INTRODUCTION
Macrophagous scavenging is a common trophic strategy in all marine habitats (Biernbaum & Wenner, Reference Biernbaum and Wenner1993). Britton & Morton (Reference Britton, Morton and Morton1993, Reference Britton and Morton1994) reviewed scavenging in the sea and concluded that almost all phyla have necrophagous representatives, although the existence of obligate necrophages is more debatable (Stockton & DeLaca, Reference Stockton and DeLaca1982; Smith, Reference Smith1985; Sainte-Marie, Reference Sainte-Marie1986a; Sainte-Marie et al., Reference Sainte-Marie, Percy and Shea1989; Kaiser & Moore, Reference Kaiser and Moore1999; Morton & Yuen, Reference Morton and Yuen2000; Ruxton & Houston, Reference Ruxton and Houston2004). Carrion usually serves, therefore, as a dietary supplement (Britton & Morton, Reference Britton and Morton1994).
Bait trapping has been used to catch marine fisheries invertebrates, such as crabs and lobsters, for centuries (Busdosh et al., Reference Busdosh, Robilliard, Tarbox and Beehler1982; Biernbaum & Wenner, Reference Biernbaum and Wenner1993). Secondary arrivals to baited traps after epibenthic macrofaunal and fish exclusion have also been studied for more than a century (Chevreux, Reference Chevreux1900) and have shown that lysianassoids often account for 100% of all organisms drawn to such traps in boreal deep-sea environments (Paul, Reference Paul1973; Dahl, Reference Dahl1979; Jannasch et al., Reference Jannasch, Cuhel, Wirsen and Taylor1980; Ingram & Hessler, Reference Ingram and Hessler1983; Sekiguchi & Yamaguchi, Reference Sekiguchi and Yamaguchi1983; Smith & Present, Reference Smith and Present1983; Hargrave, Reference Hargrave1985; Nishida et al., Reference Nishida, Kikuchi and Toda1999).
An increasing number of studies in shallow coastal waters of hyperbenthic species attracted to baited traps, however, have identified a gradual diminishing lysianassoid importance from high to low latitudes. As in great depths, lysianassoids virtually monopolize the necrophagous fauna of shallow waters at high latitudes. In the Beaufort Sea, they account for >99% of organisms attracted to bait (Busdosh et al., Reference Busdosh, Robilliard, Tarbox and Beehler1982). Their numerical importance at lower latitudes is more variable and gives way progressively to various other organisms (Sekiguchi et al., Reference Sekiguchi, Yamaguchi and Kobayashi1982), but still mostly crustaceans. For instance, 75% of all trapped scavengers were cirolanid isopods along the west coast of South Africa (34°S) (Griffiths et al., Reference Griffiths, Mayfield and Branch2000) and 34% of total captured organisms comprised leptostracans off the coast of California (32°N) (Biernbaum & Wenner, Reference Biernbaum and Wenner1993).
The aims of this study were therefore: (i) to capture and identify necrophagous hyperbenthic organisms in the study area using baited traps with other benthic macrofauna and fish excluded; (ii) identify spatial variations in the composition of any captured species within the study area; and (iii) compare the so-identified community with previous regional and local studies, notably an earlier investigation of temporal variations in scavenging within the Cape d'Aguilar Marine Reserve (Lee & Morton, Reference Lee and Morton2004).
MATERIALS AND METHODS
Site and station descriptions
The Cape d'Aguilar Marine Reserve is situated on the south-eastern tip of Hong Kong Island (22°12′N 114°15.5′E) and is regarded as having the cleanest waters locally (Environmental Protection Department HKSAR, 2008) (Figure 1A). The reserve encompasses the shallow Lobster Bay, which has two openings into it from the sea: to the east and the south-west. Because the Cape d'Aguilar Marine Reserve is located in the south-eastern wave-exposed quadrant of Hong Kong it is semi-exposed (Morton & Wu, Reference Morton and Wu1975), but is partly sheltered by the island of Kau Pei Chau. The local tidal regime is generally unequally semi-diurnal at the times of spring tides with a regular transition to diurnal at the times of neaps. Tidal currents and wave action have created a complex topographic pattern of gullies and sandbanks within the reserve and at low spring tides two ramparts are unveiled within it: one to the east, the other to the south-west (Morton & Harper, Reference Morton and Harper1995) and these demarcate Lobster Bay at the centre of the reserve.
(A) A map of Hong Kong showing the location of the Cape d'Aguilar Marine Reserve; (B) the depth contours. The cross-hatched areas represent ramparts. Scale bar: 10 km.
Six stations were designated in the present study (Figure 1B). Stations L1 and L2 were situated at a depth of ~–1.5 m Chart Datum (CD) inside Lobster Bay. Because these two stations abut the narrow eastern gap, they are constantly under the impact of waves from the east, particularly during the north-east monsoon in winter. Water movements here average 0.19 m·sec.−1, ranging between 0.06 and 0.63 m·sec.−1 (Clark, Reference Clark1997). Station L3 was also located along the south-western rampart of Lobster Bay at a depth of –0.5 m CD. Stations M1 and M2 were located at depths of ~–6 m CD to the south-west of Lobster Bay. These five sampling stations were over sand and a deeper mud station (D1) was established at a depth of ~–17 m CD yet further to the south-west but still inside the perimeter of the marine reserve. Details of the six stations and their respective granulometries are described by Lee (Reference Lee2004).
Trap design
Traps were made from 250-ml, transparent, bottles with an opening of 5 mm to exclude gastropods, fish, crabs and prawns from the traps because they are numerous in the study area and would have blocked the entrance (Morton & Chan, Reference Morton, Chan and Morton2000) and were not, anyway, the objects of interest.
The bait comprised 10 g pieces of crab tissue, as suggested by Sainte-Marie (Reference Sainte-Marie1986b), in this case Charybdis feriatus (Linnaeus, 1758) (Crustacea: Decapoda). This bait weight was also chosen because at the end of each trial, there were still large pieces of tissue remaining in the traps to ensure an ample supply of food for any trapped animals. The bait was put into 4.0 mm-mesh bags, large enough for scavenging hyperbenthos to access and burrow into the contained bait, but prevent the bait from detaching and clogging the mesh at the trap aperture. Empty mesh bags were placed in control bottles.
Trapping experiments
Two sets of five traps (comprising three baited and two un-baited controls) were deployed at a distance of between 20 and 30 cm from each other at each of the six stations for three days to avoid any catch biases arising from tidal conditions (Morritt, Reference Morritt2001) and the day–night cycle (Sainte-Marie, Reference Sainte-Marie1986a, Reference Sainte-Marieb, Reference Sainte-Marie1987). This experiment was repeated three times at time intervals of between two weeks and one month. Details of the trap design and trapping experiments are further elaborated upon in Lee (Reference Lee2004).
Statistical analyses
UNIVARIATE ANALYSES OF CATCHES
The numbers of individuals of each species captured per trap on each sampling occasion were calculated for every station. Data from the three baited bottles and two un-baited controls in each set of five traps were averaged separately to give a better mean and to avoid pseudo-replication. Total catches and the catches of each dominant taxon were examined for both normality and heterogeneity using Shapiro–Wilk and Bartlett's tests, respectively. If either were not normally distributed or had unequal variances, data were (x + 0.5)1/2-transformed, as suggested by Moore & Wong (Reference Moore and Wong1995, Reference Moore and Wong1996) and Morritt (Reference Morritt2001). Catches were subsequently compared between the six sampling stations using a two-way analysis of variance at a significance level of P = 0.05 using the SAS Release 8.02 software. If significant spatial distributions were identified, a posteriori pair-wise multiple comparisons using the Student–Newman–Keuls correction were carried out, again at P = 0.05.
MULTIVARIATE ANALYSES OF THE COMMUNITY OF ORGANISMS TRAPPED
Both classification and ordination techniques were employed to identify groups of biotically similar samples. An ascending hierarchical clustering analysis was conducted using group-average sorting with Bray–Curtis similarities upon fourth-root transformed data. This helped define communities spatially. Subsequently, non-metric multidimensional scaling was carried out to consolidate the results of the clustering dendrogram. Two-way crossed analysis of similarity (ANOSIM) was also accomplished to address statistically significant differences between the six stations and the two trap (baited versus un-baited) types. The species mainly responsible for any community difference were also identified using the SIMPER routine. All the multivariate analyses were performed using the PRIMER 5.2.0 statistical package.
RESULTS
Catch composition
An array of animals was captured over the course of the study (Table 1). A species of Tisbe (Harpacticoida: Tisbidae) accounted for 72.0% of the total number of all animals. The second major taxon was amphipods, dominated overwhelmingly by a species of Ceradocus (Gammaridea: Melitidae) (22.7%). These two crustaceans were most abundant not only in baited traps but also in the empty controls.
A list of hyperbenthos captured from the Cape d'Aguilar Marine Reserve, their abundances per trap and relative proportions over the course the study.
Other frequent but less numerous visitors to the traps included a species of Nebalia (Leptostraca: Nebaliacea) currently being described by Lee & Bamber (in preparation) (2.0%), benthic ostracods (1.0%) and Neanthes cricognatha (Ehler, 1904) (Polychaeta: Nereidae) (0.5%). In terms of species diversity, amphipods were the highest among all trapped taxa (>11 species). Among all retained visitors to the traps, only lysianassoids, Nebalia sp., N. cricognatha and benthic isopods occurred exclusively in baited traps.
Univariate analyses of catches
The presence of bait largely determined the total catches (Table 2). Baited traps captured significantly more organisms (mean ± SD = 45 ± 33 individuals·trap−1) than controls (10 ± 12 individuals·trap−1). Notwithstanding, the numbers of individuals trapped in the presence of bait at Stations M2 and D1 numerically exceeded the others (Figure 2). Further analysis of the data revealed that a spatial pattern was established only if empty control traps were used (Table 2). Among the six stations, the controls deployed at L1 and L2 within Lobster Bay significantly retained more organisms than L3 on the south-western rampart and the deeper-water stations (M1, M2 and D1).
Average faunal catches from (A) baited and (B) un-baited control traps during the spatial heterogeneity experiment in the Cape d'Aguilar Marine Reserve.
Statistical analyses (two-way analysis of variance (ANOVA) in the first row and one-way ANOVA in the second and third row) of possible spatial patterns in total catches for Stations L1, L2, L3, M1, M2 and D1, based on ([x + 0.5]1/2)-transformed data. Significant results at P <0.01 and 0.001 are indicated as two and three asterisks, respectively. Stations not significantly different from each other are underlined.
As the species of Tisbe, Ceradocus and Nebalia together accounted for >96% of the total number of organisms caught over the course of the study, more detailed data analyses were separately carried out for them. Data for Tisbe sp. have already been reported upon (Lee, Reference Lee2004). Catches of Nebalia sp. resembled Tisbe sp. in terms of both spatial distribution and difference between trap types (Figure 2) because both were attracted significantly to baited ones and had a preference for the south-western deeper-water stations (M1, M2 and D1) (Table 3).
Statistical analyses (two-way analysis of variance (ANOVA)) of catches of species of Ceradocus in the first row and (one-way ANOVA) of catches of Nebalia in the second row from the Cape d'Aguilar Marine Reserve (Stations L1, L2, L3, M1, M2 and D1). Data were ([x + 0.5]1/2)-transformed. Significant results at P < 0.001 are indicated by three asterisks. Stations not significantly different from each other are underlined.
1, as no Nebalia sp. individuals were captured in this study, only one-way analyses of catches across the six sampling stations were carried out.
No Nebalia sp. individuals were recorded from empty controls (Table 1), so that a spatial pattern was identified only for baited traps. Among catches from baited traps, Nebalia sp. was not recorded from the shallower Lobster Bay (L1 and L2) and the rampart station (L3) (Figure 2). Within the south-western deeper-water stations, numbers of Nebalia sp. caught at D1 were significantly greater than at M1 and M2.
Opposite to Nebalia sp., the catches of Ceradocus sp. did not differ between the two trap types (Table 1; Figure 2) and a reverse spatial distribution was identified. Catches were significantly higher at the two Lobster Bay Stations L1 and L2, followed by the shallow rampart Station L3 but much lower at the south-western deeper-water Stations M1, M2 and D1, irrespective of the presence or absence of bait.
Multivariate analyses of the organisms trapped
Results from ANOSIM showed that species composition differed significantly between baited and control traps (Table 4).
Statistical multivariate analysis (two-way crossed analysis of similarity) of the taxa trapped at Stations L1, L2, L3, M1, M2 and D1 using baited and un-baited control traps. This test examined for any possible similarity between stations and between trap types. The pair-wise tests identify the possible differences between any two stations if both baited and empty control traps are taken into account. Significant results at P < 0.01 are indicated by two asterisks.
According to the results of the SIMPER analyses, the dissimilarity index between them was 70.5% and Tisbe sp. accounted mainly for this difference. Table 4 also shows that if both baited and empty control traps were considered, two major station groups were demarcated, that is: (i) shallow Stations L1, L2 and L3; and (ii) Stations M2 and D1. The species composition at M1 was significantly different from all the shallow water stations and D1, but not M2. Benthic ostracods and Tisbe sp. contributed to community differences between M1 and M2–D1 because they were more abundant at the latter. Results from SIMPER analyses, however, show that the dissimilarity index between the shallow (Lobster Bay) sites and the station group of M2–D1 (55.9%) approximated those between the shallow ones and M1 (53.9%). This was only 39.0% between M1 and the other two deeper stations. This means the species composition of catches among the six stations could be allocated mainly to two major groups, that is: (i) the shallow Lobster Bay (L1, L2) and rampart (L3) stations; and (ii) the south-western deeper-water ones (M1, M2 and D1).
If only baited traps were considered, the same two clusters were also formed with one of them comprising all the deeper-water stations (M1, M2 and D1), the other the shallower water stations (L1, L2 and L3) (Figure 3A). Similarly, two clusters are identified in Figure 3B with the shallow Lobster Bay and rampart stations on the left, and the south-western deeper-water stations on the right. Both Ceradocus sp. and Nebalia sp. accounted for the substantial community differences identified between the shallow and deeper waters. Ceradocus sp. dominated the former station group, while Nebalia sp. dominated the trapped fauna in the latter.
Multivariate analyses of the taxa caught using baited traps in the spatial heterogeneity experiment conducted in the Cape d'Aguilar Marine Reserve. (A) A dendrogram of ascending hierarchical agglomerative clustering analysis. Data were fourth-root transformed to compute the Bray–Curtis similarities using group-average linking; (B) a two-dimensional multidimensional scaling (MDS) configuration. The stress value is indicated at top right. Symbols are presented as station name followed by first ‘1’, second ‘2’ or third ‘3’ times of sampling. For the empty control traps, samples from Station M1 and sometimes Station M2 acted as outliers, and community composition here appeared to be more variable and, hence, dispersed virtually throughout the MDS configuration (Figure 4).
This could be partly attributed to low catches, as shown in Figure 2. Apart from the outliers, a demarcation in trapped hyperbenthic community composition was obvious. The dissimilarity index between the shallow and deeper-water station groups varied between 81.21% and 83.05%. Ceradocus sp. mainly accounted for community differences between the shallow and the deep-water stations with faunal dominance of the former station group. No short-term temporal effects over the course of the study period could be identified (Figures 3 & 4).
Multivariate analyses of the faunal community retained using un-baited control traps in the spatial heterogeneity experiment conducted in the Cape d'Aguilar Marine Reserve. (A) A dendrogram of ascending hierarchical agglomerative clustering analysis. Data were fourth-root transformed using group-average linking to calculate Bray–Curtis similarities; (B) a two-dimensional multidimensional scaling configuration. The stress value is indicated at top right. Notations are the same as in Figure 3. Two data points were missing because both controls on a trap-set at Stations M2 and D1 did not capture any organisms.
DISCUSSION
Species composition and comparisons with other locations
The results of the present study provide further evidence in support of the hypothesis of Biernbaum & Wenner (Reference Biernbaum and Wenner1993) that at lower latitudes a coastal necrophagous assemblage in baited traps is not necessarily dominated by lysianassoids. This study has demonstrated that, in a decreasing order of importance, Tisbe sp., Nebalia sp., benthic ostracods, Neanthes cricognatha and Lepidepecreum sp. are important scavenging members of the hyperbenthic fauna of the Cape d'Aguilar Marine Reserve, if fish and macrofana are excluded, because their catches using baited traps exceeded those of controls. The trapped lysianassoids accounted for only ≤0.5% of the total captured fauna and have not been trapped locally before. It should be stressed that the trapped fauna herein reported upon is likely to represent secondary arrivals, following larger, more agile fish, crabs, prawns and gastropods (Britton & Morton, Reference Britton, Morton and Morton1993, Reference Britton and Morton1994). This study, however, provides the first information on a hyperbenthic community of scavenging crustaceans and polychaetes in the shallow coastal waters of Hong Kong and the wider Indo-West Pacific subtropics.
Nishida et al. (Reference Nishida, Kikuchi and Toda1999) reported that the trapped fauna might be related to trap entrance aperture size. A 5.5 mm diameter opening allowed the entrance of both lysianassoid amphipods and copepods to baited traps in the deep sea off Japan, leading to intensive predation of the latter by the former. A 2.0 mm aperture size, on the other hand, allowed copepods to pass through but prevented lysianassoids from entering, resulting in a predominance of the former in the traps. Both lysianassoids and copepods were, therefore, suggested to be necrophages exhibiting a mutually exclusive and competitive relationship. Opposite results in this study using traps with a 5.0 mm opening were obtained with a numerical preponderance of Tisbe sp. over the melitid amphipod Ceradocus sp. Although Ceradocus sp. appeared to have a negative correlation with Tisbe sp., insignificant differences in catches between baited and control traps indicated that the former was attracted to the structural habitat complexity provided by the trap and not the carrion in it nor the potential prey value of the copepod already inside it. Ceradocus sp. is, hence, not a scavenger.
Spatial distribution of necrophagous communities in Hong Kong
This study shows that all identified scavenging hyperbenthic species, including Tisbe sp. (Lee, Reference Lee2004), were generally caught in increasing numbers with increasing depth. Conversely, Ceradocus sp. replaced the other scavenging invertebrates in the shallows, regardless of the presence or absence of bait. Small-scaled spatial segregations in amphipod community composition (Sekiguchi & Yamaguchi, Reference Sekiguchi and Yamaguchi1983) have been usually attributed to various physical factors, such as substratum type (Kirkwood & Burton, Reference Kirkwood and Burton1988), particle size (Robertson et al., Reference Robertson, Hall and Eleftheriou1989), water depth, seabed topography, shoreline characteristics and current speed (Ingram & Hessler, Reference Ingram and Hessler1983; Arntz et al., Reference Arntz, Brey and Gallardo1994). In addition, the fact that no significant differences were identified between trap types for Ceradocus sp. indicates that it is not attracted to the bait but to the substratum complexity offered by the traps. Ceradocus sp. therefore appears to favour topographic heterogeneity within the shallow, spatially heterogeneous, Lobster Bay because boulders and cobbles, even the epilithic fauna, create many niches into which it can retreat. Those traps herein deployed thus acted solely as ‘habitat traps’ for this species. Griffiths et al. (Reference Griffiths, Mayfield and Branch2000) reported that lobsters entering baited traps were not attracted to the bait but to isopods also drawn to and concentrated there, another example of the bias of ‘habitat’ trapping. However, this does not seem to be the case for Ceradocus sp. (this study).
Low catches of scavengers in baited traps in the shallows may result from the exposed nature of Lobster Bay. The eastern gap between the Cape d'Aguilar Peninsula and island of Kau Pei Chau exposes the enclosed Lobster Bay to strong wave action, particularly in winter. Turbulent waters would act to disperse any chemical cues emanating from carrion and reduce the ability of scavengers to detect potential food sources (Britton & Morton, Reference Britton, Morton and Morton1993). However, the scavenging gastropods Ergalatax contractus (Reeve, 1846) and Nassarius pauperus (Gould, 1850) were caught in ‘their hundreds’ using baited traps in Lobster Bay (Morton & Chan, Reference Morton, Chan and Morton2000). The spatial segregation of the trapped fauna might also be affected by differences in the amount of organic carbon between shallows and deeper-waters. Opportunistic scavengers are thought to supplement their natural diets with carrion (Britton & Morton, Reference Britton, Morton and Morton1993, Reference Britton and Morton1994; Kaiser & Moore, Reference Kaiser and Moore1999). Significantly higher organic content at stations beyond the western ramparts of Lobster Bay, particularly at Stations M2 and D1, might support more hyperbenthos in the absence of carrion. Composition of the fauna caught by traps in Lobster Bay contrasted sharply with that in the water column, and only Tisbe sp. has been obtained locally using plankton nets, where it accounts for <1% of total zooplankter abundance (Lee & Chen, Reference Lee, Chen and Morton2003).
The trapped hyperbenthic fauna also differed markedly from the macrobenthic community of the study area identified by Morton & Harper (Reference Morton and Harper1995, Reference Morton, Harper and Morton1997), as well as in the nearby Tai Tam Bay (Ong Che & Morton, Reference Ong Che and Morton1991, Reference Ong Che and Morton1995). This may be partly attributable to the different sampling methods employed. In the present study, >10 species of gammarideans were identified, but only one, Maera pacifica Schellenberg, 1938, has been reported from Hong Kong previously (Moore, Reference Moore and Morton1990). Although knowledge of Hong Kong gammaridean amphipods (Hirayama, Reference Hirayama and Morton1990a, Reference Hirayama and Mortonb, Reference Hirayama1991; Moore, Reference Moore and Morton1990, Lowry, Reference Lowry2000; Horton, Reference Horton2008) is fragmentary, resulting in the novel species recorded from this study, the dramatic difference between the benthic community and the trap samples supports the contention of Shulenberger & Hessler (Reference Shulenberger and Hessler1974) and Nishida et al. (Reference Nishida, Kikuchi and Toda1999) that trapping with bait is an efficient means of collecting dispersed, swimming and elusive hyperbenthic species obtained only rarely by other methods.
ACKNOWLEDGEMENT
The authors would like to thank Dr B. Sainte-Marie for advice on trap design.
