The enteropneust hemichordate Glandiceps hacksi inhabits the muddy bottoms of the intertidal to subtidal zones of Koguno-shima Island, located in the central part of the Seto Inland Sea of Japan. Monthly collections from October 2005 to September 2007 revealed that their spawning occurs once a year, in the latter half of May. Parameters such as density and sex ratio, as well as the type of sediment, were also examined. Worm behavior and type of burrows revealed that G. hacksi are infaunal burrowers. Autotomy and regeneration of their posterior regions, and swimming behavior were also observed in an aquarium environment. This is the first comprehensive study on the biology of G. hacksi, the swimming acorn worm.
Enteropneusta is a class of the phylum Hemichordata, accommodating ca. 80 species assigned to four families, Harrimaniidae, Spengelidae, Ptychoderidae, and Torquaratoridae (Nishikawa, 1986; Osborn et al., 2012). They are commonly known as acorn worms, mainly soft, benthic animals burrowing in the mud or sand bottom of the sea. Their morphological and developmental characters, such as gill slits, stomochord, and dipleurula-type tornaria larvae, have often been regarded as ancestral characters inherited in deuterostomes. Molecular studies on Harrimaniidae and Ptychoderidae have revealed some homologies and analogies to chordates or echinoderms (Ogasawara et al., 1999; Tagawa et al., 2001; Lowe et al., 2003, 2006). The interclass or family relationships in Hemichordata were also determined by molecular phylogenetic analysis (Cannon et al., 2009; Osborn et al., 2012).
The family Spengelidae is a small group composed of about 20 species including Glandiceps qingdaoensis, recently discovered in China by An and Li (2005). In contrast to the families Harrimaniidae and Ptychoderidae, the ecology and behavior of the Spengelidae are less known. Glandiceps hacksi was discovered in Yokohama, in middle Japan (Marion, 1885), followed by further illustrating in datails (Marion, 1886), and findings of one large individual in Nagoya and two small one elsewhere in Japan (Spengel, 1893). When six small individuals were discovered from Surabaya (Java) by the Siboga-expedition (1899–1900), Spengel (1907) described them as G. malayanus and distinguished the small individuals previously collected from Japan as G. eximius which he had considered to be a younger stage of G. hacksi. Recently, many individuals of Glandiceps sp. were collected from Shodo-shima Island in the Seto Inland Sea of Japan (Fig. 1) (Yoshimatu and Nishikawa, 1999).
A large-scale swarming of G. hacksi in the water column was observed off Sensui-jima Island (originally described “Sesuijima” was probably an erratum), located in the central part of the Seto Inland Sea of Japan (Fig. 1), predawn on September 3 to 5, 1907 (lkeda, 1908). He reported anecdotal evidence from local fishermen that such unique behavior of G. hacksi had taken place under calm sea conditions, during nocturnal periods, from August to the beginning of September. The swarming individuals were characterized by dorsoventrally flattened bodies, dorsoventrally varying body color, and mud-free intestines, suggesting that they may be benthic creepers with their dark dorsal side up, rather than a sediment burrower (lkeda, 1908). Recently, swarming of Glandiceps sp. was similarly recorded from Harima-nada in the Seto Inland Sea of Japan (Fig. 1), from the middle of August to the beginning of October (Yoshimatu and Nishikawa, 1999). On the other hand, Spengel (1909) reported on the swarming behavior of Glandiceps malayanus in Surabaya (Java) during the months of April to June 1899. However, the purpose of their swimming behavior and lifestyle remained uncertain.
In 2005, a highly dense habitat of G. hacksi was found on the mudflat of Koguno-shima (local name Kokuno-jima) Island, located in the central part of the Seto Inland Sea of Japan. Owing to current limited knowledge of the biology of the species, apart from the bizarre swarming behavior in the water column, a biological investigation of this species was undertaken.
MATERIALS AND METHODS
Monthly collections of G. hacksi were carried out in the lower intertidal and shallow subtidal muddy sediments over a defined area, the southeastern side of the Koguno-shima, Takehara, Hiroshima, Japan: 34°18′17.0″ N, 132°59′04.2″ E (Fig. 1), from October 2005 to September 2007. Koguno-shima is a small, uninhabited, island (0.1 km2) located in the central part of the Seto Inland Sea of Japan, where the maximum tidal range is about 3.5 m. Individuals were obtained by digging into intertidal and shallow subtidal muddy sediment by hand at low tide. The subtidal zone outside the collection area was also investigated using a 0.04 m2 Ekman-Berge bottom sampler on January 24, 2006, to determine the presence or absence of the acorn worm. Some individuals from the intertidal area were photographed and then either studied in the laboratory, or released back into their natural habitat. The size of living individual was measured at their collar in length and width. The density of burrows and individuals within an area of 1 m2 by 20 cm depth was measured on 12 April, 21 June, and 21 September 2006. The maturity of gonads was identified by artificial insemination followed by cleavage and blastula stage, or estimated by trunk color.
To observe worm burrowing, behavior, and regeneration, ten or more individuals were cultured in an aquarium (temperature approximately 21 to 29°C, fed with a unicellular algae, Chaetoceros gracilis (I.S.C Co. Ltd.), and using an 1 L plastic bottle filled with mud from their natural habitat for each one or two individuals) at Takehara Marine Science Station, Hiroshima University (Takehara MSS, Fig. 1), or Marine Biological Laboratory, Hiroshima University (Mukaishima MBL, Fig. 1). One individual was cultured in a flat plastic case to observe its burrow. Some individuals were kept in constant light conditions to suppress their sensitivity to light. Swimming action of an intact individual about 13 cm in total body length was observed in a vat of seawater (30 × 20 × 4 cm) without mud and recorded by a video camera (OLYMPUS C-770). Two individuals were fixed in Bouin's solution and sectioned.
Sediment samples were taken from the surface of four sites of tidal flats in Hiroshima prefecture: the habitat on Koguno-shima and three non-habitats, the Zostera belt in Koguno-shima, Yoshina, and Jigozen (Fig. 1). Triplicate samples at each site were sieved over the range of 4.75 mm down to 0.075 mm with sedimentation analysis (according to Japanese Industrial Standard A 1204: 2000), followed by loss-on-ignition analysis to estimate the amount of organic content. Surface water temperature was simultaneously recorded at the collecting site.
Habitat and density
All collected acorn worms were identified as adult G. hacksi, with the exception of one small Saccoglossus sp., collected on June 9, 2006. Their collar size, the smallest was 3.4 mm in length, 7.2 mm in width, largest was 8.7 mm in length, 13.8 mm in width and the average was 6.22 mm in length, 9.47 mm in width (n = 26), could be distinguished from that of G. eximius, 1 to 1.75 mm in length, 2 to 3.5 mm in width. Sections showed the absence of proboscis septum in the front part of the acorn, unlike G. qingdaoensis.
All individuals were found as infaunal burrowers within 10 cm of the benthic surface. They were found every month during our study period, except February 2007, without any significant changes to their morphology except for the development of gonads. The number of individuals collected is shown in Fig. 3. Estimations of individual density, per habitat, in the intertidal zone, are shown in Table 1. The density ranged from 11 to 24 ind/m2 with a mean density of 16.7 ind./m2 (n = 3). The density in the subtidal zone was 12.5 ind/m2, estimated by an Ekman-Berge bottom sampler on January 24, 2006. The eelgrass Zostera japonica grew in some parts of the mudflat, but few individuals were collected from this location. Small numbers of other benthic animals were also collected from the habitat of G. hacksi. The decapod Asthenognathus inaequipes, belonging to the family Pinnotheridae, a possible symbiote with molluscs, polychaetes, echinoderms, and enteropneustes (Ross, 1983), was also collected around the burrow of G. hacksi on April 12, June 9, 21, and July 21, 2006, although the possibility of a symbiotic relationship to G. hacksi was not determined.
The area consisted of well-sorted mud and fine sand containing a small amount of weakly carbonized vegetable debris. The results of a grain-size distribution curve for the sediment are shown in Fig. 4. The median diameter of sediments, obtained from Fig. 4, was 0.08 mm for the Kogunoshima habitat, 0.06 mm for the Zostera belt in Kogunoshima, 0.40 mm for Yoshina, and 0.43 mm for Jigozen. The ignition losses of sediments were 5.32% for the Koguno-shima habitat, 4.51% for the Zostera belt in Koguno-shima, 1.73% for Yoshina, and 1.69% for Jigozen.
Breeding and sex ratio
Artificial insemination allowed us to determine the sex of matured individuals by their trunk color, i.e. pink ovaries for a female, white testes for a male (Fig. 2). Gonads filled the lateral body wall and usually ranged from the branchial region to the hepatic region. During the breeding season of 2006, mature individuals identified by artificial insemination were found in January 2006. Nearly all individuals collected on April 12 and May 11 developed gonads estimated by trunk color, and then, the gonads of all individuals emptied until June 21. No mature individuals were collected from the summer to winter. During the breeding season of 2007, mature individuals identified by artificial insemination were found in December 2006, and 19 out of 20 individuals collected on May 11 developed gonads estimated by trunk color. The gonads of all individuals emptied on May 31, and no mature individuals were collected until September 2007. Surface water temperature was approximately 10 to 15°C in maturing periods and increased to about 18°C in spawning periods in both years (Fig. 3, Table 1).
Collection data from 1 m2 × 20 cm depth of muddy area in the Koguno-shima intertidal zone.
As shown in Table 1, the M : F sex ratio was 1 : 0.92 on April 12, 2006. A significant bias in the sex ratio was not detected (χ2 test, P > 0.05).
Autotomy and regeneration
The soft trunk of worms, especially the hepatic and caudal region, was easily damaged and amputated during handling. In addition, the artificial change of seawater temperature, up to 28°C from 18°C for example, could trigger autotomy, which occurred simultaneously at many cutting points in the trunk. These points seemed to appear at certain constant intervals from the posterior (Fig. 5A). In aquaria, the antero-most piece regenerated its lost posterior part within 10 to 30 days after cutting (Fig. 5B), but the posterior piece did not regenerate. Regeneration proceeded by a morphallaxis process to form the caudal region from the branchial and/or postbranchial region within 10 days, followed by an epimorphosis process to form the posterior trunk, including the hepatic region (Fig. 5B).
Burrows and behavior
On the sediment surface in the Koguno-shima habitat, many holes, approximately 5 to 8 mm in diameter, were observed (Fig. 6A). These holes opened slopingly and were liable to collapse. Because individuals in aquaria made holes similar to these (Fig. 6B), it was inferred that these were their burrow openings. Results of the estimation of the number of openings of burrows within 1 m2 quadrat are shown in Table 1. No fecal mounds were identified in the field, although individuals in aquaria sometimes made a mound consisting of fecal pellets that was fragile, cylindrical, and 2 or 3 mm in width (Fig. 6B). Usually, the gut of an individual contained some grayish mud and/or fragile blackish feces, clearly distinguishable from their burrowing mud.
When individuals were put in an aquarium, some individuals crawled on the surface of the sediment for a few minutes with their dorsal side upward, but all individuals finally burrowed into sediment approximately 5 cm in depth (Fig. 6C). Burrows were relatively stable and possessed an additional opening (Fig. 6D). Fecal pellets were discarded at a place apart from the opening (Fig. 6B, D). In addition, an extended posterior region and a crawling trace from an opening were sometimes observed (Fig. 6E). Under dark conditions, individuals sometimes extended an anterior region of the body from the opening and retracted it by light stimulation. Some individuals cultured in artificial constant light condition for half a year also extended and moved the body horizontally around the opening with their posterior end hooked into the burrow (Fig. 6E).
Although we did not observe any swimming individuals of G. hacksi around Koguno-shima, collected individuals quite often exhibited a body wiggle and readily engaged in swimming when removed from the sediment. Swimming behavior was often induced by artificial light or physical stimulation. Individuals could crawl out from their burrow and swim away within several seconds. This swimming ability was observed irrespective of seasons or size of individuals.
During swimming, the proboscis forewent peristalsis movement and remained fixed in a pear shape. In addition, the trunk became more flattened dorsoventrally, and individuals swam forward by wiggling the trunk (Fig. 6F), with an intermittent clockwise rotation. Swimming usually finished in a short time, followed by the body returning to its benthic form. Four episodes of swimming recorded under aquaria conditions revealed that swimming speed and wiggling cycle was 45 to 55 mm/sec and 1.80 times/sec, respectively. The average swimming time was 71 seconds and the longest was 165 seconds. Under the bright conditions in the field, released individuals immediately began to burrow into mud after reaching the sediment, even if they had started swimming.
Glandiceps species have been usually collected from depths of 10 m or more, e.g. G. talaboti from about 350 m (Marion, 1885) and G. abyssicola from 4572 m (Spengel, 1893). In this study, the estimated habitation density of G. hacksi in intertidal zone is comparable to previous reports that show the density of ptychoderid and harrimaniid acorn worms could amount to a few tens of individuals/m2 (Nishikawa, 1977; Larsen, 1979). The results also show the restriction of the habitat to a small area containing a high ratio of silt and fine sand in comparison with the neighboring typical tidalflats. This suggests a muddy sediment environment is the typical habitat for this species or genus. Because rootstalks of eelgrass seemed to impede the burrowing of G. hacksi, other unknown large habitat may extend to subtidal zone.
As reviewed by Hadfield (1975), and Nishikawa (1986), previous studies on the complete development and regeneration of acorn worms have been limited to Ptychoderidae and Harrimaniidae. The clarification of the breeding season and regeneration of G. hacksi will serve as useful information in future developmental studies. Our data clearly show that the spawning season of G. hacksi is once a year, in the latter half of May. The spawning may be correlated to changes in seawater temperature, such as the sudden rise in seawater temperature during May, from 15 to 18°C, as shown in Fig. 3. Inducing a change to seawater temperature has been a useful method for inducing artificial spawning in other acorn worms (Colwin and Colwin, 1962; Tagawa et al., 1998; Urata and Yamaguchi, 2004; Miyamoto and Saito, 2007). However, it is still unclear what triggers the induction of spawning in nature. Asexual reproduction was previously reported in some ptychoderid species and the whole process of asexual reproduction, fragmentation, and subsequent morphogenesis, was recently described in Balanoglossus simodensis (Miyamoto and Saito, 2010). We observed body fragmentation by autotomy in G. hacksi, but did not identify subsequent morphogenesis. The regeneration is well described in ptychoderid acorn worms, mainly on the regeneration from a posterior piece to the anterior (Rychel and Swalla, 2008). Miyamoto and Saito (2010) showed that the trunk piece of B. simodensis could regenerate posterior parts, while Humphreys et al. (2010) and Tweedell (1961) also documented the occurrence of posterior regeneration in Ptychodera flava and the harrimaniid acorn worm Saccoglossus kowalevskii, respectively. Our observations reveal a regeneration process in Spengelidae for the first time, with its remarkable ability to regenerate the posterior similar to these reports, suggesting that it may be a common feature of acorn worms. As shown in Fig. 6E, the caudal region, which regenerated immediately, functioned as an anchor during the extension of their anterior body. Therefore, the foremost purpose of the regeneration for G. hacksi may be a recovery mechanism in response to the stresses of daily body action and feeding.
As shown in Fig. 3, we performed monthly collection without a drastic seasonal change in the number of individuals following breeding or a previously reported swimming season. If G. hacksi is a stable benthic animal, the swimming swarms may not be an annual migration event, but rather caused by a temporary change to environmental conditions, for example, a decrease in dissolved oxygen content that is not uncommon in muddy sediment environments. The constant swimming ability and the dual mode (burrowing and swimming) potential of G. hacksi also suggest a quick response to changes in environmental conditions. On the other hand, the identification of their spawning season in the latter half of May is completely disparate from the previously reported swimming season of G. hacksi, from August to the beginning of September. Therefore, as lkeda (1908) suggested by their sexual immaturity, it is clear that swimming behavior was unrelated to the breeding season. Our observations of their swimming pattern support and expand upon previous studies (lkeda, 1908; Yoshimatu and Nishikawa, 1999). However, the purpose of their swimming behavior in nature requires further investigation.
As described in the Introduction, lkeda (1908) suggested that G. hacksi is a benthic creeper. Contrary to his expectation, our investigation supported that they are infaunal burrowers, as proposed by Spengel (1909). However, the observations of their body extension, crawling traces, and their sloping burrow openings suggest that they occasionally crawl out from their burrows (Fig. 6). The observation of anterior body extension behavior under simulated nocturnal conditions, suggests this behavior in nature occurs primarily at night. Feeding is likely to be the primary purpose of this epifaunal mode owing to their blackish feces consisting of microorganisms such as diatoms and dinoflagellates (lkeda, 1908), which suggests filter and/or deposit feeding. Cameron (2002) investigated the feeding behavior of the acorn worm Harrimania planktophilus and suggested that the filter-feeding pharynx is evolved prior to the evolutionary divergence of the hemichordate-echinoderm clade from the chordates. He also mentioned G. hacksi as a potential filter-feeder. Interestingly, many acorn worms are found crawling on deep-sea floors (Osborn et al., 2012). It is now recognized that crawling, or gliding, is one life mode of the acorn worm. G. hacksi has, as it were, triple mode that was partially shared with the deep-sea acorn worms.
We are grateful to Dr. K. Torigoe for the first discovery of the habitat. We thank Drs. T. Nishikawa, M. Yamaguchi, N. Miyamoto for valuable suggestions on our work, Dr. R. Ishii for analysis of sediments, Dr. D. Ueno for taking the photograph of burrows, Drs. K. Yasui, K. Tagawa, N. Yamaguchi and students at Hiroshima Univ. for their help in collecting animals.