Diagnosis. Edwardsiid with very tiny column, not differentiated into the capitulum, scapus, and physa. Surface of long column smooth, lacking nemathybomes or tenaculi. Tentacle sixteen, in two cycles, arranged octamerously, with eight axes of symmetry on the tentacular circle; inner cycle tentacles comparatively longer than outer ones. There is no siphonoglyph. Sphincter muscle not present. Aboral end tapered or rounded but not differentiated into a physa. Inhabits only in homoscleomorph sponge symbiotically and never lives independently. Cnidae: spirocysts, basitrichs, holotrichs, and microbasic p-mastigophores.

Etymology. Tempura is a deep-fried, batter-coated nugget of seafood and/or vegetables in Japanese cuisine. This word comprises the first half of the Japanese name of the type species of this genus, as the shape of the actiniarian when embedded in a sponge tissue resembles shrimp tempura. The siffix -actis is commonly used in actiniarian genus names, meaning radiation of sunshine in Greek. The new genus name is feminine in gender.

Remarks. Within the Edwardsiidae, Tempuractis gen. nov. morphologically resembles the valid genera Edwardsiella Andres, 1881, Drillactis Verrill, 1922, Nematostella Stephenson, 1935, and Metedwardsia Carlgren, 1947 in possessing a smooth scapus with no nemathybomes or tenaculi. The following genera are distinguishable from this new genus: Edwardsia de Quatrefages, 1842, Scolanthus Gosse, 1853, and Edwardsianthus England, 1987 have nemathybomes (Gosse, 1853; Carlgren, 1949; England, 1987); Paraedwardsia Nordgaard, 1905 and Synhalcampella Carlgren, 1921 have tenaculi on the scapus, and the former generally adhere grains of sand on the column; and Halcampogeton Carlgren, 1937 has 12 longitudinal rows of solid papillae (Carlgren, 1937). The most prominent difference between Tempuractis and Edwardsiella is periderm; species in Edwardsiella bear periderm on the scapus, but the column in Tempuractis rinkai is naked and has no periderm. In addition, tentacular arrangement is useful for reference; in contrast with Tempuractis rinkai which has Edwardsia-like tentacular arrangement, Edwardsiella species possess three or more cycles of tentacles that are hexamerously arranged, with the innermost cycle being the longest (Daly et al., 2013). Drillactis species are most similar to Tempuractis rinkai, but there are several differences between them: holotrichs are abundant in T. rinkai, but are absent in Drillactis (Carlgren, 1949, 1954). As a reference, all Drillactis species are distinguishable from Tempuractis by the difference of characters as below; tentacles in Drillactis species has vertical rows of white spots (Carlgren, 1954; Verrill, 1880) while there are no patterns on tentacles of T. rinkai; the two Drillactis species have far larger bodies than T. rinkai (e.g., the body lengths of T. rinkai are even much smaller than the tentacle lengths of Drillactis pallida) (Verrill, 1922; Carlgren, 1954; Fautin, 2013); T. rinkai inhabits only in a homoscleomorph sponge, an extraordinary place for the sea anemone, with symbiotic ecology, while both Drillactis species live in sand, very ordinary habitat for edwardsiids, and without symbiosis (Verrill, 1922). Nematostella is characterized by having nematosomes, the spherical structures, 15–45 μm in diameter, and flagellated bodies containing nematocysts (Hand and Uhlinger, 1992), which are structures that are present only in this genus in Edwardsiidae (Hand, 1994) and are the origin of the name of this genus. However, there is no structure like a nematosome observed both from the outside when they were living and in their coelenteric cavity on transversal sections in T. rinkai. Therefore, this new species does not belong to Nematostella. Metedwardsia is monotypic for Metedwardsia akkeshi (Uchida, 1932). This species, the only edwardsiid without nemathybomes in Japan, is obviously distinguished from all other edwardsiids by the distribution of microcnemes; microcnemes of M. akkeshi are elongated from distal to proximal end (Carlgren, 1947; Uchida, 1932), while all other edwardsiids' microcnemes are limited to the distal end. This is the most unique character of Metedwardsia, so T. rinkai, in which the elongation of microcnemes is the same that in other edwardsiids, also cannot be included this genus.

Given the above, it is inappropriate to include this new species in existing genera. To begin with, this species has several peculiar morphological features for Edwardsiidae; T. rinkai has large holotrichs, which is one of the most recognizable characteristics of this genus; there is no description of holotrich in any recent genus diagnosis of Edwardsiidae (e.g., Carlgren, 1949; Daly and Ljubenkov, 2008; Daly et al., 2013; Gusmão et al., 2016). Tempuractis rinkai, especially in the column, is rich in prominently large holotrichs, so the cnidom can be said to be a unique character of this genus. In addition, the new species also possesses a few microbasic p-mastigophores and spirocysts in its column, which is peculiar to Edwarsiidae. Moreover, the small size of this species, less than 5 mm in whole body length even for adults, is prominent in this family. Ultimately, the habitat of this species is very unique for not only edwardsiids but also for sea anemones in general; only Spongiactis japonica Sanamyan, Sanamyan and Tabachnick, 2012 is recorded to be symbiotic with sponges (Sanamyan et al., 2012). Tempuractis rinkai is quite different from S. japonica in morphological features and, of course, belongs to a quite different taxon at the family level.

Therefore, we establish the new genus Tempuractis to accommodate the newly identified sea anemone, which is characterized by possessing holotrichs, a smooth body surface without nemathybomes or cuticles, s tenticular arrangement in two octamerous cycles, a quite tiny body and inhabiting symbiotically in a sponge.

Type species. Tempuractis rinkai sp. nov. Izumi, Ise and Yanagi 2017 fixed by original designation.

Material examined. Holotype. NSMT-Co 1573. One specimen cut into three parts and prepared for nematocyst observation, collected by wading on 7 June 2013 from the intertidal zone of Aburatsubo, Misaki, Kanagawa, Japan by Yuji Ise. Paratypes. CMNH-ZG 08969. Whole specimen extracted from a sponge; CMNH-ZG 08970. Whole specimen left inside a sponge; CMNH-ZG 08971. Series of histological longitudinal sections; CMNH-ZG 08972. Series of histological cross sections, all specimens collected by wading on June 7, 2013 from the intertidal zone of Aburatsubo, Misaki, Kanagawa, Japan by Yuji Ise; CMNH-ZG 08973. Histological sections. Collected by SCUBA diving on 3 October 2013 at a depth of 8 m in Shukunegi, Sado Island, Niigata, Japan by Yuji Ise; CMNH-ZG 08974. Whole specimen left in a portion of the sponge, collected by wading on August 22, 2013 from the intertidal zone of Tohama, Toba City, Mie, Japan by Takeya Moritaki.; CMNH-ZG 08975. Whole specimen, collected by snorkeling on 13 October 2014 at a depth of 2 m in Sugashima, Toba City, Mie, Japan by Yuji Ise.

Note. Series of histological sections were prepared from a specimen from the same host sponge from which the holotype and paratypes were collected.

Etymology. The species epithet is dedicated to marine biological stations around Japan. The first specimens of this species were collected from a rocky shore in front of the Misaki Marine Biological Station (the University of Tokyo). This station is called “Misaki rinkai jikkenjo” in Japanese (“rinkai” means seaside and “jikkennjo” means research facility). Other specimens were collected during a subsequent faunistic survey in collaboration with other marine biological stations: Sugashima Marine Biological Laboratory (Nagoya University) and Sado Marine Biological Station (Niigata University).

Description. External anatomy. Column naked, smooth, very small, ca. 3.0–5.0 mm in length (3.0 mm in the holotype) and 0.7–1.2 mm in width (0.8 mm in the holotype), and pipe-like in form (Fig. 2B). The surface of column simple, no nemathybomes or tenaculi. Epiderm adhesion with endopinacocytes of Oscarella sp. very tight. Aboral end tapered, not differentiated from scapus, but more or less adherent (see Ecological remarks below). Tentacles slender, without acrospheres, but bearing white patches on each tip, ca. 2.5–4.0 mm in length in living and 1.0–2.0 mm (1.2–1.8 mm in the holotype) in fixed specimens, longer than diameter of oral disk, but well contractible. Tentacles 16 in number, arranged in two concentric cycles of eight inner and eight outer ones positioned alternately (Fig. 3), inner tentacles a little longer than outer ones. Oral disk 0.7–1.2 mm in diameter (0.8 mm in the holotype). Column and tentacles pale orange or pale pink, semitransparent so that mesenterial insertion visible in upper part when alive, no pattern or patches. Area around mouth and actinopharynx white. Body completely surrounded by the tissue of host sponge Oscarella sp. except tentacles and capitulum (Fig. 2A, C–F), so the color of column unidentifiable in alive.

Internal anatomy. Mesenterial arrangement as typical as that of Edwardsia. Eight macrocnemes, four dorsal and ventral directives, and the other four lateral mesenteries (Figs. 3, 4G). All macrocnemes perfect, present along whole length of body (extending from oral to aboral end). Retractor muscle of lateral mesenteries all facing ventrally (Fig. 4G). Eight tiny microcnemes, without muscles, only in distal-most part of column, extending about 30 μm under the base of tentacles (Fig. 4C). Four micronemes between dorsal directives and dorso-lateral mesenteries, two between the dorso- and ventro-lateral mesenteries and two between the ventrolateral mesenteries and ventral directives (Fig. 3). Each tentacle between either exo- or endocoelic. Retractor muscles comparatively weak, restricted or reniform in upper part, but diffused in lower part (Fig. 4G, I, J). Muscle processes mostly simple, around 10 in each muscle pennon. Parietal muscles of macrocnemes not distinct. Actinopharynx short, without distinct siphonoglyphs (Fig. 4A). Tentacular circular muscle endodermal and longitudinal muscle ectodermal (Fig. 4B, D–F). Mesoglea in body wall, mesenteries, and actinopharynx thin, < 10 μm in thickness (Fig. 4G). Marginal sphincter muscle and basilar muscle absent. All parts of body wall, except capitulum, tightly adhered to endopinacocytes of Oscarella sp. (Fig. 4A, H). Many cilia from epiderm of sea anemone, invaginating into endopinacocytes of the sponge, this structure may strengthen the adhering between epiderm of T. rinkai and endopinacocytes of Oscarella sp. (Fig. 5). No mature gametes in holotype. Mature testes observed in paratype (CMNH-ZG 08973; Fig. 4K).

Cnidom. Spirocysts (in tentacles and column), basitrichs (in every tissue), holotrichs (in tentacle and column) and microbasic p-mastigophores (in actinopharynx, column and filament; Table 3, Figs. 6, 7).

Ecological remarks. Colonies of Tempractis rinkai sp. nov. were always found in the host sponge Oscarella sp., and no living individual was found independently outside of a sponge. The position of T. rinkai sp. nov. inside the sponge was unique, as the oscula of the sponge opens beside the oral disc of sea anemone (Fig. 2C), which means that the sea anemones do not utilize the spongocoel or central cavity of the host sponge as do other temporary visitors. Each polyp of T. rinkai sp. nov. was isolated from other polyps and was completely buried in the sponge body, exhibiting a bunch-like shape (Fig. 2A). The epidermis of T. rinkai sp. nov. was strongly adherent to the endopinacocytes of Oscarella sp. Although the majority of individuals were completely buried in the sponge, some sea anemones were piercing their body through Oscarella sp. and adhered to the substrate by their aboral end. Thus, the aboral end of this sea anemone is more or less adhesive.

Tempuractis rinkai sp. nov. and the host sponges were found in cryptic habitats, such as the underside of overhangs, underside of rocks, or interstices of beach rocks. This phenomenon is probably the result of the habitat preference of the host sponge, rather than that of T. rinkai sp. nov., and allows these sea anemones to live in habitats that differ from those of other edwardsiids, which are usually buried in sandy or muddy sea bottom.

Tempuractis rinkai sp. nov. elongated its tentacles outside its host sponge when relaxed, and when exposed to various external stimuli, such as being touched by something, being exposed to a strong current, or when a shadow of something fell on them, they retracted their tentacles and hid themselves inside the sponge (Fig. 8). Tentacles of T. rinkai sp. nov. were sometimes observed to be in contact with those of other polyps because colonies of T. rinkai sp. nov. were densely distributed (Fig. 2A, C, F); however, they did not retract their tentacles or attack each other.