Crustacean y-larvae (subclass Facetotecta) were first described from marine plankton in the late 1800s 12, and they have since been recorded from the arctic to the tropical waters of all oceans 34. The adult organisms have never been identified, and the Facetotecta is the only crustacean group with a formal taxonomy based solely on larval stages 56. At our study site, Sesoko Island near Okinawa, Japan, the enigmatic y-larvae form a significant and diverse component in the plankton. Based on previous extensive sampling at Sesoko Island 7 and on an extensive collection of mounted final naupliar exuviae (obtained from individual rearing surveys from 1996 to 2004) we found that more than 40 morphological types of y-larvae occur there, often abundantly and all representing undescribed species. Ontogenetic, morphological and molecular evidence all point to a taxonomic relationship with the crustacean class Thecostraca. These comprise the well-known barnacles (Cirripedia) and the Ascothoracida, but the Facetotecta forms a separate subclass and is not an ingroup in any of these two taxa 589. Thecostracan crustaceans have sessile adults and a development comprised of a series of nauplius larvae and a final stage, the cyprid, specialized for attachment 10. Similarly, the developmental sequence hitherto known for y-larvae (Facetotecta) comprises a series of (five) naupliar instars and a succeeding stage, the y-cyprid (Additional file 1), which is obviously adapted for settlement 511. Among the related groups, the Ascothoracida are parasites in anthozoans and echinoderms while the Cirripedia, mostly comprised of suspension feeding organisms, also include the Rhizocephala, one of the most specialized groups of parasites within Crustacea 1213. This has prompted speculations that the Facetotecta are also parasitic 10, whence a study of the metamorphosis of their larvae could yield crucial information about the unknown adults. However, until now it has not been possible to rear y-larvae past the cyprid stage 3.

Exposure to 20-HE at concentrations within the effective range (1.04–2.08 μM) induced y-cyprids to metamorphose into a new and unexpected instar (Figure 1). This 300–400 μm long juvenile has a greatly simplified morphology and is surrounded by an extremely thin (<5 nm) cuticle. We documented this metamorphosis from several 'species' of y-cyprids by still photography, video and transmission electron microscopy (TEM); see Figures 1, 2, 3, 4, 5 and Additional files 1, 2, 3, 4, 5. Following exposure to the hormone, metamorphosis begins after 12–24 hours by a retraction of all, or nearly all, cyprid tissues into a compact body anteriorly, which is then surrounded by the newly secreted, thin cuticle. While still inside the y-cyprid, the body of the juvenile begins to exhibit contractions and bending motions. Aided by these motions it exits from the y-cyprid after 31–72 hours through a hole between the bases of the antennules in a process lasting 2–4 hours (Figure 1A and Additional file 2). The molted juvenile has been named the 'ypsigon', which refers to 'ypsilon' (the Greek letter y) and 'gonos' (Greek for 'larva'). The ypsigon is unsegmented, slug-like and lacks appendages (Figures 1B and 1C). The body motions persists after escape and allow the ypsigon to crawl on the bottom of the culture vessel and move several body lengths away from the spent y-cyprid within a few minutes (Additional files 3 and 4).

Like the preceding y-nauplius and y-cyprid stages, the ypsigon lacks a functional gut, but an elongated, dark mass of cells filled with fat globules and extending through most of the body may represent a vestige of this organ system (Figures 1B and 2C). Within the ypsigon body the muscles and compound eyes of the preceding y-cyprid stage are in a state of advanced degeneration (Figure 2A).

The ultrathin cuticle and the decomposing compound eyes are the only morphological indications of its arthropod affiliation (Figures 2A and 2B). According to the species tested, the resulting ypsigon could differ in morphology from being elongated to having a rather plump shape, but it always exhibited the same simplified structure (Figures 3, 4, 5). Ypsigons kept alive in culture for more than 24 hours passed through a molt that yielded another, second juvenile stage (Figures 2C and 2D). This second ypsigon stage had the same morphology as the first and continued to exhibit the same bending motions until they were preserved 48 hours after emerging from the y-cyprid (Additional file 5).

Within Arthropoda, the extreme morphological reduction seen in the new ypsigon stage is matched only by the early endoparasitic 'vermigon' stage recently discovered in certain parasitic barnacles, the Rhizocephala Kentrogonida 1415. Both the vermigon and the ypsigon are formed by comparable metamorphic molts, in which the epidermis withdraws from the old cuticle, closes around all organs in the body and thereby forms the new slug-shaped stage. In addition, both the vermigon and the ypsigon exhibit vigorous body movements after their escape from the cypris, and they have virtually the same reduced morphology, being surrounded only by an exceedingly thin and very pliable cuticle and lacking segmentation, appendages and sensory organs. Metamorphic stages of rhizocephalans, similar to those emerging from naturally settled larvae 1316, have been produced when chemically inducing cyprids to metamorphose in vitro 17. By analogy with the rhizocephalan vermigon 1518 and our observation that the ypsigon does not change morphology at the first molt after escape, all evidence appears consistent with the ypsigon also being endoparasitic, eventually giving rise to a parasitic adult. This conclusion is also supported by comparison with the similarly apodous commiform larvae in the more distantly related entoniscid Isopoda 1920. However, while rhizocephalans are incontestable members of the barnacles (Cirripedia), facetotectans diverge as a separate subclass at the base of the thecostracan phylogenetic tree and must represent a wholly separate evolution of extremely specialized parasites 589. The evolution of very similar, slug-shaped endoparasitic stages (vermigon, ypsigon) in two different lineages (Rhizocephala, Facetotecta) is therefore a stunning example of convergent evolution. Analogy with the Rhizocephala also leads to the prediction that the adult facetotectan will turn out to have a highly simplified structure and this may help to explain why they have escaped observation. Future efforts will focus both on culturing the ypsigon in vitro to more advanced stages and screening of the local fauna to identify the host animals.

While we cannot yet identify the adult y-organism or its host, the new, highly aberrant and obviously endoparasitic stage in the life history of the Facetotecta takes us a major step towards the solution of an enigma that has puzzled zoologists for more than 100 years 2410212223. It has been documented that parasites can significantly affect the structure of ecosystems, especially when they are parasitic castrators and/or are affecting host behavior 24. Our observation of more than 40 facetotectan 'species' at a single study site indicates that a hitherto unknown and diverse fauna of advanced endoparasites remain hidden and could play important roles in many marine habitats.

JTH conceived the project, which was jointly planned in detail by JTH and MJG. MJG and YF were responsible for project logistics. Initial investigations were carried out jointly by JTH, MJG and YF. HG designed and conducted the experiments using 20-HE. JTH managed still and video photography and TEM techniques and assisted with the 20-HE experiments. YF managed larval culture and MJG the mounting of larvae and identification of 'species'. All authors participated in the analysis of the results. The manuscript was drafted by HG and JTH and all authors contributed jointly to the completed version.