Introduction

Triassic crinoids gained scientific interest (Agricola 1546) long before their echinoderm nature was known (Rosinus 1719). Encrinus liliiformis, the stone lily of the eighteenth century naturalists, was a favorite object in art cabinets. Moreover, crinoidal limestones from the Muschelkalk were named by early stratigraphers using Agricola’s term trochites (wheelstones) for the cylindrical encrinid columnals. The Trochitenkalk (Schlotheim 1820) has nowadays the rank of a formation. Since then, Encrinus liliiformis has become one of the best known crinoids in morphological, palaeoecological and taphonomic respects, and the Trochitenkalk became a model for crinoidal limestone formation. Although several Triassic crinoid species and genera have been added subsequently (for an overview see Biese 1934 and Simms 1990b), their true diversity has long been disregarded, as many taxa were later included into the synonymy of Encrinus or Pentacrinus. Other taxa are based on poorly preserved and fragmentary material. This is particularly true for the family Isocrinidae, whose Triassic record consists mainly of isolated columnals and cirrals. They are found in Lagerstätten, in which autotomized and subsequently disarticulated distal stem parts have been accumulated.

This is the state of the art as reflected in the crinoid volumes of the Treatise (Moore and Teichert 1978) which reported only nine genera and five families for the entire Triassic. In the meantime, many publications have increased the number of well-defined taxa. They shed new light on the post-Palaeozoic crinoid phylogeny (Klikushin 1992; Simms 1990a, b; Simms and Sevastopulo 1993; Hagdorn 1995; Simms 1999).

Due to Hans Hess’s experience and effort, the Articulata volume of the revised Crinoid Treatise has been finished and awaits publication (Hess 2010a). It includes all post-Palaeozoic crinoids and lists 33 Triassic genera in 12 well-defined families and 5 orders; the systematic position of 4 additional families is uncertain (Fig. 1). Unlike the 1978 Treatise that assigned Encrinidae (with only one valid genus) to the subclass Inadunata, this clade is now regarded as a separate order Encrinida of the subclass Articulata and comprises the families Encrinidae, Dadocrinidae, Traumatocrinidae, and Ainigmacrinidae. The true diversity of Triassic crinoids is probably even broader, because there are new finds from the early Carnian Cassian Formation of the Dolomites (Italy) and from the middle Carnian Hanwang Formation of Sichuan (Central China) that still await description.

Fig. 1
figure 1

Triassic crinoid diversification. Genus and higher taxonomic units are based on Hess (in press). Numbers indicate valid species as referenced in the Appendix. Artefacts may originate from the limited fossil record. Conservation Lagerstätten with diverse crinoid palaeocommunities such as the Anisian Muschelkalk or the late Ladinian/early Carnian Cassian Formation may be overrepresented compared to early Ladinian or early Norian sediments that have yielded relatively few crinoids. Moreover, the diversity of Isocrinidae recorded in countless nominal “Isocrinus” species based on isolated columnals (Simms 1990a, b) is certainly higher than in this chart. The species concept of Osteocrinus (ca. 15 nominal species) and of the benthic microcrinoids is also based on isolated sclerites showing minimal differences, many of which may have resulted from the etching process during preparation. In this paper 5 well characterized Osteocrinus species were selected. Nominal species attributed to the parataxonomic genus Entrochus were excluded. A distinct faunal change caused by the “Mid Carnian Wet Intermezzo” is indicated by the extinction of order Encrinida, while Isocrinida and Millericrinida persisted into the Norian and later Mesozoic. Thus, four Triassic benthic crinoid associations can be distinguished (cf. Fig. 11), (1) an Early Triassic Holocrinida association of extremely low diversity, (2) a diverse Middle Triassic Holocrinida/Encrinida association, (3) an early Late Triassic Isocrinida/Encrinida association, (4) a late Late Triassic Isocrinida/Millericrinida association. The major turnover happened in Carnian times after the overall Triassic diversity had reached its peak

Full size image

The present paper is based on the classification that Hans Hess used in the new Treatise. It does not deal with the Palaeozoic ancestors and their extinction (Simms and Sevastopulo 1993), nor with the Early Triassic echinoderm bottleneck, which has only recently been summarized by Twitchett and Oji (2005). Its aim is rather to give a preliminary and concise overview of radiations and morphodynamics during the Triassic recovery, in particular (1) crinoid diversity through the Triassic stages and substages, (2) character evolution and functional morphology with a focus on convergent adaptations, (3) ecological niches. With regard to the entire class Echinodermata, post-Palaeozoic radiation has been dealt with by Simms (1990a), Hagdorn (2000) and Twitchett and Oji (2005). Due to their poor fossil record and fragmentary preservation, the Quingyanocrinidae (Stiller 2000) and the Triassic microcrinoid families Tulipacrinidae, Lanternocrinidae and Leocrinidae (Kristan-Tollmann 1990) have also been excluded. They are taken into account, however, in the diversity chart (Fig. 1).

Evolutionary trends and morphodynamics

The fossil record of the less derived Early and Middle Triassic crinoids contains a set of ancestral characters shared by the orders Holocrinida and Encrinida. Probably, these characters were already present in a Late Palaeozoic or Early Triassic “proto-Articulata” (Fig. 2) that is still to be found among advanced cladids (Simms and Sevastopulo 1993). The characters of this hypothetical ancestor should include:

Fig. 2
figure 2

Reconstruction of the hypothetical “proto-Articulata”

Full size image
  1. 1.

    a dicyclic cup with circles of infrabasals (visible in lateral view), basals, and radials;

  2. 2.

    no anal plate;

  3. 3.

    muscular distal radial facets;

  4. 4.

    double axial canal in radials and proximal arms;

  5. 5.

    the arms branching at the second, axillary primibrachial with muscular facets;

  6. 6.

    proximal arms uniserial but distally becoming biserial with wedge-shaped brachials;

  7. 7.

    muscular and ligamentary articulations in proximal arms;

  8. 8.

    ligamentary articulations in distal arms;

  9. 9.

    brachials pinnulate, with the second pinnular longer than the first;

  10. 10.

    proximal stem pentagonal, distally becoming circular;

  11. 11.

    proximal nodals with cirri;

  12. 12.

    terminal discoid holdfast.

During the Triassic, most of these characters underwent significant changes. Thereby, the morphospace expanded with the derived character sets being distinctive for the more advanced later clades. Ancestral features, however, may be preserved in juveniles. Examples are the uniserial arms in advanced encrinids or rudimentary cirri in some encrinids, or even discoid holdfasts in extant comatulids. The living Proisocrinus has retained a holdfast even in the adult stage and reduced the cirri to the proximal stem.

The rich fossil record of the Encrinida and the less continuously documented lineage leading from Holocrinida to Isocrinida provides a more detailed view of this change. As the morphological remodeling was connected with niche splitting, Mesozoic crinoid phylogeny opens a fascinating field of convergent morphological changes. This has been demonstrated by Seilacher and Hauff (2004) and Hagdorn et al. (2007) who compared the convergent adaptation of Pentacrinitidae and Traumatocrinidae to driftwood-dwelling.

Remodeling of the cup

The dicyclic cup with the infrabasals visible in side view is a constant character among all Anisian holocrinids (Fig. 7) and in an unidentified, generically uncertain, five-armed dadocrinid (see Appendix; Fig. 8a). In Dadocrinus, the dicyclic cup may be present in parts of a population (Dadocrinus grundeyi; Fig. 4a) or not at all (Dadocrinus gracilis, Dadocrinus kunischi). Holocrinids have a barrel-shaped or cylindrical, dicyclic cup with extremely thick plates and a narrow lumen (Figs. 3, 5a). The zygosynostosial basal-radial and interradial facets make the holocrinid cup inflexible. In dadocrinids, the basals and radials of the conical cup are synostosial with deep ligament pits and are much thinner and thus enclose a wider lumen. In both groups, the radial facets are directed upwards (perpendicular to the long axis). However, the small holocrinid radial facets cause a constriction between cup and arms that must have been fragile and sensitive to the autotomy of complete arms. Accordingly, the holocrinid fossil record has yielded many specimens with regenerating arms, as well as crowns with all arms autotomized (Hagdorn and Baumiller 1998).

Fig. 3
figure 3

Cup remodeling in Holocrinida/Isocrinida and Encrinida. Arrows indicate the orientation of the radial facets

Full size image

In the middle Anisian 20-armed dadocrinid Carnallicrinus, formerly attributed to the Encrinidae, the cup has become a low cone with the infrabasals concealed in the stem pit and with long basals still visible in side view (cryptodicyclic cup; Fig. 4b). The radials are much wider than in other dadocrinids and their facets are directed obliquely outward. Thus, there is more space for the proximal arm fan, which has become much wider at the level of the second branching (secundibrachial 2).

Fig. 4
figure 4

Encrinida cups. a Dadocrinus gracilis, lateral view, Lower Gogolin Formation, Gogolin (Poland), MHI 1284/2; ventral view, Formazione à gracilis, Recoaro (Italy), MHI 1283/29. b Carnallicrinus carnalli, Lower Muschelkalk, Freyburg/Unstrut (Germany), MHI 1602. c Encrinus liliiformis, Upper Muschelkalk, Mistlau/Jagst (Germany), MHI 1214/1 (H. Hess ded.; same specimen as in Fig. 8c). d Cassianocrinus varians, ventral side with radial facets, Cassian Formation, St. Kassian, NHMW2006 z 0261/0001; synostosial interradial facets, basals, with two columnals in stem pit, NHMW 2006 z 0261/0004. e Zardinicrinus granulosus, dorsal side with large columnal, lateral side with radial facets, Cassian Formation, St. Kassian, NHMW 2006 z 0265/0005. f Traumatocrinus hsui, radials with facets exposed, Xiaowa Formation, Guanling (China), MHI 1850. g Traumatocrinus hsui, dorsal side of radials, primibrachials, interbrachials, note fossulae openings along interbrachial sutures, Xiaowa Formation, Guanling (China), MHI 1882/5. Scales 5 mm

Full size image

The same trend can be observed in the family Encrinidae. The earliest known encrinids already have a cryptodicyclic, low bowl-shaped cup with the basals visible in side view (Encrinus brahli, E. robustus) and the radial facets are directed obliquely outwards. In E. liliiformis (Fig. 4c) as well as in the 20-armed Chelocrinus schlotheimi and Ch. cassianus, the basals are largely within the stem pit and only their distal parts remain visible in dorsal view. The latest encrinids, Cassianocrinus and Zardinicrinus, have even smaller basals that remain almost completely or completely concealed by the first columnal, while the high and wide radial facets are directed outwards providing space for an immediately widening arm fan (Fig. 4d, e). By a trend towards of zygosynostosial basal-radial and interradial facets the cup became more solid by extremely short ligament fibers. In Traumatocrinus, the basals are even more strongly reduced and the cup is extremely small compared to the arms. Nevertheless, their radial facets are directed outward (Fig. 4f, g). All interplate facets of Traumatocrinus are ligamentary with a system of dorsally open grooves similar to the intercolumnar fossulae (Fig. 4f, g).

The same trend from a dicyclic to a cryptodicyclic cup by shortening of the basals and widening of the radials and turning their facets outwards has convergently taken place in the lineage from Holocrinida to Isocrinida. However, the isocrinids never attained the extreme degree of cup reduction seen in the Zardinicrinus–Traumatocrinus clade.

Among the poorly documented cups of Triassic isocrinids (unpublished material MHI), two morphologies can be observed, (1) relatively large basals that form a continuous circle (Fig. 5b), and (2) very small basals that are not in contact on the dorsal side of the cup (Fig. 5c). The latter group might be regarded as ancestral to the family Pentacrinitidae.

Fig. 5
figure 5

Holocrinida and Isocrinida cups. a Holocrinus dubius, infrabasal circle, from lateral, radial circle with facets directed upward, Lower Muschelkalk, Jena Formation, Rittersdorf (Thuringia, Germany), scale 1 mm, MHI 1175/1. b Singularocrinus sp., large basals in lateral contact, radial facets directed outward, Hanwang Formation, Hanwang (China), scale 5 mm, MHI 1603/1/1, 1603/1/7. c “Isocrinus” propinquus, small basals not in lateral contact, large radial facets directed outward, Cassian Formation, Forcella di Giau (Dolomites, Italy), scale 1 mm, MHI 2050/1

Full size image

Ontogenetic sequences of Encrinidae (Hagdorn and Schulz 1996) and of Traumatocrinidae (unpublished material MHI) clearly mirror this general phylogenetic trend. The relatively large cone-shaped cup of juveniles with large and laterally visible basals becomes progressively lower and wider during ontogeny and eventually is much smaller relative to the size of the arms (negative allometry).

The highly specialized Ainigmacrinus from the early Carnian has a cup, the dorsal side of which is hidden by an extremely enlarged last nodal, the calycinodal (Hagdorn 1988). It must have been modified for a protective function at a certain time in ontogeny. At the same time the formation of new nodals must have stopped; otherwise, the calycinodal would later on have shifted away from the cup (developmental count down).

Larger and more outwardly directed radial facets presumably strengthened the muscular attachments, enlarged the tilting angle, and allowed the introduction of additional arm rays. Thus, this trend indirectly improved the filtration capacity (Fig. 3).

Increasing arm number

As a second convergent phylogenetic trend, arm numbers and arm lengths increased among Encrinida (Fig. 6) as well as Holocrinida/Isocrinida (Fig. 7). The most primitive state is represented by the small 5-armed dadocrinid genus. However, this character could also be interpreted as a secondary reduction from the primitive ten-armed state, probably arising through paedomorphosis (M. Simms, pers. comm.). Instead of being axillary, their primibrachial 2 has a muscular facet for a non-axillary third primibrachial on the outer side of the ray and a pinnular socket (with a pinnule attached) on the inner side (Fig. 8a). This character is regularly found in juveniles and adults of this dadocrinid from the early Anisian Muschelkalk (Lower Gogolin Formation). In the 10-armed genus Dadocrinus, regular arm branching occurs at the axillary primibrachial 2 (Fig. 8b). In Carnallicrinus, which here is included in the Dadocrinidae, this branching pattern is repeated once more, at axillary secundibrachials 2. As a rule, Carnallicrinus has 20 arms branching isotomously (Fig. 8c). The distal axillary facets are muscular, the proximal facets are ligamentary.

Fig. 6
figure 6

Isotomous to endotomous arm branching in Encrinida

Full size image
Fig. 7
figure 7

Crown morphology of Holocrinidae and Isocrinidae

Full size image
Fig. 8
figure 8

Increase of arm number in Encrinida. a Dadocrinidae gen. nov., the IBr2 is not axillary, but has the first pinnule attached, Lower Muschelkalk, Lower Gogolin Formation, Wojkowice (Poland), scale 5 mm, MHI 1285/1. b Dadocrinus kunischi, branching at axillary IBr2, Lower Gogolin Formation, Gogolin (Poland), scale 1 cm, MHI 1284/1. c Carnallicrinus carnalli, Lower Muschelkalk, Jena Formation, Freyburg/Unstrut (Germany), scale 1 cm, MHI 1602 (H. Hess ded.; same specimen as in Fig. 4b). d Encrinus liliiformis, Upper Muschelkalk, Trochitenkalk Formation, Schwäbisch Hall (Germany), scale 1 cm, MHI 1043/1. e Chelocrinus schlotheimi, regular second branching at axillary IIBr2, Upper Muschelkalk, Trochitenkalk Formation, Örlinghausen (Germany), scale 1 cm, MHI 1093/14. f Cassianocrinus varians, juvenile individual with endotomous branching starting at axillary IIBr2, Cassian Formation, St. Kassian (Italy), scale 1 cm, NHMW 1865/IX/130. g Cassianocrinus varians, adult individual with endotomous branching starting at axillary IIBr2, axillaries dorsally inflated, Cassian Formation, St. Kassian (Italy), scale 1 cm, NHMW 20006 z 0262/0001. h Traumatocrinus hsui, juvenile individual with armlets slightly smaller than major arms, Xiaowa Formation, Guanling (China), scale 1 cm, MHI 2051. i Traumatocrinus hsui, semiadult individual with armlets smaller than major arms, distal armlets isotomously branching at axillaries with dorsal spines (better seen in Fig. 9f), Xiaowa Formation, Guanling (China), scale 1 cm, MHI 1882/4

Full size image

Encrinidae have 10 (Encrinus, Fig. 8d) or 20 arms (Chelocrinus, Fig. 8e) that branch isotomously, paralleling the situation found in Dadocrinus and Carnallicrinus. In Chelocrinus, an occasional third division of the arms may occur in arms regenerated after traumatic events. This was the case in the type specimen of Ch. schlotheimi (Hagdorn 1982). However, such additional branchings are only found in outer arm branches. In the early Carnian Cassianocrinus (Fig. 8f, g) and Zardinicrinus, the latest known encrinids, this branching pattern is found regularly; it was obviously genetically fixed. In other words, their arms are isotomous up to the second branching and then become endotomous (Hagdorn 2004).

The Traumatocrinidae carry this trend still further. They still have four major arm branches in each ray, but total arm number is greatly enlarged by successive endotomous armlets with each arising from the inner sides of arm pairs resulting from the last isotomous branching (Fig. 8i). Moreover, in the distal arms of adult Traumatocrinus an additional isotomous branching of each armlet occurs at an axillary that bears a long dorsal spine (Fig. 9f). Before being compactionally flattened, the Traumatocrinus crown resembled a plicated coffee filter with the 20 major arms forming the exterior “ribs”. It was held together by a mechanism that Seilacher and Hauff (2004) compared to a velcro-like adhesion between neighboring major arms. Sharing the pore systems (tubuli) in the stem, Traumatocrinus and Zardinicrinus, respectively represent a pseudoplanktonic and a still benthic offshoot of a common Ladinian ancestor. The boreal Vostocovacrinus has the same endotomous branching pattern as Traumatocrinus, but its arms are uniserial and the brachials are low (Yeltysheva and Polayrnaya 1986).

Fig. 9
figure 9

Brachial arrangement and pinnulation in Encrinida: a trend from uniserial to biserial arms. a Dadocrinus kunischi, uniserial brachia, widely spaced pinnules, Lower Gogolin Formation, Gogolin (Poland), MHI 1284/1. b Encrinus liliiformis, juvenile individual with wedge-shaped uniserial brachials, Upper Muschelkalk, Trochitenkalk Formation, Mistlau/Jagst (Germany), IGPT 1525/6. c Carnallicrinus carnalli, immature biseriality of low wedge-shaped brachials, wide zigzag-suture, medium spaced pinnules, Lower Muschelkalk, Jena Formation, Freyburg/Unstrut (Germany), MHI 1602 (same specimen as in Fig. 8c). d Encrinus liliiformis, mature biseriality, narrow zigzag-suture, densely spaced pinnules, Upper Muschelkalk, Trochitenkalk Formation, Neckarwestheim (Germany), MHI 1116/5. e Traumatocrinus hsui, adult individual, 4 mature biserial major arms and densely spaced pinnules (pinnules on armlets not to be seen), Xiaowa Formation, Guanling (China), MHI 1882/5. f Traumatocrinus hsui, same individual, major arms and biserial armlets branching distally at axillaries with dorsal spines, Xiaowa Formation, Guanling (China), MHI 1882/5. Scales 10 mm

Full size image

Among the Holocrinida, the transition from 10-armed isotomous branching to an initial endotomous pattern is already to be observed in the 15-armed Moenocrinus, whose arms branch a second time, but only in the left arm of each ray. While their benthic descendants, the Isocrinidae, retained the isotomous pattern, the pseudoplanktonic Pentacrinitidae (Pentacrinites and Seirocrinus) switched to a highly advanced endotomous branching pattern before the end of Triassic times (Simms 1990a).

This morphological change to endotomous branching, seen in both the Pentacrinitidae and Traumatocrinidae, is interpreted as a convergent attempt to improve nutrition by enlarging the filter fan and the food gathering system. Compared to crinoids that are anchored on the sea bottom and are constantly provided with plankton and organic detritus by currents, pseudoplanktonic crinoids required more effective filters because their substrate drifted with the current (Seilacher and Hauff 2004; Seilacher, this volume) and because there was a requirement for rapid growth (Simms 1986).

Associated with the overall size increase of the crown and the widely splaying arm rays, the areas between the primibrachials were closed by polygonal interbrachial plates, probably of tegminal origin (Figs. 4g, 8i). Thus, the functional calyx was enlarged by integration of the proximal arms, whereas the original cup was reduced to a minor, dorsal part of the calyx. This remodeling of the calyx happened convergently in both the Encrinidae to Traumatocrinidae and the Holocrinidae to Pentacrinitidae transitions, and again in an analogous way, although without the development of endotomous branching, in the Cretaceous Uintacrinus.

Development of biserial arms

A second evolutionary strategy to improve the filtration can be observed in Dadocrinidae and Encrinidae. The early Anisian dadocrinids still have uniserial arms with relatively long brachials (Figs. 8a, b, 9a). In Carnallicrinus, the arms have low, wedge-shaped brachials with oblique articulation facets, representing lowly developed biseriality (Fig. 9c). The same is true in earlier (Pelsonian, early Illyrian) encrinids, whereas the late Illyrian Encrinus liliiformis and Chelocrinus schlotheimi have advanced biserial arms with a narrow zigzag-suture and wide horizontal articulations between brachials in one row (Figs. 8d, e, 9d). This advanced biseriality was maintained in Carnian encrinids and Traumatocrinus (Fig. 9e, f).

In the 1978 Treatise the Encrinidae were assigned to the subclass Inadunata because of their biserial arms, while the otherwise very similar Dadocrinidae with uniserial arms were included in the millericrinid Articulata. However, juvenile encrinids have uniserial arms with long brachials that become gradually lower and biserial during ontogeny (Hagdorn and Schulz 1996; Fig. 9b). As biseriality allows a greater number of pinnules than a uniserial arm (Simms 1990a) it should be considered as an advanced character and as an adaptation to improve the effectivity of the filtration fan.

However, biserially arranged brachials require ligamentary articulations that make the arms inflexible. Only in the proximal, uniserial arms of encrinids did the brachials have muscular articulations that allowed outward tilting to a certain degree. Therefore, the encrinid crown was relatively immobile (Jefferies 1989).

The Holocrinida/Isocrinida lineage maintained long and slender uniserial arms and increased the number of muscular articulations. This imparted a much greater degree of arm flexibility and even the ability to crawl away for defense against cidaroid predators (Baumiller and Messing 2007; Baumiller et al. 2009). This adaptation, however, required a non-permanent attachment, as provided by cirri. In contrast, all Encrinida were permanently fixed to their substrate by holdfasts or similar structures and were unable to re-fix themselves or to regain an upright feeding position after detachment. However, Linck (1965) demonstrated that such individuals of Encrinus liliiformis remained alive sufficiently long to close their open central canals by callus and to regenerate bulbous terminal columnals (Fig. 10b–d). However, their inflexible arms did not allow the animal to re-assume a feeding position (Baumiller et al. 2008).

Fig. 10
figure 10

Attachment structures. a Traumatocrinus hsui, terminal stem with root cirri welded to driftwood, Xiaowa Formation, Guanling (China), scale 10 mm, MHI 1882/1. b–e Encrinus liliiformis, Upper Muschelkalk, Trochitenkalk Formation. b–d terminal stems with regenerated columnals, Wollmershausen (Germany), scales 5 mm, MHI 2052/1–2052/3. e Cluster of encrusting holdfasts from a bioherm, Warburg (Germany), scale 10 mm, MHI 1255/1. f Carnallicrinus carnalli, discoid holdfasts attached to a bored hardground, Lower Muschelkalk, Jena Formation, Gutendorf (Germany), scale 10 mm, MHI 1555/1. g–i Millericrinida gen. nov., discoid holdfasts with irregular lateral callus growth, Hallstatt Limestone, Timor, scales 10 mm, IPB Bather 1a–1c. j–l Holocrinus dubius, preformed rupture point below nodals. j Noditaxis with terminal nodal and cirri attached indicating stem rupture, Lower Muschelkalk, Udelfangen Formation, Ralingen-Kersch (Germany), scale 5 mm, MHI 1276/6. k–l proximal pluricolumnal with crenulated lower nodal sutures, symplectial lower nodal facet, Lower Muschelkalk, Jena Formation, Dietershausen (Germany), scales 5 mm (l), 1 mm (k), MHI 1182/3/1, 1182/3/2. m–n Singularocrinus sp. synostosial lower nodal facet, Hanwang Formation, Hanwang (China), scale 1 mm, MHI 1603/1/2, 1603/1/4. o–p Tyrolecrinus tyrolensis, cryptosymplectial lower nodal facet, Cassian Formation, St. Kassian, scale 1 mm, o (syntype) GBA 1865/3/83 × 7, p (holotype) GBA 1865/3/83 × 6

Full size image

Remodeling of the stem and of attachment structures

Like the crown of the Triassic crinoids, their stem and its attachment to the substrate underwent substantial modification. The hypothetical “proto-Articulata” ancestor of the Triassic crinoids is inferred to have attached to hard substrates by a terminal discoid holdfast (Fig. 2). However, it must also have had cirriferous nodals as a plesiomorphic feature. These structures were prerequisite for the two Middle Triassic strategies of substrate attachment: (1) maintaining the holdfast for permanent attachment to hard substrates, (2) restructuring and strengthening the cirri as a flexible attachment tool that allowed re-attachment after stem rupture and even some degree of motility. During this transformation, the cirral articulation was modified from multiradiate to synarthrial with a strong transverse ridge. The Encrinida—and the less well known early Millericrinida—persisted with the first strategy, whereas the Holocrinida and their descendants developed the second.

Permanent fixation by holdfast

Cirri at the proximal nodals were unnecessary in a long, upright stem attached by a holdfast. Therefore, the short encrinid cirri consisting of small barrel-shaped cirrals with multiradiate articulations lost any function and occur only as rudiments in some species, e.g. in Encrinus cf. E. brahli (Hagdorn and Schulz 1996), or in some populations of Ch. schlotheimi (Hagdorn 1982). However, multiradiate cirrus scars can still be observed in some nodals of the Carnian Cassianocrinus varians. Dadocrinids have no cirri at all; neither do Millericrinidae. The cirri at the nodals of juvenile encrinids must have been shed while these nodals moved distally through ontogeny.

Cylindrical cirri with multiradiate facets also occur in Eckicrinus, a poorly known Pelsonian/Illyrian crinoid with characters that are otherwise holocrinid in nature. Its irregularly arranged cirri may become extremely thick and directed towards one side. Such cirri probably acted as props for the stem as it lay horizontally on the sea floor (Hagdorn et al. 1996, Fig. 6; Seilacher and MacClintock 2005).

The stem of the reef-dwelling Ainigmacrinus remained probably rather short, because after the designation of a cirrinodal as a calycinodal, no additional nodals could be formed. Its laterally enlarged cirrals may have protected the arms (Hagdorn 1988).

In contrast, the Traumatocrinidae modified the encrinid stem by extreme columnal production to attain lengths of more than 10 m (Hagdorn et al. 2007). Systems of longitudinal canals (tubuli) run through the entire stem and horizontal intercolumnar fossulae connect with the tubuli, reducing the weight of these pseudoplanktonic crinoids and thereby prolonging the floating time of the driftwood. Bunches of tubuli filled with mutable collagen fibers may have prevented torsion and breaking of the long stems and made them flexible or even motile. The Pentacrinitid Seirocrinus whose stem was even longer (Haude 1981; Seilacher et al. 1968; Seilacher and Hauff 2004; Seilacher, this volume) achieved flexibility and lightweight construction by lensoid intercolumnar spaces and long intercolumnar collagen fibers. Both characters are alternative adaptations to a pseudoplanktonic life style.

Intermittent fixation by cirri and motility

A major feature of Holocrinidae and Isocrinidae was their ability to shed distal stem parts at a preformed rupture point on the distal side of the nodals. This innovation originated during Anisian times. In holocrinids, this facet remains an unmodified symplectical articulation (Hagdorn 1983; Fig. 10k, l). Nevertheless, the fossil record indicates the presence of this autotomy site also among holocrinids (Baumiller and Hagdorn 1995; Fig. 10j). In isocrinids, the distal nodal facets are modified as cryptosymplexies (Fig. 10o, p) or as smooth synostoses (Fig. 10m, n). After distal parts of the stem are shed, the remaining part was automatically equipped with a cirriferous nodal that could take over the attachment function. This innovation originated during the Anisian times. The pseudoplanktonic pentacrinitids, however, required lifelong solid attachment structures and solid, flexible stems devoid of rupture points; otherwise, they would have sunk to the seafloor.

Eleutherozoic lifestyle by stem reduction

In contrast to the extremely elongated stems of pseudoplanktonic crinoids, extreme shortening of the stem led to an eleutherozoic lifestyle. Stem reduction took place twice and independently during Triassic times: (1) before the Rhaetian within the lineage Holocrinida–Isocrinida in Eocomatula and Paracomatula, i.e. in short-stemmed offshoots of the Pentacrinitidae (Hagdorn and Campbell 1993; Hess 1950; Simms 1994). Their stems are reduced to a few cirriferous nodals plus internodals devoid of cirri. Fusion into a true centrodorsal did not appear before Early Jurassic times (Kristan-Tollmann 1988); (2) before the late Ladinian within a lineage that led to the planktonic somphocrinids. So far, there is no evidence when these highly specialized “Schwebcrinoiden” (German for planktonic crinoids) originated, or from which group. The miniaturization and lightweight construction that eventually enabled them to be dispersed by currents over the entire Tethys must have fundamentally changed their morphology. This may be the reason why it is so difficult to find the ancestors. In contrast, eocomatulids and paracomatulids retained their ability to attach and detach at will by means of their cirri and could change locations by either crawling or active swimming.

Habitat occupation

The dynamic changes in stem morphology and attachment structures happened in connection with active or passive habitat changes. During the Palaeozoic, crinoids occupied a broad range of ecological niches, to which they adapted in size and morphology. Following the Permian/Triassic crisis, crinoid diversity was reduced to small unspecialized benthic animals attached by holdfasts or cirri to solid substrates. However, during the ensuing 50 million years of Triassic times, many of these niches were re-colonized by descendents of the only surviving clade. In this ecological radiation crinoids diversified again by modifying and expanding the inherited morphospace in two realms, the seafloor and the open ocean (Fig. 11).

Fig. 11
figure 11

Habitat adaptation. From Early to Late Triassic times, four subsequent crinoid associations can be identified (see also Fig. 1), which are characterized by an increasing diversity of niches. Niche adaptation caused a strong morphological change of overall size, cup, filtration fan, stem, and attachment structures. Convergent development occurred among different clades depending on the niches. The basic type was the holocrinid attachment by cirri to either hard or soft bottoms; this type remained most successful among isocrinids until today (1, 2). Stems with stout prop-like cirri are to be found in the poorly known Eckicrinus (3). The holdfast-attached encrinids settled on shells (5), on hardgrounds (6) and acted as frame builders in small bioherms (4). Their small size allowed Dadocrinus to incrust mudsticking bivalves and thus to settle on softgrounds (7). In late Ladinian/early Carnian times two lineages led to a pelagic life: the traumatocrinids by settling on driftwood (8), and the somphocrinids by planktonic drifting (9). Due to the “Mid Carnian Wet Intermezzo”, the entire order Encrinida became extinct, and their niches were re-occupied by millericrinids (4, 5) and by the pseudoplanktonic pentacrinitids (10), respectively. As a new niche, the eleutherozoic lifestyle of the actively swimming feather stars was added by the paracomatulids and the eocomatulids (11). The Norian/Rhaetian crinoid association (4) persisted until Middle Jurassic times, in its fundamental parts until today

Full size image

Benthic strategies

Most crinoids are sessile. A stalk elevates them above the seafloor into a zone of stronger currents and above other competing benthic filter feeders. In Palaeozoic crinoid communities different species occupied different tiers above the seafloor (Ausich and Bottjer 1985). Among the Triassic crinoids, the following benthic strategies can be observed.

Settling on hardgrounds and in bioherms

Attachment by a discoid holdfast that was inherited from Palaeozoic ancestors and consists of a specialized terminal columnal is the most common strategy (Fig. 10f). During the Triassic dadocrinids and encrinids, as well as the poorly known early Millericrinida, were attached by a holdfast (Fig. 10g–i; for Silesiacrinus, cf. Hagdorn et al. 1996, Pl. 6, Fig.t). In the late Anisian Upper Muschelkalk (Trochitenkalk Formation), encrinids encrusted patches of terquemiid “oysters” that were slightly elevated above the seafloor and were thus less likely to be smothered by mud. In these biohermal bodies, which reached 2–3 m in diameter and 1.5 m thickness, the encrinids acted as frame builders, because their holdfasts formed irregular, solid crusts (Hagdorn 1978, Hagdorn and Ockert 1993; Fig. 10e). Within the bioherms, the stems of adult crinoids also provided solid anchoring grounds for subsequent generations. It has to be stressed that after the post-Palaeozoic reef crisis, crinoids were—together with bivalves—among the first animals to contribute to reef formation. In the slightly older bioherms in the Upper Silesian Muschelkalk (early Illyrian Karchowice Formation), the reef building role of crinoids is secondary only to hexactinellid sponges and the earliest post-Palaeozoic hermatypic corals (Fig. 12).

Fig. 12
figure 12

Reconstruction of a typical Anisian Encrinid/Holocrinid habitat in the late Anisian (early Illyrian) sponge-coral reefs of the Karchowice Formation (Lower Muschelkalk, Upper Silesia, Poland). Encrinus aculeatus was permanently attached by a holdfast, e.g. to hexactinellid sponges. The cirrus-attached Holocrinus meyeri was able to change its location

Full size image

Secondary soft-bottom dwelling

Holdfast fixation always requires a solid substrate for anchorage. This prevents the larvae of holdfast-crinoids from settling on muddy seafloors unless they find cephalopod shells or other suitable starter substrates. The early Anisian Dadocrinus used mudsticking endobyssate bivalves, such as bakevelliids, for initial fixation. However, in turbulent water the bivalve byssus was insufficient to fix larger dadocrinid groups. The small size of Dadocrinus may be a major reason for its ecological success in the soft ground palaeocommunities of the early Anisian Lower Gogolin Formation of Poland and the contemporaneous Formazione à gracilis of the Vicentinian Alps in Italy (Hagdorn 1996).

Large holdfasts from the Norian and Rhaetian Hallstatt Limestones of the Austrian Alps and from Timor (Fig. 10g–i) showing millericrinid characters already resemble the large and heavy multilayered Jurassic millericrinid roots that anchored these crinoids on soft seafloors (Kleinert 1969; Seilacher and MacClintock 2005).

Pseudoplanktonic strategies

As has been demonstrated by Seilacher et al. (1968), Simms (1986), Wignall and Simms (1990) and Seilacher and Hauff (2004) in Liassic examples and by Hess (2010b) generally for any pseudoplanktonic crinoids, the adaptation to a pseudoplanktonic lifestyle on driftwood required fundamental morphological remodeling. Recent studies of complete Traumatocrinus driftwood-“colonies” from early Carnian black shales of South China have lead to a better understanding of the same lifestyle and corresponding morphologies (Hagdorn et al. 2004, 2007). Among these are extremely enlarged, flexible stems and multiply branching endotomous arms that provide dense pinnulation and extremely long food groove systems arranged like the roads in a banana plantation (Cowen 1981). With diameters reaching more than 40 cm, Seirocrinus and Traumatocrinus crowns by far exceed those of any known benthic crinoids.

Another requirement is a solid and flexible attachment to the driftwood. The traumatocrinids transformed the terminal holdfast of their encrinid ancestors, which consisted of a single sclerite, into a complex multielement attachment structure comprising many columnals and assisted by rootlike, irregularly branching cirri with multiradiate articulations (Fig. 10a; Hagdorn et al. 2007, Fig. 6). Such an attachment structure could grow on with the animal by integrating more distal columnals and by covering the wooden substrate with thin sheets of callus. It also remained flexible enough to buffer current drag and wave action. In contrast to encrinid cirri in the proximal stem (see above), traumatocrinid cirri occur only in the most distal stem part and do not seem to be restricted to nodals. Therefore, it is an open question whether they are homologous with true cirri.

Adult pentacrinitids were devoid of a holdfast and had to get attached by cirri. However, attachment structures of available specimens are difficult to study because they are hidden under bunches of cirrals and stem fragments. Presently, it is impossible to describe them in more detail. Seilacher (this volume) interprets the densely spaced long cirri with rhomboid cross sections of the Norian and Liassic Pentacrinites as motile enough to produce a filter current of their own. However, as no other attachment structure has been observed in this short-stemmed pentacrinitid, their primary function was probably for attachment (Simms 1986). If one considers the second function (self ventilation or filter current; Seilacher, this volume) it was primary in an ontogenetic sense.

Another peculiarity of pseudoplanktonic crinoids has to be explained. Among hundreds of Traumatocrinus and Pentacrinites crowns no regenerating arms have been observed. While benthic crinoids commonly autotomized their arms when exposed to sediment suspended by storms, or when attacked by predators, no such disturbance occurred in pseudoplanktonic filter feeders. This allowed them to keep their arms splayed continuously.

Eleutherozoic life styles

In the present oceans, the most diverse crinoid group is the order Comatulidina, members of which reduced their stems to a fused, cirriferous centrodorsal. Their long and slender arms with muscular articulations enable them to change their locations for feeding and to escape slow predators by active swimming or hiding. This eleutherozoic mode of life became the most successful adaptation of crinoids after the Mesozoic trophic revolution (Vermeij 1977). Today, comatulids inhabit reefs and variable seafloor types. The earliest approach to this adaptive peak is represented by the Late Triassic (Norian/Rhaetian) Eocomatula and Paracomatula, which are holocrinid offshoots, probably derived from the early pentacrinitid lineage (Simms 1994; Hagdorn and Campbell 1993). Both genera are better known from Jurassic representatives. The first true comatulids with a fully fused centrodorsal occur in the Early Jurassic (Kristan-Tollmann 1988).

Another eleutherozoic life style, that of the planktonic microcrinoids, occurred much earlier. Their immediate worldwide mass appearance in late Ladinian sediments (Kristan-Tollmann 1977) is poorly understood because the fossil record has provided so far no plausible ancestors. In any case, their stem became reduced to a fused centrodorsal that bears a radial circle, but no cirri. Their long uniserial arms have extremely long brachials with thickened, bonelike articulating ends (Wang et al. 2008, Fig. 11). Centrodorsals and radials are composed of a coarse meshwork that indicates a lightweight stereome modification. The Late Jurassic Saccocoma may represent still another life style within this clade or may represent an entirely separate clade derived independently from an unknown benthic ancestor (Hess 2002).

Tiering

In Palaeozoic crinoid communities recorded from the Mississippian of North America individual species occupied different tiers above the bottom. In the Upper Muschelkalk of South Germany, Encrinus liliiformis was the only crinoid and, within its palaeocommunity, was the dominant macrofossil. The lowest tier was dominated by articulate brachiopods, also monospecific, and by various bivalves. Higher tiers up to 1.5 m above the seafloor were occupied by growth stages of Encrinus liliiformis. Such an intraspecific tiering did not require competitive growth rates. It is also unlikely that E. liliiformis was specialized on nannoplankton as suggested by Jefferies (1989) because there were no other filter feeders in the higher tiers that could have competed for food.

Stem lengths show a similar intraspecific tiering on a large driftwood with many adult Traumatocrinus attached, even though, being suspended beneath the driftwood, the tiering is inverted (Hagdorn et al. 2004). The crowns of almost identical size were positioned at different depths between 1 and more than 9 m. How this intraspecific, inverse tiering was “organized” within the group is an open question. In any case, filtering at different levels reduced the competition for food.

Conclusions

After the end-Permian near-extinction of the Crinoidea, this echinoderm class recovered during the Triassic and adapted to most of the niches that had been occupied by crinoids during the Palaeozoic. However, the full diversity was not reached before the mid Anisian, followed by an even more distinct diversity peak in early Carnian times. After that, the “Mid Carnian Wet Intermezzo” (Simms and Ruffell 1990; Kozur and Bachmann 2010) caused a global decrease of marine faunas. Due to a rise of surface temperature to 30°C in the Palaeotethys (Hornung et al. 2007), the pseudoplanktonic traumatocrinids became extinct, together with the entire order Encrinida. The open driftwood niche was re-occupied in Norian times by the pentacrinitids and finally disappeared in Mid Jurassic times, when shipworms started their destructive activity on wood in marine environments (Hagdorn et al. 2007). Hence, the late Carnian through Rhaetian crinoid faunas were dominated by Isocrinida, who persisted without fundamental changes beyond the Triassic/Jurassic boundary. The niche of the holdfast-attached benthic encrinids was re-occupied by early millericrinids. As all Triassic crinoids probably derived from an unknown advanced cladid that survived the end-Permian crisis, all extant crinoids are descendants of this “proto-Articulata”. However, phylogenetic analyses are hampered by the convergent morphological trends that different clades underwent during their adaptation to similar habitats.

Encrinids are the most typical Triassic crinoids. During late Anisian times, they flourished in the Western Peritethys realm and contributed by their attachment structures to the buildup of small bioherms. In regional palaeo-highs, immense amounts of encrinid biodetritus resulted in crinoidal limestones that reached up to 16 m thickness in the German Upper Muschelkalk (Hagdorn and Ockert 1993), comparable to the Mississippian Burlington Limestone (Ausich 1999) of Missouri and Illinois. With the mid Carnian decline of the encrinids, this Palaeozoic sediment type disappeared forever. The only comeback happened in the Early Jurassic Hierlatz Limestone of the Alps, derived from millericrinid detritus, though, on a much smaller scale.

Compared to other echinoderm classes (such as the Echinoidea and Asteroidea) whose major post-Palaeozoic radiation did not take place until Jurassic times, the Triassic can be regarded as the crucial period of post-Palaeozoic crinoid phylogeny. This statement also includes the cyrtocrinids the earliest representatives of which have been recovered in the Carnian Hanwang Formation of Central China and await publication. Except for the Jurassic Cotylederma and the Cretaceous Uintacrinus, no basic innovation appeared after the Triassic, and no new crinoid habitat has been invaded since that time.

Repository of specimens—GBA Geologische Bundesanstalt Wien; IPB Institut für Paläontologie Bonn; IFGT Institut für Geowissenschaften Tübingen; MHI Muschelkalkmuseum Ingelfingen; NHMW Naturhistorisches Museum Wien.