Food habits of fishes on an exposed sandy beach at Fukiagehama, South-West Kyushu Island, Japan

Abstract

To clarify the feeding habits and major food sources of sandy beach fishes, the gut contents of 55 fish species collected on a sandy beach at Fukiagehama, South-West Kyushu Island, Japan, were examined. Ontogenetic changes in food preference were recognized in nine species (Hypoatherina valenciennei, Lateolabrax japonicus, Trachurus japonicus, Sillago japonica, Sphyraena japonica, Paralichthys olivaceus, Heteromycteris japonica, Paraplagusia japonica, and Takifugu niphobles). A cluster analysis based on dietary overlaps showed that the sandy beach fish assemblage comprised six trophic groups (mysid, amphipod, zooplankton, juvenile fish, terrestrial insect, and mollusk feeders). Of these, the first three groups were the most abundantly represented, whereas the last two were represented by only a single species. These results indicated that epibenthic macrofauna, such as mysids and gammaridean amphipods, and zooplankton, were important food resources for the fish assemblage at the study site, but infaunal macrobenthos, such as polychaetes and bivalves, being relatively unimportant.

Introduction

Studies of sandy beach fish assemblages in many parts of the world have indicated that sandy beaches are an important habitat for a number of fishes (Lasiak 1986; Suda et al. 2002; McLachlan and Brown 2006), apparently because sandy beaches provide abundant food and refuges from predation (e.g. McIvor and Odum 1988; Ruiz et al. 1993).

In a review of resource partitioning in fish assemblages, Ross (1986) concluded that trophic separations are generally important in understanding the structure and function of fish assemblages. However, most studies on sandy beach fish assemblages have concentrated on temporal and spatial variations in assemblage structure (Gibson et al. 1993; Clark et al. 1996; Layman 2000; Suda et al. 2002). There is only a limited number of studies on the trophic relationships of sandy beach fishes so far (Robertson and Lenanton 1984; DeLancey 1989; Beyst et al. 1999).

In addition, the predation refuge theory is based on the premise that few piscivorous fishes occur in shallow beach habitats (e.g. Boesch and Turner 1984; Paterson and Whitfield 2000). Recent studies have suggested the importance of small and occasional or ‘minor’ piscivores such as sparids and small juvenile carangids that are previously overlooked in shallow habitats (Baker and Sheaves 2009 and Sheaves 2001). However, studies on the feeding habits of all species within such an assemblage have rarely been conducted (but see Inoue et al. 2005; Baker and Sheaves 2005). Thus, it is unclear how many piscivore species are in fact present on sandy beaches.

In this study, we examined the patterns of food resource use within a fish assemblage on an exposed sandy beach at Fukiagehama, with the aim of (1) describing the feeding patterns of the fish species within the assemblage and (2) identifying the feeding guild structures in the assemblage by determining the degree of dietary overlap among species.

Materials and methods

Study area

The study was carried out at Fukiagehama (31°28′N, 130°18′E), situated on the South-West coast of Kyushu Island, temperate zone of Japan, and opening broadly to the northern part of the East China Sea (Fig. 1). The sandy beach was approximately 45 km long, with three distinct beach zone represented by different beach types (northernmost zone—reflective beach, southernmost zone—dissipative beach, and middle zone—intermediate beach) (Fig. 1). The dissipative beach was characterized by a wide, low-inclination surf zone across which breaking waves gradually dissipated their force. In contrast, the reflective beach had a relatively steep slope and no surf zone, waves breaking abruptly near the shoreline. The intermediate beach was located between the dissipative and reflective types, being best characterized as a ridge-runnel system.

Fig. 1

Map of the sandy beach of Fukiagehama, South-West Kyushu Island, Japan, showing sampling sites (filled black circle) in three coastal segments representing different beach types (reflective, intermediate with a ridge-runnel system, and dissipative)

Fish sampling

Because of the environmental differences among the three beach types, fish were sampled from each using a beach seine net (4 mm square mesh, 16 m wide and 1.5 m deep), which was pulled parallel to the shore line in a water depth of approximately 0.5 m. Sampling was conducted at low tide between 0900 and 1600 h in May, August, and November 2006 and 2007. Each specimen was immobilized on ice immediately after collection. The body cavity was subsequently injected with concentrated formalin and the specimens preserved in 10% formalin. Very small individuals (too small for injection) were placed directly into 10% formalin after immobilization. In the laboratory, specimens were identified to species level following Nakabo (2002) and Okiyama (1988) and their standard lengths (SL) measured to the nearest 0.1 mm.

Gut content analysis

A total of 2,357 individuals representing 55 species were examined (Table 1). Some studies have suggested that fewer than five individuals containing food are inadequate for a realistic food item representation of a species (Nakamura et al. 2003; Inoue et al. 2005). However, all species with fewer than five individuals in this study were subjected to gut content analyses because of the value of dietary information for rare species (e.g. Platax boersii and Kuhlia mugil). Food items in the stomach contents or anterior half of the gut (if the fish lack a stomach) of each specimen were identified to the lowest possible taxon, and the percentage volume of each food item in the diet visually estimated under a binocular microscope, as follows: initially, gut contents were squashed on a 1 mm × 1 mm grid slide to a uniform depth of 1 mm and the area covered by each item measured. The measured area was then divided by the total area of the gut contents in order to calculate the percentage volume of that item in the diet (Horinouchi and Sano 2000; Kanou et al. 2004). Food resource use was expressed as mean percentage composition of each item by volume, which was calculated by dividing the sum total of the individual volumetric percentage for the item by the number of specimens examined (Sano et al. 1984; Sano 1989). We have compared food habit of each species among different size class after stomach content inspection, and dealing as different size class group if significant difference was recognized by this analysis. Specimens with empty guts (13.0%) were excluded from this analysis.

Table 1 Number of specimens and standard length (SL) of each species used for gut content analyses

Data analyses

Percentage data of food items of each species were pooled for each year, season, and beach site, because the aim of the study was to describe the feeding patterns of the fish species within the assemblage as a whole. Because of individual size variations in some species, different size classes were examined in order to assess any dietary changes with growth. Since the percentage volume of each major food item in the diet of each species changed rather abruptly at a critical size, the lengths at which such changes occurred were taken as transitional points between size classes. Because the assumption of homogeneity of variance for parametric analysis was not met, even for transformed data, non-parametric Kruskal–Wallis and Mann–Whitney analyses were employed to test whether or not size class differences in the percentage volume of each major food item existed. Nemenyi test was performed when significant differences appeared in the Kruskal–Wallis analyses.

In order to sort the sandy beach fishes into groups that took similar food, dietary overlaps were calculated and a cluster analysis applied. For calculation of dietary overlaps, prey items were grouped in mutually exclusive categories (Table 2). Calculation of the dietary overlap between all species pairs was based on mean percentage volumetric composition of each prey category. For the species with size class differences in feeding habits, each size class was regarded as a separate unit in the analysis. The degree of similarity of fish assemblages among the sites on each sampling occasion was calculated using the Bray-Curtis similarity coefficient, based on the number of individuals of each species. The overlap data were subjected to an average linkage clustering method in order to generate a diet similarity phenogram for the assemblage. This clustering algorithm was used so as not to unduly distort multivariate species (Jaksić and Medel 1990). We arbitrarily adopted a level of 40% similarity (intermediate overlap value) as a basis for dividing the fishes into feeding groups.

Table 2 Gut content components of fishes collected on the sandy beach at Fukiagehama

Results

Diets

The diets of the 55 species sampled are shown in Fig. 2. Ontogenetic changes in feeding habitat were recognized in nine species: smatran silverside (Hypoatherina valenciennei), Japanese seabass (Lateolabrax japonicus), Japanese jack mackerel (Trachurus japonicus), Japanese sillago (Sillago japonica), Japanese barracuda (Sphyraena japonica), bastard halibut (Paralichthys olivaceus), bamboo sole (Heteromycteris japonica), black cow-tongue (Paraplagusia japonica), and grass puffer (Takifugu niphobles). These ontogenetic changes in the diets are described below for each species. The feeding habits of the remaining species are not dealt with specifically in the text because their dietary composition is included in Fig. 2.

Fig. 2

Dendrogram obtained from dietary overlap data and mean percentage volume of food items (%V) of each species. Abbreviations given in Table 2. *, Fish with ontogenetic dietary shift; M, items composing less than 3% of gut content volume of each species; SL, standard length; n, number of fish examined containing food. At the 40% similarity level, the assemblage was divided into six trophic groups: A mysid feeders, B amphipod feeders, C terrestrial insect feeder, D zooplankton feeders, E fish feeders, and F mollusk feeder

Hypoatherina valenciennei—Smaller fish (18–63 mm SL) consumed predominantly planktonic animals, such as ostracods and calanoid copepods, although their importance decreased significantly in the diet of larger individuals (64–97 mm SL) (Mann–Whitney analysis, P < 0.001 for ostracods and calanoid copepods, respectively). Conversely, the contribution of swimming mysids to the diet increased with fish body size (Mann–Whitney analysis, P < 0.001).

Lateolabrax japonicus—In the smaller size class (22–112 mm SL), swimming mysids were the most important food item. Their contribution to the diet decreased with increasing body size (Mann–Whitney analysis, P < 0.001), whereas the percentage volume of juvenile fishes and crabs increased in larger individuals (113–133 mm SL) (Mann–Whitney analysis P = 0.005 for fish, P < 0.001 for crabs).

Trachurus japonicus—The major food item of smaller fish (27–37 mm SL) was harpacticoid copepods but this item decreased in importance in the larger size class (38–79 mm SL) (Mann–Whitney analysis, P < 0.001). Conversely, swimming and sand-burrowing mysids became more important with growth (Mann–Whitney analysis, P < 0.001 for both swimming and sand-burrowing mysids).

Sillago japonica—This species fed on several food categories but their relative importance differed among three size classes (14–23, 24–62, and 83–134 mm SL) (Kruskal–Wallis analysis, P < 0.001 for each prey item). Smaller individuals (14–23 mm SL) took chiefly planktonic animals, such as ostracods and calanoid copepods, comprising 82% of the stomach contents by volume, while middle-sized fish (24–62 mm SL) fed mainly on gammaridean amphipods. The contribution to the diet of swimming mysids, shrimps, gammaridean amphipods, polychaetes, and juvenile fishes increased in larger individuals (83–134 mm SL) (Nemenyi test, P < 0.05).

Sphyraena japonica—The diet of the smallest size class (33–38 mm SL) consisted mostly of ostracods but their share decreased with body size (Kruskal–Wallis analysis, P < 0.001). Middle-sized individuals (39–41 mm SL) consumed mainly swimming mysids, with larger individuals (46–67 mm SL) taking mostly juvenile fishes, along with swimming mysids (Kruskal–Wallis analysis, P < 0.001 for swimming mysids, P = 0.003 for fishes; Nemenyi test, P < 0.05).

Paralichthys olivaceus—Swimming mysids constituted the major food of smaller fish (26–105 mm SL) but decreased in importance in larger individuals (109–181 mm SL) (Mann–Whitney analysis, P < 0.001). The contribution of juvenile fishes, especially Sillago japonica, increased with body size (Mann–Whitney analysis, P < 0.001).

Heteromycteris japonica—Smaller individuals (15–24 mm SL) took mostly gammaridean amphipods, whereas the larger size class (25–96 mm SL) fed predominantly on sand-burrowing mysids and mysid larvae. The relative importance of gammaridean amphipods, sand-burrowing mysids, and mysid larvae in the diet differed significantly between the size classes (Mann–Whitney analysis, P < 0.001 for, gammaridean amphipods and mysid larvae, respectively, P = 0.03 for sand-burrowing mysids).

Paraplagusia japonica—The diet of the smallest size class (16–50 mm SL) consisted largely of harpacticoid copepods and mysid larvae. However, their relative importance decreased with increasing body size (52–109 and 126–266 mm SL) (Kruskal–Wallis analysis, P < 0.001 for both harpacticoid copepods and mysid larvae). The largest individuals (52–109 and 126–266 mm SL) fed mainly on gammaridean amphipods, gastropods, swimming mysids, bivalves, and polychaetes (Kruskal–Wallis analysis, P < 0.001 for gammaridean amphipods, gastropods, swimming mysids, and bivalves P = 0.01 for polychaetes; Nemenyi test, P < 0.05).

Takifugu niphobles—Smaller fish (15–65 mm SL) preyed primarily on gammaridean amphipods and swimming mysids, but the relative importance of these items decreased in the diet of larger individuals (82–118 mm SL) (Kruskal–Wallis analysis, P < 0.001 for gammaridean amphipods, P = 0.01 for swimming mysids; Nemenyi test, P < 0.05). In the middle- (69–78 mm SL) and large- (82–118 mm SL) sized classes, on the other hand, the contribution to the diet of juvenile fishes, decapods (such as callianassid shrimps and crabs) and bivalves increased significantly (Kruskal–Wallis analysis, P < 0.001 for each food item; Nemenyi test, P < 0.05).

Feeding guild

As mentioned earlier, successive changes in food preference by size class were recognized in nine species, each size class of each species therefore being regarded as a separate unit for the cluster analysis. Cluster analysis, based on the dietary overlap among species, showed that the sandy beach fish assemblage was divided into six trophic guilds at a level of 40% similarity (Fig. 2).

Mysid feeders (guild A in Fig. 2) comprised 26 species or size groups. Members of this group preyed mostly on swimming mysids (e.g. Acanthomysis nakazatoi and Acanthomysis pseudomitsukurii), sand-burrowing mysids (e.g. Archaeomysis vulgaris and Iiella oshimai) being consumed by only a limited number of species.

Amphipod feeders (guild B) were represented by 20 species or size groups. Although gammaridean amphipods were the most dominant food item, other items, such as swimming mysids, were also important.

Terrestrial insect feeder (guild C) included only one species, Hyporhamphus sajori, which captured primarily terrestrial insects (e.g. ants and dipterans).

Zooplankton feeders (guild D) comprised 17 species or size groups, most of which took mainly calanoid copepods and ostracods. Other items, such as mysid larvae and harpacticoid copepods, were also important for some group members. Zooplankton consumed by this guild was autochthonous.

Fish feeders (guild E) included three species or size groups. All members of this guild exclusively or chiefly captured juvenile fishes, including Engraulis japonicus, Matsubaraea fusiforme, and Sillago japonica, the later species being consumed extensively. Platycephalus sp. and Acanthopagrus latus also took other items, such as callianassid shrimps and crabs.

Mollusk feeder (guild F) was represented by a single species only, Dasyatis akajei, which fed exclusively on venerid bivalves.

Important food

Based on the cumulative percentage volume values of all fish units for each food item and the percentage of fish units consuming each item, swimming mysids comprised the most important food item for the sandy beach fish assemblage at Fukiagehama (Table 3). Gammaridean amphipods and ostracods were also consumed by many fish units (50 and 29% of total, respectively), being the second and third most important items by cumulative percentage volume, respectively. Ostracods were largely consumed by small juvenile fishes (e.g. Paraplagusia japonica, Hypoatherina valenciennei, Sillago japonica, and Sphyraena japonica), whereas gammaridean amphipods tended to be taken by larger individuals (e.g. Paraplagusia japonica and Sillago japonica) (Fig. 2). Fishes and calanoid copepods were the fourth and fifth most important items by cumulative percentage volume, respectively, although consumed by about 20% of the fish units.

Table 3 Cumulative percentage volume (%V) for each food item (listed in decreasing order) and percentage of fish units consuming each item (%U)

Discussion

In general, fish species in the sandy beach study areas had essentially similar diets to those described in other sandy beach habitats. In some species, however, differences in food habits were found between the present and previous studies. For example, Plectorhinchus cinctus (11–55 mm SL) on a sandy beach at Sanrimatsubara, Fukuoka, southern Japan, was recognized as a benthic or epibenthic crustacean feeder, taking mostly mysids (Inoue et al. 2005), and Paraplagusia japonica (56–116 mm SL) on Yanagihama Beach, Nagasaki, southern Japan, fed mainly on copepods (Noichi et al. 1993). At Fukiagehama, however, the major food item of these species, having similar body sizes as those in the aforementioned studies, comprised gammaridean amphipods. Furthermore, ostracods were the most important food item of Sillago japonica (14–24 mm SL) at Fukiagehama, although the species (14–17 mm SL) at Tateyama Bay, Chiba, central Japan, consumed primarily copepods (Arayama et al. 2003). In a sandy beach at Higashihama, Kumamoto, southern Japan, Takifugu niphobles (40–105 mm SL) took mostly callianassid shrimps (Yamahira et al. 1996). In contrast, the present study found that it fed dominantly on gammaridean amphipods, along with swimming mysids. Such food item differences may be partly related to geographic variations, including differences in food availability and/or habitat characteristics (i.e., beach exposure and type) at different locations.

Ontogenetic changes in food preference were recognized in at least nine species, including the four most dominant (Sillago japonica, Trachurus japonicus, Takifugu niphobles, and Hypoatherina valenciennei). Each of these had relatively wide range of fish length compared with other fishes in this study. Such ontogenetic differences in food habits, at least for some species, may be related to increasing gape size with growth, as a greater mouth width enables the capture of longer or larger prey (Checkley 1982; Schmitt 1986; Boubée and Ward 1997; Lukoschek and McCormick 2001; Kanou et al. 2005). The harder teeth of larger individuals enable the crushing of armored animals, which juveniles are unable to do (Gosline 1971; Wainwright 1988; Horinouchi et al. 1996). Other mechanisms, such as greater catching ability with growth (Horinouchi and Sano 2000), changes in intestinal microbial community (Luczkovich and Stellwag 1993), niche partitioning (Holbrook and Schmitt 1992), and increasing alimentary canal length (Arayama et al. 2003) may also be responsible for such dietary changes.

Fishes collected in this study fell into six trophic groups. Among these, the most abundant feeding guild by unit number was mysid feeders (26 species or size groups). Mysids are known to be extremely common inhabitants on sandy beaches throughout the world (Cockcroft et al. 1988; Takahashi and Kawaguchi 1995; McLachlan and Brown 2006). Lasiak and McLachlan (1987) suggest that opportunistic utilization of such super-abundant prey items may be of immense importance to fishes frequenting highly dynamic environments, such as surf-exposed beaches. In addition to mysid feeders, amphipod feeders (which took predominantly gammaridean amphipods) and zooplankton feeders (which fed mostly on ostracods and calanoid copepods) were the second (20 species or size groups) and third (17 species or size groups) most numerous, respectively. Gammaridean amphipods and zooplankton have also been found to be important food items on other sandy beaches (McFarland 1963; Ross 1986; Inoue et al. 2005), indicating that epibenthic crustaceans (mysids and gammaridean amphipods) and zooplankton are important as food both for the present sandy beach fish assemblage and those on sandy beaches at other geographic locations (Lasiak 1986; McLachlan and Brown 2006). These epibenthic crustaceans and zooplankton are also largely consumed by many fishes in other coastal habitats, including tidal mudflats, reed belts, and seagrass beds (Horinouchi and Sano 2000; Nakamura et al. 2003; Kanou et al. 2004; Horinouchi et al. 2008).

Infaunal macrobenthos, such as polychaetes and mollusks, on the other hand, were important food items for only a limited number of fish species (only Dasyatis akajei) in the present study, the lessened importance of infaunal macrobenthos to fish diets consisting with the findings of a previous study on another temperate sandy beach (Inoue et al. 2005).

On the present sandy beach, only three species (Platycephalus sp., Strongylura anastomella, and Acanthopagrus latus) were fish feeders. However, several small fishes, such as Paralichthys olivaceus, Takifugu niphobles, and Terapon jarbua, often consumed juvenile fish (Fig. 2), although they were not included into the fish feeder guild. Such species may be threat to juvenile and small fish.

In this study, piscivores were hunted mainly juvenile fishes, including Sillago japonica, Engraulis japonicus, Trachurus japonicus, and Matsubaraea fusiforme. This food category, however, was consumed in varying quantities by several species or had a high cumulative percentage volume value (Fig. 2; Table 3). Nakane et al. (2009) found that most juveniles of Sillago japonica used as prey in tethering experiments at the present study beach were consumed within a short time (<10 min) of the experiments more commencing. These findings indicate that juvenile fishes comprised one of the more significant food items in the overall diet of the sandy beach fish assemblage.