Morphological description of comparable features of specimens B–H

Specimen B (ZMUC-CRU-8681) (Fig. 5):

Shield. Shield length about 8.5 mm (measured with rostral spine). Maximum width (measured without spines) about 3.5 mm (about 40 % of shield length).

Rostral spine. Anterior part slightly bent upwards. About 45 % of the shield length.

Telson. Anterior and posterior rim slightly convex. The lateral rim on each side slightly convex, width slowly increasing from anterior to posterior. Rim of telson more or less curly brace-shaped, with a rounded tip. Telson armed with two spines on distal rim as protrusion of lateral rim on each side.

Specimen C (ZMUC-CRU-8680) (Fig. 5):

Shield. Shield length about 7.5 mm (measured with rostral spine). Maximum width (measured without spines, about 3.4 mm (45 % of shield length).

Rostral spine. Anterior part slightly bent upwards. About 35 % of the shield length.

Telson. Anterior rim slightly convex, posterior rim slightly concave. The lateral rim on each side slightly convex, width slowly increasing from anterior to posterior. Rim of telson more or less curly brace-shaped, with a rounded tip. Telson armed with two spines on distal rim as protrusion of lateral rim on each side.

Specimen D (ZMUC-CRU-8683) (Fig. 5):

Shield. Shield length about 5 mm (measured with rostral spine). Maximum width (measured without spines, about 2.3 mm (45 % of shield length).

Rostral spine. Anterior part slightly bent upwards. About 35 % of the shield length.

Telson. Anterior rim slightly convex, posterior rim slightly concave. The lateral rim on each side slightly convex, width slowly increasing from anterior to posterior. Rim of telson more or less triangular-shaped from dorsal view, with a flattened tip. Telson armed with two spines on distal rim as protrusion of lateral rim on each side.

Specimen E (MNHN-IU-2014-5468) (Fig. 5):

Shield. Shield length about 7.4 mm (measured with rostral spine) and about 3 mm (measured without rostral spine). Maximum width (measured without spines, about 2.5 mm (30 % of shield length).

Rostral spine. Anterior part strongly bent downwards. About 120 % of the shield length.

Telson. Anterior rim slightly convex, posterior rim slightly concave. Lateral rim difficult to recognize; apparently slightly convex, probably with lobate structure. Telson width suddenly increasing after about 45 % from anterior to posterior rim. Rim of telson more or less triangular-shaped from dorsal view, with a slightly flattened tip. Telson probably armed with two spines on distal rim as protrusion of lateral rim on each side.

Specimen F (Mu_267) (Fig. 5):

Shield. Shield length about 8.3 mm (measured with rostral spine). Maximum width (measured without spines), about 4 mm (50 % of shield length).

Rostral spine. No bending visible. Not documented from lateral, or ventro-lateral view. About 50 % of the shield length

Telson. Anterior rim slightly convex, posterior rim slightly concave. Lateral rim slightly convex, with lobate structure. Telson width suddenly increased after about 40 % from anterior to posterior rim. Rim of telson more or less triangular-shaped from dorsal view, with a flattened tip. Telson armed with two spines on distal rim as protrusion of lateral rim on each side.

Specimen G (ZMUC-CRU-8684) (Fig. 5):

Shield. Shield length about 5.2 mm (measured with rostral spine). Maximum width (measured without spines), about 2.5 mm (50 % of shield length).

Rostral spine. No bending visible. Not documented from lateral, or ventro-lateral view. About 40 % of the shield length.

Telson. Anterior rim slightly convex, posterior rim slightly concave. The lateral rim on each side slightly convex, width slowly increasing from anterior to posterior. Rim of telson more or less triangular-shaped in dorsal view, with a flattened tip. Telson armed with two spines on distal rim as protrusion of lateral rim on each side.

Specimen H (ZMUC-CRU-8682) (Fig. 5):

Shield. Shield length about 9 mm (measured with rostral spine). Maximum width (measured without spines, about 3.6 mm (40 % of shield length).

Rostral spine. Anterior part slightly bent upwards. About 40 % of the shield length.

Telson. Anterior rim slightly convex, posterior rim slightly concave. The lateral rim on each side slightly convex, width slowly increasing from anterior to posterior. Rim of telson more or less curly brace-shaped, with a rounded tip. Telson armed with two spines on distal rim as protrusion of lateral rim on each side.

Identification of the specimens

The specimens described here show some inter-individual differences, but are sufficiently similar to be discussed together. The overall morphology immediately identifies them as reptantian zoea larvae; the embryonic-like posterior thoracopods identify them as an ingroup of Meiura (cf. e.g. [3, 23]). Most meiuran zoeae possess a forked telson (e.g. [3, 23–25]), while the posterior rim of the telson of the here described specimens has a roughly convex shape in dorsal view (with additional lobate protrusions in some specimens). This shape is known from zoea larvae of hippidan species (Figs. 5, 6). Hippidan larvae, similar to the present specimens, have been described with a more or less spherical shield with a rostral spine and two large spines on each postero-lateral margin, but no postero-dorsal spine. Finally, representatives of Hippidae achieve significantly larger sizes as zoea larvae [3, 4] compared to other meiurans. This is also true for the specimens described here (Figs. 2, 5).

After metamorphosis to the megalopa stage in hippidans, antennula and antenna include a long, setose flagellum, the mandible is divided into two parts, thoracopods are divided into elements and largely resemble the setae-bearing juvenile and adult ones; the pleopods as well as the exopods of the uropods bear setae [4]. The new specimens are therefore interpreted as zoea larvae of species of Hippidae. A late zoea stage is indicated by a differentiation of the non-setae-bearing uropods into an endopod and exopod, by more than three tiers of aesthetascs on the antennula, comparably large maxillipeds, and the presence of primordial thoracopods and pleopods. The details of the appendages of specimen A also strongly resemble known late zoea features (besides their size) [26]. Although the other specimens have not been documented in detail, all are interpreted as representing rather late or final zoea stages.

This is supported in particular by a notable difference to already known zoea larvae of Hippidae: in all specimens the sixth pleomere is already set off from the telson (Figs. 2, 4, 5, 6). Hippidan species for which the larval sequence is known are considered to have the sixth pleomere conjoined to the telson in all zoea stages (forming a pleotelson), becoming finally articulated in the megalopa stage [27].

Hence, in the specimens described here, we probably have ultimate zoea larvae, which could be also interpreted as early megalopae [28], retaining some (in fact most) zoea characters, but already having some megalopa characters. Hippidans are able to vary the number of larval stages (e.g. [29]); these specimens could represent such a case of a prolonged pelagic phase, most likely due to the lack of a settling trigger, which would induce transformation to the megalopa.

Larvae of mole crabs – what is known so far

Hippoidea includes the groups Blepharipodidae, Albuneidae and Hippidae, the latter two representing sister groups (e.g. [30]). Larval representatives of Albuneidae and Hippidae feature a more or less spherical shield equipped with one prominent lateral spine on each side (absent in larvae of blepharipodids) and one elongate rostral spine [25, 31, 32]. Despite obvious general similarities due to their close relationships, larvae of Albuneidae and Hippidae differ in many aspects [4, 26, 31, 33, 34], assuming the here described specimens are hippidans. A specimen observed by Gurney [25] was referred to as Albunea sp. It strongly resembles one of the specimens described herein (Fig. 5h), and is probably also a hippidan larva.

Within Hippidae, there are currently only three species groups recognised: Emerita, Hippa and Mastigochirus (e.g. [4, 26, 33]). Species of Hippa and Emerita have very similar zoea larvae, with only a few differences. For comparison we refer to the last zoea stage before metamorphosis as the megalopa, since the detailed described larva (specimen A, ZMUC-CRU-8679) is most likely a very late stage zoea.

In larvae of Hippa species the rostral spine is slightly curved downwards, but upwards in larval representatives of Emerita. Also, the former bear fewer aesthetascs on the antennula, a shorter flagellum on the antenna, fewer setae on the exopod of the maxilla, and fewer setae on the telson. Additionally, larval forms of Emerita (as far as known) do not achieve the impressive size of larval forms of Hippa. The specimens in the 6^th zoeal stage of Hippa can reach a shield length of up to 6 mm, whereas zoea larvae in the same stage of Emerita only reach a shield length of about 2 mm (see [4, 26, 33]).

Grouping of material described here

Based on gross morphological aspects, the specimens described here can be sorted into four more or less distinct groups or morphotypes. This will allow an easier comparison with existing descriptions instead of treating each specimen separately.

Morphotype 1 includes specimens A and F. Both share a more spherical shield, a straight rostral spine, a telson with lobate extensions laterally and a flattened tip.

Morphotype 2 includes specimen D and probably G. Both possess a more elongate shield, a straight rostral spine and a telson lacking a lateral protrusion.

Morphotype 3 includes specimens B, C and H. They all possess a shield comparable to morphotype 2, yet the shape of the posterior margin of the telson is not simple and convex (as in morphotype 2), but with a curled, brace-shaped distal margin with a small and rounded lobate tip.

Morphotype 4 is represented by specimen E, which features a more or less spherical shield with a curved rostral spine. The telson of the specimen features lateral protrusions and a flattened tip as in morphotype 1.

Comparison to the new material

Specimen A (morphotype 1) is quite large, almost reaching the size of a specimen described by Martin and Ormsby [4], which was interpreted as representing the larva of a species of Hippa. The rostral spine of the latter specimen is curved downwards, whereas in specimen A the rostral spine is curved slightly upwards.

The antennula of specimen A bears six, rather than the typical five, tiers of setae. It is divided into strongly curved peduncle and flagellum, instead of being not subdivided into a peduncle and flagellum (Fig. 3). The antenna also differs in many aspects. The basipod bears one spine on its distal rim, and there is no flagellum developed as described by Martin and Ormsby [4]. A marked difference constitutes the pointed and strongly curved endopod without setae and the presence of a well-developed paddle-shaped exopod with numerous plumose setae, instead of an endopod with spines and no exopod on the described species of Martin and Ormsby [4] (Fig. 3).

The maxilla has two coxal and two basipodal endites and no endopod, instead of no endites and a setae-bearing endopod. The bilobed exopod and the number of setae of the specimen described here largely resemble the description of Martin and Ormsby [4] (Fig. 3).

The number of telson setae is about twice the number described by Martin and Ormsby [4]. Additionally, the telson is equipped with two lobate structures on each lateral rim (Fig. 4).

Concerning the large size, one might assume that the observed larva is a representative of Hippa. However, the considerable morphological differences between the described structures of the specimen of Martin and Ormsby [4] and specimen A (and morphotype 1 in general) does not support its interpretation as a representative of Hippa.

Specimen A also differs in many aspects from larvae of species of Emerita described by Knight [26]. Concerning the larger number of aesthetascs on the antennula, and the higher number of setae on the exopod of the maxilla and the telson, specimen A resembles larvae of an Emerita species. Additionally, the rostral spine of specimen A is bent upwards, as presented in the drawings of Knight [26].

In specimen A, there is no flagellum on the antennula, the antenna bears a paddle-shaped exopod, the telson differs morphologically concerning the lobate structure on the lateral rim, and most strikingly, larvae of Emerita are significantly smaller; they achieve a mean size of only about 2 mm shield length [26].

Specimen E (morphotype 4), due to the downwards curvature of the rostral spine and the size of about 3 mm (Fig. 5), matches earlier descriptions of larvae of Hippa (e.g. [4]). Hence, specimen E is likely a larval representative of Hippa.

The additionally documented specimens, although not investigated in detail, do not show many similarities with the known larvae of Hippa or Emerita either. They also differ among each other (see also below). While based on the size we can hypothesize that the specimens represent late, but different larval stages, not all specimens can be attributed to a single developmental sequence.

Since all specimens described here, except specimen E, differ greatly from earlier described species (Fig. 5), we are unable to determine whether they are larvae of a species of Emerita, Hippa or Mastigochirus. These few specimens indicate that there is still an unknown morphological diversity within larval hippidans. The morphological differences are probably not caused by ontogenetic factors. As discussed above, the larvae most likely either represent ultimate zoea stages or even early megalopa stages. Therefore, they probably do not represent subsequent stages of a single species. The unexpected diversity may indicate that these larvae are individuals of species for which larvae are wholly unknown, differing more significantly than expected from other larval sequences. Yet, it is also possible that they represent special cases of developmental plasticity, which means that they are morphological variations of already known larvae caused by specific environmental conditions. In any case, the morphological diversity of hippidan larvae appears to be higher than that of hippidan adults.

Enrollment

The specimens described here have not been observed when alive. Still, we can make inferences about their original behavior based on their functional morphology, as recently suggested by Haug and Haug [7]. The basic idea is to employ approaches from paleontology; e.g., recording the preserved position of specimens, identifying specialized morphological features, and comparing these to those of animals with similar features in which behavior can be directly observed. Using these approaches, while any conclusions remain a matter of conjecture, they nonetheless represent an important tool for understanding. So far it has been impossible to breed giant hippidan larvae or to observe them directly in the field. Hence, the approach discussed here is currently the only possibility. Ideally the prediction made here can be tested in the field (see [7] for more details and a comparable example, also corroborated by field observation).

All specimens were originally preserved in an enrolled position, indicating the possibility of the animals to achieve this position. Further morphological adaptations are given for each morphotype separately.

Morphotype 1 (Figs. 2, 5, 6): Specimen A (ZMUC-CRU-8679), and specimen F (Mu_267). In these specimens the shield appears more or less spherical and the posterior gape of the shield has the same width as the pleonal tergites. This combines a spherical shield with maximum mobility of the pleon. The pleon can be stretched out or flexed forward, without any limitations. The large and lid-like telson features a lobate structure on each lateral margin, which ends up in a spine on the distal rim, and the rim of the telson is more or less triangular-shaped in dorsal view and has a slightly flattened tip. The width of the telson and the length of the pleon appear not to be entirely adapted to the ventral gape of the shield. The telson is broader, and due to a comparatively short pleon, the telson does not reach the anterior rim of the shield, even if the pleon is fully flipped forward. Therefore, antennula and antenna as well as the compound eyes and the distal parts of the maxillipeds are not fully concealed.

Morphotype 2 and morphotype 3 (Fig. 5, 6): Specimen D (ZMUC-CRU-8683) and probably specimen G (ZMUC-CRU-8684); respectively specimen B (ZMUC-CRU-8681), specimen C (ZMUC-CRU-8680) and specimen H (ZMUC-CRU-8682). The two morphotypes differ from each other only in the shape of the posterior rim of the telson, but otherwise are quite similar and therefore treated together.

Similar to morphotype 1, both morphotypes feature a posterior gape which has the same width as the pleon. In both morphotypes, the width of the telson is not adapted to the ventral gape of the shield. The telson is wider than the ventral gape and has a different shape. Yet, when flipped forward, the ventral gape is entirely covered across its width.

The large telson has slightly convex lateral margins, which end up in a spine on each side, and the telson width increases slightly from anterior to posterior. Additionally, due to a longer pleon the larvae are able to flex the telson far anteriorly so that the ventral gape of the shield is entirely closed and almost all parts of the appendages are covered by the telson, which perfectly protects the entire body. Only the compound eyes and the distal parts of antennula and antenna remain exposed.

Morphotype 4 (Fig. 5, 6): Specimen E (MNHN-IU-2014-5468) mainly resembles morphotype 1 in relation to the more or less spherical shield shape, the ventral gape and the shape of the telson. As in morphotypes 2 and 3, the pleon is flexed far anteriorly. Additionally, the telson is positioned inside the shield and reaches the anterior rim (Fig. 5). All appendages, except the eyes and the distal tip of the antennulae are protected by the shield, pleon and telson. The fully enrolled specimen has the appearance of a compact ball, armed with spines.

Here the position of the telson is most important. The telson appears to perfectly fit into the shield, with “rail-like” protrusions of the shield keeping it in place. In this position, the telson is tightly locked in place, and the dorsal area of the pleon perfectly closes the posterior gape. With this, the lateral rim of the telson and the ventral rim of the shield apparently form coaptative structures that tightly enclose the enrolled body. The animal probably achieves this position by 1) flexing its pleon forward, 2) pressing it towards its ventral side right in front of the mouthparts and maxillipeds, and 3) sliding it back. During the last step, the lid-like telson is pulled inside the ventrally curved rims of the shield. As a result, the animal secures the enrolled position and achieves full protection of the ventral appendages. Interestingly, this mechanism appears to be arranged “the-other-way-round” as compared to the coaptative structures in stomatopod larvae where the telson is pushed forward in order to lock it [7].

So do giant hippidan larvae perform defensive enrollment?

Hence, for morphotype 4 there should be little doubt about whether it was able to perform enrollment. For the other three, the stronger argument of the coaptative structures cannot be used. Yet, as in the discussion about different morphotypes of stomatopod larvae [7], the presence of only some of the adaptations cannot be used as an argument to exclude this behavior. Also among other animals which are known to perform defensive enrollment coaptative structures appear to be absent (e.g. polyplacophorans; stonefly larvae of Pteronarcys dorsata [37]).

With this, we consider it likely that all specimens were able to perform defensive enrollment, but to differing degrees of specialization. The ventral “softer” part of the body is in all cases concealed by the spine-bearing shield and the sclerotized pleonal tergites. Yet, in morphotype 1 (specimens A, F) there is a larger unconcealed region remaining up to the anterior margin of the ventral gape of the shield (Fig. 5; similar to a stomatopod larva morphotype, [7]). Morphotypes 2 (specimens D, G) and 3 (specimens B, C, H) achieve the same defensive effectiveness since they flex the telson further anteriorly, and the ventral gape of the shield is almost entirely closed (Fig. 5). Morphotype 4 (specimen E) achieves the most effective degree of defense since there is no unconcealed region left up to the anterior margin of the ventral gape, and due to the position of the telson inside the shield (Fig. 5).

Comparison to Brachyura and Stomatopoda

Martin and Ormsby [4] have stated that hippidan larvae appear very similar to brachyuran larvae. This seems to be largely attributable to the shield morphology. Brachyuran zoeae also feature a spherical shield; yet here the rostral spine is directed more ventrally (e.g. [38]) instead of being anteriorly directed as in most hippidan larvae (Fig. 5). Additionally, brachyuran zoea larvae have no uropods [39, 40], whereas hippidan larvae feature uropods from their early zoeal stages onwards [26] (Figs. 2, 4, 5). Brachyuran zoea larvae have two lateral spines and the rostral spine (as hippidan zoeae), but additionally a long postero-dorsal spine, which is absent in hippidan larvae (Figs. 2, 5). Also the telson differs morphologically. There is a pronounced furca with a medial cleft in brachyurans (at least in early zoea stages; [23, 34]. Hippidan larvae never have a forked telson (Figs. 2, 3, 4 and 5), but the posterior rim is convex.

Most strikingly, brachyuran larvae do not achieve the giant size of hippidan larvae; brachyuran larvae with small body lengths seem to be very common (e. g. [41, 42]). Discrete lengths are rarely stated for brachyuran larvae. A shield length (without the rostral spine) of 0.88 mm in an advanced zoeal stage has been reported [34]. The largest late stage zoea in our investigation (specimen F, Mu_267) reaches a shield length of about 5.5 mm (Fig. 5); 6 mm has been reported [4].

There have been no reports to date of defensive enrollment in brachyuran zoea larvae, nor do we see morphological adaptations for it, but that may change if such features are searched for directly. Currently, only a roughly spherical shield with three spines seems to be similar between brachyuran and hippidan zoeae (Fig. 2).

Stomatopods (mantis shrimps) are also discussed here in light of their similarities to hippidan zoea larvae, since some mantis shrimp larvae can also tightly enroll their bodies [7] (Fig. 1). Interestingly, the specimens described here were found between mantis shrimp larvae in two museum collections (see Methods for details; also in other collections, pers. obs.) in roughly pre-sorted samples. This shows nicely how similar mantis shrimp and hippidan larvae appear at first sight.

Also here especially the shield appears similar, even more similar than to brachyuran larvae as stomatopod larvae, like hippidan zoea larvae, lack the pronounced postero-dorsal spine of brachyuran zoeae (Figs. 1, 5). A future, more intensive functional comparison of hippidan and stomatopod larvae could reveal the evolutionary mechanisms leading to the morphological adaptation coupled to defensive enrollment.