Ophiuroid ontogeny

The here used terminology for the ophiuroid skeleton follows Sumida et al. [51], and Stöhr et al. [54]. In the species that have so far been examined, development follows a similar pattern, which we postulate here as the typical ophiuroid stages of juvenile development, based on our own examinations and other published data [21,41,47,48,51,55,56]. In most ophiuroid species, the dorsal disc of the smallest postlarva is formed by six plates, a centrodorsale and five radial primary plates (Fig 1), together known as the primary rosette, although exceptions where the rosette is lacking have been reported [41]. Radial shields and secondary disc scales appear at different stages during growth [41,51]. Dorsal arm plates are at first absent and develop gradually, starting on the proximalmost joints. Of the ventral arm plates the first one is always present from the youngest stages as part of the mouth frame, while the following develop gradually, but usually before the corresponding dorsal plate. Arm segments are relatively longer in juveniles than in adults, but adult distal arms are similar to juvenile arms, because the arms grow at the tip, proximal to the terminal plate, and thus the distal part is younger and less developed than the proximal part. Growth occurs first lengthwise, later widthwise. Tentacle scales are at first absent and appear at a rather late stage, after dorsal and ventral plates and independent of arm spine development.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Skeletal morphology of Ophiuroidea postlarvae, Ophiura sarsii.

(A) dorsal aspect, (B) ventral aspect. AS, adoral shield, ASS, adoral shield spine, BS, buccal scale, CPP, central primary plate, D, dental plate, DAP, dorsal arm plate, IR, interradial secondary disc scale, J, jaw = oral plate, LAP, lateral arm plate, M, madreporite, OS, oral shield, RPP, radial primary shield, S, arm spine, TPo2, second oral tentacle pore, T, tooth, VAP, ventral arm plate. Scale bars 0. 1 mm.

https://doi.org/10.1371/journal.pone.0164562.g001

In the oral frame of Ophiuridae and Ophiolepididae, at first the dental plate at the apex of the jaw bears a single tooth, and usually, a wide low lateral papilla termed buccal scale [55] is found at the oral plate. A second oral papilla is usually present in the adoral shield spine, at first usually spine-shaped and situated at the disc edge, next to the second oral tentacle pore, which is in early stages placed far from the mouth angle, into which it opens in adult Ophiuridae, but into which it is completely incorporated in Ophiolepididae. The oral shields are absent in the smallest postlarvae, except for the madreporite in some species. The madreporite in young stages can often be recognized by a cone-shaped protruding hydropore. Oral shields (including the madreporite) first appear at the dorsolateral edge of the disc [41,45,51] and move towards the mouth during ontogeny. The adoral shields that in adults frame the proximal and often lateral edges of the oral shields are present in early stages, since they are homologous to lateral arm plates. They obtain their specific shape and position during later ontogeny. In their juvenile state, the adoral shields reach around the oral shield, separating it from the arm, and the distal part of the adoral shields is placed at the lateral arm. During development the adoral shields shorten and in their most advanced state they are completely removed from the lateral arm, only bordering the proximal edges of the oral shield. The second tentacle pore is in young stages far from the mouth, but moves closer during ontogeny. It is associated with the adoral shields and the adoral shield spine, which turns into the distalmost oral papilla.

The order in which teeth appear seems to vary between taxa. Hendler [55] reported a series of teeth appearing at the same time in Amphioplus abditus (Verrill, 1871), Clark [46] described the first tooth to be the dorsalmost one, subsequent teeth forming ventralwards, whereas Stöhr [41] interpreted the first tooth as the ventralmost one, with subsequent teeth forming dorsalwards. Consequently, the dental plate may grow in either ventral or dorsal direction and tooth papillae in the Ophiotrichidae and Ophiocomidae may possibly form either in situ or by fragmentation of the ventral teeth. Clearly, this subject needs additional attention. Nevertheless, the dental plate in postlarvae is more or less round and cup-shaped, compared to the elongated, flat adult condition. Tooth sockets are simple openings or depressions in juveniles, but larger holes bordered by ridges or knobs in well developed adults.

Taxon selection

We selected 40 species with more or less paedomorphic appearance from the families Ophiuridae (26 species) and Ophiolepididae (14 species) (Table 2). Most of the samples are housed in the collections of the Zoological Museum, Moscow State University (ZMMU), and the Swedish Museum of Natural History (SMNH). We also borrowed type specimens of paedomorphic species from Museum of Comparative Zoology, Harvard, Natural History Museum, London, US National Museum, Washington D.C., Zoological Museum, Amsterdam, Zoological Museum, Berlin, and Zoological Museum, University of Copenhagen, for additional data and as reference for correct identification of our material (Table 2). From type material we could gather mostly external characters, but from Ophiomastus bulufonica Tommasi, 1976, Ophiomastus perplexus Koehler, 1904 and Ophiolepis impressa Lütken, 1859 we received permission to dissect small arm pieces. To determine juvenile character states, we examined and dissected postlarvae of Ophiura sarsii Lütken, 1855 of the family Ophiuridae, Ophiomusium lymani Wyville-Thomson, 1873 of the family Ophiolepididae, and Amphipholis squamata (Delle Chiaje, 1828) of Amphiuridae. Also for several paedomorphic species, juveniles were available for examination. For comparison with a species with few or no paedomorphic character states, Ophiocoma (Breviturma) krohi Stöhr et al., 2013 was chosen, of which all skeletal structures have already been documented [57].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. List of studied species.

https://doi.org/10.1371/journal.pone.0164562.t002

Preparations and measurements

Whole specimens were lightly bleached with household bleach NaOCl diluted 1:1 with tap water, to remove the outer integument. The dried specimens were then mounted on aluminium stubs with spray glue, examined in a Hitachi FE-SEM S4300 scanning electron microscope (SEM) and digitally photographed. Additional specimens were dissolved in concentrated bleach, the skeletal elements washed in tap water, and jaws, vertebrae, dental plates and lateral arm plates were mounted on aluminium stubs as above, examined and photographed in the SEM.

We analysed 29 characters from material specifically prepared during this study as well as from our image archives, and classified them as postlarva (= early postmetamorphic), juvenile (= intermediate between postlarva and adult) or adult (Table 3). Thus, the states as described in Table 3 are based on the respective ontogenetic stages in non-paedomorphic species, e.g. a primary rosette is not obvious (= same size as disc scales) in a fully developed dorsal disc, obvious but not dominating the disc (= larger than disc scales) in juveniles and dominating the disc (= no/few additional scales) in postlarvae. Juvenile state was assessed based on the conditions found in postlarvae with additional elements or shape changes, adult state was based on the terminal state of the ontogenetic trajectory of a well developed species from a different family than the studied group, because we suspect that all Ophiuridae and Ophiolepididae possess some paedomorphic characters. As an example, the medium sized postlarvae in Fig 1 have a primary rosette dominating the disc, weakly developed radial shields, separated dorsal arm plates along the entire arm, no ventral disc scales, a madreporite that differs from the other oral shields (by a clearly visible hydropore), the second oral tentacle pore far from the mouth slit. Weakly defined states are observed in early ontogeny when a structure has begun to form and its typical shape has not manifested itself yet. In these early stages, the edges of the skeletal element are fuzzy, any depressions, knobs or flanges are barely noticeable. Each state was assigned an integer value as a "paedomorphosis score", with zero denoting a non-paedomorphic adult state and higher numbers assigned to increasingly juvenile-looking characters. We follow the classification of the World Ophiuroidea Database [58].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Character states used for scoring developmental/paedomorphic status.

https://doi.org/10.1371/journal.pone.0164562.t003

The "paedomorphosis score" is admittedly subjective since continuously developing structures are difficult to delimit into well defined stages. Therefore we also developed a mathematical approach that can easily be reproduced. We selected arm vertebrae (from the proximal arm), oral plates, and dental plates, for morphometric measurements. These skeletal elements vary in their precise shape with flanges and knobs protruding to various degree, which makes it difficult to measure length and height or width at exactly the same spot (Fig 2A). Therefore, we decided to develop a standardized approach. We fitted an ellipse to these skeletal elements on the SEM images in the software programme ImageJ [59], measured the minor and major axes and calculated the quotient Major/Minor. This is a simple procedure in which the scale bar of the SEM image is used to calibrate the tool, then a polygon is drawn around the outline of the object (Fig 2B), a programme function is used to fit the ellipse and then the programme measures the axes of the ellipse (Fig 2C). In most cases, the major axis was the length of the structure, if not, we corrected for that. The resulting ellipses were rarely accurately parallel to the horizontal axis of the object, and we compared the values to measurements of straight lines placed on the object by eye. The absolute numbers between these methods differed, but the results in ordering the species according to the geometry of their skeletal elements were comparable. The ellipse method can more reliably be reproduced than the placement of a straight line arbitrarily across an irregularly shaped object. Thus we deemed the ellipse method more objective.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. SEM images of a proximal arm vertebra of Ophiozonella falklandica, dorsal view.

(A) Selecting a place to measure the length (horizontal line) and width (vertical lines) is difficult due to the complicated shape of the ossicle. (B) For a more standardized approach, in the software ImageJ a polygonal line is drawn around the shape of the ossicle. (C) Using a built in function of ImageJ an ellipse is then fitted to the ossicle; major (here length) and minor (here width) axes are measured automatically. Scale bar 0.1 mm.

https://doi.org/10.1371/journal.pone.0164562.g002

Paedomorphosis and Phylogeny

Based on the above description of ophiuroid skeletal development, we can now analyse the status of adult morphologies. The primary rosette is present in most ophiuroids at an early stage, but in many species it is obscured during growth and often no longer distinguishable in adults. Hendler [55] commented for Amphiuridae that the primary plates are not regenerated when the disc is damaged (in contrast to radial shields and other structures), and concluded that the primary plates must be an ancient character that is retained in early ontogeny, but not necessary for disc formation in modern ophiuroids. The absence of typical primary plates in postlarvae of several extant species [41] supports this interpretation. A primary rosette was present in the earliest ophiuroids, as Hotchkiss [60] showed in a juvenile specimen of the extinct Palaeozoic order Oegophiurida (absent in adults), and retained through all of ophiuroid evolution, as evidenced by Jurassic postlarvae [52]. Thus, the existence of primary plates as such is a plesiomorphy that can be traced back to the very origin of the Ophiuroidea, but their presence in adult ophiuroids is a paedomorphic state. This is an important and still underestimated conclusion for the theory and practice of phylogenetic systematics (cladistics) since such an apparently plesiomorphic feature may reappear in heterochrony-driven descendant taxa many times, thus forming an apomorphic pseudoplesiomorphy [21]. Interestingly, primary plates seem to have been lost completely (not just in adults), several times independently on species level in several families [41]. This cannot be explained by heterochrony since secondary scales, and usually radial shields, are formed in these species, but possibly genes were switched off. If primary plates are not regenerated after disc damage, the question arises, how do strongly paedomorphic species that lack secondary scales repair their disc? Once their ontogeny has passed the stage of primary plate formation, do they keep the ability to switch the respective genes on again? Or do they switch on genes that are responsible for the formation of secondary scales, but which are not active during undisturbed growth? Since many of these species are found only at great depth (>2000 m) they are rarely collected and chances of finding regenerated specimens are slim; keeping them alive for laboratory experimentation is hardly possible. However, importantly, according to the available present and previous data [30,32,35] the paedomorphic morphology becomes evident in ophiuroids at depths below 1500–2000 m. This may be linked to low nutrient and energy potential of the bathyal and abyssal environments and corroborated by data from other groups, e.g. deep sea fishes [61]. The gene control behind ophiuroid development has barely been studied yet within a large comparative taxonomic framework, but may shed some light on these questions in the future since some pioneer studies on ophiuroid gene regulatory networks have appeared most recently [62].

Already Matsumoto [32] was aware of partially unfused vertebrae in some species and suggested a paedomorphic origin. The almost completely unfused vertebrae of adult Perlophiura may be evidence of extreme paedomorphosis in which the development is terminated earlier than the state found in typical Ophiuridae postlarvae. Fused vertebrae evolved from the left and right ambulacrals of Palaeozoic ophiuroid stem representatives [63] and therefore the unfused state could be plesiomorphic. Fused vertebrae occurred in the Ordovician to Devonian Oegophiurida [64]. The vertebrae of the Lower Carboniferous genus Onychaster were recently re-evaluated and classified as a special type of articulation, different from modern types [65]. Interestingly, Onychaster had quite short vertebrae, similar to adult Ophiura sarsii, suggesting accelerated development. Short vertebrae are also known from Oegophiurida [64], suggesting that fused, well developed vertebrae are a plesiomorphic state for Ophiuroidea. The long, lightly fused vertebrae of Perlophiura show the typical zygospondylus proximal and distal articulation faces of modern ophiuroids in the family Ophiuridae, which suggests that they are not descendants of a plesiomorphic lineage that never evolved fused vertebrae, but more likely development is indeed abbreviated by paedomorphosis. By the same argument, the partially fused vertebrae of other species examined here are also affected by paedomorphosis to various degrees.

Matsumoto [32] observed that in many species the lateral arm plates meet in the dorsal and/or ventral midline, the dorsal and ventral plates (if present at all) only covering part of the arm segment. He argued that the joined state of the lateral arm plates must be an advanced state, since the extinct Oegophiurida had an open ambulacral groove, which was covered over by ventral plates in later evolution. The lateral arm plate pairs were widely separated in these early forms. In juveniles the lateral arm plates are always joined, but may separate during later development, the opening being covered by ventral and dorsal plates. From this evidence we conclude that the joined state of lateral arm plates and widely separated or absent dorsal and ventral arm plates are not ancient, but paedomorphic states.

The position of the second tentacle pore inside the mouth is evidently the developed state, its position outside the mouth in adult specimens is paedomorphic. This suggests that the whole family Ophiuridae has a paedomorphic origin, whereas the genera (e.g. Ophiolepis, Ophiozonella, Ophiomusium) currently grouped in Ophiolepididae are not paedomorphic for this character. However, the genus Ophiomusium possesses several paedomorphic characters, such as presence of a primary rosette, few ventral disc scales, absence of most ventral arm plates, small dorsal arm plates, joined lateral arm plates, tentacle scales obvious only on few proximal joints, and elongated oral plates. These characters distinguish it clearly from the type genus Ophiolepis and a paedomorphic origin of the whole family Ophiolepididae is unlikely. More likely it is necessary to re-evaluate this family and possibly separate it into strongly paedomorphic (e.g. Ophiomusium, Ophiosphalma) and well developed species (e.g. Ophiolepis, Ophiozonella). The genus Ophiomusium was considered as a member of the Ophiuridae by Martynov & Litvinova [66], underlain by an at the time not yet completely published analysis of the spine articulation ridges and other micromorphological features (Martynov, 2010), but most recently its position as a close relative to the family Ophiuridae s.l. has been confirmed by a large transcriptomic analysis [67] and by a novel morphological phylogenetic analysis [68]. The present study thus provides additional evidence for a polyphyletic origin of Ophiuridae and Ophiolepididae as currently understood. There is however at least one species of Ophiolepis that may have evolved through progenesis, because its adults are considerably smaller than its congeners and their morphology corresponds to the juveniles of larger species [69]. The adults of several species of Ophiozonella examined in this study are moderately paedomorphic, but the generic type O. longispina has one of the lowest paedomorphosis scores.

The madreporite is lateral in early Ophiuroidea and in small postlarvae of modern forms, but ventral in all adults. The developmental trajectory seems to move from a clearly distinguishable plate with a cone shaped protrusion to a plate that more and more resembles a regular oral shield. Thus, a madreporite that can be distinguished from the other oral shields is a paedomorphic character.

Various, sometimes conflicting, classifications of heterochrony have been suggested and the usage and definition of terms such as neoteny and progenesis have changed over time, with more terms such as hypomorphosis being introduced [4,12,39]. Ideally, developmental rate changes should be compared between a species and its ancestor but ancestor-descendant pairs are rarely available for study and none are known for ophiuroids so far. The suggestion that paedomorphic ophiuroids are progenetic [34] was based on the small size of paedomorphic adults in comparison to non-paedomorphic adults. This approach is flawed, because there is no "general size" of non-paedomorphic ophiuroids. Both recent (e.g. Amphipholis squamata) and extinct faunas (Hotchkiss, personal communication) include diminutive species with well developed skeleton. Instead, we compare the size of individuals of similar developmental stage among closely related taxon groups. The evolutionary relationship between Ophiura and Perlophiura can be established by similarities of the arm spine articulations, lateral arm plates, and juvenile vertebrae. Terminal stages of Perlophiura reach a maximum disc diameter of 5 mm, whereas most species of Ophiura reach 10–15 mm disc diameter. The hypothetical non-paedomorphic ancestral taxa can be expected in the upper size range of Ophiura. Perlophiura possesses mostly postlarval characters. At a size of 2.2 mm dd it has longer and narrower arm joints than postlarvae of O. sarsii at 0.86 mm dd. In most species, postlarvae of more than 2 mm dd possess a greater number of skeletal elements and can be identified using keys for adult characters [41]. The so far largest early postlarva known is that of Asteronyx loveni [40,41], but it has a disproportionately large disc and short arms with short joints. Perlophiura thus appears to be an overgrown postlarva in which development is much slowed down, while growth continues. The same may be true for Ophiotypa simplex, and the here studied strongly paedomorphic species all seem quite large in relation to the status of their skeleton. The direct ancestor of these genera is not known and we cannot know if these species are larger than their ancestors, qualifying them for Gould's [4] definitions of neoteny or any of the later refinements of terms. Thus, our study provides additional evidence that rigid traditional distinction of several fixed heterochronic modes needs to be reconsidered, and using the more general term paedomorphosis is much more useful.

Limitations

The approach used here successfully identified paedomorphic characters, but it has some limitations that may be approved upon with different techniques. The simple method of adding up score values was somewhat biased by the fact that not all target characters could be evaluated for all species. However, since it was not our intention to find the exact position of a species on a trajectory of heterochrony, the scores still suffice to show general trends.

The morphometric method was affected by the shape of the ossicles. The proximal vertebrae turned out to be the best indicator of juvenile/paedomorphic conditions. Oral plates (jaws) are probably undervalued by our method, because the fitted ellipse takes into account only the outline of an object, which works best with straight and relatively flat structures. Oral plates are strongly curved in young stages and a more correct measurement should at least measure the length along the curve, whereas we measured in a straight line from one end to the other. Measuring curves is difficult though, since SEM images are two-dimensional and curvatures or elevations are difficult to capture. The small size of the objects makes orientating them in a specific position extremely difficult, which may have lead to slight differences in positioning and measuring. The same applies to the dental plates that are cup-shaped in young stages and often almost round in outline. Even in vertebrae, there are projecting parts that affect the outline to variable degree depending on the exact orientation of the vertebra. For a statistically sound analysis, a larger number of specimens from each species should be examined. Methods of geometric morphometrics might be tried, but the aim of this study was not to find exact values and we do not believe that it is meaningful or possible to define the degree of paedomorphosis in that way. All skeletal elements have specific structures such as depressions, protrusions, flanges, excavations and other details. These may change shape during development, but are generally already present in young/paedomorphic specimens. They are also species specific and may obscure the developmental signals common to all species. Length to width ratios seem a simpler, but more robust measurement for the purpose of this study.

We cannot know the direct ancestors of the species studied here, and strictly speaking we cannot know if their development is really abbreviated in some mechanistic terms in comparison to the ancestral forms. Instead we propose that the observed morphologies are the result of complex developmental processes which may include a complicated mosaic of retardation and acceleration of morphological development underlain by changes in the gene regulatory networks. But ophiuroid development follows certain common paths as we and previous authors have demonstrated, and some species follow these paths much farther along than others to such a degree that the application of the term “paedomorphosis” for the majority of the studied cases is completely clear and unambiguous because not only some external features but also internal ones show such paedomorphic trajectories. These paedomorphic trajectories allow then the almost direct comparison of juvenile features of non-paedomorphic taxa with adult characters of paedomorphic ones, thus making the general pattern of phylogenesis less random and more predictable than has traditionally been concluded. Paedomorphic lineages may be quite old and indeed Ophiuridae and Ophiomusium are probably some of the oldest groups of Ophiuroidea. Thus one might argue that younger groups are peramorphic, displaying accelerated development. The fossil record however, seems to disprove this idea, since well developed forms are known from the oldest groups. Obviously, there is no limit for the appearance of paedomorphic taxa in a given group at any geological time and this is in concordance with the general conclusions on the evolution in the class Ophiuroidea [70].