Zootaxa 4415 (3): 452–472 ISSN 1175-5326 (print edition) Article ZOOTAXA http://www.mapress.com/j/zt/ Copyright © 2018 Magnolia Press ISSN 1175-5334 (online edition) https://doi.org/10.11646/zootaxa.4415.3.3 http://zoobank.org/urn:lsid:zoobank.org:pub:EB29389D-F8EC-41FE-80B5-1D1B948DD9F6 Description of Tottonophyes enigmatica gen. nov., sp. nov. (Hydrozoa, Siphonophora, Calycophorae), with a reappraisal of the function and homology of nectophoral canals P. R. PUGH1,4, C.W. DUNN2 & S.H.D. HADDOCK3 1National Oceanography Centre, Empress Dock, Southampton, Hants, SO14 3ZH, UK. 2Department of Ecology and Evolutionary Biology, Yale University, New Haven, Connecticut, USA. 3Monterey Bay Aquarium Research Institute, 7700 Sandholdt Road, Moss Landing, California 95039, USA. 4Corresponding author. E-mail: [email protected] Abstract A new species of calycophoran siphonophore, Tottonophyes enigmatica gen. nov, sp. nov., is described. It has a unique combination of traits, some shared with prayomorphs (including two rounded nectophores) and some with clausophyid diphyomorphs (the nectophores are dissimilar, with one slightly larger and slightly to the anterior of the other, and both possess a somatocyst). Molecular phylogenetic analyses indicate that the new species is the sister group to all other diphy- omorphs. A new family, Tottonophyidae, is established for it. Its phylogenetic position and distinct morphology help clar- ify diphyomorph evolution. The function and homology of the nectophoral canals and somatocyst is also re-examined and further clarification is given to their nomenclature. Key words: Siphonophora, Calycophorae, Tottonophyes enigmatica, description, systematics, Tottonophyidae Introduction Calycophorae is a sub-order of siphonophores characterized by loss of the gas-filled pneumatophore, reduction in the number of nectophores, and simplification of cormidial structure (Totton, 1965). Calycophorae includes the majority of described siphonophore species, but key questions regarding the evolution of important traits in this clade remain poorly understood. Phylogenetic analyses of siphonophores (Dunn et al. 2005) have helped frame and, in some cases, answer these questions. However, there is still a need to characterize the morphology of poorly known members of the group in order to understand the implications of these relationships for trait evolution. Here we describe a new species of siphonophore that has a unique set of characters. Its phylogenetic position sheds important light on one of the still uncertain evolutionary transitions within the Calycophorae – the origin of diphyomorph siphonophores. Calycophorae has been divided into two groups, referred to as the prayomorphs and the diphyomorphs (Leloup, 1965; Mackie et al. 1987). Most prayomorphs have two mature nectophores that are rounded, arranged in apposed positions, and are very similar to each other. The diphyomorphs, in contrast, have one or two angular nectophores. (The sole exception is the family Sphaeronectidae, which belong to the diphyomorph grouping as molecular phylogenetic data (Dunn et al., 2005) have shown, that retains a single rounded larval nectophore in the mature colony.) When two diphyomorph nectophores are present, one is located anterior to the other and they are highly differentiated. The diphyomorphs include many small species that are abundant in shallow waters of the oceans and that are among the most frequently encountered siphonophores. Molecular phylogenetic analyses (Dunn et al. 2005) support the monophyly of Calycophorae. They indicate, though, that the prayomorphs are paraphyletic with respect to the monophyletic diphyomorphs. This indicates that features shared by prayomorphs were present in the most recent common ancestor of Calycophorae, whereas unique features of diphyomorphs are derived traits that are specific to the clade. Those analyses also provided two 452 Accepted by B. Bentlage: 7 Mar. 2018; published: 1 May 2018 other important insights into diphyomorph evolution. Firstly, they suggested that Clausophyidae (of which two species were sampled in the analysis) is sister to the remaining described diphyomorph species. Secondly, they provided strong support for placing an undescribed calycophoran species, tentatively labelled "Clausophyid sp 1" (Dunn et al., 2005) due to several similarities shared with described clausophyids, outside the clade containing all other diphyomorphs. This unique species, which we describe here, has a combination of traits, some shared with prayomorphs (including two almost apposed rounded nectophores) and some with Clausophyidae (including the course of the lateral radial canals of the nectophores, and the presence of a somatocyst in both), that help to clarify the evolution of diphyomorph origins. Because of this uniqueness, we establish a monotypic family for it, the Tottonophyidae. Terminology. Before describing the new species, and in the light of recent publications, it is necessary to reconsider the terminology for the various parts of the canal system in the nectophores of both physonect and calycophoran siphonophores. Haddock et al. (2005) gave a detailed analysis of the morphological nomenclature that is applied to the various axes of orientation, and those definitions are adhered to here. However, with regard to the canal system in the nectophores their discussion mainly considered the arrangement in prayid calycophoran siphonophores, and particularly with the differing usage of the term pallial canal, and even somatocyst, by certain authors, e.g. Totton (1965) and Pugh (1992). Since then Mapstone (2009) has given another alternative interpretation and, thus, we feel that it is necessary to re-address the matter in order to provide a consistent framework for describing these structures based on the roles that they play. FIGURE 1. Schematic representations of canal systems of physonect nectophores, in lateral view. A. Halistemma sp. (based on Pugh & Baxter, 2014, Figure 14 in partim), B. Bargmannia elongata (based on Pugh, 1999b, Fig. 2C), C. Apolemia uvaria (based on Totton, 1965, Fig. 15). lrc. lateral radial canals; mca, mcd. ascending and descending mantle canal, respectively; mll, mlu. lower and upper parts of muscular attachment lamella, respectively; np. nectosomal palpons; nst. nectosomal stem; pce, pci. external and internal pedicular canal, respectively. Pedicular canal: There is little dispute that the main nectophoral canal, the pedicular canal, is homologous in all siphonophores that possess nectophores. This canal was clearly defined by Haddock et al. (2005, p. 705) when they said "The pedicular is considered to be the entire canal that runs from the stem to the hydroecial wall, TOTTONOPHYES ENIGMATICA, A NEW SPECIES OF SIPHONOPHORA Zootaxa 4415 (3) © 2018 Magnolia Press · 453 penetrates the mesogloea, and connects to the radial canals of the nectosac. The portion of the pedicular canal from the stem to the nectophore can be termed the external pedicular canal [pce, Figures 1 & 2], while the portion passing through the mesogloea to the nectosac is the internal pedicular canal [pci, Figures 1 & 2]". Haeckel (1888) appears to have been the first to use the term pedicular canal, although probably it was just a direct translation of Stielkanal or Stielgefäss previously used, on occasion, by 19th century German authors. In all physonect species, the pedicular canal always travels directly from the nectosomal stem to the nectosac of the nectophore (Figure 1), and can be divided into the external portion, between the stem and the nectophore, and the internal portion, within the nectophore, connecting to the radial canals on the nectosac. However, its course is more complicated in some calycophorans. Within the paraphyletic calycophoran family Prayidae, a dichotomy appeared with regard to the presence or absence of what Haddock et al. (2005) referred to as a disjunct portion of the pedicular canal (Figure 2A, pcd). Instead of the pedicular canal running directly to the nectosac, upon entering into the nectophore it first runs longitudinally and posteriorly below the median surface of the hydroecium, for a variable distance, before penetrating into the mesogloea and continuing to the nectosac where it gives rise to the radial canals (Figure 2A). This character basically is coincident with the rough division of the prayine genera into "cylindrical" or "conoid" forms (Pugh & Harbison, 1987), but is more precise. The pedicular canal of the cylindrical forms is disjunct, while in the conoid forms it is direct. Mantle canals: While it is relatively straightforward to define the pedicular canal, the nomenclature regarding the blind canals that can arise from it is considerably more complicated. Haddock et al. (2005, p. 705) noted, "Historically, the term pallial canal has been used to describe a variety of gastrovascular extensions in siphonophore nectophores. In calycophorans, particularly prayines, it has been used to describe various parts of the somatocyst and segments of the pedicular canal, including, perhaps mistakenly, the portion giving rise to the radial canals (Totton, 1965; Pugh, 1992). In physonects, it has consistently referred to the ascending and descending branches of the pedicular canal that run along the proximal surface of the nectophore. It is probable that the pallial canals of physonects are homologous to the somatocyst and descending branch of the pedicular canal in calycophorans. Nonetheless, because of these uncertainties and the many ways that the term has been applied, we have avoided using pallial canal in the present manuscript, and await detailed examination of the homology of these structures between calycophorans and physonects". Thus, Haddock et al. (2005) considered the ascending branch and somatocyst as homologous, and the descending branch as an independent extension of the pedicular canal. FIGURE 2. Schematic representations of canal systems of calycophoran siphonophores. A. Rosacea sp., B. Chelophyes appendiculata. For annotations see Figure 1. h, hydroecium; ml. muscular attachment lamella; pcd, disjunct pedicular canal; so, somatocyst. 454 · Zootaxa 4415 (3) © 2018 Magnolia Press PUGH ET AL. With regard to these definitions, Mapstone (2009, p. 10) considered that "somatocyst along the hydroecium, the name used by Haddock et al. (2005a) for that diverticulum from the pedicular canal which extends anteriorly along the nectophore surface, could be applied only to calycophorans, since physonects have neither a somatocyst nor a hydroecium". She used that reasoning to suggest that it warranted a new name, ascending surface diverticulum, and that there should be a corresponding descending surface diverticulum. However, this nomenclature has its own problems that will be discussed further. Huxley (1859, p. 16) originally defined the term hydroecium, in Praya, as "a sort of chamber or hydroecium, into which the cœnosarc can be retracted, as into a house". He then went on to describe its presence in several other calycophorans, except Hippopodius it appears. However, he did not confine his definition to calycophoran siphonophores but (ibid. p. 80), stated that the overlapping opposed nectophores of the physonect species, Physophora sp. "enclose the cœnosarc in a sort of hydroecial canal". Although the structure of the hydroecium is often an important taxonomic character in calycophoran species, it has little such value in physonects and so the term is rarely applied to them, although there are a few instances such as Carré (1971), Grossmann et al. (2013), and Pugh (2016). As for referring to the ascending canal as being part of the somatocyst then, if this were proved the case, there would be no reason why the homologous canal in physonects should not be referred to in the same way. If we are to avoid calling them pallial canals then do they need new names? If one considers the literature then clearly this is not the case, as was briefly discussed by Pugh & Baxter (2014). The term pallial canal first appeared in Haeckel (1888) and it appears likely that the person who translated the original German text derived it from Mantelgefässe, via the Latin pallium, meaning mantle. The term Mantelgefässe was first used by Leuckart (1853, 1854) who, indeed, was the first person to recognize the presence of these canals in physonect nectophores. Thus Leuckart (1854, p. 322–31) noted that "The vascular apparatus of our nectophores presents a very complete state of development, with mantle and nectosac canals. The former, overlooked by Kölliker, as well as Mr Vogt, loop upwards and downwards in the median plane of the nectophore, as we have already found in Hippopodius and Praya. They arise from the central [pedicular] canal immediately after it enters the mantle and are, above and below, developed in a quite consistent manner". The term Mantelgefässe became widely used by German scientists, e.g. Gegenbaur (1859, Claus (1860), Chun (1891) and Schneider (1896), and was also recognised as an appropriate name by Huxley (1859). The term mantle canals has clear precedence over pallial canals, but it fell into virtual disuse after Haeckel's (1888) Challenger Monograph, although Totton's (1965, p. 35) definition of these canals as "Upper and lower diverticula of the pedicular canal at the point of entry into a nectophore or gonophore" is entirely appropriate. We therefore recommend that the term mantle canals be reinstated to describe those blind canals that can arise from the pedicular canal. The evolution of the mantle canals: While the ascending and descending mantle canal are both blind-ending diverticula that arise from the pedicular canal, their phylogenetic distribution and, therefore, evolutionary history are quite distinct. Cystonects lack a nectosome, but do possess nectophores within their gonodendra. The arrangement of the pedicular canal in these nectophores is not as simple as in physonect nectophores, as the internal part of the pedicular canal does not connect directly with the external part of the canal, but there is an extensive disjunct section running along the median distal side of the hydroecium, connecting the two (see Totton, 1960, Plate XXIV, fig. 6; PRP personal observation). Nonetheless, neither a descending nor an ascending mantle canal is present. Thus, mantle canal evolution would be a feature of the Codonophora, the clade that includes the paraphyletic Physonectae and the monophyletic Calycophorae. The phylogenetic distributions suggest that all ascending mantle canals are homologous to each other, and that all descending mantle canals are homologous to each other. The ascending mantle canal has a much broader phylogenetic distribution than the descending mantle canal. It is found in all physonects that possess nectophores, and is present and variously elaborated in prayomorph calycophorans, as well as some clausophyid species (Fig. 3) (see also Table 3 in Haddock et al., 2005). This distribution is consistent with a single gain of the ascending mantle canal along the stem of Codonophora, making 1. Original text: "Der Gefässapparat unserer Schwimmglocken zeigt eine sehr vollständige Entwickelung, Mantelgefässe und Schwimmsackgefässe. Die ersteren, die sowohl von Kölliker, als auch von Herrn Vogt übersehen sind, verlaufen (Fig. 13) in der Medianebene der Schwimmglocke bogenförmig nach oben und unten, wie wir es schon bei Hippopodius und Praya gefunden haben. Sie entspringen aus dem Centralkanale so gleich nach dem Eintritte desselben in den Mantel und sind oben, wie unten, in ganz übereinstimmender Weise entwickelt." TOTTONOPHYES ENIGMATICA, A NEW SPECIES OF SIPHONOPHORA Zootaxa 4415 (3) © 2018 Magnolia Press · 455 it a synapomorphy of the clade, followed by one or more losses within the diphyomorphs. It is a simple, unbranched caecum in all physonects, but it takes on a much greater variety of forms within the Calycophorae. The descending mantle canal is found in a subset of those siphonophores that have an ascending mantle canal. For the physonects, it is absent in the Apolemiidae and genera of the "dioecious" group of physonects, as designated by Pugh (2006a), but present in his "monoecious" group. Among the calycophorans, it is present in the Prayidae and some Clausophyidae, but absent from all others. Thus this distribution is consistent with a single gain of the descending mantle canal, roughly coincident with the shift from dioecy to monoecy (Dunn et al. 2005; Pugh 2006a), and then a single loss within Calycophorae such that it is missing in the diphyomorphs that are the sister group to the Clausophyidae and to the species herein described. The distribution of the descending mantle canal within the Clausophyidae is more complicated than in any other siphonophore taxa. In two of the genera, Kephyes and Crystallophyes, a descending mantle canal is present in the anterior nectophore, while in the three others it is absent. However, for three of the genera a descending mantle canal is present in the posterior nectophore, while in the genus Clausophyes it is absent (PRP personal observation). The posterior nectophore of the species of the genus Heteropyramis has yet to be described. For the diphyid sub-family Sulculeolariinae, Mapstone (2009) considered that the descending mantle canal had disappeared in both nectophores, while the anterior nectophore retained a remnant of the ascending mantle canal, as Totton (1965) had suggested. However, Haddock et al. (2005) have already noted that this was actually not a canal but scar tissue left after the detachment of the muscular attachment lamella. The same applies to her illustration (see figure 5H of Mapstone, 2009) of Lensia conoidea (Keferstein & Ehlers). The function of the various nectophore canals: The phylogenetic distribution and variation in morphology of the mantle canals may provide some insight into their function. In the case of the pedicular canal, it seems clear that its basic function is to provide nutrition to the muscular walls of the nectosac and to the velum surrounding the ostium; thereby allowing the nectophore to fulfil its function of jet propulsive swimming. The function of the mantle canals rarely has been considered in the past. However, Leuckart (1853, p. 12)1, when first describing them, stated, "The vessels, which are intended for the nutrition of the mantle, develop rather late, after the latter has reached a considerable size". On the other hand, Mapstone (2009, p. 10) stated that: "The present author considers that this 'pallial canal' canal [sic] may fulfill an important function in many species by facilitating the shedding of nectophores during autotomy". Nevertheless, it seems unlikely that a canal should facilitate autonomy, as there is no known mechanism by which its presence would weaken or provide a breakpoint for nectophore attachment. However, Mapstone (2009, p. 10) did draw attention, in her Figure 5, to the "relationship between the nectophoral muscular lamella and the median gastrovascular canal on the proximal surface of the nectophore". We propose that there is a functional connection between the two, in that the canal provides nutrition to the muscular lamella. This refines Leuckart's suggestion that the canals do provide nutrition, but simply to the muscular attachment lamella, rather than the whole mantle. The lamella is a highly contractile structure that plays a critical role in orientating the nectophore during swimming (Costello et al, 2015), and may have a high metabolic activity. Within the full diversity of physonect and calycophoran siphonophores, it is found that most have mantle canals that lie in close proximity to the attachment lamella, throughout most or all of its length. In species with a long lamella, such as Rosacea (see Figures 2A, 3A), these canals are extensive, but in species with a reduced lamella, such as the diphyids, the subtending canals are also reduced. Interestingly, though, the particular canals that lie in close proximity to the lamella vary from species to species. These canals can include the ascending mantle canal, descending mantle canal, disjunct pedicular canal, and even the lower radial canal on the nectosac. Within the Physonectae, as noted above, those species that are included in the "monoecious" group have an attachment lamella both anterior and posterior to the pedicular canal and it is subtended entirely by the ascending and descending mantle canals (Figure 1A). For the species of the "dioecious" group, the attachment lamella is present only anterior to the external pedicular canal and is subtended by the ascending mantle canal, and there is no descending canal (Figure 1B). However, the Apolemiidae is exceptional in that a lamella is present both anterior and posterior to the external pedicular canal but, while the anterior portion is subtended by the ascending mantle canal, the role of nutrition for the part of the lamella posterior to the pedicular canal falls to the lower radial canal that closely subtends it (Figure 1C), as Totton (1965) described. 1. Original text "Die Gefässe, die für die Ernährung des Mantels bestimmt sind, entstehen erst ziemlich spät, nachdem der letztere bereits eine ansehnliche Grösse erreicht hat". 456 · Zootaxa 4415 (3) © 2018 Magnolia Press PUGH ET AL. This pattern of canals paralleling the attachment lamellae is even more apparent in Calycophorae, where both the course of the canal systems and the length of the lamella are more variable (Figures 2, 3). Within Prayidae there is great diversity in the extent of both the ascending and descending mantle canals and the pedicular canal. For the mantle canals, there was generally a reduction eventually to nothing; for the pedicular canal, it mainly concerned the extent of the disjunct portion. Thus, in Rosacea, the attachment lamella is subtended by the ascending mantle, disjunct pedicular, and descending mantle canals (Figures 2A, 3). As Table 3 in Haddock et al. (2005) showed the descending mantle canal was eventually lost and species of only three of the nine known genera of prayine siphonophores retain it. The most extreme reduction in the canal systems occurs in the adult nectophores of Lilyopsis and Gymnopraia species. There, not only has the descending mantle canal completely disappeared, but also there is no disjunct portion to the pedicular canal. In addition, the ascending mantle canal lies entirely within the mesogloea of the nectophore, such that there are no superficial canals subtending the muscular lamella. The tendency for all, or just the anterior part, of the ascending mantle canal to branch off into the mesogloea of the nectophore is common amongst prayine species, occurring in six out of the nine current genera. It can range from a very short inflection, as in Desmophyes haematogaster, that may swell up immensely as in D. annectens (Figure 4); through multiple, more extensive and complexly branched apical inflections, together with transverse branches from the mantle canal itself, as in Praya dubia; to the situation, as discussed above, where the ascending mantle canal lies entirely within the mesogloea. FIGURE 3. Schematics showing the diversity in the arrangement of the various canals in the nectophores (Rosacea and Vogtia) or anterior nectophore (Chuniphyes and a diphyid), and the region of attachment of the muscular lamella (shaded). Annotations as for Figures 1 & 2; abs, "ascending branch of somatocyst". The three canals arising from the internal pedicular canal are, from top to bottom the upper, lateral and lower radial canals on the nectosac, except for the diphyid where the bottom canal is the ostial ring canal as the lower canal is virtual. It remains to be determined whether the disjunct or internal portion of the pedicular canal is always present in diphyomorphs. Modified and adapted from Mapstone (2009) Figures 5 & 6. There is also great diversity in the internal branching of the ascending mantle canal. While the lamella, nectosac, and ostium are the primary muscular structures of the nectophore and likely have the greatest metabolic needs, the region where the ascending mantle canal is found varies greatly in volume and the canal system may co- vary so that no part of this structure is too far from the nutritional supply of a canal (see Mapstone, 2009, fig. 28D). In the Prayidae, there is much variation in this branching but, with one major exception, the branches are usually simple canals and are not swollen. In this regard, Amphicaryon, Rosacea and Craseoa are the simplest, as the ascending mantle canal does not penetrate into the mesogloea, and it is neither branched nor swollen (Totton, 1965). In Maresearsia, the ascending mantle canal is enlarged considerably, but it does not penetrate into the mesogloea. In Desmophyes haematogaster, Mistoprayina, Prayola and Gymnopraia, the ascending canal does penetrate the mesogloea, but it is neither swollen nor branched. In Praya, Stephanophyes, Lilyopsis, Nectopyramis and Nectadamas the canals are branched but not extensively swollen, although their distal ends may be slightly so TOTTONOPHYES ENIGMATICA, A NEW SPECIES OF SIPHONOPHORA Zootaxa 4415 (3) © 2018 Magnolia Press · 457 (Pugh. 1992; Haddock et al., 2005). However, for all siphonophores, there is bound to be at least one exception to the rule and that is the case in D. annectens where the internal extension of the ascending mantle canal is greatly inflated (Figure 4). In some hippopodiid species there is a short apical extension of the ascending mantle canal within the mesogloea, which Mapstone (2009, e.g. Fig. 40) referred to this as an "ascending branch of the somatocyst". We consider that, as in other prayomorphs, it represents an extension of the mantle canal system, with the same nutritive function. Within the Calycophorae, there was a shift from apposed to superimposed nectophores, i.e., one nectophore came to be located to the anterior of the other (see Figure 4 in Mapstone, 2009). This coincided with a reduction in the extent of the attachment lamella and its associated canals. The descending mantle canal, already lost in many prayids, disappeared completely in Hippopodiidae and all diphyomorph calycophorans, with the exception of some members of Clausophyidae, e.g. Kephyes (Pugh, 2006b). The ascending mantle canal also became reduced to the point of total elimination so that in many diphyid species the attachment lamella is only partially subtended by a canal, the disjunct portion of the pedicular canal (Figure 3 Diphyid). However, the area of attachment is relatively small and, presumably, sufficient nutrition is derived from the external and disjunct portions of the pedicular canal. FIGURE 4. Desmophyes annectens Haeckel, 1888. Thus, our basic contention is that the mantle canals arose at separate stages in the evolution of the Codonophora in order to provide nutrition to the muscular lamella attaching the nectophore to the stem and were subsequently lost in some clades where the extent of the lamella was greatly reduced. Among physonect siphonophores, with the exception of the Apolemiidae, the presence of a descending mantle canal is directly associated with the attachment lamella extending posterior to the external pedicular canal. In the Apolemiidae, as Totton (1965) points out, it is the lower radial canal that fulfils the same function. Within the Calycophorae, it would appear that both the ascending and descending mantle canals were originally present and that there was an extensive attachment lamella between the two apposed nectophores extending both to the anterior and posterior of the external pedicular canal. Defining the somatocyst. The term somatocyst has been applied inconsistently by different authors, although it has always been applied only to calycophorans and never to physonects. It has variously been defined as the entire ascending mantle canal system (regardless of structure), or portions of that canal system, or only to the anterior, swollen portion that lies entirely within the mesogloea. Huxley (1859, p. 5) was the first to define the term somatocyst, as "In the Calycophorid ... the proximal end of the cœnosarc dilates a little, and becomes ciliated internally, forming a small chamber, which gives off the ducts, by whose intermediation the systems of canals, which embrace the cavities of the organs of locomotion, are brought 458 · Zootaxa 4415 (3) © 2018 Magnolia Press PUGH ET AL. into communication with the somatic cavity. At its upper end, this chamber is slightly constricted, and so passes by a more or less narrowed channel into a variously shaped sac whose walls are directly continuous with its own, and which will henceforward be termed the somatocyst. The endoderm of this sac is ciliated, and it is generally so immensely vacuolated as almost to obliterate the internal cavity and give the organ the appearance of a cellular mass. "The somatocyst very commonly contains large, strongly refracting globules of an apparently albuminous matter, of precisely the same character as those which may be observed occasionally to pass through the pyloric valves of the polypites, into the somatic cavity; and I do not doubt that the globules result from the accidental accumulation of such products of digestion." Thus Huxley (1859) considered that the somatocyst could function as a storage organ. Although, he also applied the term somatocyst quite broadly, including the longitudinal median, i.e. mantle, canals of a prayid species. Haddock et al. (2005, p. 705) defined the somatocyst as "any blind branch of the gastrovascular system that runs anteriorly from the external pedicular canal at the point it reaches the hydroecial wall. The somatocyst may penetrate into the mesogloea, either immediately or after extending along the hydroecial wall". They noted (ibid) that "this terminology, as opposed to the previous terminology used in prayines, is consistent with that of diphyid and abylid calycophorans, in the sense that our definition of a somatocyst accommodates the way that term is usually applied in those groups". Mapstone (2009, p. 12) disagreed with Haddock et al. (2005) in that she did not regard her "ascending surface diverticulum" to form part of the somatocyst, and considered that a new definition of the somatocyst was required; namely "those diverticula from the pedicular canal or its ascending surface diverticulum that penetrate the mesogloea". This definition, thus, includes all the internal branches of the mantle canals found in some prayomorph species as well as the somatocyst in diphyomorphs. For the prayomorphs, Mapstone (2009, e.g. fig. 6) then refers to the anterior canal, penetrating into the mesogloea, as the "ascending branch of the somatocyst". This would, thus, appear to indicate that the somatocyst and the ascending surface diverticulum are one and the same, thereby reflecting the original definition given by Haddock et al. (2005). The great variation in the structure of the ascending mantle canal system makes such distinctions difficult to define and apply universally, and they may be somewhat arbitrary. Thus, we have decided to refine our definition, somewhat along the line suggested by Mapstone, and now consider the somatocyst as an extended portion of the ascending mantle canal that lies entirely within the mesogloea of only diphyomorph calycophorans. We chose this definition because it is consistent with most historical usage; it is relatively simple; and it probably corresponds to novel evolutionary function (see below). For all other diphyomorphs, where the descending mantle canal and the portion of the ascending mantle canal along the hydroecium are absent, the somatocyst arises directly from the external pedicular canal that penetrates into the anterior, or larval in the case of the Sphaeronectidae, nectophore. The arrangement in Clausophyidae is more complicated. In some clausophyid species, the ascending mantle canal runs along the hydroecium and then penetrates into the mesogloea as the somatocyst, in both the anterior and posterior nectophores (PRP personal observation). However, for the posterior ones, although it is difficult to be certain of the point of insertion of the external pedicular canal, it appears that the latter canal gives rise directly to the somatocyst and to its disjunct portion within the nectophore. Thus, the ascending mantle canal along the hydroecium already has been suppressed completely. For four of the clausophyid genera the descending mantle canal is still present, but it is absent from Clausophyes species. Similarly, a short ascending mantle canal along the hydroecium appears to be retained in the anterior nectophore, with the exception of Clausophyes species, where the somatocyst arises directly from the external pedicular canal. However, for Chuniphyes moserae, the ascending mantle canal along the hydroecium is very short. While nourishment of the lamella provides a clear hypothesis for the function of the canals that branch from the ascending mantle canal in prayomorph siphonophores, it does not provide a functional explanation for the ascending structures that penetrate into the mesogloea in many diphyomorph calycophorans, which are unbranched, but vary greatly in the degree to which they are swollen. There are at least two possible functional implications: They may provide for nutritional storage, as previously hypothesized by Huxley (1859). Alternatively, they may serve for buoyancy control. Buoyancy would be further refined into two categories – gross buoyancy, which could influence location in the water column, and trim, which could influence the resting orientation of the nectophore in the water column. For instance, the nectophores of Lensia conoidea, in situ, are TOTTONOPHYES ENIGMATICA, A NEW SPECIES OF SIPHONOPHORA Zootaxa 4415 (3) © 2018 Magnolia Press · 459 seen to be horizontal when at rest, with the somatocyst in the anterior nectophore uppermost. However, in several diphyids the somatocyst is greatly reduced and it would be difficult to ascribe a buoyancy function to it. Buoyancy in the prayomorph calycophorans is mainly provided by the large volumes of mesogloea, and active exclusion of heavy sulphate ions (Bidigare & Biggs, 1980) also would help to reduce the overall of the nectophore. Thus, we suggest that a change in function of the internal extension of the apical end of the ascending mantle canal occurred during the evolutionary transition from apposed to superimposed nectophores. Description Family Tottonophyidae fam. nov. Tottonophyes gen. nov. Monotypic genus for Tottonophyes enigmatica sp. nov. Tottonophyes enigmatica sp. nov. Diagnosis. With two rounded, almost spherical, dissimilar nectophores with one slightly larger and slightly to the anterior of the other. Extensive nectosac and hydroecium occupying most of each nectophore. No flaps on lateral walls of hydroecium; no mouth plate on posterior nectophore; ascending and descending mantle canals present, with internal somatocyst forming a conical structure. No disjunct portion of the pedicular canal in either nectophore. Material examined. Four specimens of Tottonophyes enigmatica sp. nov. have been examined. All were collected by the MBARI ROVs, namely Tiburon (T), Ventana (V), or Doc Ricketts (DR), whose dive details are shown in Table 1. They were preserved in 5% buffered formalin in sea water. Unfortunately, only one of the nectophores from the DR0105 and DR0552 dives was preserved. A fifth specimen (T0398) was collected and photographed in the ship-borne laboratory, then part was frozen for molecular analyses. Five other specimens (unemboldened) in Table 1 have been identified from in situ frame grabs. The specimen from Tiburon Dive 897 has been designated as the holotype, and will be deposited at the Smithsonian National Museum of Natural History. The remainder will be deposited with the Peabody Museum of Natural History at Yale University. TABLE 1. List of observed specimens of Tottonophyes enigmatica sp. nov. For the specimens collected (emboldened), "D" means that they were collected with a "Detritus sampler". * Specimens not examined for present study. ROV Depth of collection/ Date Latitude Longitude Water Depth (m) Dive Observation (m) T0364 3092 2–Oct–2001 36°20.25’N 122°53.96’W 3496 T0398–D1* 2836 23–Feb–2002 36°34.37’N 122°31.20’W 3036 T0438 2368 12–Jun–2002 36°34.81’N 122°25.18’W 2421 T0762 3384 16–Nov–2004 36°19.25’N 122°53.29’W 3451 T0897–D4 1342 23–Sep–2005 36°39.31’N 122°09.52’W 1356 V3005–D8* 963 15–May–2007 36°41.90’N 122°03.93’W 1741 DR0026 3281 27–May–2009 36°06.88’N 122°45.09’W 3415 DR0105–D3 2469 13–Dec–2009 36°04.08’N 122°17.89’W 2476 DR0327 3539 4–Dec–2011 35°55.97’N 122°55.99’W 3612 DR0552–D1 2222 20–Nov–2013 36°4.11’N 122°17.77’W 2481 DR0965–D1 910 14–Jun–2017 36°41.96’N 122°01.98’W 1555 460 · Zootaxa 4415 (3) © 2018 Magnolia Press PUGH ET AL. Description: An in situ frame grab from Doc Ricketts Dive 105 (Figure 5A) shows that the larger of the two nectophores of Tottonophyes enigmatica sp. nov. is only partially superimposed over the smaller one. This superimposition (i.e. positioning of one nectophore to the anterior of the other) is much greater than seen in prayomorphs (Figure 2A), where they are apposed, and much less than for other diphyomorphs, even the clausophyids (Figure 5B, C). However, unlike prayids, there are clear differences in the two nectophores of T. enigmatica sp. nov., and for ease of comparison with the diphyomorph species the nectophores will be referred to as anterior (larger) and posterior (smaller). The holotype specimen from Tiburon Dive 987 was chosen as the best preserved of all the specimens collected up until the time that this description was begun. Very recently a further specimen was collected during Doc Ricketts Dive 965. This proved to be the largest specimen so far collected, with the anterior nectophore measuring 13.5 mm in length and 9 mm in width, and the posterior one 11.5 by 7 mm. Nonetheless, the general structure of both nectophores showed no major variations from the norm, and there was no evidence there to suggest the unlikely eventuality that we were dealing with a different species. However, there appeared to be a minor difference in the special nectophores in the cormidia that is discussed below. Anterior nectophore: (Figures 6 & 7). The larger, anterior nectophore of the holotype specimen measured, in its preserved state, 7.6 mm in height, 7.6 mm in width and 4.6 mm in breadth. The proximal two-fifths of the nectophore was almost entirely occupied by the hydroecium (Figure 7), which was open at its lower end, but at its upper end its median wall curved over and ran to the proximal wall of the nectophore a short distance from its apex. The distal three-fifths of the nectophore were mainly occupied by the extensive nectosac; its ostial opening, which included a large velum, was directed obliquely toward the upper side of the nectophore, such that outer side of the nectophore was only about two-thirds the height of the inner or proximal side. All four radial canals arose from the short and direct internal pedicular canal. In its preserved state the upper part of the nectosac was more extensive laterally than in the mid-line, such that the upper radial canal ran through a median depression. However, in life this depression was less marked (Figure 6A). The upper and lower radial canals ran directly to the ostial ring canal. The lateral radial canals, however, followed a course somewhat similar to that found in Clausophyes species. From their origin, they extended across the nectosac, with only a very slight upward curve. Close to the outer side of the nectosac they curved through 90° and ran directly toward the ostium, before again bending through 90° and running back across the nectosac, with a slight upward curve as they approached the proximal side. A further curve through 90° again directed them toward the ostium, but before reaching it, they curved through c. 60° and then ran obliquely down to join with the ostial ring canal. FIGURE 5. A. In situ Frame Grab of Tottonophyes enigmatica sp. nov. taken during Doc Ricketts Dive 105; Specimens of (B) Kephyes hiulcus Grossmann & Lindsay, 2017. and (C) Clausophyes moserae from Pugh, 2006b. Scale bars 2 mm. TOTTONOPHYES ENIGMATICA, A NEW SPECIES OF SIPHONOPHORA Zootaxa 4415 (3) © 2018 Magnolia Press · 461