Introduction

Sheltered, submarine limestone caves on reef slopes are inhabited by unique invertebrate communities that include “living fossil” species and relatives of deep-sea taxa. A number of workers have addressed the ecological and evolutionary significances of submarine cave communities (Jackson et al., 1971; Jackson and Winston, 1982; Kobluk, 1988, Hayami and Kase, 1992, 1993; Kase and Hayami, 1992, 1994; Kase and Kano, 2002; Reitner and Gautret, 1996; Wörheide, 1998; Tabuki and Hanai, 1999; Kano and Kase, 2000; Kano et al., 2002; Motchurova-Dekova et al., 2002; Lozouet, 2004) and seasonal to annual changes in fauna and environmental conditions in submarine caves (Gili et al., 1986; Fichez, 1990a, b, 1991; Harmelin, 1997; Lejeusne and Chevaldonné, 2005). For example, Kase and Hayami (1992) examined cavernicolous bivalves from many submarine caves of the Ryukyu Islands, southern Japan and reported the following common characteristics: 1) very small adult size (usually less than 5 mm in length), 2) unusually large prodissoconch I and absence of prodissoconch II in many species, indicating non-planktotrophic development, 3) persistent denticles of the provinculum are retained until the adult stage in many pteriomorph species, implying significant paedomorphosis by progenesis. According to Kase and Hayami (1992), these characteristics indicate that a relatively small number of larvae are produced, a predominantly K-selected reproductive strategy which is considered the optimal adaptation for living in stable or predictable environments. Although the evolutionary and ecological significance of cryptic organisms and seasonal/annual environmental fluctuations within submarine caves have been discussed, there is no published study of millennia-scale variations in the organisms and environments, excluding Kitamura et al. (2007). They examined a sediment layer (43 cm thick) in Daidokutsu cave and found that there were no remarkable temporal changes in grain-size distribution, in the components of the sediment, or in the species composition of the bivalve fauna during the past 2,000 years. In the present study, we collected surface and cored sediments from within Daidokutsu cave and examined the temporal and spatial distribution of both the sediments and cavernicolous bivalves during the last 5,000 years.

Study area

Daidokutsu cave is situated in the northeastern corner of Ie Island, which is located ca. 10 km west of the Motobu Peninsula on Okinawa Island (Figure 1). The cave's entrance lies about 20 m below sea level, and is about 1 m high and 2 m wide. Daidokutsu cave is 40 m long, very dark and deepens abruptly inward to its deepest point of 31 m. The floor is covered by more than 1.4-m-thick muddy deposits. Kitamura et al. (2003, 2007) reported that the deposits consist mainly of carbonate debris, benthic foraminifera and spicules of both sponges and didemnids. They documented a depositional rate that ranges between 20 cm/1,000 yrs and 40 cm/1,000 yrs. Water temperatures within the cave range from 30°C (August) to 20°C (February) and its seasonal pattern is similar to that at 30 m deep around Okinawa (Kitamura et al., 2007). In addition, the patterns of daily changes in temperature are synchronous with the tidal cycle. This indicates that the alternation of water masses within the cave is caused by the tidal cycle. Based on the sea-level curve of Fairbanks (1989), Bard et al. (1996) and Toscano and Macintyre (2003), the cave was submerged at ca 9,000 yr BP, and even the entrance might have been completely submerged at about 8,000 yr BP.

Figure 1.

Location of Daidokutsu submarine cave on Ie Island, off Okinawa Island, Japan.

Methods

We collected surface sediments (5 cm thick) at six sites (a to f) situated at 5 m intervals along a transect (Figure 2). During sediment sampling, we found air bubbles rising from the sea floor (2 m depth) above Daidokutsu cave. The mud content was determined using standard sieves. All microbivalves were picked and counted from the > 0:5 mm fraction of each sample. The weight of whole sediment in each sample was 23–26 g. We also obtained cave sediment by hand with a coring tube 5 cm in diameter at two sites (cores 04 and 06) (Figure 2). The sediment cores were split and described. Each core was sliced into 1-cm-thick samples to measure the mud and carbonate content. We picked and counted all micro-bivalves from the > 0:5 mm fraction of each sample. The mud content was determined using standard sieves. The carbonate content of 500 mg of sediment was determined by gentle removal with 10% acetic acid. Kitamura et al. (2006) discovered a sediment layer containing pumice grains (Daidokutsu pumice) which were deposited between BC 440 ± 40 and AD 640 ± 80. The pumice grains were transported to this area by the Kuroshio Current, since there are no active volcanoes around Ie Island or Okinawa. Kitamura et al. (2006) reported that the stratigraphic distribution of Daidokutsu pumice can be identified by a low value in the carbonate content.

Figure 2.

Simplified transverse section and horizontal projection of Daidokutsu cave. We observed air bubbles rising from the sea floor (2 m depth) above Daidokutsu cave during sediment sampling. Core 01 was examined by Kitamura et al. (2003, 2006, 2007).

Bivalves from the surface and cored sediments exhibit excellent preservation despite being mostly disarticulated. The prodissoconch can be observed in many individuals. These factors imply that the shells represent a life assemblage. Taxonomic identifications are based on Hayami and Kase (1993). For disarticulated shells, a separate valve was counted as one individual.

The radiocarbon ages of 14 well preserved mollusc shells from two cores were conducted by Beta Analytic Inc., using accelerator mass spectrometry (Table 1). Calibrated age ranges were calculated according to Method A of Stuiver et al. (1998), after applying a local correction for the northwestern Pacific of 355 years (ΔR = 35 ±25) (Hideshima et al., 2001).

Table 1.

Results of mollusc ¹⁴C-dating. All samples were analyzed with an accelerator mass spectrometry by Beta-Analytic Corporation.

Results

Sediment and Bivalvia of the surface sediments

An analysis of grain distribution shows that three surface-sediment facies are present (Figure 3). Facies 1 is gray calcareous sand and occurs at sites a to c. Facies 2 is gray calcareous mud and occurs at sites d and e. Facies 3 is restricted to around site f and is calcareous sand with skeletons of partly encrusted coralline sponges, referred to as coralline sponge calcareous sand.

Figure 3.

Histograms of grain distribution and facies at each site.

The dominant species are Cosa kinjoi, Cosa waikikia, Parvamussium crypticum, Cyclopecten ryukyuensis, Chlamydella tenuissima, Malleus sp., Carditella iejimensis and Hiatella sp. aff. H. orientalis. Cosa waikikia, Malleus sp. and Hiatella sp. aff. H. orientalis mainly inhabit the area near the entrance, while both Cosa kinjoi and Parvamussium crypticum mainly inhabit the innermost part of the cave (Figure 4, Table 2). The distributions of Chlamydella tenuissima, Cyclopecten ryukyuensis and Carditella iejimensis do not exhibit distinct patterns.

Figure 4.

Spatial distributions of bivalve species in Daidokutsu cave. Percentage is given relative to total individuals of all bivalve species (Table 2).

Table 2.

List of bivalve species in surface sediments within Daidokutsu cave. A: articulated shells, R: right valve, L: left valve.

Sediment and Bivalvia of the cored sediments

Core 04.—The 84-cm-thick sediment consists of gray calcareous mud (Figure 5). The mud content of the sediment is 78.0 ± 3.0%. Except for the horizon from 75 to 34 cm depth, the carbonate content is 93.4 ± 2.2% (Figure 5). However, the content decreases to 73.5% between 75 and 34 cm depth. This horizon corresponds to the layer containing Daidokutsu pumice. The sediment depth versus ¹⁴C age diagram in Figure 5 implies that the average sedimentation rate is 26.5 cm/1,000 yrs and that the cored sediment covers 3,000 yrs. Cosa kinjoi, Parvamussium crypticum and Carditella iejimensis are predominant and occur throughout the cored sediments (Figure 5, Table 3).

Figure 5.

Columnar diagrams of submarine-cave sediment core 04, showing stratigraphic changes in mud and carbonate content and dominant bivalve species, with depositional rates inferred from ¹⁴C ages of molluscs.

Table 3.

List of bivalve species in core 04. A: articulated shells, R: right valve, L: left valve.

Table 3.

Continued.

Table 3.

Continued.

Core 06.—The 148-cm-thick sediment is divided into lower and upper parts. The lower part consists of yellow calcareous mud (148–126 cm depth) with a mud content of 60.1 ±6.5%, while the upper part consists of gray calcareous mud (126–0 cm depth) with a mud content of 66.7 ±4.4% (Figure 6). The boundary between the two parts is very sharp. The carbonate content of the yellow calcareous mud is consistent at 97.6 ±0.6%. Except for the horizon from 37 to 21 cm depth, the carbonate content of the gray calcareous mud is consistent at 96.8 ±0.6%. Between 37 and 21 cm depth, the carbonate content decreases to 82.3% (Figure 6). We regard this horizon as the deposit containing Daidokutsu pumice. The sediment depth versus ¹⁴C age diagram in Figure 6 implies that the average sedimentation rate of the upper part is 21.1 cm/1,000 yrs in this part of the cave. Although the ¹⁴C ages of all four samples from the lower part fall within a narrow range from BC 3,110 to 4,030, they exhibited a nonlinear relationship with depth. This may have resulted from bioturbation and/or a relatively high sedimentary rate such as sediment gravity flow processes. From these ¹⁴C data, the boundary between the lower and upper parts is estimated to be about BC 3,500.

Figure 6.

Columnar diagrams of submarine-cave sediment core 06, showing stratigraphic changes in mud and carbonate content and dominant bivalve species, with depositional rates inferred from ¹⁴C ages of molluscs.

The bivalve fauna is dominated by Bentharca tenuis, Cosa kinjoi, Cosa waikikia, Parvamussium crypticum, Cyclopecten ryukyuensis, Chlamydella tenuissima, Malleus sp., Carditella iejimensis and Hiatella sp. aff. H. orientalis (Figure 6, Table 4). The relative abundances of Cosa kinjoi and Parvamussium crypticum increase rapidly just above the boundary, while Bentharca tenuis, Cosa waikikia, Chlamydella tenuissima, Malleus sp. and Hiatella sp. aff. H. orientalis decrease at about 47 cm depth, which corresponds to about BC 250. Since the sedimentation rate was constant throughout the upper part, the upward decrease in bivalve abundance implies that their population density dropped over time. Both Cyclopecten ryukyuensis and Carditella iejimensis occur continuously throughout the cored sediments (Figure 6). Kitamura et al. (2007) examined the stratigraphic distribution of bivalves in a core 01 collected close to core 06 (Figure 2). The result showed that Parvamussium crypticum, Cosa kinjoi, Carditella iejimensis and Cyclopecten ryukyuensis predominated during the last 2,000 years. This is consistent with our result.

Table 4.

List of bivalve species in core 06. A: articulated shells, R: right valve, L: left valve.

Table 4.

Continued.

Table 4.

Continued.

Table 4.

Continued.

Table 4.

Continued.

Discussion and conclusion

Based on our analyses of sediments and bivalves, a generalized stratigraphic section of Daidokutsu sediments is shown in Figure 7. Isochronous lines were drawn in this section based upon the sedimentation rates. In terms of the surface sediments, calcareous mud occurs in the inner flat area, whereas calcareous sand occurs on a slope with a 45-degree angle. We therefore think that the distribution of surface sediment is mainly related to a combination of topographic features and the distance from the entrance. Encrustations observed on skeletons of coralline sponges suggest that sediment starvation occurred at site f. A companion study examining the grain composition of the sediments is underway and will document the detailed sedimentary environment and hydrodynamic conditions within the cave.

Figure 7.

Stratigraphic sections of sediments in Daidokutsu cave, showing sediment facies and bivalves of 1st and 2nd groups.

Except for the change in color between the lower and upper parts of core 06, there are no changes in sedimentary characters in the studied cores, such as sedimentary structure or mud and carbonate content. The continuous deposition of mud-sized particles implies that still-water conditions prevailed for at least 5,000 yrs.

Eight bivalve species are dominant in the surface sediments, and they are divided into three groups based on their horizontal distribution (Figure 4). The first group mainly inhabits the innermost cave area and includes Cosa kinjoi and Parvamussium crypticum. The second group dwells near the entrance and includes Cosa waikikia, Malleus sp. and Hiatella sp. aff. H. orientalis. The third group includes Chlamydella tenuissima, Cyclopecten ryukyuensis and Carditella iejimensis and does not exhibit a distinct distributional pattern. According to Kase and Hayami (1994), Cosa kinjoi, Parvamussium crypticum and Carditella iejimensis are dominant in the inner area, while Cosa waikikia and Chlamydella tenuissima are common in the area near the entrance of many caves. This pattern is consistent with our results.

The stratigraphic distribution of the bivalves shows that the first group increased rapidly just above the boundary between yellow and gray calcareous muds (ca. BC 3,500) in the central area (core 06) (Figure 7). This group occurs continuously throughout core 04, from the innermost cave area, and covers the past 3,000 yrs. The second group decreased at about BC 250 in the central area. The relative abundance of this group is very low throughout core 04. Except for Chlamydella tenuissima, there were no significant changes in either the temporal or spatial distribution of the third group (Cyclopecten ryukyuensis and Carditella iejimensis).

It is widely known that progressive reduction of the number of phyla, species, and biomass occurs towards the interior of submarine caves (e.g., Harmelin et al., 1985; Gili et al., 1986). This faunal pattern has been discussed by many authors. Gili et al. (1986) and Harmelin (1997) suggested that the diminution of food inputs below a critical level within the wall boundary layer may lead to decrease in the biomass of benthic organisms. Fichez (1990a, b, 1991) measured suspended chloropigments and particulate organic matter and sedimentation rates in a submarine cave and concluded that the decrease in particulate organic matter input results in increasingly oligotorophic conditions with distance from the cave entrance. All of the eight species noted above are suspension feeders that are sustained by suspended nutrients which flow from the open sea into the cave. We therefore think that the zonation of the first and second groups in the fine surface deposits relates to nutrient levels, although there is no data showing that nutrient levels decline towards the innermost part of Daidokutsu cave. When this interpretation is applied to the spatial and temporal changes seen in the present bivalve samples, it is evident that the environmental conditions of the innermost cave area gradually spread to the entrance of the cave. In other words, a deficiency in nutrition has spread from the innermost cave to the area near the cave entrance over the past 5,000 yrs or more.

Temperature records show that the fluctuation of water masses within the cave is related to the tidal cycle (Kitamura et al., 2007). In addition, we found air bubbles rising from the sea floor above Daidokutsu cave during sediment sampling. These observations indicate that water exchange between inside and outside the cave takes place through the entrance and many small cavities. Kobluk (1988) noted that most cavities in reefs eventually fill up with surface-derived sediments, cement, debris, and in situ skeletons of cryptobionts. We believe that the infilling of cavities has caused a decrease in the exchange of water between the interior and the exterior of the cave, and consequently caused nutritional deficiency to spread within Daidokutsu cave. If this interpretation is correct, the long-term changes in the environments observed in Daidokutsu cave may be common to many submarine caves. An examination of the cause of the change in color of the sediments at ca. BC 3,500, will be the subject of future work.