INTRODUCTION
The Gnathiidae is an isopod family including over 170 species worldwide. The larvae (pranizae) are ectoparasites of fish and have attracted much interest from researchers. In coral reefs, larval gnathiids are among the most common ectoparasites of fish and appear to be a major food item of cleaning fish such as Labroides dimidiatus. Thus, gnathiids may play an important role in cleaning interactions (Grutter, Reference Grutter1995, Reference Grutter1996, Reference Grutter1999; Arnal & Côté, Reference Arnal and Côté2000; Arnal & Morand, Reference Arnal and Morand2001). The pranizae of Gnathia africana are vectors of fish blood parasites (Davies & Smit, Reference Davies and Smit2001). However, the biology of the Gnathiidae itself is not fully understood because of the complex life history that is typical of most species.
Gnathiids are highly unusual isopods, with a biphasic life cycle that includes a parasitic larval phase and a non-feeding adult phase (Smith, Reference Smith1904; Mouchet, Reference Mouchet1928; Stoll, Reference Stoll1962). Larval growth proceeds via several alternating periods of ectoparasitic feeding on host fish, followed by moulting in benthic habitats, before individuals reach maturity (Mouchet, Reference Mouchet1928; Stoll, Reference Stoll1962; Wägele, Reference Wägele1988; Klitgaard, Reference Klitgaard1991; Tanaka & Aoki, Reference Tanaka, Aoki, Watanabe and Fusetani1998; Smit et al., Reference Smit, Basson and Van As2003). Larval gnathiids metamorphose into non-feeding adults with significantly different morphology from juveniles (Mouchet, Reference Mouchet1928; Stoll, Reference Stoll1962; Wägele, Reference Wägele1988; Klitgaard, Reference Klitgaard1991; Tanaka & Aoki, Reference Tanaka, Aoki, Watanabe and Fusetani1998). The collection of gnathiids is often difficult because they spend much time in cryptic habitats. Taxonomic studies have been conducted on male specimens (e.g. Cohen & Poore, Reference Cohen and Poore1994), but the identification of larval and female gnathiids is very difficult.
Larval gnathiids are morphologically very different from conspecific adults and thus have been described separately in taxonomy. However, after Hesse (Reference Hesse1864) first observed metamorphosis from larvae to adult males in Paragnathia formica, the life cycles of six species have been described (Mouchet, Reference Mouchet1928; Stoll, Reference Stoll1962; Wägele, Reference Wägele1988; Klitgaard, Reference Klitgaard1991; Tanaka & Aoki, Reference Tanaka, Aoki, Watanabe and Fusetani1998; Smit et al., Reference Smit, Basson and Van As2003). Population dynamics of the Afro-European P. formica and the Japanese Elaphognathia cornigera have been studied (Amanieu, Reference Amanieu1963; Upton, Reference Upton1987a; Tanaka & Aoki, Reference Tanaka and Aoki2000; Tanaka, Reference Tanaka2003) and harem-formation in benthic habitats was individually reported for P. formica, Caecognathia abyssorum, C. calva and C. robusta (Upton, Reference Upton1987a, Reference Uptonb; Wägele, Reference Wägele1988; Klitgaard, Reference Klitgaard1991, Reference Klitgaard1997; Barthel & Brandt, Reference Barthel and Brandt1995). Except for the mud-burrowing P. formica, these studies were mainly conducted on gnathiids collected from sponges. Information on the association between gnathiids and other benthic substrata is restricted to fragmented descriptions in the taxonomic literature.
In Shizugawa Bay, Miyagi Prefecture, Japan, a single species of larval gnathiids is frequently found on benthic fish. A haphazard survey of benthic substrata, including sediment, algae, sponges, dead barnacles and sea squirts, revealed that these gnathiid larvae cohabit with adults within the tubes of terebellid polychaetes. Adult males from these collections were identified as Elaphognathia discolor (Nunomura, Reference Nunomura1988). Other than this paper, the original taxonomic description of E. discolor (Nunomura, Reference Nunomura1988), based on a single adult male specimen, is the only other existing document on this species. To address the lack of detailed ecological information on E. discolor, here we describe aspects of its breeding biology and habitat use based on bimonthly samples collected over nearly one year.
MATERIALS AND METHODS
Study area
Field sampling was conducted along a sandy coast in southern Shizugawa Bay, Miyagi Prefecture, Japan (38°38.519′ N 141°29.020′ E). Water depth at the study area was 1–5 m, and most of the coast was covered with sand and gravel interspersed with boulders and rocks. Sea grass beds consisting of Zostera marina, Z. caulescens and Z. caespitosa occurred about 30 m offshore and extended along the coast. Algal assemblages were distributed in patches of various thicknesses.
Periodic sampling
Terebellids are sessile polychaetes, generally tube- and burrow-building, with some species lying unprotected in the sediment. Tubes of terebellids are soft, composed mainly of sediment and lined with a membrane secreted by the worms. They occur abundantly under boulders, in rock crevices and in sea grass beds in the study area.
Samples were collected every two months between June 2003 and April 2004 around a buoy used to anchor fishing boats 160 m offshore (3–5 m deep). Terebellid tubes attached to the underside of boulders were haphazardly collected by SCUBA diving, and each tube was stored in a 3 × 12-cm plastic vessel in seawater. Samples were transported to the laboratory and dissected under a binocular microscope. The presence or absence of worms in each tube was recorded and empty tubes were excluded. When terebellids were found in their nests, worms and their cohabitants were separated. Worms were identified, and gnathiids from each sample were sorted and counted. Among the terebellids collected, Nicolea gracilibranchis was most frequently collected throughout the investigation; large individuals constructed tubes approximately 4–6 mm in diameter and 15–20 cm in length. The prevalence of gnathiids on N. gracilibranchis tubes was high. Thus, data on Nicolea-associated gnathiids were analysed in this paper.
Although pre-feeding larval gnathiids (pranizae) show apparent segmentation on the dorsal region, their thoraces are dilated when they ingest host fluid. Swollen pranizae become segmented larvae in the next stage or adults by moulting (Mouchet, Reference Mouchet1928; Stoll, Reference Stoll1962; Wägele, Reference Wägele1988; Klitgaard, Reference Klitgaard1991; Tanaka, 1998, Reference Tanaka2003; Smit et al., Reference Smit, Basson and Van As2003). Therefore, gnathiids collected from N. gracilibranchis were classified as unfed pranizae (segmented), satiated pranizae (swollen), adult males and females after fixation with 10% formalin in seawater. Furthermore, some large swollen pranizae were recognized as premature females because ovaries were visible dorsally through the transparent body walls, as reported for Gnathia africana (Smit et al., Reference Smit, Basson and Van As2003). These premature females were treated as females in the analyses described below. After identification of life stages, the body length of each individual (from the frontal margin of the cephalosome to the tip of the pleotelson) was measured. First larvae (in segmented form, N = 4) observed in the marsupium of a female collected in June were also measured to determine the minimum size of pranizae.
Distribution of gnathiids
To assess the spatial distribution of E. discolor, the standardized Morisita index of dispersion (I p; Krebs, Reference Krebs1989) was calculated for adult males, females and larvae in each sample. The index ranges from −1 to +1, with zero indicating a random distribution, a positive value indicating a contagious distribution and a negative value indicating a uniform distribution, with 95% confidence limits at +0.5 and −0.5. In most months, adult male gnathiids never coexisted with other males, and I p was −1 regardless of the number of male specimens. In such cases, the probability of a uniform distribution was directly calculated as follows:
where N and n are the number of polychaete worms and male gnathiids collected in each sampling month, respectively.
For each sample, the prevalence (proportion of polychaete tubes inhabited by gnathiids) and abundance (number of gnathiids per polychaete tube) of pranizae and females were compared to those of males. However, because sample sizes were not large, no statistical test was performed to examine the effect of male presence on the distributions of pranizae and females. Therefore, the effect of male presence on prevalence was tested for larvae and females by Fisher's exact test.
Spearman's rank correlation was used to assess the relationships between male gnathiid body length and the number of coexisting females. In February 2004, we found one terebellid tube in which two E. discolor males coexisted with no females. In this case, the number of females was zero for each of the two males.
RESULTS
Seasonal changes in gnathiids in terebellid tubes
During the bimonthly sampling, 183 terebellid polychaetes belonging to nine species in at least seven genera were collected (Table 1). Throughout the study, Nicolea gracilibranchis was the most frequently found polychaete, with 132 individuals collected. Elaphognathia discolor was observed in 93 N. gracilibranchis tubes. The bimonthly prevalence (proportion of polychaete tubes inhabited by gnathiids) ranged from 57.1% (February 2004) to 80.0% (April 2003). Gnathiids were also found in tubes of Thelepus japonica, Neoleprea japonica and Pista sp. 1, although only one larval specimen was observed in each of these cases.
Table 1. Species and number of terebellid polychaetes collected in bimonthly samples.
* The number of terebellids inhabited by Elaphognathia discolor is shown in parentheses.
Both larvae and adults of E. discolor were observed in tubes of N. gracilibranchis throughout the sampling period (Figure 1). Larvae were abundant in June 2003. The abundance (number per polychaete tube) decreased from 3.73 in June 2003 to 0.67 in February 2004 and then increased again in April. Female abundance showed a similar trend to larval abundance and decreased from 1.87 in June 2003 to 0.33 in February 2004. The abundance of males was almost constant, ranging from 0.38 in February 2004 to 0.53 in June 2003, and no apparent seasonal pattern was observed.
Fig. 1. Seasonal fluctuation in the abundance of Elaphognathia discolor collected from Nicolea gracilibranchis tubes. Closed circles, adult males; open circles, adult females; solid line, larvae; broken line, water temperature.
The body lengths of segmented and swollen larvae ranged from 1.18 to 3.52 mm and 1.40 to 5.81 mm, respectively. Their size histograms were polymodal, and a peak of small swollen larvae was detected at 1.5–2.0 mm in June, October and December 2003 and April 2004 (Figure 2). The body length of premature females (swollen pranizae) ranged from 4.27 to 5.80 mm and widely overlapped with that of adult females (3.31–5.98 mm). Twenty-four first larvae (segmented) were collected from a female brood pouch in June 2003. Their body length ranged from 1.04 to 1.30 mm with a mean (±SD) of 1.20 (±0.07) mm.
Fig. 2. Size distribution of Elaphognathia discolor collected from Nicolea gracilibranchis tubes in each sample.
Distribution of gnathiids in terebellid tubes
In total, 60 male gnathiids were obtained from 59 N. gracilibranchis tubes, i.e. a single male gnathiid existed in each tube, except for one case in which two male specimens coexisted (Table 2). I p for males was −1, except in February 2004, and the probability of this distribution was significantly lower than 0.05 in August and December 2003 (Table 2).
Table 2. The standardized Morisita index of dispersion (Ip) for males, females and larvae of Elaphognathia discolor inhabiting Nicolea gracilibranchis tubes in each sample. N, number of polychaete tubes; n, number of gnathiids.
* The probability of a uniform distribution of males calculated when Ip is minus 1 is shown in parentheses.
In contrast to males, several female gnathiids, including premature individuals, often coexisted (3.00 ± 1.96 individuals, 41 aggregations), and a maximum of nine females was found in a N. gracilibranchis tube collected in June 2003. I p was >0.5, except in April 2004. Aggregations of females were frequently attended by a single adult male (85.4%, N = 41). Thus, the prevalence and abundance of female gnathiids in each sample were always higher in polychaete tubes containing a male gnathiid than in tubes with no male (Figure 3A; Table 3). Fisher's exact tests showed that the probability of female appearance significantly differed depending on male presence/absence throughout the study period, except in February 2004 when both mature and premature females were scarce. The number of females was positively related to male body size in August 2003. However, Spearman's correlation coefficients were not significantly different from zero in the other months (Figure 4).
Fig. 3. Frequency distribution of (A) the numbers of females and (B) larvae of Elaphognathia discolor in each polychaete tube in relation to male presence.
Fig. 4. Relationships between male body length and the number of females cohabiting with each male. The Spearman's rank coefficient (r s) and the probability (p) are shown.
Table 3. Prevalence and abundance of female gnathiids (including premature females) in Nicolea gracilibranchis tubes, in relation to the absence/presence of male gnathiids.
Results of Fisher's exact test on prevalence are shown. N, number of polychaete tubes; n.s., not significant.
Larval gnathiids, excluding premature females, were found in 63 N. gracilibranchis tubes, and their prevalence ranged from 38.1% in February 2004 to 73.3% in June 2003. More than two individuals were often observed in a single polychaete tube and I p was >0.5, except in February 2004 when larvae were scarce. Male gnathiids were sometimes found in larval aggregations. However, larvae often occurred in polychaete tubes without a male gnathiid (Figure 3B). No significant relationship existed between male presence and larval presence/absence at any sampling time (Table 4).
Table 4. Prevalence and abundance of larval gnathiids (except premature females) in Nicolea gracilibranchis tubes, in relation to the absence/presence of male gnathiids.
n.s., not significant.
DISCUSSION
Since gnathiid isopods are fish ectoparasites as larvae and non-feeding as adults (Monod, Reference Monod1926; Wägele, Reference Wägele1988; Klitgaard, Reference Klitgaard1991; Tanaka & Aoki, Reference Tanaka, Aoki, Watanabe and Fusetani1998; Smit et al., Reference Smit, Basson and Van As2003), their benthic habitats are thought to serve primarily as refuges from predation. A variety of benthic substrata (e.g. rock crevices, mud burrows, sponges, dead barnacle shells and sea anemones) are used by gnathiids as habitats (Monod, Reference Monod1926; Holdich & Harrison, Reference Holdich and Harrison1980). However, some species occupy only one or a few particular habitats (e.g. sponges or mud burrows, Table 5) and use them as resting or moulting places between ectoparasitism for larvae or as breeding habitats for non-feeding adults (Upton, Reference Upton1987a; Wägele, Reference Wägele1988; Klitgaard, Reference Klitgaard1991; Tanaka & Aoki, Reference Tanaka, Aoki, Watanabe and Fusetani1998; Smit et al., Reference Smit, Basson and Van As2003). Here both larvae and adults of Elaphognathia discolor were frequently found in tubes of terebellid polychaetes, particularly Nicolea gracilibranchis.
Table 5. Habitats and the number of females cohabiting with a male occupying a ‘cavity’ in six well-studied gnathiid species.
1, Klitgaard (Reference Klitgaard1991); 2, Klitgaard (Reference Klitgaard1997); 3, Wägele (Reference Wägele1988); 4, Barthel & Brandt (Reference Barthel and Brandt1995); 5, Barnard (Reference Barnard1914); 6, Smit et al. (Reference Smit, Van As and Basson1999); 7, Nunomura (Reference Nunomura1992); 8, Tanaka & Aoki (Reference Tanaka, Aoki, Watanabe and Fusetani1998); 9, Tanaka (Reference Tanaka2003); 10, Monod (Reference Monod1926); 11, Upton (Reference Upton1987a).
The lower body length limit for segmented larvae was almost the same as the body length of larvae released from female brood pouches, and the body lengths of larger swollen larvae overlapped with those of adults. Therefore, E. discolor is thought to use terebellid tubes throughout its life cycle. Although the factors affecting the habitat preferences of gnathiids have not been detailed, gnathiids are often found in ‘cavities’ (Monod, Reference Monod1926; Wägele, Reference Wägele1988; Klitgaard, Reference Klitgaard1991, Reference Klitgaard1997; Barthel & Brandt, Reference Barthel and Brandt1995), suggesting that such structures are important for these isopods. The availability and stability of potential habitats are also crucial for successful settlement in the benthos and for continuous use by gnathiids. Terebellid tubes may fit the above criteria and are thought to be an important potential habitat in unstable sandy bottoms such as our study site.
Elaphognathia discolor was principally found in N. gracilibranchis tubes, although it occasionally used tubes of other polychaete species. This type of habitat use is similar to that of the porcelanid crab Allopetrolisthes spinifrons, where adults inhabit sea anemones but smaller individuals use other organisms such as limpets (Baeza & Stotz, Reference Baeza and Stotz2001). In A. spinifrons, which acquires food from its habitat organisms (Baeza & Stotz, Reference Baeza and Stotz2001), a single adult often monopolizes and protects a sea anemone, resulting in the use of alternative habitats by smaller crabs (Baeza et al., Reference Baeza, Stotz and Thiel2001). In contrast, E. discolor do not rely on their habitat for food and were often found in aggregations, although males rarely coexisted with other males. This may indicate that the gnathiids are attracted by conspecific individuals, as is known in the mud-burrowing Paragnathia formica (Upton, Reference Upton1987a, Reference Uptonb). The use of alternative terebellid species occurred in rare cases and is regarded as accidental or limited when favourable habitat is scarce. However, habitat preferences and communication among conspecific individuals of E. discolor should be confirmed in future studies.
Although the adaptive reason why E. discolor uses one primary habitat is unclear, aggregating in a particular habitat would increase the rate of conspecific encounters and may be important for adult reproduction. In many isopods, females do not store sperm, and copulation is followed quickly by egg-laying and fertilization (Wilson, Reference Wilson, Bauer and Martin1991; see also Jormalainen, Reference Jormalainen1998). Furthermore, gnathiids move between fish hosts and benthic habitats several times (Stoll, Reference Stoll1962; Wägele, Reference Wägele1988; Klitgaard, Reference Klitgaard1991; Tanaka & Aoki, Reference Tanaka, Aoki, Watanabe and Fusetani1998; Smit et al., Reference Smit, Basson and Van As2003) and their life cycle is thought to involve the risk of not encountering mates. Therefore, particularly in females that are known to be semelparous, increasing the conspecific encounter rate may decrease the risk of reproductive failure. Although some females were found not cohabiting with males, this may be due to the loss of males during sampling or the death of males after females settle in their tubes.
Encounter probabilities between the sexes have strong effects on the evolution of mating systems (Emlen & Oring, Reference Emlen and Oring1977). As the probability of male–female encounters increases, the risk of sperm competition among males increases. In such situations, males may engage in behaviours that reduce this risk, such as guarding receptive females, as suggested for the symbiotic isopod Iais pubescens (Thiel, Reference Thiel2002). In E. discolor, males rarely coexisted with other males and females tended to be concentrated in tubes with an adult male. This distribution may be the result of a male mating strategy preferring aggregating females to ensure copulation.
Similar male–female associations are known in other gnathiid species (Table 5) and are often referred to as ‘harems’. The association of semelparous females in breeding aggregations in which males may mate with multiple females but females are likely to mate with one or, at most, a few males can be classified as semelparous harem polygynandry (Shuster & Wade, Reference Shuster and Wade2003). Males of P. formica are well-documented harem-formers, and a single male has been observed with up to 25 females in a mud burrow (Upton, Reference Upton1987a). Wägele (Reference Wägele1988) reported ‘harems’ consisting of one male with up to 43 females in Caecognathia calva inhabiting hexactinellid sponges in the Antarctic. In C. robusta, a male was found with four females in a preoscular cavity of the demosponge Geodia mesotriaena in north-eastern Greenland and five females in company with a male were observed in a G. mesotriaena collected in northern Iceland (Barthel & Brandt, Reference Barthel and Brandt1995). Here, a male E. discolor coexisted with up to nine females (including a premature individual). Although the number of females accompanied by a male E. discolor was much lower than that observed in P. formica and C. calva, the E. discolor male can be regarded as a harem-former (Upton, Reference Upton1987a; Wägele, Reference Wägele1988; Barthel & Brandt, Reference Barthel and Brandt1995).
The distribution pattern of male E. discolor may be attributed to competition among males. Wägele (Reference Wägele1988) observed intraspecific fights leading to the death of one rival when two mature C. calva males were in a tube under laboratory rearing conditions. Such male–male competition may explain the uniform distribution of E. discolor males. However, the habitat selection of immature males may also be important in determining the distribution of adult males. Therefore, the behaviour of both immature and adult males should be investigated to better understand the mechanisms behind the male distribution and, further, harem maintenance.
ACKNOWLEDGEMENTS
We thank the staff of the Shizugawa Nature Center, Y. Yokohama, A. Dazai, O. Abe and M. Sato, for generous assistance throughout the study. Many thanks to S. Shuster for his helpful comments and language support on a previous draft of the manuscript. We are also indebted to N. Nunomura of the Toyama Science Museum for loaning type material during the identification of gnathiids.
