Revue suisseZool. Tome99 Fase.4 p. 755-770 Genève,décembre 1992 Evolutionary aspects of development, life style, and reproductive mode in incirrate octopods (Mollusca, Cephalopoda)* by Sigurd v. BOLETZKY ** With7figures Abstract The incirrate octopods are defined as amonophyletic group by anumberofcharacters unknown inthecirrate octopods orin othercephalopods. Thebiologicallymost significant incirrate feature is the incubation of eggs by the female. The morphological and behavioural characters underlying this special mode are analysed with regard to developmentalprocesses andtheirmodificationrelatedtolife styleevolution. INTRODUCTION 1. Common inshore species ofthe genus Octopus Lamarck, 1798 are popularmodels in comparative invertebrate biology (Young, 1971; Wells, 1978); in such acontextthey can be viewed as 'typical cephalopods'. When attention is focussed on diversity within the class Cephalopoda, however, the common octopuses turn out to be rather 'special' cephalopods. This observation raises the question ofrespective systematic positions: what is special aboutwhich groupofoctopods ? 1.1. SYSTEMATICS Within the cephalopod subclass Coleoidea Bather, 1888 (=Endocochleata, Dibran- chiata), the order Octopoda Leach, 1818 is a well defined taxon. It contains two suborders, the Cirrata Grimpe, 1916 (better known as 'finned octopods') and the Incirrata TravailprésentéàZoologia92. *C.N.R.S.,URA 117,LaboratoireArago,F-66650Banyuls-sur-Mer(France) 756 SIGURD V. BOLETZKY Grimpe, 1916, to which the common octopuses belong. Young (1989) proposed a new classification, with an infraclass Octobrachia containing two orders, the Cirroctopoda (=Cirrata) and the Octopoda (=Incirrata). For practical reasons, the old classification is used here, especially to avoid confusion between Octopoda Leach, 1818 and Octopoda Young, 1989 (cf. quotationfrom Young, loc. cit. inDiscussion). 1.2. Phylogeneticbackground To appreciate the evolutionary significance of incirrate characters with regard to functional adaptation andphyletic conservation ofpatterns, the incirrates mustofcourse be viewedincomparisontotheirsupposedlyclosestrelatives,thedeap-seacirrateoctopods. The respective positions ofthe fossil Proteroctopus ofthe Middle Jurassic and Palaeoctopus of the Late Cretaceous are not discussed here, because the available morphological data are insufficient (see Engeser, 1988 for a review). The characters of living octopods can nevertheless be scrutinized in the greater framework ofcoleoid phylogeny, starting out from the Vampyromorpha (cf. Young, 1989). This taxon was formerly included in the Cirrata because ofthe great similarity in arm morphology. When the so-calledretractile filaments of Vampyroteuthh were recognized as an additional pair of rudimentary arms, the Vampyromorphaweremadeanorderofitsown(Pickford, 1939, 1949). The evolutionary history of endocochleate cephalopods can be traced back to the lower Devonian (Bändel et al. , 1983; Bändel & Boletzky, 1988). For the present purpose it is sufficient to consider the extant coleoids in relation to those fossil coleoids that had ten arms ofsimilar length andstructure, a crucial feature that Naef (1923) used for the definition of a hypothetical 'Protodibranchus'. This arm pattern disappeared with theextinction ofthe belemnites attheendofthe Cretaceous. The 'belemnoid' arm crown morphology is important for our understanding of coleoid phylogeny because it is the only one from which one can derive the respective patterns of 1) the decapodan cuttlefishes and squids, which have thefourth arm pair modified as tentacles, and 2) a group provisionally named 'Vampyropoda' (=Octopodi- formes Berthold & Engeser, 1987; name preoccupied) to include the Vampyromorpha Pickford, 1939 with their modified second arm pair and the Octopoda Leach, 1818 (lack ofprobably the second arm pair). The importantpointhere is thatthese two modifications must have occurred independently, i. e. at two different speciation events, because a simultaneous occurrence of mutually exclusive modifications is inconceivable (Fig. la). Prerequisite to this phylogenetic deduction is the existence ofunambiguous identities and positional relationships of brachial appendages allowing one to recognize modifications for a given pair of arms (see Boletzky, 1992 for a review). These criteria are concerned withthe 'integration level' ofdistinct appendages; they donotinvolvethe specialisationsat the next lower level which comprises the armature of the arms and tentacles (suckers, hooks,cirri). Given the situation described above, any discussion of phylogenetic systematics of the Coleoidea has to cope with (only) two possibilities: either the decapods and the belemnoids are sister groups for which the 'Vampyropoda' are the outgroup, or the 'Vampyropoda' and the belemnoids are sister groups, the decapods being the outgroup (Fig. lb). The question ofwhich one ofthese arrangements is true lies outsidethe scope of thispaper. . EVOLUTIONARYASPECTS OFOCTOPODS 757 INCIRRATE CHARACTERS 2. In addition to the absence of brachial cirri warranting the name of the taxon, and hectocotylization ofoneoftheventro-lateral arms, theIncirratashow aseries ofcharacters unknown in the Cirrata or in any other coleoid cephalopod (cf. Naef, 1923, 1928); the mostconspicuousofthese are: 1 absenceofmuscularfins ('finlessness') 2. presenceofKölliker's organs inhatchling skin 3. partial modificationofeggcase (chorion stalk) 4. reducedencapsulationofegg case (chorion stalkonly) 5. egg-carebehaviour(orovovivipary) inthe female. Although none of these characters is known in any other cephalopod, the question remains whether they are uniquely derived (apomorphic) characters of the Incirrata, or whether they (or some of them) could be autapomorphic characters of the octopodan ancestorthat were subsequently eliminated in the Cirrata (so they wouldbe plesiomorphic at the level ofthe Incirrata). So farnothing seems to indicatethatthe absence ofthe above characters inthe Cirratacouldbetheresultofsuchan elimination. An equally important question is how closely related these characters are to one another. Here one has to consider several variables; a behavioural one, namely the post- hatching life style as compared to the adult mode of life, and two morphometrical variables, egg size and body proportions. The behavioural features necessarily lead to the question ofhow to define the ancestral life style from whichthe modes ofliving incirrates must be derived. Before this question can be addressed, the incirrate characters listed abovehavetobe scrutinizedin somedetail. Ad 1. The absence of muscular fins in the Incirrata is not total if embryonic development is taken into account. Fin rudiments do appearduringorganogenesis, inclose positional relation to the shell sac as is typical forthe coleoids (Fig. 2). The incirrate shell sac is very small from the beginning; it becomes drawn out laterally during early development (AppelOf, 1898). The resulting transverse tube finally splits into two indépendant tubes which become embedded in the musculartissue ofthe mantle (and can finally disappear, as inArgonauta). During this shell sac differentiation the fin rudiments gradually smooth out (Naef, 1928). Whatmight be taken as rudimentary fins in preserved hatchlings viewed in the scanning electron microscope are fixation artefacts due to shrinkage (Fig. 3); although the position of these ostensible 'buds' corresponds to the location ofthe shell stylets, notraceoffintissue is histologically detectable. The character 'absence ofmuscularfins' could be rephrasedtoemphasize the rudimentation ofthe whole fin-shell complex, assuming homology of fin rudiments in Cirrata and Incirrata (Boletzky, 1982 a). When viewed against the pattern ofan unpaired transverse shell sac and associated fins (as present in cirrates), the bipartite shell sac of incirrates suggests a correlation between two apomorphic characters, namely subdivision of the shell sac and truncation offin differentiation, i. e. 'finlessness'. Ad 2. The majority of incirrate hatchlings have special tegumentary organs; they were observed in Argonauta embryos by KOlliker (1844) and fully described in other incirrates by Querner (1927) and more recent authors (e. g. Fioroni, 1962) (Fig. 3). The absence ofKölliker's organs in Octopus briareus (Fig. 7) and O. maya (Boletzky, 1973) 758 SIGURDV. BOLETZKY can be viewed as the result of total suppression in specifically modified integument morphogeneses. So far no trace of these organs has been found in embryos of cirrate octopods (Boletzky, 1982b). Ad 3. During vitellogenesis, cephalopod oocytes may take on a markedly elongate form, but it is only in the Incirrata that this elongation leads to the differentiation of a distinct chorion stalk (Figs 4, 5). The elongate form of the incirrate chorion is also meaningful withregard to embryonic movements that occur in all incirrates so farstudied, with the exception of Argonauta this feature could be a side effect of the primary modification oflate oogeneticproces;ses (Boletzky &Fioroni, 1990) (Fig. 4). Ad 4. In the incirrates the 'cement' secreted by the oviducal glands (Froesch & Marthy, 1975) 'encapsulates' only part of the chorion stalk (Fig. 5). This 'partial egg encapsulation' contrasts withthecomplete encapsulationofthecirrate eggs; itthus appears asanapomorphic characterlinked with the formation ofachorion stalk (character3). Ad 5. Although visual stimulation by otheregg masses can induce spawning inmany coleoid cephalopods, post-spawning egg care exists only in the incirrate octopods (Boletzky, 1986). This unique protective behaviour must be related to the absence of protective encapsulation (character 4). In other words, characters 3, 4 and 5 are clearly connected, forming an apomorphic complex of features named the "incubating mode" of reproduction. In the pelagic genus Ocythoe, incubation exists in the 'pure' form of ovovivipary (Naef, 1923). For two families, the Alloposidae and the Amphitretidae, egg- care isnotyetdocumented (Hochberg, pers. comm.). Before approaching the question ofpossible relations between the above 'incubation complex' and the characters 1 and 2 (see C), it is necessary A. to review life styles and morphometries at different stages of the incirrate life cycle, and B. to see whether characters 1 and2 are correlated. A. LIFESTYLES AND MORPHOMETRICS A.l. Life styles a) Adultlifestylesintheincirrata Eight incirrate families are recognized ifthe Idioctopodidae Taki, 1962 are included in the Amphitretidae Hoyle, 1886 (cf. Hochberg et al., 1992; disregard the erroneous statementin Boletzky, 1978-79,p. 107). Only the Octopodidae Orbigny, 1840 are clearly benthic at the adult stage. R. E. Young (pers. comm. to Hochberg et al., 1992) suggests that adult Alloposus mollis Verrill, 1880 (of the monotypic family Alloposidae Verrill, 1882) may also be benthic. The remaining six families are pelagic; these are the Argonautidae Tryon, 1879, the Tremoctopodidae Tryon, 1879, the Ocythoidae Gray, 1849 (these three families were grouped with the Alloposidae in a tribe called Argonautida by Robson, 1932), the Vitreledonellidae Robson, 1930, and the two "ctenoglossan" families Bolitaenidae Chun, 1911 andAmphitretidae Hoyle, 1886. EVOLUTIONARY ASPECTSOFOCTOPODS 759 In all these families, reproduction takes place in midwater, and the eggs apparently remain with the female until the young hatch out. The most elaborate mode ofegg care is achieved by the female Argonauta which produces a calcified "brood shell". However, along withhousingtheeggmass, this pseudoconch serves as afloater; theanimalkeeps an airbubble in the apex and thus obtains neutral buoyancy (Boletzky, 1983). Moreoverthe brood shell supports the brachial membranein afooddetective function (Young, 1960). A much simpler form of egg carrier is produced by Tremoctopus ; as in Argonauta, the calcified structures are secreted by the dorsal arms (Naef, 1923). In Eledonellapygmaea (family Bolitaenidae), the whole arm crownofthefemaleforms abroodchamber(Young, 1972). Male sexual behaviour in pelagic incirrates can be only partly inferred from the structure ofthe copulatory arm (hectocotylus). Inthe "Argonautida" sensu Robson (1932), the morphologically and morphometrically extreme differentiation of the hectocotylus seems correlatedwiththecapacitytoautotomize. In the benthic Octopodidae, females always spawn on the bottom. Generally single eggs or egg strings are cemented to the wall or ceiling ofthe den occupied by the female. In a few octopodid species, the females carry egg masses loose and thus can move about while brooding theeggs (see Hochbergetal., 1992). b) Post-hatchinglifestyles intheincirrata As far as is known (cf. Hochberg etal., 1992), thejuveniles ofpelagic families live in midwater (including the Alloposidae; see above). In the benthic Octopodidae, the representatives of the subfamily Bathypolypodinae Robson, 1931 probably stay on the bottom throughout their life (cf. Bathypolypus arcticus, as observed by O'Dor & Macalaster, 1983). If the new arrangement proposed by Voss (1988) is accepted, the new subfamilies Graneledoninae and Pareledoninae are entirely holobenthic (Hochberg., pers. coram.). The subfamilies Octopodinae Grimpe, 1921 and Eledoninae Gray, 1849 include numerous species characterized by the same 'holobenthic' mode as Bathypolypus, while others have a planktonic post-hatching phase; the mode oflife ofthe latter species can be named 'merobenthic'. Their young animals are actively foraging carnivores that feed on both living planktonic prey and drifting food items (facultative scavenging). They generally remain in midwater until they have grown larger. In some species, newly- hatched animals show temporary settling between phases ofactive swimming (Boletzky, 1977). A.2. Morphometries a) Bodyproportionsofincirratehatchlings The hatchlings of pelagic incirrates are characterized by short arms (generally less than 1/3 oftotal length) with few suckers. This feature again appears in the newly hatched animals ofmerobenthic octopodids (Fig. 6), although inthe largerhatchlings each ofthese relatively short arms maycarry up to 15 suckers. In contrast, the hatchlings ofholobenthic octopodids have arms at least as long as the rest of the body, with more than 20 suckers perarm (Fig. 7). 760 SIGURDV. BOLETZKY The body proportions of planktonic hatchlings gradually change due to the positive allometric growth of the arms. In the young merobenthic octopodids, body proportions thus become similar to those of the 'crawl-away' hatchlings of holobenthic species. In Octopus vulgaris (and probably in the majority of merobenthic octopodids) the young animals, having reached these body proportions, gradually change from continuous swimming to the adult-type bottom life, which includes only occasional excursions into the water column (Itami et al, 1963). This drastic change contrasts with the condition of pelagic incirrates, in which juvenile arm growth is not accompanied by a thorough modificationoflife style. b) Incirrateeggsizes andhatchlingfeatures Withinthe Incirrata, the size ofasingle ovum varies from 0.8 mm inArgonauta spp. to 35 mm in Graneledone sp. (Hochberg et al, 1992). Among the pelagic incirrates, the variation spans only from 0.8 to about 4 mm, however. In contrast, egg sizes vary from mm mm about 1.5 to 35 in the Octopodidae. Robson (1932, p. 25) once expressed egg lengths as percentages of adult mantle- lengths, buthis erroneous egg index forEledone cirrosa preventedhim from realizing the great difference between Eledone moschata (egg length ca 15% of adult mantle length) and E. cirrhosa (ca 5%; cf. Fig. 5); adults of the two species are similar in size. This index becomes particularly interesting when absolute egg sizes are similar among species with very different adult sizes. There are several octopodid species that produce eggs measuring about 5 to 8 mm; in the larger species the embryos become planktonic hatchlings with short arms and less than 15 suckers per arm, whereas in the smaller species embryos ofthe same size end up as benthic hatchlings with long arms and more than 20 suckers per arm (cf. A.2.a). In fact, the ostensibly 'intermediate' egg sizes fall under the same categories as the 'very large' and the 'very small' eggs and are distinguishable by the relative egg size, or egg index (Boletzky, 1974, 1977). An index smaller than 10% is indicative of the merobenthic mode, whereas an index greater than 10% reflects holobenthic conditions (occasional behavioural peculiarities innewlyhatched animals notwithstanding). An exception is Octopusfitchi, a very small species in which the eggs (ca 5 mm) are large relative to the adult mantle-length (ca 30 mm: egg index ca 16); the arms are stout and almost as long as the rest of the body like in hatchlings of holobenthic species, but each arm carries less than 17 suckers, and the post-hatching life style isclearly planktonic (Hochberg etal., 1992). B. CORRELATION OFCHARACTERS 1 AND 2 a) FlNLESSNESS ANDTHEBENTHICLIFESTYLE That the incirrates lack fins was long interpreted as a result ofadaptation to benthic life (Naef, 1923). The absence offins in the pelagic incirrates was then naturally viewed as a condition conserved from a finless benthic ancestor, assuming that fins lost in that ancestor were not "reinvented" in its pelagic descendants. However, a causal relationship between the benthic life style of an octopus and the absence of muscular fins has never been shown to exist. In fact, cuttlefish and sepiolid squids demonstrate that fins may be indispensable even on the bottom as can be seen when these animals bury in soft EVOLUTIONARYASPECTS OFOCTOPODS 761 substrates: at the outset of burying they can remain on the spot only because the fin movements counteract the propulsive effect of the funneljet by which substrate particles are blown up. Fins can be expected to have disappeared from the morphogenetic program only ifthey were incompatible with the functional morphology corresponding to a given life style. There is no indication ofsuch incompatibility in relation to the benthic life style as itappears inthe Octopodidae. b) FlNLESSNESSANDTHEPELAGICLIFESTYLE Among the pelagic incirrates, one condition may appear incompatible with the presence offins; this is the presence ofa brood shell in femaleArgonauta. However, one cannot reasonably assume that this highly elaborate female structure represents an ancestral incirrate condition. A feature really incompatible with the presence offins could be character 2 ofour list, i. e. presence ofKölliker's organs in thejuvenile skin. The way these organs function, especially when they evaginate and spread the setal tufts, suggests that they would interfere with fin activity if fins still existed along with them. Provided that tuft spreading occurs under higher nervous control and has a parachute effect in midwater when the animal remains motionless (Boletzky, 1978-79), the establishment of Kölliker's organs can be considered in relation to the pelagic life style. At the level ofthe incirrate ancestor, this of course holds only for small body sizes at which the tufts can generate enough drag to slow sinking. In otherwords, the formation ofKölliker's organs is likely to reflect an originally juvenile adaptation to pelagic life. Finlessness thus appears as the obligatory counterpart of the juvenile 'setaceousness', in other words characters 1 and2ofourlist areprobablycorrelated. C. CORRELATION OFCHARACTER COMPLEXES Nothing so far mentioned provides an indication of any relationship between the complex ofcharacters 1 and 2 and the complex ofcharacters 3, 4 and 5 ofourlist. Such a link appears only when hatching is considered. In the merobenthic and pelagic incirrates, Kölliker's organsplay anessential, though passive, role during hatching (Boletzky, 1978- 79). The setal cores ofthese organs provide a "shingle" structure to the hatchling skin and thus prevent its slipping back into the chorion when the animal makes the stretching movements necessary to work itself through the hatch opening (which is produced by enzymes released from the hatching gland). Notwithstanding exceptions like Scaeurgus unicirrhus where short arms are used during hatching (Boletzky, 1984), the role ofthe setal cores seems essential in the small young having very short arms that remain passive during hatching (in contrast to the holobenthic octopodids where the crawl-away hatchlings always use their long arms to work themselves out of the chorion). However, the shingle structureofthe skin is effectiveonly ifthehatch openinghas asolidedge. This condition isfulfilledbytherelativelythick, stiffchorionofincirrateeggs. Another question is whether this particular function during hatching is the primitive function of Kölliker's organs. The complex structure of these organs, especially the elaborate musculature that permits repeated spreading and retraction of the tufts during post-hatching life, and the fact that the organs cover also the arms where they are not needed for hatching, suggest that their function during hatching is a secondary adaptation 762 SIGURDV. BOLETZKY superimposed on a primary function related to the post-hatching mode of life. A prerequisite of this secondary adaptation must have been the modification of the encapsulationprocess, which changedfrom thecomplete encapsulation seen inthecirrates topartial encapsulation ofthe egg. This evolutionary transformation is conceivable only in combination with a special timing of egg release allowing the follicular chorion attachment to be drawn out into a distinct chorion stalk. Internal fertilization, a likely prerequisite, was already achieved in the octopodan ancestor, as demonstrated by the Cirrata (cf. Villanueva, 1992); Vampyroteuthis appears to have external fertilization similartothedecapods (Pickford, 1949). DISCUSSION 3. Starting from the feature 'finlessness' through the related 'setaceousness' and its implications in both the post-hatching life style and the hatching mechanism, our survey arrives at the question of the evolutionary origin of the incirrate mode of reproduction. Considering the constraints placed on egg shaping and timing ofegg release, attention is naturally drawn to ovovivipary, the special incubation mode ofOcythoe. Could this mode represent the primitive condition from which the post-spawning egg care was derived? Is the inverse process more likely? Or is the 'intermediary' condition ofArgonauta, where eggs are released only after the first cleavage stages (Naef, 1928), closer to the primitive conditionfrom whichthe othertwo were derived? These questions inevitably raise the problem of the ancestral life style under which incubation became established. Given that seven of the eight living incirrate families are pelagic, it appears likely that the 'most generalized' life style represents the ancestral condition. But this hypothesis remains very vulnerable as long as it is only based on the respective numbers ofextant families representing the pelagic orthe benthic life style. An indication supporting the above hypothesis could be the existence of a pelagic juvenile phase in many octopodids (merobenthic species). This juvenile phase is likely to be a conserved feature that stems from a pelagic ancestor. Advantages of this conservation could have been greater availability of small prey animals in midwater (Boletzky, 1977, 1981) andlow selectivepressure in arelatively 'simple' open waterenvironmentwherethe limited behavioural repertoirs ofvery smalljuveniles suffice (Boletzky, 1987). Thus the planktonic juvenile phase would have been eliminated in the holobenthic species. It is indeed easier to imagine an evolutionary parallelism resulting from convergent suppressions of the pelagic phase than the inverse, namely independently emerging pelagic juvenile phases. With the latter hypothesis, it would be particularly difficult to explain why the planktonic hatchlings tend to be so similar, and why they resemble so closelythose ofthe pelagic incirrates. One may of course argue that perhaps the pre-octopodid ancestor was already characterized byjuvenile life style switching; this wouldhave allowedthe holopelagic life cycle of the majority of incirrates to emerge through a paedomorphic 'abbreviation' by suppression of the ancestral adult mode. This could be the hypothesis underlying the comment of Young (1989, p. 235-236) on a cryptic incirrate character related to the receptor system ofthe statocyst: "The division ofthe crista into nine sections is a unique apomorphic feature of the order Octopoda; it is not present in Cirroctopoda, which presumably neverpossessed it. Thefeature was possiblydevelopedtoprovide forthe wide EVOLUTIONARY ASPECTS OFOCTOPODS 763 range offrequency of turning during walking and swimming. It is surprising to find that the crista is still so divided in all the pelagic octopods examined". The question can of course be reversed: is it surprising to find that the crista is so divided in all the pelagic incirrates ? Not if one assumes that the benthic octopodids are derived from a pelagic ancestor. Efforts should now be concentrated on the identification of sister group relationships within the incirrates. Dothe octopodidshavean immediatecommon ancestor with one ofthe other incirrate subgroups, or is the octopodid lineage derived from a basic dichotomy sothatthe Octopodidae werethe sistergroupofall otherincirrates ? Two variants of a peculiar behavioural feature in some pelagic incirrates deserve special attention. One is the use of the own brood shell as a buoyant device by female Argonauta (cf. 2.A.l.a), the other is the use that male Ocythoe make of empty tests of doliolids and salps as drifting 'homes' (Naef, 1923). The great similarity of these behavioural patterns suggests that the typical arm posture ofabenthic octopodid sitting in its denis homologous totherespective attitudes offemaleArgonauta and male Ocythoe in their pelagic 'homes'. It is conceivable that the behaviour pattern corresponding to such a 'rafting' mode of life provided the initial condition for the establishment of an adult benthic mode. The inverse process seems conceivable only if the supposed benthic ancestoralreadyhadaplanktonicjuvenilephase. In conclusion, the most generalized life style in incirrates is characterized by active swimming and drifting; ontogenetically this is an elaboration ofa pelagicjuvenile phase. This phase has probably been eliminated in many species of the benthic family Octopodidae. To derive the wide variety of incirrate modes from a holobenthic ancestor, through repeated 'invention' of the pelagic juvenile phase, seems rather problematic. Egg incubation in incirrates may thus be surmised tohave emerged in the adaptive context ofa bentho-pelagic orpelagic life style. SUMMARY This paperreviews the common features ofthe octopodan subgroup Incirratafrom an evolutionary point ofview,raising questions offunctional adaptation andco-adaptation of morphological and behavioural characters. The most conspicuous difference between incirrate octopods and other cephalopods is the 'incubating mode' of reproduction (post- spawning egg care or ovovivipary). As the cirrate octopods, which are the likely sister group ofthe incirrates, show no signs of incubation, the evolutionary origin ofthis novel mode ofincirrate reproduction can only be 'reconstructed' through careful weighing ofthe relative importance of characters that are more or less closely related to reproduction. Developmental features provide particularly interesting cues (e. g. truncation of fin development, formation of special tegumentary organs) allowing one to approach the question of the ancestral life style from which the pelagic and benthic modes of extant incirrates mustbederived. Acknowledgments I sincerely thank Dr. F. G. Hochberg (Santa Barbara Museum) forhis critical reading ofthemanuscriptandforhis stimulating suggestions. 764 SIGURDV. BOLETZKY Decapoda Vampyropoda (Vampyromorpha + Octopoda) Belemnoidea ext. Decap. Belemn. Vampyrop. Decap. Belemn. Vampyrop. Fig. 1 a. PhylogeneticrelationshipsbetweenthelivingcoleoidgroupsDecapoda(fivearmpairs,fourthpair modified)andVampyropoda(fivearmpairs, secondpairmodified)viatheextinctBelemnoidea(five arm pairs without distinct modifications). The trichotomy is shown unresolved in terms of sister group relationships, b. The two conceivable sister group relationships (the theoretical third one, supposing Decapoda + Vampyropoda with Belemnoidea as the outgroup, is inconceivable formorphologicalreasons,asexplainedinthetext). Fig. 2 Lateral view ofalive embryo ofOctopus vulgaris in its chorion, the large outeryolk sac (atright) andthechorionstalk(atleft)arenotshown.Atthisadvancedorganogeneticstage(stageXIIofNaef, 1928),onecanrecognizetheorgan complexes surrounding thedarkyolkmass: avoluminousbuccal mass (b), the stubby arms (a), the funnel tube (f), and the cap-like mantle with the fin rudiments (arrow)overlyingtherudimentaryshellsac(arrowhead). Fig.3 PreservedhatchlingofOctopusvulgaris incaudo-dorsalview(SEM),withtheposteriormantleapex at the lower left. The elevations marked by arrowheads correspond roughly to the position of fin rudiments in decapod embryos, butthey are fixation artefacts dueto shrinkage (seetext). The small arrows point at some ofthe small elevations producingthe "shingle" structure ofthe hatchling skin; thetipsofthesetalcoresofKölliker'sorganshavebrokenthroughtheskinsurfaceonlyinthenuchal region(upperright).