciphering chaetognath relationships and morphological evolution with any certainty. Such a conclusion stresses the need for more data from molecular markers as well as from histo- and cytoarchitecture of the muscular apparatus  and the neurosensorial system ,. Because of the scarcity of novelties at the anatomical level, it becomes needed to explore their body plan variations from the tissue level to the cell level. Considering the pivotal phylogenetic position of Chaetognatha within bilaterians, it is of primary importance to reconstruct their ground pattern. Future studies need to focus on a set of new characters based on a broad range of taxa including specimens belonging to meso-bathypelagic and deep benthoplanktonic genera. The use of an expanded taxonomic dataset combined with appropriate observation (i.e., with transmission electron microscopy, immunohistochemistry combined with confocal laser scanning microscopy and Next Generation Sequencing) will be crucial in improving the understanding of Chaetognathas diversification processes.
Collection, species diagnosis and taxonomic sampling
Details of the sequences and GenBank accession numbers obtained from this study
SSU rRNA I
SSU rRNA II
LSU rRNA I
LSU rRNA II
DIVA3 St ME 791/540.1
TARA St 64
TARA St 76
TARA St 132
Florida/CRER 2St 91
Florida /CRER 2St 42
Tulear/ M. Pagano
Florida/CRER 2St 42
France/ Y. Perez
TARA St 34
TARA St 34
TARA St 65
MSN 14/1St 1159
TARA St 64
TARA St 16
France/ Y. Perez
TARA St 66
TARA St 16
TARA St 79
TARA St 65
TARA St 66
TARA St 23
TARA St 15
TARA St 86
TARA St 86
MSN 14/1St 1159
TARA St 52
TARA St 64
Florida CRER2 St 42
TARA St 18
TARA St 15
TARA St 34
TARA St 64
Aidanosagitta cf. oceania
Aidanosagitta cf. septata
TARA St 58
TARA St 109
TARA St 130
TARA St 130
TARA St 50
TARA St 32
TARA St 32
TARA St 66
TARA St 64
TARA St 86
TARA St 85
TARA St 100
TARA St 125
TARA St 125
Ferosagitta cf. tokiokai
TARA St 100
TARA St 17
TARA St 15
TARA St 85
Details of the sequences and GenBank accession numbers obtained from previous studies and used in the present analysis
SSU rRNA I
SSU rRNA II
LSU rRNA I
LSU rRNA II
DNA extractions, PCR amplifications and sequencing
All specimens were placed in 80% ethanol for preservation. The genomic DNA was then extracted using the DNAeasy kit (Qiagen, Valencia, CA) from pieces or entire individuals dried on filter paper and devoid of alimentary bolus to prevent contamination.
Because the whole ribosomal cluster in Chaetognatha is duplicated , two sets of specific primers for each paralogous SSU rRNA gene were used to amplify sequences of approximately 1800bp (class I: 18SCI5? TTGATGAAACTCTGGATAACTC and 18SCI3? GGACCTCTCTACATCGTTCG) and 1200bp (class II: 18SCII5? TCGTCGGGGTCTCATCC and 18SCII3? AGATACCTCGCAAAATCG). As we concentrated our efforts on the class I of LSU rRNA gene, which is the most represented class in public databases, only one set of primers previously described in  was used to amplify a fragment of approximately 500 pb: 28S5? AAAGGATCCGATAGYSRACAAGTACCG and 28S3? CCCAAGCTTGGTCCGTGTTTCAAGAC. Most of the sequences obtained with this couple of primers belonged to the class I but we also amplified LSU rRNA class II genes for several species.
PCRs were performed according to  in 50?L volumes with the following reagents: 1 PCR buffer (Taq PCR core kit, Qiagen), 0.2mM of each dNTPs mix, 0.5mM of each primer, 2 to 4?l (depending on DNA concentration) of extracted genomic DNA, and 1U of Taq polymerase. The PCR cycling parameters for amplification of LSU rRNA were: 95C, 3min, then 35cycles of 95C for 1min, 50C for 1min, 72C for 2min.
For SSU rRNA, we used the following PCR program: 2min at 92C; 5cycles of 92C for 30s, 48C for 45s, 48C to 72C for 80s (ramp rate of 0.3C/s) and 72C for 90s; 30cycles of 92C for 45s, 48C for 45s, 72C for 90s; and a final extension time of 72C for 7min.
After amplification, all PCR fragments were purified with Wizard SV Gel and PCR Clean-Up System (Promega, Madison, WI), cloned into pGemT-easy vector (Promega, Madison, WI) and sequenced in both directions using the T7 and SP6 primers with a ABI 96-capillary 3730XL sequencer at Eurofins genomics. 145 sequences have been deposited in GenBank under the following accession numbers:
SSU rRNA class I: KM519789 - KM519853
SSU rRNA class II: KM519854 - KM519901
LSU rRNA class I: KM519902 - KM519931
LSU rRNA class II: KM519932 - KM519933.
Molecular phylogenetic analysis
Five data sets were used for molecular analyses: dataset 1?=?LSU rRNA Class I and II, dataset 2?=?SSU rRNA Class I and II, dataset 3?=?SSU rRNA Class I, dataset 4?=?SSU rRNA Class II and dataset 5?=?SSU rRNA Class I, SSU rRNA II and LSU rRNA Class I concatenated sequences. The sequence alignments were established using CLUSTALX  and Muscle  implemented in Mega 5 and then further improved manually. The MODELTEST v3.0b4 program  was used to identify the best model of DNA evolution for each of our dataset based on maximum likelihood (ML) and using Bayesian information criterion (BIC). We used Aliscore , to test the impact of highly heterogeneous sites that could negatively affect the phylogenetic reconstruction. We used the following parameters -N and -N r w4 to remove heterogeneity sites. A Maximum Likelihood tree was estimated using the Nearest-Neighbour-interchange (NNI) option with Mega 5. A random starting tree was generated using the Neighbour-Joining method with the partial deletion option selected (75% site coverage cut-off). Topological robustness was investigated using 1000 non-parametric bootstrap replicates. Branches with bootstrap values higher than 70% were considered well supported . We also performed Bayesian phylogenetic analyses using MrBayes 3.0b4 . Each analysis consisted of 2.107 generations with a random starting tree, default priors, the same set of branch lengths for each partition, and four Markov chains (with default heating values) sampled every 1000 generations. Adequate burn-in was determined by examining a plot of the likelihood scores of the heated chains for convergence on stationarity as well as the effective sample size (ESS) of values in Tracer 1.5 .
To test the impact of potential noisy sites we computed maximum likelihood phylogenetic analyses using PHYML aBayes 3.0.1 beta programme , on the LSU rRNA data set. We calculated two non-parametric branch supports (Bootstrap and SH-aLRT) and two parametric branch supports (aBayes and approximative likelihood ratio test, aLRT) as developed in ,. We used bootstrap (bv) and aLRT (aLRTv) values and posterior probabilities (pp) to establish a criterion of quality. If bv was low but the other two were high then we considered a potential false negative support; if bv was high but the other two were low then we considered a potential false positive support.
We sometimes included sequences that were highly divergent, for instance from Aidanosagitta crassa, Eukrohnia fowleri, Serratosagitta tasmanica and, less frequently, from Mesosagitta decipiens because our primary goal was to accommodate the widest taxonomic and molecular dataset possible. This will provide a good foundation for future studies on chaetognaths evolution, but it may negatively impact our phylogenetic reconstruction. We removed some of these sequences in the concatenated dataset 5 to test whether they could produce artefacts in phylogenetic reconstructions.
The data set considered here is constituted of the 34 species used in the molecular analysis. The following 32 characters were chosen on the basis of their traditional importance as key characters and their use in species diagnosis. These variables are a mixture of different types (see the full list below): 23 qualitative (binary and polytomic) and 9 quantitative (e.g., lengths). These data are coming from a compilation of original descriptions and reviews ,,,,-. Characters coding is presented in Additional file 2.
Qualitative characters (n = 23)
C1- Body type: Flaccid =1; Rigid =2
C2- Number and type of lateral fins: one long pair of lateral fins extended on the tail as well as on the most part of the trunk =1; one short pair on the trunk and tail, the anterior end at the level of the caudal septum =2; one short pair on the trunk and tail, with the anterior and the posterior ends at equal distance from the caudal septum =3; Two pairs of lateral fins =4
C3- Tegumentary bridge connecting the anterior and posterior lateral fins: absent?=?1; present =2
C4- Phragms (transverse muscles): absent?=?1; present in the trunk only =2; present in the trunk and tail =3
C5- Type of phragms: supercontraction =1; normal contraction =2
C6- Type of longitudinal muscles: only B fibre =1; A and B fibres =2
C7- Type of secondary muscles: heterosarcomeric secondary muscles (He S) =1; homosarcomeric secondary muscles (Ho S) =2
C8- Organisation of RFamide-like neurons: Type A (absence of D6 and X posterior neurons, absence of caudal loop) =1; Type B (presence of D6 and X posterior neurons with caudal loop) =2
C9- Intestinal diverticula: absent?=?1; present =2
C10- Vacuolated intestinal cells: absent?=?1; present =2
C11- Type of hooks: gently curved =1; gently curved and serrated =2; abruptly curved =3
C12- Type of seminal vesicles: elongated with a lateral opening =1; elongated and an anterior protruding part usually roundish =2; roundish or slightly oval with a lateral opening =3; elongated with an anterior opening =4; presence of small indentations =5; oval with bulb-like shape =6
C13- Position of seminal vesicles (in respect to lateral and tail fins): touching neither lateral fins nor tail fin but closer to lateral fins =1; touching neither lateral fins nor tail fin but closer to tail fin =2; touching, or close to, lateral fins and well separated from tail fin =3; touching, or close to, tail fin and well separated from lateral fins =4; touching both lateral fins and tail fin =5
C14- Ocular type: inverted =1; everted =2
C15- Pigmented cell in the eye: absent?=?1; present =2
C16- Secretory ventral gland: absent?=?1; present =2
C17- Gelatinous masses in the lateral fins: absent?=?1; present =2
C18- Adhesive papillae: absent?=?1; present on the ventral side of the body and fins =2; concentrated on adhesive appendages =3
C20- Number of teeth rows: one anterior row =1; one posterior row =2; two rows =3;
C21- Epidermal glandular structures: Glandular structure on the body surface scarcely developed =1; numerous glandular structures on the body surface =2
C22- Type of teeth: stout teeth arranged in fan shape =1; slender teeth arranged in comb-shaped =2
C23- Ray less zone in the lateral fins: absent?=?1; present =2
Quantitative characters (n = 9)
C24- Trunk/tail length ratio (minimum value)
C25- Trunk/tail length ratio (maximum value)
C26- Position of the ventral nerve centre (in respect to the trunk length)
C27- Minimum number of anterior teeth
C28- Maximum number of anterior teeth
C29- Minimum number of posterior teeth
C30- Maximum number of posterior teeth
C31- Minimum number of hooks
C32- Maximum number of hooks
Qualitative morphological data were analysed using Paup* 4.0b10 under maximum parsimony (MP) with a heuristic search with 10 random taxon addition replicates followed by tree bisection and reconnection (TBR) branch swapping. All characters were treated as unordered and unweighted. ACCTRAN (accelerated transformation) and DELTRAN (delayed transformation) character optimization were both used to map the character changes and resolve ambiguous nodes. The g1 statistic was obtained using 1 000 000 random trees. Clade frequencies were obtained by 50% majority-rule consensus trees.
As many chaetognath lineages have been defined by the lack of a given structure, we also conducted a phenetic approach to integrate quantitative data and to estimate the degree of overall similarity information available (i.e., the absence of a character as valuable phylogenetic information). In order to include all variables in a common analysis, we chose to treat the data set as quantitative by replacing each qualitative variable by its disjunctive table. Such a table contains as many columns as modalities: each column defining a binary variable of 1 if the modality is observed and 0 otherwise. After this operation we ended up with a table of 34 lines (taxa) and 59 columns (original variables for quantitative characters or binary score corresponding to a modality for qualitative ones). We considered the Euclidian distance between two taxa after scaling each column to one and performed a non-supervised hierarchical clustering using the Ward algorithm. As the number of columns is slightly higher than the number of taxa it is especially important here to access the uncertainty of the relationships obtained after the classification procedure. We did that using the re-sampling procedure implemented in the pvclust package  of R version 3.0.1 . At each bifurcation of the classification, the variables significantly different between the two classes were identified. This was made by performing t-test for quantitative variables and chi-square test for qualitative variables that allowed us to characterize classes of the topology. Although the bootstrap probability test is very useful for tree selection, it is biased. The selection bias comes from comparing many trees at the same time and often leads to overconfidence in the wrong trees. So, we chose the approximately unbiased (au) test for assessing the confidence of tree selection - a method less biased than other methods such as bootstrap probability test .
We carried out a geometric morphometric approach  to explore body shape variations among the species used in the molecular analysis. The aim of this method was to test two different primary homology hypotheses on the evolution of structure and number of lateral fins and to correlate body shape patterns in relation to different locomotor and environmental behaviours (i.e., benthic versus pelagic) in Chaetognatha. Over the last three decades, systematic studies have often been complemented by geometric morphometrics, allowing the computation and visualization of global shape changes in organs or organisms. Procrustes superimposition is the most effective method for creating spatial graphical representations of shape variations .
The variation of the body shape patterns was statistically studied from morphotypes belonging to the six traditional clades identified in the molecular analysis: Heterokrohniidae, Eukrohniidae, Spadellidae, Pterosagittidae, Krohnittidae and Sagittidae. Illustrations of representative species used in this pilot study mostly come from the publications of Alvario , who provided the most accurate drawings of chaetognaths with respect to their body shape proportions and the position of their ventral nerve centre, lateral fins and seminal vesicles. Other sources were Tokiokas illustrations of Aidanosagitta crassa, Casanovas illustrations of Paraspadella gotoi and Xenokrohnia sorbei, the description by Dallot and Ducret  of Parasagitta megalophthalma as well as pictures of specimens belonging to the Spadella genus (Spadella ledoyeri, Spadella cephaloptera and Spadella valsalinae) by the authors of the present study.
Digital images were obtained with a flat bed scanner. Then, 20 landmarks were digitalized using TPSdig2  (Additional file 5). When the depicted specimens were not straight, we used the following procedure to get landmark coordinates of straightened specimens: we first calculated the mid points between homologous points on the right and left side of the specimens. This series of points was then aligned on the x-axis, and the relevant landmark points repositioned with a Y coordinate of half the distance between the left and right points (Additional file 5). Only one side of the individual specimens was then used for further analyses. The shape variation was analyzed by the generalized Procrustes method using the R shape package . Two primary homology hypotheses (PH) were tested: the anterior end of the posterior lateral fin in two fin species is homologous to the anterior end of the unique lateral fin in one-fin species (PH1), or the anterior end of the anterior lateral fin in two fin species is homologous to the anterior end of the unique lateral fin in one fin species (PH2). The full Procrustes distances between the conformations of each possible pair of species were computed in the two homology hypotheses. Dendrograms of landmark conformation similarities were computed by UPGMA. These hypotheses were tested to establish which of the two dendrograms from the morphometric data shows the higher agglomerative coefficient, and whether they were congruent with molecular data.
SG and YP identified the specimens. SG carried out the molecular experiments. YP and GN designed the geometric morphometrics study and drafted the corresponding part of the manuscript. SG, NP, AG and YP did the phenetic and cladistic analysis of morphological characters. SG, AG and YP did the molecular phylogenetic analyses. AG, YP and ST drafted the manuscript. YP conceived the study and its design. AG and YP supervised the study. All authors read, amended and approved the final manuscript.
We are deeply grateful to the Tara schooner and crew for collecting plankton samples all over the World Oceans during three years. Our special thanks go to Gaby Gorsky and Christian Sardet who facilitated our integration in the Tara Oceans consortium. We are also keen to thank the following people for providing us with planktonic samples: Jean-Marc Pagano from the Institut de Recherche pour le Dveloppement (IRD), John Lamkin and Akihiro Shiroza from the Southeast Fisheries Science Centre (SEFSC), Cdric Guigand from the Rosentiel School of Marine and Atmospheric Science (RSMAS) and Fredrika Norrbin from the Department of Arctic and Marine Biology (The Arctic University of Norway). Our manuscript profited from stimulating discussion with Daniel Papillon who also provided English corrections. This article is the contribution no. 19 of the Tara Oceans Expedition 2009/2012.
- Bone Q, Kapp H, Pierrot Bults AC: Introduction and Relationships of the Group. The Biology of Chaetognaths. Edited by: Bone Q, Kapp H, Pierrot-Bults AC. 1991, Oxford University Press, Oxford, 1-5.Google Scholar
- Casanova JP: Chaetognatha. South Atlantic Zooplankton. Edited by: Boltovskoy D. 1999, Backhuys Publishers, Leiden, 1353-1374.Google Scholar
- Vannier J, Steiner M, Renvois E, Hu SX, Casanova JP: Early Cambrian origin of modern food webs: evidence from predator arrow worms. Proc Biol Sci. 2007, 274: 627-633. 10.1098/rspb.2006.3761.PubMedPubMed CentralView ArticleGoogle Scholar
- Slabber M: Natuurkundige Verlustigingen, beheizende microscopise Waarneemingen van in-en Uitlandse Wateren Land-dieren. Haarlem: J. Bosch; 1778Google Scholar
- Perez Y, Mller CHG, Harzsch S: The Chaetognatha: An Anarchistic Taxon Between Protostomia and Deuterostomia. Deep Metazoan Phylogeny: The Backbone of the Tree of Life: New Insights from Analyses of Molecules, Morphology, and Theory of Data Analysis. Edited by: Wgele JW, Bartholomaeus TW, Misof B. 2014, Walter de Gruyter, Berlin, 49-77.Google Scholar
- Edgecombe GD, Giribet G, Dunn CW, Hejnol A, Kristensen RM, Neves RC, Rouse GW, Worsaae K, Srensen MV: Higher-level metazoan relationships: recent progress and remaining questions. Org Divers Evol. 2011, 11: 151-172. 10.1007/s13127-011-0044-4.View ArticleGoogle Scholar
- Harzsch S, Mller CH: A new look at the ventral nerve centre of Sagitta: implications for the phylogenetic position of Chaetognatha (arrow worms) and the evolution of the bilaterian nervous system. Front Zool. 2007, 4: 14-10.1186/1742-9994-4-14.PubMedPubMed CentralView ArticleGoogle Scholar
- Rieger V, Perez Y, Mller CH, Lacalli T, Hansson BS, Harzsch S: Development of the nervous system in hatchlings of Spadella cephaloptera (Chaetognatha), and implications for nervous system evolution in Bilateria. Dev Growth Differ. 2011, 53: 740-759. 10.1111/j.1440-169X.2011.01283.x.PubMedView ArticleGoogle Scholar
- Marltaz F, Martin E, Perez Y, Papillon D, Caubit X, Lowe CJ, Freeman B, Fasano L, Dossat C, Wincker P, Weissenbach J, Le Parco Y: Chaetognath phylogenomics: a protostome with deuterostome-like development. Curr Biol. 2006, 16: R577-R578. 10.1016/j.cub.2006.07.016.View ArticleGoogle Scholar
- Matus DQ, Copley RR, Dunn CW, Hejnol A, Eccleston H, Halanych KM, Martindale MQ, Telford MJ: Broad taxon and gene sampling indicate that chaetognaths are protostomes. Curr Biol. 2006, 16: R575-R576. 10.1016/j.cub.2006.07.017.PubMedView ArticleGoogle Scholar
- Marltaz F, Gilles A, Caubit X, Perez Y, Dossat C, Samain S, Gyapay G, Wincker P, Le Parco Y: Chaetognath transcriptome reveals ancestral and unique features among bilaterians. Genome Biol. 2008, 9: R94-10.1186/gb-2008-9-6-r94.View ArticleGoogle Scholar
- Dunn CW, Hejnol A, Matus DQ, Pang K, Browne WE, Smith SA, Seaver E, Rouse GW, Obst M, Edgecombe GD, Srensen MV, Haddock SH, Schmidt-Rhaesa A, Okusu A, Kristensen RM, Wheeler WC, Martindale MQ, Giribet G: Broad phylogenomic sampling improves resolution of the animal tree of life. Nature. 2008, 452: 745-749. 10.1038/nature06614.PubMedView ArticleGoogle Scholar
- Hejnol A, Obst M, Stamatakis A, Ott M, Rouse GW, Edgecombe GD, Martinez P, Bagu J, Bailly X, Jondelius U, Wiens M, Mller WE, Seaver E, Wheeler WC, Martindale MQ, Giribet G, Dunn CW: Assessing the root of bilaterians animals with scalable phylogenomic methods. Proc Biol Sci. 2009, 276: 261-270. 10.1098/rspb.2009.0896.View ArticleGoogle Scholar
- Nielsen C: Animal Evolution: Interrelationships of the Living Phyla. 2001, Oxford University Press, Oxford New York, 2Google Scholar
- Nielsen C: Larval and adult brains. Evol Dev. 2005, 7: 483-489. 10.1111/j.1525-142X.2005.05051.x.PubMedView ArticleGoogle Scholar
- Bieri R: Systematics of the Chaetognatha. The Biology of Chaetognaths. Edited by: Bone Q, Kapp H, Pierrot-Bults AC. 1991, Oxford University Press, Oxford, 122-136.Google Scholar
- Bieri R: Six new genera in the chaetognath family Sagittidae. Gulf Res Rep. 1991, 8: 221-225.Google Scholar
- Papillon D, Perez Y, Caubit X, Le Parco Y: Systematics of Chaetognatha under the light of molecular data, using duplicated ribosomal 18S DNA sequences. Mol Phylogenet Evol. 2006, 38: 621-634. 10.1016/j.ympev.2005.12.004.PubMedView ArticleGoogle Scholar
- Von Ritter-Zahony R: Revision der Chaetognathen. Dtsch Sdpolar-Exped. 1911, 13: 1-71.Google Scholar
- Hyman LH: Phylum Chaetognatha. The Invertebrates. Volume 5. 1959, McGraw Hill Book Company, New York, 1-71.Google Scholar
- Tokioka T: The taxonomical outline of chaetognaths. Publ Seto Mar Biol Lab. 1965, 12: 335-357.Google Scholar
- Tokioka T: Supplementary notes on the systematics of Chaetognatha. Publ Seto Mar Biol Lab. 1965, 13: 231-242.Google Scholar
- Alvario A: Quetognatos epiplanctonicos del Mar de Cortes. Rev Soc Mex Hist Nat. 1963, 24: 97-202.Google Scholar
- Casanova JP: Description de lappareil gnital primitif du genre Heterokrohnia et nouvelle classification des chaetognathes. CR Acad Sci Paris. 1985, 301: 397-402.Google Scholar
- Dallot S, Ibanez F: Etude prliminaire de la morphologie et de lvolution chez les chaetognathes. Inv Pesq. 1972, 36: 31-41.Google Scholar
- Salvini-Plawen LV: Systematics notes on Spadella and on the Chaetognatha. J Z Zool Syst Evol. 1986, 24: 122-128. 10.1111/j.1439-0469.1986.tb00620.x.View ArticleGoogle Scholar
- Telford MJ, Holland PWH: Evolution of 28S ribosomal DNA in chaetognaths, duplicate genes and molecular phylogeny. J Mol Evol. 1997, 44: 135-144. 10.1007/PL00006130.PubMedView ArticleGoogle Scholar
- Jennings RM, Bucklin A, Pierrot-Bults A: Barcoding of arrow worms (Phylum Chaetognatha) from three oceans: genetic diversity and evolution within an enigmatic phylum. PLoS One. 2010, 5: e9949-10.1371/journal.pone.0009949.PubMedPubMed CentralView ArticleGoogle Scholar
- Park JK, Rho HS, Kristensen RM, Kim W, Giribet G: First molecular data on the phylum Loricifera: an investigation into the phylogeny of ecdysozoa with emphasis on the positions of Loricifera and Priapulida. Zoolog Sci. 2006, 23: 943-954. 10.2108/zsj.23.943.PubMedView ArticleGoogle Scholar
- Farris J: Estimating phylogenetic trees from distance matrices. Amer Nat. 1972, 106: 645-667. 10.1086/282802.View ArticleGoogle Scholar
- Tokioka T, Pathansali D: Another new chaetognath from Malay waters, with a proposal of grouping some species of Sagitta into subgenera. Publ Seto Mar Biol Lab. 1963, 11: 119-123.Google Scholar
- Harzsch S, Mller CH, Rieger V, Perez Y, Sintoni S, Sardet C, Hansson B: Fine structure of the ventral nerve centre and interspecific identification of individual neurons in the enigmatic Chaetognatha. Zoomorphology. 2009, 128: 53-73. 10.1007/s00435-008-0074-4.View ArticleGoogle Scholar
- Casanova JP, Duvert M: Comparative studies and evolution of muscles in chaetognaths. Mar Biol. 2002, 141: 925-938. 10.1007/s00227-002-0889-3.View ArticleGoogle Scholar
- Casanova JP: A new genus and species of deep-benthic chaetognath from the Atlantic: a probable link between the families Heterokrohniidae and Spadellidae. J Nat Hist. 1996, 30: 1239-1245. 10.1080/00222939600770671.View ArticleGoogle Scholar
- Dallot S: Lanatomie du tube digestif dans la phylognie et la systmatique des chaetognathes. Bull Mus Natl Hist Nat. 1970, 42: 549-565.Google Scholar
- Furnestin ML: Chaetognathes et zooplankton du secteur atlantique marocain. Rev Trav Inst Pech Mar. 1957, 21: 1-356.Google Scholar
- Alvario A: The Chaetognatha of the NAGA Expedition (19591961) in the South China Sea and the Gulf of Thailand. Part I. Systematics. NAGA report. 1967, 4: 1-197.Google Scholar
- Alvario A: Los quetognatos del Atantico: Distribucion y notes essenciales de sistimatica. Trab Inst Esp Oceanogr. 1969, 37: 1-290.Google Scholar
- Szaniawski H: Cambrian chaetognaths recognized in Burgess Shale fossils. Acta Palaeontol Pol. 2005, 50: 1-8.Google Scholar
- McLelland JA: Eukrohnia calliops, a new species of Chaetognatha from the northern Gulf of Mexico with notes on related species. Proc Biol Soc Wash. 1989, 102: 33-44.Google Scholar
- Casanova JP, Nair VR: A new species of Sagitta (Chaetognatha) from a Laccadive lagoon (Indian Ocean) having fan-shaped anterior teeth: phylogenetical implications. J Nat Hist. 2002, 36: 149-156. 10.1080/00222930010009345.View ArticleGoogle Scholar
- Casanova JP, Moreau X: Calispadella alata n. gen., n. sp., the first chaetognath recorded from a hydrothermal vent site (Mid-Atlantic Ridge). J Plank Res. 2005, 27: 221-225. 10.1093/plankt/fbh163.View ArticleGoogle Scholar
- Stebbins GL: Mosaic evolution, mosaic selection and angiosperm phylogeny. J Linn Soc Bot. 1984, 88: 149-164. 10.1111/j.1095-8339.1984.tb01568.x.View ArticleGoogle Scholar
- Knoll AH, Niklas KJ, Gensel PG, Tiffney BH: Character diversification and patterns of evolution in early vascular plants. Paleobiology. 1984, 10: 34-47.Google Scholar
- Kearney M, Stuart BL: Repeated evolution of limblessness and digging heads in worm lizards revealed by DNA from old bones. Proc R Soc Lond B. 2004, 271: 1677-1683. 10.1098/rspb.2004.2771.View ArticleGoogle Scholar
- Winter W, Oxnard CE: Evolutionary radiations and convergences in the structural organization of mammalian brains. Nature. 2001, 409: 710-714. 10.1038/35055547.PubMedView ArticleGoogle Scholar
- Lazzari V, Charles C, Tafforeau P, Vianey-Liaud M, Aguilar JP, Jaeger JJ, Michaux J, Viriot L: Mosaic convergence of rodent dentitions. PLoS One. 2008, 3: e3607-10.1371/journal.pone.0003607.PubMedPubMed CentralView ArticleGoogle Scholar
- de Queiroz K: The ontogenetic method for determining character polarity and its relevance to phylogenetic systematics. Syst Zool. 1985, 34: 280-299. 10.2307/2413148.View ArticleGoogle Scholar
- Bone Q, Brownlee C, Bryan GW, Burt GR, Dando PR, Liddicoat MI, Pulsford AL, Ryan KP: On the differences between the two `indicator species of chaetognath, Sagitta setosa and S. elegans. J Mar Biol Ass UK. 1987, 67: 545-560. 10.1017/S0025315400027284.View ArticleGoogle Scholar
- Bone Q, Duvert M: Locomotion and Buoyancy. The Biology of Chaetognaths. Edited by: Bone Q, Kapp H, Pierrot-Bults AC. 1991, Oxford University Press, Oxford, 32-44.Google Scholar
- Kapp H: Some aspects of buoyancy adaptations of chaetognaths. Helgol Meeres. 1991, 45: 263-267. 10.1007/BF02365646.View ArticleGoogle Scholar
- Perez Y, Casanova JP, Mazza J: Degrees of vacuolation of the absorptive cells of five Sagitta (Chaetognatha) species: possible ecophysiological implications. Mar Biol. 2001, 138: 125-133. 10.1007/s002270000438.View ArticleGoogle Scholar
- Corse E, Rampal J, Cuoc C, Pech N, Perez Y, Gilles A: Phylogenetic analysis of Thecosomata Blainville, 1824 (holoplanktonic opisthobranchia) using morphological and molecular data. PLoS One. 2013, 12: e59439-10.1371/journal.pone.0059439.View ArticleGoogle Scholar
- Mller CHG, Rieger V, Perez Y, Harzsch S: Immunohistochemical and ultrastructural studies on ciliary sense organs of arrow worms (Chaetognatha). Zoomorphology. 2014, 133: 167-189. 10.1007/s00435-013-0211-6.View ArticleGoogle Scholar
- Thompson JD, Higgins DG, Gibson TJ: CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 1994, 22: 4673-4680. 10.1093/nar/22.22.4673.PubMedPubMed CentralView ArticleGoogle Scholar
- Edgar RC: MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res. 2004, 32: 1792-1797. 10.1093/nar/gkh340.PubMedPubMed CentralView ArticleGoogle Scholar
- Posada D, Crandall KA: MODELTEST: testing the model of DNA substitution. Bioinformatics. 1998, 14: 817-818. 10.1093/bioinformatics/14.9.817.PubMedView ArticleGoogle Scholar
- Misof B, Misof K: A Monte Carlo approach successfully identifies randomness in multiple sequence alignments : a more objective means of data exclusion. Syst Biol. 2009, 58: 21-34. 10.1093/sysbio/syp006.PubMedView ArticleGoogle Scholar
- Kck P, Meusemann K, Dambach J, Thormann B, von Reumont BM, Wgele JW, Misof B: Parametric and non-parametric masking of randomness in sequence alignments can be improved and leads to better resolved trees. Front Zool. 2010, 31: 10-10.1186/1742-9994-7-10.View ArticleGoogle Scholar
- Hillis DM, Bull JJ: An empirical test of bootstrapping as a method for assessing confidence in phylogenetic analysis. Syst Biol. 1993, 42: 182-192. 10.1093/sysbio/42.2.182.View ArticleGoogle Scholar
- Huelsenbeck JP, Ronquist F: MRBAYES: Bayesian inference of phylogenetic trees. Bioinformatics. 2001, 17: 754-755. 10.1093/bioinformatics/17.8.754.PubMedView ArticleGoogle Scholar
- Rambaut A, Drummond AJ: Tracer: MCMC Trace Analysis Tool. 2003, University of Oxford, OxfordGoogle Scholar
- Guindon S, Dufayard JF, Lefort V, Anisimova M, Hordijk W, Gascuel O: New algorithms and methods to estimate maximum-likelihood phylogenies: assessing the performance of PhyML 3.0. Syst Biol. 2010, 59: 307-321. 10.1093/sysbio/syq010.PubMedView ArticleGoogle Scholar
- Anisimova M, Gil M, Dufayard JF, Dessimoz C, Gascuel O: Survey of branch support methods demonstrates accuracy, power, and robustness of fast likelihood-based approximation schemes. Syst Biol. 2011, 60: 685-699. 10.1093/sysbio/syr041.PubMedPubMed CentralView ArticleGoogle Scholar
- Anisimova M, Gascuel O: Approximate likelihood-ratio test for branches: a fast, accurate, and powerful alternative. Syst Biol. 2006, 55: 539-552. 10.1080/10635150600755453.PubMedView ArticleGoogle Scholar
- Fowler GH: Biscayan plankton collected during a cruise H.M.S. Research 1900, III. The Chaetognatha. Trans Linn Soc Lond (Zool). 1905, 10: 55-87. 10.1111/j.1096-3642.1905.tb00184.x.View ArticleGoogle Scholar
- Fowler GH: The Chaetognatha of the Siboga expedition with a discussion of the synonymy and distribution of the group. Siboga Exped Rep. 1906, 21: 1-86.Google Scholar
- Frazer JH: Chaetognatha. Fiches didentification du zooplancton. Zooplankton. 1957, Intern Counc Explo Sea, Charlottenlund, 1-6.Google Scholar
- Tokioka T: A new chaetognath (Sagitta crassa n. sp.) from Ise Bay. Zool Mag (Tokio). 1938, 50: 349-351.Google Scholar
- Casanova JP:Spadella ledoyeri, un chaetognathe nouveau de la grotte sous-marine obscure des Trmies (Calanques de Cassis). Rapp Comm Mer Mdit. 1986, 30: 196-197.Google Scholar
- Casanova JP: A new species of Paraspadella (Chaetognatha) from the coastal waters of Japan. Proc Biol Soc Wash. 1990, 103: 907-912.Google Scholar
- Casanova JP: A new genus and species of deep-sea chaetognath from the Bay of Biscay with a strange ventral secretory gland. J Nat Hist. 1993, 27: 445-455. 10.1080/00222939300770211.View ArticleGoogle Scholar
- Dallot S, Ducret F: Un chaetognathe msoplanctonique nouveau Sagitta megalophthalma sp. n. Beaufortia. 1969, 17: 13-20.Google Scholar
- Winkelmann C, Gasmi S, Gretschel G, Mller CHG, Perez Y: Description of Spadella valsalinae sp. nov., a neo-endemic benthic chaetognath from Northern Adriatic Sea (Croatia) with remarks on its morphology, phylogeny and biogeography. Org Div Evol. 2013, 13: 189-202. 10.1007/s13127-012-0108-0.View ArticleGoogle Scholar
- Suzuki R, Shimodaira H: Pvclust: an R package for assessing the uncertainty in hierarchical clustering. Bioinformatics. 2006, 22: 1540-1542. 10.1093/bioinformatics/btl117.PubMedView ArticleGoogle Scholar
- R Core Team: R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria; 2013 , [http://www.r-project.org/]
- Shimodaira H: An approximately unbiased test of phylogenetic tree selection. Syst Biol. 2002, 51: 492-508. 10.1080/10635150290069913.PubMedView ArticleGoogle Scholar
- Claude J: Morphometrics with R. 2008, Springer, New YorkGoogle Scholar
- Rohlf FJ, Slice D: Extensions of the Procrustes method for the optimal superimposition of landmarks. Syst Biol. 1990, 39: 40-59.Google Scholar
- Rohlf FJ: TpsDig Version 2.00. Department of Ecology and Evolution, State University of New York. New York: Stony Brook; 2004 , [http://life.bio.sunysb.edu/morph/]
- Dryden IL: Shapes package. R Foundation for Statistical Computing, Vienna, Austria. Contributed package. , [http://www.r-project.org]
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.