The effects of seagrass (Zostera japonica) canopy structure on associated fauna: a study using artificial seagrass units and sampling of natural beds

Journal of Experimental Marine Biology and Ecology 259 Ž2001. 23–50 www.elsevier.nlrlocaterjembe

The effects of seagrass žZostera japonica/ canopy structure on associated fauna: a study using artificial seagrass units and sampling of natural beds S.Y. Lee a,) , C.W. Fong a , R.S.S. Wu b a

The Swire Institute of Marine Science and Department of Ecology and BiodiÕersity, The UniÕersity of Hong Kong, Cape d’Aguilar, Hongkong, People’s Republic of China b Department of Biology and Chemistry, City UniÕersity of Hong Kong, Tat Chee AÕenue, Kowloon, Hongkong, People’s Republic of China Received 20 July 2000; received in revised form 8 February 2001; accepted 9 February 2001

Abstract The importance of seagrass canopy to associated fauna was assessed by comparing the species richness, abundance and diversity of the epi- and infaunal macroinvertebrate assemblages in a seagrass Ž Zostera japonica Ascherson and Graebner. bed and the adjacent unvegetated area in Hong Kong. Seagrass cover had significant effects on the composition and abundance of the associated fauna and the amount of detritus accumulated on the sediment surface. Detritus abundance was significantly higher in the seagrass bed, and was positively correlated with both the above- and belowground biomass of Z. japonica. Both the abundance and species richness of the epi- and infauna were significantly positively correlated with the belowground biomass of the seagrass and detritus standing crop. Macrofaunal species richness was higher Ž118. in the seagrass bed than the adjacent unvegetated areas Ž70., with a higher degree of similarity between the infauna than the epifauna of the two habitats. While all species recorded from the unvegetated areas were found in the seagrass bed, 48 species occurred only in the seagrass-covered areas. Species richness of epifauna was significantly higher in the seagrass bed, but there was no difference between infaunal species of the two habitats. On the contrary, faunal Žepi- and infauna. abundance was significantly higher in seagrass areas. The seagrass bed also supported species of

)

Corresponding author. Present address: School of Environmental and Applied Sciences, Griffith University Gold Coast, PMB 50, Gold Coast Mail Centre, Queensland Qld 9726, Australia. Tel.: q61-7-55528886; fax: q61-7-55528067. E-mail address: [email protected] ŽS.Y. Lee.. 0022-0981r01r$ - see front matter q 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 2 2 - 0 9 8 1 Ž 0 1 . 0 0 2 2 1 - 0

24

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

small tellinid bivalves previously not recorded from Hong Kong. Artificial seagrass units ŽASUs, 0.2 m2 . with four combinations of leaf density and leaf length and a control Žbare sand. were placed at short distances from natural patches of Z. japonica. The composition, abundance and biomass of the epibenthos associated with the ASUs and the control were recorded after 3 months in the field. While species richness did not differ among the treatments, total abundance of epibenthos was significantly higher in the high density–long leaves ŽHL. treatment than in the control. Results of a discriminant analysis using log-transformed abundance data suggest that the gastropod Clithon oualaniensis, the mussel Musculista senhousia and the crab Thalamita sp. were important species distinguishing the assemblages in the various treatments. All the three species were significantly more abundant in the HL treatment than in the low density–short leaves ŽLS. treatment and the control. By contrast, there was no significant difference in the biomass of the epifauna, but discriminant analysis again separated the five treatments based on the composition of the biomass, with the same three species identified as the most important discriminating species. The species richness and abundance of the epifauna associated with the ASUs were similar to the adjacent unvegetated areas, but significantly lower than in the Zostera patches. The physical canopy structure of Z. japonica beds increased the abundance of the epibenthos, potentially through provision of canopy and indirectly through trapping of detritus. q 2001 Elsevier Science B.V. All rights reserved. Keywords: Zostera japonica; Canopy structure; Artificial seagrass units; Associated fauna

1. Introduction Species diversity and abundance of animals are usually greater in seagrass beds than in unvegetated habitats ŽHutchings et al., 1991; Kirkman et al., 1991; Orth, 1992; Edgar et al., 1994.. Compared with intertidal mud-flats, sand-flats and salt marsh habitats, seagrass beds often accommodate higher species richness, abundance, biomass and production of macroinvertebrates ŽHeck et al., 1995.. Many workers have shown that associated fauna are not distributed randomly in seagrass habitats, but that their abundance is correlated with seagrass bed structure, namely, shoot density ŽWebster et al., 1998 and references therein; see also Attrill et al., 2000. and biomass ŽAttrill et al., 2000 and references therein.. The presence of seagrasses may increase the abundance of organisms by increasing Ži. the amount of physical structure usable as living space; Žii. the number of microhabitats; Žiii. sediment deposition and stabilisation; Živ. food resources and Žv. protection from predators. Seagrasses may also reduce hydrodynamic forces ŽLewis, 1984.. The ability of seagrass beds to fulfil the majority of these roles is well documented ŽFonseca et al., 1982; Robertson, 1984; Walker and McComb, 1985; Bell and Westoby, 1986; Edgar, 1990; Walker et al., 1991; Edgar et al., 1994; Koch, 1996; Komatsu, 1996., through both biological Že.g. seagrass production. and physical Že.g. protection offered by canopy structure. modifications of the habitat by seagrass presence. Removal of seagrass canopy, alterations to seagrass canopy height and density are common experimental manipulations performed to assess the effects of seagrass physical structure on faunal distribution and abundance Že.g. Edgar and Robertson, 1992; Connolly, 1995; Connolly and Butler, 1996; Horinouchi and Sano, 1999.. Although field

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

25

manipulations of seagrass can help investigate the relationship between seagrasses and their associated fauna, there are disadvantages of this approach ŽBell et al., 1985.. For example, the duration of the manipulative experiment could be limited by the growth rate of the seagrass and repeated removal of the canopy may be necessary. In most cases, the reduced seagrass density and canopy height will recover within a week through leaf regrowth. Artificial seagrass units can, however, be used to overcome these problems. Artificial seagrass units mimicking the physical structure of real meadows have been used previously in manipulative experiments, especially those studying the colonization of seagrass-associated fauna and epiphytic algae Že.g. Barber et al., 1979; Bell et al., 1985; Virnstein and Curran, 1986; Almasi et al., 1987; Lethbridge et al., 1988; Lee and Low, 1991; Edgar and Robertson, 1992; Sogard and Olla, 1993; Ceccherelli and Cinelli, 1999; Kenyon et al., 1999.. Artificial seagrass is often used in field experiments to investigate the relationship between associated fauna and the physical structure of the seagrass, especially canopy structure. This approach eliminates the potentially confounding factor of biochemical interactions between damaged seagrasses and the colonising assemblage. Artificial seagrass mimics can also provide large flexibility in experimental design, as seagrass presence, canopy density and height can be easily manipulated depending on the experimental requirements. However, an obvious disadvantage of the artificial seagrass approach is the inability to provide a comprehensive evaluation of seagrass Žbiotic and physical. influence, as significant biotic interactions can occur between seagrass and their associated flora and fauna Že.g. Pinckney and Micheli, 1998.. Zostera japonica is a small-sized Žleaf blades length - 20 cm; width - 3 mm. intertidal seagrass common on the West Pacific coast of Asia ŽPhillips and Menez, ´ 1988; Shin and Choi, 1998.. Compared to other seagrasses in the Indo-Pacific, the value Žtrophic versus physical protection. of Z. japonica to coastal consumer assemblages is poorly known ŽFong, 1998.. Animals associated with seagrass beds can benefit both from the food resources available, i.e. detritus and algae, or the protection the aboveand belowground seagrass structure provide towards predators and extreme environmental conditions. This study assessed the importance of aboveground seagrass structure Žshoot density, canopy height. on the diversity and abundance of the associated fauna of a Z. japonica bed in Hong Kong. Sampling of the natural seagrass bed and artificial seagrass units were employed to evaluate the importance of seagrass canopy to the associated biota.

2. Materials and methods 2.1. Site description The study was conducted at Lai Chi Wo, located in Crooked Harbour, northeast Hong Kong Ž114816X E; 22832X N.. Refer to Fong et al. Ž2000. for a detailed description of the site.

26

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

2.2. Associated fauna of Z. japonica The associated epifaunal and infaunal organisms of the seagrass bed at Lai Chi Wo were sampled using a 10-cm diameter corer. Sampling was conducted at Lai Chi Wo from March 1995 to March 1996 during spring low tide periods. June 1995 was not sampled because of bad weather. The corer was pushed into the sediment to a depth of 10 cm to obtain the samples. The aboveground components of the seagrass was included in the core as they were cut through by the corer, no effort was made to include all leaves that were originated from ramets enclosed by the corer. Fifteen samples were taken randomly from within the Z. japonica bed every month, and another 15 samples were taken from an adjacent unvegetated area bimonthly. The samples were washed over a 0.5-mm sieve with running seawater to collect the macrofauna. Detrital material present in the samples was collected for biomass measurements. The biomass of above- and belowground portions of the seagrass were determined separately and as described in Fong et al. Ž1998.. The dry and ash-free dry weight values of detritus retained on a 0.5-mm sieve were obtained by drying at 808C for 48 h and combustion at 5508C for 4 h, respectively. The fauna was fixed in 10% buffered formalin immediately after sieving. To increase the efficiency of faunal sorting, all samples were stained with rose Bengal for at least 24 h. Animals were sorted, identified and counted under a dissecting microscope. Epifaunal and infaunal animals were distinguished in the samples based on a knowledge of their biology. Diversity index Ž H X , Shannon–Wiener index; Shannon, 1948., evenness ŽPielou, 1966. and Sorensen’s index of similarity ŽSorensen, 1948. values for the associated faunal community of the seagrass bed and adjacent unvegetated areas were calculated. Qualitative ŽQA% s percentage of species belonging to a taxonomic group. and quantitative ŽQI% s percentage of individuals belonging to a taxonomic group. dominance values for the community were also calculated. These community parameters Ždiversity index, evenness, similarity index, qualitative and quantitative dominance. were calculated for each monthly sample. 2.3. Experiment employing artificial seagrass units (ASU) Artificial seagrass units ŽASUs. were used to mimic a small area of Z. japonica and to evaluate the importance of physical canopy structure to associated fauna abundance. The shoot density, number of leaves per shoot, width and length of leaves of the ASUs were set at pre-determined levels, based on the results of a preliminary survey. The numbers of shoots and leaves per shoot in three relatively dense 0.2 m2 patches of Z. japonica were counted in situ. Fifty leaves were clipped at the base from each patch for the measurement of leaf width and length. Based on the results of the survey, four treatments using ASUs, namely: Ža. high density–long leaves ŽHL.; Žb. high density– short leaves ŽHS.; Žc. low density–long leaves ŽLL.; Žd. low density–short leaves ŽLS., and a control ŽC., were established for the experiment. The density of artificial seagrass leaves in treatments HL and HS Ž2025 shootsr0.2 m2 . was about 85% that of dense natural seagrass patches Ž2395 shootsr0.2 m2 . at Lai Chi Wo. The low-density

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

27

Table 1 Results of a two-way ANOVA to investigate the effect of time Žmonth. and habitat on the number of species of fauna Žepi- and infauna. within and outside the seagrass bed at Lai Chi Wo Source of variance

df

SS

MS

F

P

Number of species of epifauna Habitat 1 Time 6 Interaction 6 Error 197

121.3 233.3 94.7 1337.7

121.3 38.88 15.78 6.79

18.77 3.28 7.89

- 0.0001 0.0004 0.0879

Number of species of infauna Habitat 1 Time 6 Interaction 6 Error 197

25.5 741.3 86.4 2193.2

25.5 123.6 14.4 11.13

2.41 6.36 3.17

0.1223 - 0.0001 0.1672

treatments LL and LS had only 1r4 the density of the HL and HS treatments, i.e. 506 shootsr0.2 m2 , equivalent to ca. 21% of the density in dense natural patches. Natural low-density patches of Z. japonica are not common at Lai Chi Wo, but are found in more marginal habitats in Hong Kong, e.g. Lee Ž1997.. The length of long artificial seagrass leaves used was 15 cm, i.e. slightly longer than the mean Ž13.1 " 4.4 cm. of the natural leaves. Half of the length of the long artificial seagrass leaves, i.e. 7.5 cm, was set as the size for the short leaves treatment. The base frame of the ASUs was constructed using a 45 = 45 cm Žs 0.2 m2 . plastic mesh of 1-cm ‘pore’ size, to which evenly spaced artificial shoots were sewn. A 1-mm mesh size net was added to the underside of the base frame to prevent the loss of fauna during retrieval of the units after the experiment. The fine mesh was only loosely fitted to the plastic mesh, allowing sand to freely fill the space among individual artificial seagrass shoots. Each ‘seagrass’ shoot consisted of three strips of 1.5–2 mm wide green nylon ribbon similar to Z. japonica leaves in width and arrangement. The control units had the base plastic mesh only. Each treatment had five replicates, and a total of 25 ASUs were constructed. The ASUs were deployed in December 1996 in an unvegetated area at about 5 m seaward from the boundary of the Z. japonica bed. Due to the gentle gradient of the shore, the tidal position of the AUSs was - 5 cm lower than that of the natural beds. U-shaped steel rods were used to anchor the ASUs into the substratum and the base of the ASUs was buried in about 5 cm of sand, leaving the ribbons above the sand surface. The experiment was a completely randomised block design, with five treatments grouped together in a block, and the position of the treatments orientated differently in the five blocks. The blocks were placed ca. 15 m apart from each other. To examine the influence of seagrass canopy density and height on the epifaunal assemblage, the species number and biomass of associated epifauna were compared. The ASUs were retrieved after 3 months in the field, which should allow enough time for colonisation of the macrobenthos Žfor example, Kenyon et al., 1999 were able to record significant

28

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

settlement of crustacean larvae in ASUs in 24 h.. Upon retrieval, each unit, with the sediment on the frame, was placed in a large plastic bag and transferred to the laboratory immediately. All animals in the ASUs were fixed in 10% buffered formalin and then sorted, identified and counted under a dissecting microscope. The biomass of the epifauna was obtained by drying the samples at 808C for 48 h. 2.4. Statistical analysis 2.4.1. Associated fauna in natural seagrass beds The biomass of detritus, species richness and abundance of the epifauna and infauna collected from the seagrass bed at Lai Chi Wo and an adjacent unvegetated area were compared using a two-way ANOVA with month and habitat as factors. The data on the numbers of individuals of epifauna and infauna were log 10 Ž y q l., transformed prior to analysis in order to satisfy the assumptions of normality and homoscedasticity. Factors detected to be significant by ANOVA were further analysed using a posteriori Student– Newman–Keuls ŽSNK. tests set at the 5% significance level. The quantitative relationships between species richness and faunal abundance Žseparately for epifauna and infauna. and detritus, above- and belowground seagrass biomass were explored by forward stepwise regression. Data were pooled from the seagrass bed and adjacent unvegetated areas for analysis, to provide data points covering a wide range of seagrass conditions. Log 10 transformation of regression variables was performed, if necessary, to meet the regression assumptions. Discriminant analysis was used to describe the major difference in abundance and biomass of epifaunal species among the five treatments of the artificial seagrass units. The data on the abundance and biomass of epifauna were log 10 Ž y q 1., transformed before analysis so as to reduce the influence of extreme values. The most important variables Žepifaunal species. for treatment discrimination, as suggested by the discriminant analysis, were used to test the difference among the five treatments by a one-way ANOVA. Significant differences between treatments were further identified using SNK tests. Transformed biomass of detritus satisfied the regression assumptions. The total abundance and biomass of the animals in the five treatments were compared by a one-way ANOVA, after ensuring that there was no violation of the assumptions. Significant differences between treatments were further identified using SNK tests.

3. Results 3.1. Species richness The macrofaunal species richness in the seagrass bed was greater than in the adjacent unvegetated area. A total of 118 species was collected from the seagrass bed, compared to 70 from the adjacent unvegetated area. All the faunal species recorded from the unvegetated area were also found in the seagrass bed. The epi- and infauna of the

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

29

Fig. 1. Monthly changes in the number of epifaunal and infaunal species associated with the seagrass and adjacent unvegetated areas at Lai Chi Wo. Data presented are mean"S.E. Ž ns15.. The diameter and area of the corer are 10 cm and 78.5 cm2 , respectively.

30

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

Fig. 2. Monthly changes in the abundance of epifaunal and infaunal species associated with the seagrass and adjacent unvegetated areas at Lai Chi Wo. Data presented are mean"S.E. Ž ns15.. The diameter and area of the corer are 10 cm and 78.5 cm2 , respectively.

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

31

Table 2 Result of a two-way ANOVA to investigate the effects of time Žmonth. and habitat on the number of individuals of fauna Žepi- and infauna. within and outside seagrass beds at Lai Chi Wo Source of variance

df

F

P

Number of indiÕidual of epifauna (log 10 y q l transformed) Habitat 1 4.70 4.7015 Time 6 2.01 0.3350 Interaction 6 0.63 0.1050 Error 195 13.58 0.0656

SS

MS

71.67 2.79 2.01

- 0.0001 0.0021 0.1053

Number of indiÕidual of infauna (log 10 y q l transformed) Habitat 1 1.71 1.7140 Time 6 5.76 0.9600 Interaction 6 4.94 0.8233 Error 195 20.28 0.104

17.49 5.35 4.36

- 0.0001 - 0.0001 0.0610

seagrass bed comprised similar species numbers, at 62 and 56, respectively. Epifaunal species Ž n s 31. were fewer than infaunal ones Ž n s 39. in the unvegetated area. The total number of animal species collected in the seagrass bed was low during summer Ž; 60. but higher in late winter and early spring Ž) 70.. The total number of species recorded from the unvegetated area fluctuated within narrow limits, except for March, and ranged from 33 to 38. The dominant groups of associated fauna in both habitats were the Polychaeta, Bivalvia and Gastropoda. Their mean qualitative dominances were 35%, 18% and 19% in the seagrass bed and 43%, 19% and 20% in the unvegetated area, respectively. These taxonomic groups comprised nearly 80% of the number of individuals of the associated fauna in both habitats. The number of epifaunal species showed significant differences with time Žtwo-way ANOVA: df s 1r6, F s 3.28, P s 0.0004. and between habitats Ž df s 1r6, F s 18.77, P - 0.0001; Table 1., being significantly higher in the seagrass bed than in the adjacent unvegetated area ŽSNK test, P - 0.05; Table 1 and Fig. 1.. The number of infaunal species showed significant differences with time Žtwo-way ANOVA: df s 1r6, F s 6.36, P - 0.0001., but not between habitats Ž df s 1r6, F s 2.41, P s 0.1223; Table 1.. The number of epifaunal and infaunal species was not significantly affected by the interac-

Table 3 Value of Sorensen’s index of similarity for the epifauna and infauna between seagrass and adjacent unvegetated areas

March 1995 May July September November January 1996 March

Epifauna

Infauna

0.566 0.571 0.632 0.421 0.545 0.571 0.591

0.828 0.745 0.830 0.720 0.878 0.740 0.902

32

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

tion between time and habitat. Species richness was low for both epifauna and infauna in summer ŽSNK test, P - 0.05; Table 1 and Fig. 1.. 3.2. Faunal abundance Total faunal abundance was much higher in the seagrass area than in the unvegetated area throughout the study period. The density of epifauna in the seagrass bed and the unvegetated area ranged from ; 5000 to 13,000 individuals my2 and from ; 2500 to 3500 individuals my2 , respectively. Infaunal density ranged from ; 12,000 to 40,700 individuals my2 in the seagrass bed and from ; 3200 to 22,300 individuals my2 in the unvegetated area. A relatively lower total faunal abundance was generally recorded from both habitats in summer ŽFig. 2.. The numerically dominant taxa in both habitats were the Polychaeta, Oligochaeta and Gastropoda, their mean quantitative dominance being 42%, 27% and 15% in the seagrass bed and 65%, 9% and 14% in the unvegetated area,

Fig. 3. Seasonal variation in the Shannon–Wiener diversity index and evenness values for the seagrass and unvegetated areas at Lai Chi Wo.

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

33

Fig. 4. Monthly changes in dry weight ŽDW. and ash-free dry weight ŽAFDW. Žmean"S.E., ns15. of detritus in the seagrass and adjacent unvegetated areas at Lai Chi Wo.

respectively. These groups together made up ) 80% of the number of individuals in both habitats. The number of epifaunal individuals was significantly different at different sampling times Žtwo-way ANOVA: df s 1r6, F s 2.79, P s 0.0021. and between habitats Ždf s 1r6, F s 71.67, P - 0.0001; Table 2.. The number of infaunal individuals also showed significant differences in time Žtwo-way ANOVA: df s 1r6, F s 5.35, P 0.0001. and between habitats Ž df s 1r6, F s 17.49, P - 0.0001; Table 2.. Both the numbers of epifaunal and infaunal individuals were significantly higher in the seagrass bed than in the adjacent unvegetated area. The abundance of both groups was also higher in winter ŽSNK test, P ) 0.05; Table 2 and Fig. 2.. Table 4 The results of a two-way ANOVA on the effect of time of sampling and habitat on the biomass of detritus accumulated per unit area Biomass of detritus Source of variance Habitat Time Interaction Error

df 1 11 11 155

SS 11.57 2.08 1.67 32.95

MS 11.572 0.190 0.152 0.159

F 72.7 1.19 0.94

P - 0.0001 0.2951 0.3152

34 Table 5 Mean abundance Ž N . Žindividuals my2 , mean"1 S.D.. and species richness Ž S . of the associated epifauna collected in artificial seagrass units ŽASUs, with four treatments and one control.

Batillaria zonalis Cerithidea cingulata Cer. coralium Clithon oualaniensis C. sowerbiana Helice sp. Hermit crab Lunella coronata Macrophthalmus sp. Musculista senhousia Nassarius festiÕus Onchidium Õerraculatum Polinices didyma Thalamita sp. Uca sp. Total mean N Total number S

Zostera bed areaa

Adjacent unvegetated areas a

246.2"341.9 25.5"71.4 17.0"65.7 3285.0"2248.6 25.5"71.4 0 34.0"75.6 0 0 1808.0"3237.3 0 0 0 0 0

67.9"64.4 280.1"416.5 0 280.1"426.1 50.9"113.9 0 0 0 8.5"19.0 577.2"1267.1 25.5"56.9 0 0 0 0

1005.4"441.0 267.7"304.3 644.9"805.9 522.5"239.7 73.1"142.4 1.0"2.2 26.7"36.6 23.7"22.7 3.0"6.6 654.8"289.0 67.2"112.0 4.9"8.6 3.0"4.4 8.9"6.4 1.0"2.2

744.7"463.1 l86.7"106.7 264.7"251.5 392.1"125.3 14.8"25.7 2.0"4.4 6.9"8.3 6.9"15.5 0 581.7"316.2 12.8"17.7 1.0"2.2 1.0"2.2 5.9"4.1 3.0"6.6

1214.8"594.1 308.1"413.8 242.0"246.5 237.0"132.4 39.5"49.0 1.0"2.2 14.8"25.7 11.9"11.4 4.9"11.0 414.8"188.9 27.7"26.3 2.0"4.4 1.0"2.2 1.0"2.2 2.0"2.7

953.1"746.4 338.8"461.6 141.2"132.1 216.3"132.4 63.2"119.7 0 17.8"24.6 11.9"16.6 9.9"22.1 251.9"67.3 73.1"120.1 2.0"2.7 0 0 2.0"4.4

459.3"415.5 175.8"156.5 118.5"100.2 81.0"42.7 7.9"17.7 0 15.8"18.3 6.9"6.6 3.0"6.6 53.3"48.8 54.3"38.6 0 0 0 0

11204.5"7922.3 28

3896.1"4424.9 16

3307.7"781.4 15

2224.2"527.8 14

2522.5"973.6 15

2081.0"970.4 12

975.8"505.3 10

ASUs high density, long leaves ŽHL. b

High density, short leaves ŽHS.

Low density, long leaves ŽLL.

Data on the same taxa from Z. japonica bed and adjacent unvegetated areas at Lai Chi Wo are included for comparison. a Samples collected in March 1996 Ž ns15.. b Samples collected in March 1997 Ž ns 5..

Low density, short leaves ŽLS.

Control ŽC.

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

.

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

35

3.3. DiÕersity index and similarity Higher values of similarity based on Sorensen’s index were recorded for the infauna Ž S ranged from 0.72 to 0.90. than for the epifauna Ž S ranged from 0.42 to 0.63., between the seagrass and adjacent unvegetated area ŽTable 3.. The lowest similarity between the two habitats occurred in September for both the infauna and epifauna. The Shannon–Weiner index was relatively low in both habitats Ž H X - 1.3.. Low values were recorded in summer, but higher values were recorded in winter in the seagrass bed. The reverse pattern was recorded in the adjacent unvegetated area. This resulted in a higher diversity index in the seagrass bed than in the unvegetated area in winter, but a similar value in summer. Evenness in the two habitats was similar ŽFig. 3.. 3.4. Relationship between the associated fauna, detritus abundance and seagrass biomass The biomass of detritus fluctuated greatly in the adjacent unvegetated area, but less so in the seagrass bed ŽFig. 4.. The mean biomass of detritus ranged between 1295.2 " 2466.0 and 601.5 " 532.0 g dry weight my2 Žmean " S.D.. in the seagrass bed and between 253.7 " 148.0 and 61.55 " 33.32 g dry weightP my2 in the adjacent unvegetated area during the study period. The mean biomass of detritus was significantly different between the two micro-habitats Žtwo-way ANOVA df s 1r11, F s 72.7,

Fig. 5. A comparison of the total abundance and biomass of epifauna collected in the various treatments of the artificial seagrass experiment at Lai Chi Wo Žmean"S.E., ns 5..

36

Ža. Summary statistics for the first canonical discriminant function Žfunction 1. Function

Eigenvalue

Percent of variance

Canonical correlation

Wilks’ l

x2

df

P

1

34.6964

80.87

0.9859

0.0016

89.946

60

0.0074

Žb. Wilks’ lambda ŽU-statistic., the significance of the univariate F-ratio Žwith 4 and 20 degrees of freedom., standardized canonical discriminant function coefficients ŽSCDFC. and pooled within-groups correlations ŽPWGC. between discriminating variables for the first discriminant function. Only variables that are statistically significant Ž P - 0.05. are presented. Key to species: Ts— Thalamita sp.; Mus— M. senhousia; Pd— P. didyma; Clo— C. oualaniensis

Wilks’ l F-ratio SCDFC PWGC

Ts

Pd

Clo

0.2842 0.0000 3.5851 0.2021

– – 2.2797 –

0.4082 0.0009 y2.6742 0.2267

Only statistically significant Ž P - 0.05. values are shown.

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

Table 6 Results of a discriminant analysis on the artificial seagrass experiment based on the abundance of epifauna

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

37

P - 0.0001; Table 4., but not between different sampling times Žtwo-way ANOVA df s 1r11, F s 1.19, P s 0.295; Table 4.. There was no significant interaction between the two factors. The mean biomass of detritus was significantly higher in the seagrass bed than in the adjacent unvegetated area Ž P - 0.05.. The biomass of detritus was significantly correlated with the biomass of the above- and belowground components of Z. japonica. A significant linear regression equation relating detritus biomass ŽD., to biomass of aboveground ŽA. and below-ground ŽB. Žall in g dry weight my2 . components of Z. japonica was obtained, as follows: log 10 Ž D q 1 . s y0.0591 q 0.625 A q 0.370 B ; r 2 s 0.507, P - 0.0001, n s 195 The amount of detritus was related to the presence of the seagrass so that an increase in seagrass biomass resulted in an increase in detritus biomass. The correlation between numbers of species and individuals of associated fauna Žboth epi- and infauna. and aboveground seagrass biomass was non-significant. In contrast,

Fig. 6. Abundance of epifauna in the artificial seagrass experiment at Lai Chi Wo. Plot of the standardized discriminant score for the first two discriminant functions between the treatments of high density–long leaves ŽHL., high density–short leaves ŽHS., low density–long leaves ŽLL., low density–short leaves ŽLS. and control ŽC..

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

38

these values were significantly positively correlated to the belowground biomass of Z. japonica and detritus biomass, as follows: log Ž Epifauna. s 1.39 q 0.184 B q 0.053 D; log Ž Infauna. s 1.76 q 0.122 B q 0.034 D;

r 2 s 0.255, P - 0.001; n s 195 r 2 s 0.321, P - 0.001; n s 195

where Epifaunas the number of epifaunal individuals, Infaunas number of infaunal individuals, B s below-ground seagrass biomass, D s detritus biomass. 3.5. Abundance of epifauna associated with the artificial seagrass units Fifteen species of epifauna were recorded in the experimental artificial seagrass units ŽASUs.: nine gastropods, five brachyurans and one bivalve. The number of species recorded for the treatments ŽHL, HS, LL, LS. and control ŽC., were 15, 14, 15, 12, and 10, respectively ŽTable 5.. The total epifaunal abundance in the HL treatment was significantly greater than in the control ŽTable 5 and Fig. 5; SNK, P - 0.05.. Batillaria zonalis was the most abundant component of the epifauna in the five treatments,

Table 7 The results of a one-way ANOVA to investigate the relationship between the abundance of C. oualaniensis, M. senhousia and Thalamita sp. ŽindividualsPmy2 . and different treatments of artificial seagrass units at Lai Chi Wo Source of variance

df

SS

C. oualaniensis Treatment Error SNK

MS

4 20

23,863.4 18,050.0

M. senhousia Treatment Error SNK

4 20

49,191.8 37,078.8

Thalamita sp. Treatment Error SNK

4 20

13.4 10.4

F

P

5965.9 902.5

6.61

0.0015

12,297.9 1853.9

6.63

0.0014

6.42

0.0017

3.340 0.520

High density–long leaves ŽHL., high density–short leaves ŽHS., low density–long leaves ŽLL., low density– short leaves ŽLS. and control ŽC.. SNK results are given when the null hypothesis was rejected. Means joined by the same line are not significantly different.

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

39

contributing 30.4%, 33.5%, 48.0%, 45.8% and 47.0% to the total abundance in the treatments HL, HS, LL, LS, and control C, respectively. In addition, the mussel Musculista senhousia, and the gastropods Cerithium coralium and Clithon oualaniensis also comprised large proportions of the total abundance. Four discriminant functions were generated. Only the first function is considered here, as only it was significant at P - 0.05 ŽTable 6a. and accounted for 80.9% of the total variance. The species C. oualaniensis, M. senhousia and Thalamita sp. had the lowest Wilks’ lambda and their P-values were also - 0.05 for the first discriminant function, which indicated that treatment means were different ŽTable 6b.. To further assess the contribution of a variable to the discriminant function, an examination of the standardized coefficients and pooled within-groups correlation coefficients was carried out. For the first discriminant function, C. sowerbiana Žy2.67., M. senhousia Ž2.60., P. didyma Ž2.28. and Thalamita sp. Ž3.59. had higher standardized coefficients ŽTable 6b., while C. oualaniensis Ž0.23., M. senhousia Ž0.24. and Thalamita sp. Ž0.20. had higher pooled within-group correlation coefficients ŽTable 6b.. Combining the results of Wilks’ lambda, standardized coefficients and pooled within-groups correlation coefficients, C.

Fig. 7. Biomass of epifauna in the artificial seagrass experiment. Plot of the standardized discriminant score for the first two discriminant functions between the treatments of high density–long leaves ŽHL., high–density short leaves ŽHS., low–density long leaves ŽLL., low–density short leaves ŽLS. and control ŽC..

40

Ža. Summary statistics for the significant canonical discriminant function Žfunction 1. Function

Eigenvalue

Percent of variance

Canonical correlation

Wilks’ l

x2

df

P

1

24.2782

83.77

0.9800

0.0030

81.25

60

0.0353

Žb. Wilks’ lambda ŽU-statistic. and the significance of the univariate F-ratio Žwith 4 and 20 degrees of freedom., standardized canonical discriminant function coefficients ŽSCDFC. and pooled within-groups correlations ŽPWC. between discriminating variables and canonical discriminant function 1. Species abbreviations follow those of Table 6.

Wilks’ l F-ratio SCDFC PWC

Ts

Mus

Clo

0.3603 0.0003 2.6463 0.1209

0.5443 0.0127 – 0.1688

0.5847 0.0241 – 0.2662

Only statistically significant values are shown.

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

Table 8 Results of a discriminant analysis on the artificial seagrass experiment based on the biomass of epifauna

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

41

oualaniensis, M. senhousia and Thalamita sp. were highlighted as important for discriminating the five treatments. The plot of epifaunal abundance along the first and second discriminant functions shows distinct clusters for the five treatments ŽFig. 6.. Abundance of epifauna in the low density–long leaves ŽLL. and low density–short leaves ŽLS. treatments, and the control ŽC. were separated from the treatments of high–density long leaves ŽHL. and high density–short leaves ŽHS. along the first discriminant function axis ŽHL ) HS ) LL s C s LS.. The abundance of epifauna in the control was well separated from the treatments by the second function. From the results of the one-way ANOVA, the abundance of C. oualaniensis and M. senhousia were significantly higher in treatment HL than in LS and the control ŽTable 7.. The abundance of Thalamita sp. in treatment HL was significantly higher than in LL, LS and C. 3.6. Biomass of associated epifauna There was no difference among the total epifaunal biomass of the five ASU treatments Žone-way ANOVA, P ) 0.05.. The biomass of epifauna in treatments HL and HS were well separated from the other treatments ŽLL, LS and C. by the first discriminant function ŽHL s HS ) LL s C s LS, Fig. 7.. As with abundance, only the first discriminant function was significant in the analysis of the biomass of epifauna Ž P - 0.05, Table 8a.. This first function described 83.8% of the total variance.

Table 9 The results of a one way ANOVA to investigate the relationship between the biomass of C. oualaniensis, M. senhousia and Thalamita sp. ŽgPmy2 . at Lai Chi Wo and different treatments of artificial seagrass units Source of variance

df

SS

MS

F

P

C. oualaniensis Treatment Error

4 20

14.9 36.9

3.71 1.85

2.01

0.1315

0.0772 0.0204

3.78

0.0190

6.562 0.992

6.61

0.0015

Musculista senhousia Treatment Error SNK

4 20

Thalamita sp. Treatment Error SNK

4 20

0.309 0.409

26.2 19.8

High density–long leaves ŽHL., high density–short leaves ŽHS., low density–long leaves ŽLL., low density– short leaves ŽLS. and control ŽC.. SNK results are given when the null hypothesis was rejected. Underlined means are not significantly different.

42

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

The species C. oualaniensis, M. senhousia and Thalamita sp. showed the lowest Wilks’ lambda in function 1, which has the highest discriminating effect ŽTable 8b.. When the standardized coefficients and pooled within-groups correlation coefficients are considered for the first function, the species Thalamita sp. Žstandardized coefficientss 2.65 and correlation coefficientss 0.266., C. oualaniensis Žcorrelation coefficientss 0.121. and M. senhousia Žcorrelation coefficientss 0.169. were the most important in distinguishing between the different treatments ŽTable 8b.. From the results of a one-way ANOVA, the biomass of Thalamita sp. was significantly higher in treatment HL and HS than LL, LS and C ŽTable 9.. A higher biomass of M. senhousia was observed in treatment HL than in LS and C, while there was no difference in the biomass of C. oualaniensis among the five treatments.

4. Discussion 4.1. Effect of seagrass canopy on detritus accumulation The detritus pool in the seagrass bed at Lai Chi Wo can comprise both allochthonous Žterrestrial plant material carried to the shore by incursing freshwater streams and litter exported from the adjacent mangroves. and autochthonous Žseagrass. sources. More allochthonous detritus can presumably accumulate in the Zostera bed when the freshwater streams flood after storms in summer. Conversely, the supply of allochthonous detritus from external ecosystems is presumably reduced in the cool and dry winter. A higher production of Z. japonica noted in winter, on the other hand, can supply more detritus to the pool. Macroalgal detritus was also only present in the winter months because of the favourable temperatures for growth. Therefore, the portion of the standing detritus biomass contributed by Z. japonica and the associated macroalgae showed significant seasonal variations. The standing crop of detritus was significantly higher in the seagrass bed than in the adjacent unvegetated area at Lai Chi Wo and was directly proportional to the biomass of Z. japonica. This supports the notion that the seagrass canopy influences water current flow pattern and, hence, detritus accumulation rate. Kenyon et al. Ž1999. observed that fine detritus particles were visibly accumulated on ASUs with 24 h of deployment. Hydrodynamic features such as current and flux reduction, shear stress at the canopy level and turbulence intensity are correlated positively with seagrass abundance ŽGambi et al., 1990.. Seagrasses can help trap organic detritus and sediment by modifying water flow and reducing currents and waves ŽFonseca et al., 1982; Walker and McComb, 1985; Gambi et al., 1990; Walker et al., 1991.. The standing crop of detritus could, therefore, be predicted adequately by the above- and belowground biomass of Z. japonica. 4.2. Structure of the faunal assemblage associated with Z. japonica The faunal community of the Lai Chi Wo Z. japonica bed had a relatively low Shannon–Wiener diversity index Ž- 1.3. compared with other seagrass beds Že.g.

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

43

Collett et al., 1984; Bostrom ¨ and Bonsdorff, 1997.. However, diversity index values are usually sample size-dependent, and direct comparisons should be interpreted with caution. The low values of the Shannon–Wiener index at Lai Chi Wo indicate that the site was dominated by only a few species. The polychaetes, bivalves and gastropods comprised ) 80% of the total number of species. Even though faunal abundance in the adjacent unvegetated area was lower than in the seagrass bed, the composition of the infauna was similar. Two of the three most abundant species in the unvegetated area, namely, the Oligochaeta, and the polychaetes Prionospio cirrifera and Polydora maculata, scored the same level of importance in the seagrass bed. As oligochaetes and polychaetes are infaunal organisms, the species number and abundance of infauna in both habitats were greater overall than those of the epifauna. Some tellinid bivalves, such as Moerella iridescens, Semele cordiformis and Macoma spp. appear unique to the Zostera beds of Lai Chi Wo, being, so far, not recorded from other sand flats in Hong Kong Žsee Morton and Morton, 1983.. Virnstein and Howard Ž1987. demonstrated that many faunal species are common in adjacent seagrass beds dominated by different species, but not in beds of the same seagrass species occurring in different environmental regimes. Seagrass faunas are rarely associated with particular seagrass species but respond to a restricted set of physical environmental parameters ŽHoward et al., 1989.. The occurrence of specific tellinid bivalves in the Hong Kong Z. japonica bed also raises the question of whether these species are unique to the Hong Kong Z. japonica bed. It should be noted that, however, Moerella iridescens and Macoma spp. are usually also found in intertidal muddy or sandy shores without seagrasses ŽYou et al., 1990; Vincent et al., 1992.. The ‘uniqueness’ of these tellinid bivalves to the Z. japonica bed may, thus, simply be because very few studies at the scale used in this study have been made on soft shores in Hong Kong. Including those organisms noted above, most of the Z. japonica associated fauna could be found in other local habitats, except the tellinid bivalves. Even though these tellinid bivalves have been recorded in Hong Kong only from the Z. japonica bed at Lai Chi Wo, this may represent more the distribution of sampling effort on soft shores in Hong Kong than a unique association of the bivalves with Zostera beds. 4.3. Significance of seagrass presence to the associated fauna The abundance and numbers of epifaunal and infaunal species were significantly higher in the Z. japonica bed than in the adjacent unvegetated area Žtwo-way ANOVA, Tables 1 and 2.. All the animal species recorded from the unvegetated area could also be collected from the seagrass area. In contrast, a total of 48 species inhabited only the seagrass bed. This finding is not surprising and concurs with those obtained in previous studies, e.g. Hutchings et al. Ž1991., Kirkman et al. Ž1991., Orth Ž1992., Edgar et al. Ž1994. and Jenkins and Sutherland Ž1997., that seagrass areas support richer and more diverse faunal communities. The abundance of epifauna and infauna were significantly positively correlated with both detritus and belowground seagrass biomass, but not aboveground biomass. Seagrass canopy may therefore enhance faunal abundance and species richness through the indirect effect of increased detritus accumulation.

44

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

Research on the influence of belowground seagrass biomass on the structure and abundance of the infauna has resulted in two different schools of thought. Orth Ž1977. obtained a positive relationship between infaunal density and belowground biomass and attributed this to protection offered by the dense rhizome mat. Stoner Ž1980., however, observed a low abundance of deposit feeders in heavily vegetated sediments and concluded that increased rhizome materials inhibited burrowing and tube-building activities. One reason why detritus and belowground biomass were more influential may be the fact that most species recorded in this study were small invertebrates associated with the epibenthic detritus layer. The benefit of the aboveground biomass was, therefore, only indirectly effected through increased detritus accumulation. The results of this study indicate that the number of infaunal species and individuals were directly proportional to detritus standing crop and belowground seagrass biomass, which may increase protection from predators and provide food to the detritivore-dominated associated fauna. Based on knowledge of the local species, at least 40% of the associated fauna recorded are classified as detritivores Žunpublished data.. Edgar et al. Ž1994. suggested that the presence of seagrasses may enhance benthic production by fueling the detritivorous food web through the input of decaying plant material. Once the rhizomes die, decomposition occurs in situ and provides food and nutrients directly to the consumers. This trophic contribution probably partly explains why the belowground biomass of Z. japonica was positively correlated with faunal Žepi- and infauna. abundance. Dense belowground seagrass structure may also help support burrows of infauna. Apart from reducing predation pressure, the seagrass canopy and detritus may also act as a shelter from extreme environmental fluctuations during low tide periods. Physical factors such as temperature, salinity, turbidity, oxygen concentration and water movement appear to be important in determining epifaunal abundance and composition than the seagrass species on which these animals live ŽEdgar, 1990.. The epifauna associated with different seagrass species are, therefore, subjected to the same environmental conditions and are often more similar than the epifauna associated with the same species of seagrass in different areas ŽEdgar, 1990.. Low species richness values are generally reported from seagrass habitats which undergo extreme environmental changes ŽJackson, 1972.. This is probably demonstrated by the intertidal Z. japonica beds at Lai Chi Wo, which experienced wide fluctuations in environmental parameters, such as temperature and irradiation. In summer, the seagrass bed was always immersed during mid afternoon. High temperature not only suppressed the Z. japonica biomass, but may have also decreased the diversity and abundance of the associated fauna. The numbers of faunal species and individuals were relatively low in the months experiencing higher air temperatures, e.g. July–October, in both the seagrass bed and unvegetated area ŽFigs. 1 and 2.. The epifaunal species numbers and abundance in the seagrass bed were, however, relatively higher than in the unvegetated area even with a low diversity index Žtwo-way ANOVA; Tables 1 and 2.. It is suggested that most epifauna may be restricted to the Z. japonica bed during summer, where exposure to the wide environmental fluctuations is at the minimum. Less heat-tolerant epifauna Že.g. the errant polychaetes. may try to avoid the stresses of high temperature and desiccation during low tide by retreating beneath the seagrass canopy.

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

45

As protection by the seagrass canopy was limited due to the aboveground biomass Žand thus low percentage cover, Lee, 1997., some animals probably could not withstand the high temperature and desiccation effects ŽLiu and Morton, 1994. resulting in the observed lower diversity index during the summer months. There were, however, no differences observed between the two habitats in terms of the infaunal species. The smaller temporal variation in the infaunal species may be due to the relatively stable environment of the sediment at both locations. Detritus in the Z. japonica bed can also act as shelter, just as the seagrass canopy provides shelter from predators and large environmental fluctuation, apart from being a food resource. Many recent reports have recorded significantly higher diversity and abundance of faunal assemblages associated with seagrass patches than in nearby bare-sand areas ŽJenkins and Sutherland, 1997; Webster et al., 1998 and references therein.. Seagrass physical structures such as shoots, leaves and rhizomes have been demonstrated to influence the structure of associated faunal assemblages ŽHeck and Westone, 1977; Webster et al., 1998.. In this study, the importance of accumulated detritus to the faunal assemblages is also apparent. Intertidal areas with Z. japonica tend to trap large amounts of accumulated detritus and can, therefore, be expected to be faunistically more diverse and abundant. The provision of shelter and food availability may be the major factors influencing the diversity and population structure of the fauna associated with the local Z. japonica bed. The relative importance of these two factors to the associated fauna has not yet, however, been evaluated. While many researchers consider seagrasses primarily functioning as ‘structural species’ providing the physical habitat fabric for the associated fauna ŽMazzella et al., 1992., the physical structure may in turn enhance food availability through ephiphytic algal growth, which is comparatively more palatable to most invertebrate consumers than seagrass tissues. 4.4. The importance of physical seagrass canopy structure to associated fauna The results of the discriminant analysis on the artificial seagrass experiment further support the notion that the total abundance and biomass of epifauna in the high density canopy treatments were significantly higher than those in the low density canopy treatments and the control ŽFigs. 6 and 7.. Unlike the findings of Connolly and Butler Ž1996., canopy height seemed to have weaker effects than shoot density in structuring the epifaunal community. Again, this difference may be a result of the difference in the nature of the species assemblages associated with the seagrass. The most important variables Žthe species C. oualaniensis, M. senhousia and Thalamita sp.. for discriminating the treatments also indicated no difference among the control and low density canopy treatments, but significantly higher levels in the high density treatments than the others ŽTables 2 and 5.. The abundance and biomass of epifauna were influenced greatly by the seagrass canopy density, with a denser canopy supporting a higher abundance and biomass of epifauna. From the results of the SNK tests on the three important variables in the one way ANOVA ŽTables 2 and 5., the sequence of the five treatments depended on the amount of total surface areas offered, that is, HL ) HS ) LL ) LS ) C. This suggests that the abundance and biomass of epifauna associated with a seagrass bed depends not only on canopy density but, for a fixed shoot density, on the surface area

46

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

Žbiomass. offered by the canopy. Attrill et al. Ž2000. reported that there is no significant relationship between habitat complexity and biomass in a Z. marina bed, and argued that changes in associated faunal richness and abundance were basically related to increase in seagrass leaf area Žbiomass.. A simple species– Žleaf. area relationship is therefore applicable to the associated fauna assemblage. By contrast, in an experiment involving the deployment of artificial seagrass beds of different leaf widths Žwhile keeping the total leaf surface area constant., Jenkins and Sutherland Ž1997. recorded significantly higher fish abundances in the narrow-leaf than the wide-leaf treatment. Their findings were, however, confounded by the difference in spacing of ‘leaves’ in the two width treatments, with the narrow-leaf treatment offering a much more closed canopy Ž5–10 = narrower spacing. than in the wide-leaf treatment, thus probably producing a local leaf density effect. In terms of the total number of species and abundance of epifauna, the ASUs had similar values as the adjacent unvegetated areas, but significantly lower than the Zostera bed in the same season, estimated from core sediment samples ŽTable 5.. It should be noted, however, that the samples were collected using different methods Žcores for the natural bed and unvegetated areas, whole sampling unit in the ASU experiment. and different years. These differences could contribute to the differences in animal assemblages recorded using the two approaches. While this study did not aim to assess to what extent ASUs mimicked the natural beds, some observations are still noteworthy. Amphipods, microgastropods and polychaetes were common in the Zostera bed, but not in the ASUs. This difference is probably a result of the differences in biological quality between the ASUs and Z. japonica. The presence of rhizomes and a spatially complex root system, the ability to release organic carbon ŽPenhale and Smith, 1977. and influence of nutrient fluxes ŽMcRoy and McMillan, 1977; Short and McRoy, 1984. by natural seagrass all may contribute to the difference in the faunal communities between the ASUs and Z. japonica. More subtle effects of live seagrass on interaction between the associated species have also been documented. For example, Pinckney and Micheli Ž1998. documented significant differences in the species composition of epiphytic algae on live and mimics of Z. marina and Halodule wrightii. These authors proposed that the differences might be a result of direct competition between seagrass and their epiphytes andror the modification by the seagrass of competitive interactions between the epiphytes. While the associated faunal assemblage may not directly compete with Z. japonica for resources, it is likely that the differences in assemblage structure between live seagrass and artificial mimics might be strongly influenced by the difference in the quality and quantity of food Že.g. live seagrass, epiphytic algae, detritus. available. Some ecological realism is therefore traded for increases in flexibility in experimental designs employing artificial seagrass units. The differences may also be due to the relatively brief duration of the experiment or biases resulting from the different sampling devices used. With the exception of food availability, the canopy structure offered by the ASUs provides almost all the habitat requirements suggested by Lewis Ž1984. Ži.e. the degree of hydrodynamic force reduction, sediment deposition and stabilization, food resource availability, the amount of physical structure, number of microhabitats and the level of protection from predators provided by the seagrass canopy., which may have resulted in

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

47

the greater abundance of organisms in the presence of macrophytes. The denser the seagrass canopy, the greater the numbers of microhabitats and living spaces provided for organisms and, hence, more protection from predators. The function of encouraging sediment deposition and stabilization cannot be justified in the experiment because of its short duration, but the reduction of hydrodynamic forces was implicated by the recruitment of M. senhousia larvae. In addition to the food resources provided by detritus and algae, the physical structure of the canopy of Z. japonica, either through direct provision of leaf area or increased detritus trapping, can also influence the abundance and biomass of epifauna.

Acknowledgements We thank Mr. C.C. Lay, Mr. C.K. Chan and Mr. Darwin Cheung of the Agriculture and Fisheries Department for their logistical support throughout the study. This study was funded by a grant awarded to S.Y. Lee and R. Wu by the Provisional Airport Authority and administered by the Agriculture and Fisheries Department. We also thank the Director of the Agriculture and Fisheries Department for the permission to publish. [RW]

References Almasi, M.N., Hoskin, C.M., Reed, J.K., Milo, J., 1987. Effects of natural and artificial Thalassia on rates of sedimentation. J. Sediment. Petrol. 57, 901–906. Attrill, M.J., Strong, J.A., Rowden, A.A., 2000. Are macroinvertebrate communities influenced by seagrass structural complexity? Ecography 23, 114–121. Barber, W.E., Greenwood, J.G., Crocos, P., 1979. Artificial seagrass—a new technique for sampling the community. Hydrobiologia 65, 135–140. Bell, J.D., Westoby, M., 1986. Abundance of macrofauna in dense seagrass is due to habitat preference, not predation. Oecologia 68, 205–209. Bell, J.D., Steffe, A.S., Westoby, M., 1985. Artificial seagrass: how useful is it for the field experiments on fish and macroinvertebrates? J. Exp. Mar. Biol. Ecol. 90, 171–177. Bostrom, ¨ C., Bonsdorff, E., 1997. Community structure and spatial variation of benthic invertebrates associated with Zostera marina ŽL.. beds in the northern Baltic Sea. J. Sea Res. 37, 153–166. Ceccherelli, G., Cinelli, F., 1999. Effects of Posidonia oceanica canopy on Caulerpa taxifolia size in a north-western Mediterranean bay. J. Exp. Mar. Biol. Ecol. 240, 19–36. Collett, L.C., Hutchings, P.A., Gibbs, P.J., Collins, A.J., 1984. A comparative study of the macro-benthic fauna of Posidonia australis seagrass meadows in New South Wales. Aquat. Bot. 18, 111–134. Connolly, R.M., 1995. Effects of removal of seagrass canopy on assemblages of small, motile invertebrates. Mar. Ecol.: Prog. Ser. 118, 129–137. Connolly, R.M., Butler, A.J., 1996. The effects of altering seagrass canopy height on small, motile invertebrates of shallow Mediterranean embayments. Mar. Ecol. 17, 637–652. Edgar, G.J., 1990. The influence of plant structure on the species richness, biomass and secondary production of macrofaunal assemblages associated with Western Australian seagrass beds. J. Exp. Mar. Biol. Ecol. 137, 215–240. Edgar, G.J., Robertson, A.I., 1992. The influence of seagrass structure on the distribution and abundance of mobile epifauna: pattern and process in a Western Australian Amphibolis bed. J. Exp. Mar. Biol. Ecol. 160, 13–31.

48

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

Edgar, G.J., Shaw, C., Watson, G.F., Hammond, L.S., 1994. Comparisons of species richness, size-structure and production of benthos in vegetated and unvegetated habitats in Western Port, Victoria. J. Exp. Mar. Biol. Ecol. 176, 201–226. Fong, C.W., 1998. Some aspects of ecology of seagrass Z. japonica in Hong Kong. MPhil Thesis, Department of Ecology and Biodiversity, University of Hong Kong, Hong Kong. 202 pp. Fong, C.W., Lee, S.Y., Wu, R.S.S., 1998. Seasonal cycle of growth 2nd reproduction in the seagrass Zostera japonica in Hong Kong. In: Morton, B. ŽEd.., The Marine Biology of the South China Sea III. Hong Kong University Press, Hong Kong, pp. 247–260. Fong, C.W., Lee, S.Y., Wu, R.S.S., 2000. The effects of epiphytic algae and their grazers on the intertidal seagrass Zostera japonica. Aquat. Bot. 67, 251–261. Fonseca, M.S., Fisher, J.S., Zieman, J.C., Thayer, G.W., 1982. Influence of the seagrass, Zostera marina L., on current flow. Estuarine, Coastal Shelf Sci. 15, 351–364. Gambi, M.C., Nowell, A.R.M., Jumars, P.A., 1990. Flume observations on flow dynamics in Zostera marina Žeelgrass. beds. Mar. Ecol. Prog. Ser. 61, 159–169. Heck Jr., K.L., Westone, G.S., 1977. Habitat complexity and invertebrate species richness and abundance in tropical seagrass meadows. J. Biogeogr. 4, 135–142. Heck Jr., K.L., Able, K.W., Roman, C.T., Fahay, M.P., 1995. Composition, abundance, biomass and production of macrofauna in a New England estuary: comparisons among eelgrass meadows and other nursery habitats. Estuaries 18, 289–379. Horinouchi, M., Sano, M., 1999. Effects of changes in seagrass shoot density and leaf height on abundances and distribution of juveniles of three gobioid fishes in a Zostera marina bed. Mar. Ecol. Prog. Ser. 183, 87–94. Howard, R.K., Edgar, G.J., Hutchings, P.A., 1989. Faunal assemblages of seagrass beds. In: Larkum, A.W.D., McComb, A.J., Sheperd, S.A. ŽEds.., Biology of Seagrasses: A Treatise on the Biology of Seagrasses with Special Reference to the Australian Region. Elsevier, The Netherlands, pp. 536–564. Hutchings, P.A., Wells, F.E., Walker, D.I., Fendrick, G.A., 1991. Seagrass, sediment and infauna—a comparison of Posidonia australis, Posidonia sinuosa and Amphibolis antarctica in Princess Royal Harbour, southwestern Australia. In: Wells, F.E., Walker, D.I., Kirkman, H., Lethridge, R. ŽEds.., Proceedings of the Third International Marine Biology Workshop: The Marine Flora and Fauna of Albany, Western Australia. Western Australian Museum, Perth, pp. 611–633. Jackson, J.B.C., 1972. The ecology of the molluscs; the Thalassia communities, Jamaica, West Indies: II. Molluscan population variability along an environmental stress gradient. Mar. Biol. 14, 304–377. Jenkins, G.P., Sutherland, C.R., 1997. The influence of habitat structure on nearshore fish assemblages in a southern Australian embayment: colonisation and turnover rate of fishes associated with artificial macrophyte beds of varying physical structure. J. Exp. Mar. Biol. Ecol. 218, 103–125. Kenyon, R.A., Haywood, M.D.E., Heales, D.S., Loneragan, N.R., Pendrey, R.C., Vance, D.J., 1999. Abundance of fish and crustacean postlarvae on portable artificial seagrass units: daily sampling provides quantitative estimates of the settlement of new recruits. J. Exp. Mar. Biol. Ecol. 232, 197–216. Kirkman, H., Humphries, P., Manning, R., 1991. The epibenthic fauna of seagrass beds and bare sand in Princess Royal Harbour and King George Sound, Albany, southwestern Australia. In: Wells, F.E., Walker, D.I., Kirkman, H., Lethridge, R. ŽEds.., Proceedings of the Third International Marine Biology Workshop: The Marine Flora and Fauna of Albany, Western Australia. Western Australian Museum, Perth, pp. 553–563. Koch, E.W., 1996. Hydrodynamics of a shallow Thalassia testudinum bed in Florida, USA. In: Kuo, J., Phillips, R.C., Walker, D.I., Kirkman, H. ŽEds.., Seagrass Biology: Proceedings of an International Workshop, Rottnest Island, Western Australia. University of Western Australia, Perth, pp. 105–110. Komatsu, T., 1996. Influence of Zostera bed on the spatial distribution of water flow over a broad geographic area. In: Kuo, J., Phillips, R.C., Walker, D.I., Kirkman, H. ŽEds.., Seagrass Biology: Proceedings of an International Workshop, Rottnest Island, Western Australia. University of Western Australia, Perth, pp. 111–116. Lee, S.Y., 1997. Annual cycle of biomass of a threatened population of the intertidal seagrass Zostera japonica in Hong Kong. Mar. Biol. 129, 183–193. Lee, H.B., Low, J., 1991. The enhancement of fish community in the Singapore River through the use of artificial seagrass. In: Chia, L.S., Chou, L.M. ŽEds.., Urban Coastal Area Management: the Experience of

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

49

Singapore. ICLARM Conference Proceedings, vol. 25. International Centre for Living Aquatic Resources Management, Philippines, pp. 21–29. Lethbridge, R.C., Borowitzka, M.A., Benjamin, K.J., 1988. The development of an artificial, Amphibolis-like seagrass of complex morphology and preliminary data on its colonization by epiphytes. Aquat. Bot. 31, 153–168. Lewis III, F.G., 1984. Distribution of macrobenthic crustaceans associated with Thalassia, Halodule and bare sand substrata. Mar. Ecol.: Prog. Ser. 19, 101–113. Liu, J.H., Morton, B., 1994. The temperature tolerances of Tetraclita squamosa ŽCrustacea: Cirripedia. and Septifer Õirgatus ŽBivalvia: Mytilidae. on a sub-tropical rocky shore in Hong Kong. J. Zool. 234, 325–339. Mazzella, L., Blia, M.C., Gambi, M.C., Lorenti, M., Russo, G.F., Scipione, M.B., Zupo, V., 1992. Plant–animal trophic relationships in the Posidonia oceania ecosystem of the Mediterranean Sea: a review. In: John, D.M., Hawkins, S.J., Price, J.H. ŽEds.., Plant–Animal Interactions in the Marine Benthos, Systematics Association Special Volume 46. Clarendon Press, Oxford, pp. 165–187. McRoy, C.P., McMillan, C., 1977. Production ecology and physiology of seagrasses. In: McRoy, C.P., Helfferich, C. ŽEds.., Seagrass Ecosystems: A Scientific Perspective. Marcel Dekker, New York, pp. 53–87. Morton, B., Morton, J., 1983. Seashore Ecology of Hong Kong. Hong Kong University Press, Hong Kong. Orth, R.J., 1977. The importance of sediment stability in seagrass communities. In: Coull, B.C., Baruch, B.W. ŽEds.., Ecology of Marine Benthos. University of South Carolina Press, Columbia, pp. 281–300. Orth, R.J., 1992. A perspective on plant–animal interactions in seagrasses: physical and biological determinants influencing plant and animal abundance. In: John, D.M., Hawkins, S.T., Price, J.H. ŽEds.., Plant–Animal Interactions in the Marine Benthos. Systematics Association Publication Clarendon Press, Oxford, pp. 147–164. Penhale, P.A., Smith Jr., W.O., 1977. Excretion of dissolved organic carbon by eelgrass Ž Zostera marina. and its epiphytes. Limnol. Oceanogr. 22, 400–407. Phillips, R.C., Menez, ´ E.G., 1988. Seagrasses. Smithson. Contrib. Mar. Sci. 34, 1–104. Pielou, E.C., 1966. The measurement of diversity in different types of biological collections. J. Theoret. Biol. 13, 131–144. Pinckney, J.L., Micheli, F., 1998. Microalgae on seagrass mimics: does epiphyte community structure differ from live seagrasses? J. Exp. Mar. Biol. Ecol. 221, 59–70. Robertson, A.I., 1984. Trophic interactions between the fish fauna and macrobenthos of an eelgrass community in Western Port, Victoria. Aquat. Bot. 18, 135–153. Shannon, C.E., 1948. A mathematical theory of communication. Bell Syst. Tech. J. 27 Ž379–423., 623–656. Shin, H., Choi, H.K., 1998. Taxonomy and distribution of Zostera ŽZosteraceae. in eastern Asia, with special reference to Korea. Aquat. Bot. 60, 49–66. Short, F.T., McRoy, C.P., 1984. Nitrogen uptake by leaves and roots of the seagrass Zostera marina L. Bot. Mar. 28, 547–555. Sogard, S.M., Olla, B.L., 1993. The influence of predator presence on utilization of artificial seagrass habitats by juvenile walleye pollock, Theragra chalcogramma. Environ. Biol. Fishes 37, 57–65. Sorensen, T., 1948. A method of establishing groups of equal amplitude in plant sociology based on similarity of species content. K. Dan. Vidensk. Selsk., Biol. Skr. 5, 1–34. Stoner, A.W., 1980. Perception and choice of substratum by epifaunal amphipods associated with seagrasses. Mar. Ecol.: Prog. Ser. 3, 105–111. Vincent, B., Joly, D., Harvey, M., 1992. Spatial variation in growth of the bivlave Macoma balthica ŽL.. on a tidal flat: effects of environmental factors and interspecific competition. J. Exp. Mar. Biol. Ecol. 181, 223–238. Virnstein, R.W., Curran, M.C., 1986. Colonization of artificial seagrass versus time and distance from source. Mar. Ecol.: Prog. Ser. 29, 279–288. Virnstein, R.W., Howard, R.K., 1987. Motile epifauna of marine macrophytes in the Indian River lagoon, Florida: II. Comparisons between drift algae and three species of seagrasses. Bull. Mar. Sci. 41, 13–26. Walker, D.I., McComb, A.J., 1985. The decomposition of leaves from Amphibolis antarctica ŽLabill.. Sonder et Aschers and Posidonia australis Hooks: f. The major seagrass species of Shark Bay, Western Australia. Bot. Mar. 28, 407–413. Walker, D.I., Hutchings, P.A., Wells, F.E., 1991. Seagrass, sediment and infauna—a comparison of Posidonia

50

S.Y. Lee et al.r J. Exp. Mar. Biol. Ecol. 259 (2001) 23–50

australis, Posidonia sinuosa and Amphibolis antarctica in Princess Royal Harbour, southwestern Australia: I. Seagrass biomass, productivity and contribution to sediments. In: Wells, F.E., Walker, D.I., Kirkman, H., Lethridge, R. ŽEds.., Proceedings of the Third International Marine Biological Workshop: The Marine Flora and Fauna of Albany, Western Australia. Western Australian Museum, Perth, pp. 597–610. Webster, P.J., Rowden, A.A., Attrill, M.J., 1998. Effect of shoot density on the infaunal macro-invertebrate community within a Zostera marina seagrass bed. Estuarine, Coastal Shelf Sci. 47, 351–357. You, Z., Wang, Y., Zhu, X., 1990. Preliminary study on population composition and growth of clam Moerella iridescens ŽBenson.. Mar. Sci. Bull. 9, 35–40 Žin Chinese..