Mar Biol (2010) 157:1237–1250 DOI 10.1007/s00227-010-1404-x
ORIGINAL PAPER
Assessing reef Wsh assemblage structure: how do diVerent stereo-video techniques compare? Dianne L. Watson · Euan S. Harvey · Ben M. Fitzpatrick · Timothy J. Langlois · George Shedrawi
Received: 10 August 2009 / Accepted: 30 January 2010 / Published online: 14 February 2010 © Springer-Verlag 2010
Abstract Measures of Wsh abundance, assemblage composition and length were compared when sampled by baited remote underwater stereo-video (stereo BRUV) and diveroperated stereo-video transects (stereo DOV) at the Houtman Abrolhos Islands and Ningaloo Reef. Species richness counts were 40% higher on stereo BRUV than stereo DOV. Stereo BRUVs also recorded a greater number of largebodied targeted species in higher abundance than stereo DOV (e.g. Lethrinus nebulosus, Plectropomus leopardus) at the Houtman Abrolhos and at Ningaloo Reef. Many nontargeted species were also recorded in greater abundances on stereo BRUV than stereo DOV (e.g. Coris auricularis, Gymnothorax spp). Stereo DOV transects recorded a greater abundance of some small-bodied Pomacentridae, Labridae and Scaridae species than did stereo BRUV, particularly at Ningaloo Reef. This study demonstrates that choice of sampling technique for surveys of reef Wsh can lead to very diVerent biological interpretations of Wsh assemblage structure.
Communicated by D. Goulet. D. L. Watson (&) Centre for Marine Futures, The University of Western Australia, 35 Stirling Hwy, Crawley, WA 6009, Australia e-mail:
[email protected] D. L. Watson · E. S. Harvey · B. M. Fitzpatrick · T. J. Langlois · G. Shedrawi School of Plant Biology, The University of Western Australia, 35 Stirling Hwy, Crawley, WA 6009, Australia
Introduction When designing a sampling programme, it is important to consider which sampling technique may provide the most accurate results for the ecological questions posed (Andrew and Mapstone 1987; Rotherham et al. 2007). For many years, scientists have critically examined the biases and limitations of diVerent non-destructive Wsh sampling techniques (Thresher and Gunn 1986; Lincoln-Smith 1989; Willis et al. 2000; Trenkel et al. 2004) with the over-arching goal of identifying at least one technique that can provide an accurate description of the Wsh assemblage. Various studies have described the biases and limitations of such methods (e.g. Harmelin-Vivien et al. 1985; Samoilys and Carlos 2000). Research has recently turned to comparing and contrasting diVerent sampling techniques and removing or minimising the biases and limitations associated with them (Willis and Babcock 2000; Harvey et al. 2001a, b; Cappo et al. 2004; Watson et al. 2005; Harvey et al. 2007). The most common outcome of research that compares the ability of diVerent monitoring techniques is that no single technique is suitable for providing information on all Wsh species (Willis and Babcock 2000; Watson et al. 2005). This is due to any number of factors that, for the most part, centre on the behaviour of Wsh towards the techniques themselves and on selectivity of the technique/gear used. A few examples include Wsh attraction to SCUBA divers (Cole 1994; Cole et al. 2007) and the inability of Wsh traps to eVectively sample herbivorous species (Munro 1974). Most studies that have compared sampling techniques focus on examining certain aspects of a Wsh assemblage, i.e. particular species of interest (Stewart and Beukers 2000; Spencer et al. 2005), species richness (Cappo et al. 2004), Wsh abundance (Sanderson and Solonsky 1986; Willis and Babcock 2000; Watson et al. 2005; Harvey et al. 2007) or
123
1238
size of selected species (Willis and Babcock 2000; Harvey et al. 2002). By focussing on these aspects alone, the studies do not consider the overall performance of the sampling technique. For example, a technique that is most eVective at surveying the abundance of coral trout in an area may only survey large individuals, and the technique may not be suitable for obtaining abundance information for other species. By considering Wsh assemblage structure as a whole (composition, Wsh abundance and size structure), a better understanding of the advantages and limitations of diVerent sampling techniques may be obtained. We compare the structure of Wsh assemblages observed using two non-destructive Wsh census methods: baited remote underwater stereo-video (stereo BRUV) and diveroperated stereo-video transects (stereo DOV). Both techniques are being used with increasing regularity without a real understanding of their advantages and disadvantages. Field sites included the sub-tropical Houtman Abrolhos Islands and the tropical Ningaloo Reef. The objective of this study was to examine diVerences in the structure of Wsh assemblages surveyed using the two stereo-video techniques. Given the use of bait and absence of SCUBA divers, it was hypothesised that stereo BRUV would survey a greater abundance of predatory Wsh species than stereo DOV. As predatory Wsh species are often those targeted by Wshers (e.g. Serranidae), it was predicted that the surveyed assemblage would comprise a higher proportion of these species than stereo DOV, and that the size structure of assemblages would reXect this (greater number of larger individuals). It was hypothesised that stereo DOV would survey a greater abundance of site-attached species (e.g. Pomacentridae), prey and herbivorous Wsh species than stereo BRUV.
Methods Study sites Houtman Abrolhos Islands The Houtman Abrolhos and associated reefs are located approximately 60 km oVshore from the mid-west coast of Western Australia on the edge of the continental shelf between 28°15⬘S and 29°00⬘S. The Houtman Abrolhos comprise four main groups of islands that span approximately 100 km in a north–south direction. From south to north, they are Pelsaert, Easter, Wallabi and North Island. From late May to early June 2007, stereo BRUV and stereo DOV surveys were conducted at three locations within each of the three Island groups and at the North Island. Each of the three locations is open to recreational and commercial
123
Mar Biol (2010) 157:1237–1250
Wshing. Surveys were conducted on the same coral reef slopes in 8–12 m water depth. Ningaloo Marine Park Ningaloo Reef is a fringing coral reef which stretches for approximately 270 km adjacent to the semi-arid north-west cape of Western Australia between 23°48⬘S and 21°48⬘S. The Ningaloo Marine Park extends over the entire reef system, and 34% of the park is designated as no-take sanctuary zones. In July 2006, stereo BRUV and stereo DOV surveys were conducted in 1–10 m water depth within two of these no-take areas, Mandu and Osprey. At both locations, surveys were randomised to cover Wve major habitat types: inshore algal reef, branching coral, reef pass, tabulate coral, reef Xat (Mandu only) and Porite massive coral (Osprey only). Habitat locations and boundaries were determined using habitat maps, aerial photography and conWrmed with video-groundtruthing. Sampling techniques Baited remote underwater stereo-video surveys (stereo BRUV) The stereo BRUV technique used in the present study is the same as that used by Harvey et al. (2007) and Watson et al. (2007). Detailed information on the design and photogrammetric speciWcs can be viewed in Harvey and Shortis (1996, 1998). The stereo BRUV systems used comprised two SONY HC 15E video cameras mounted 0.7 m apart on a base bar inwardly converged at 8 degrees to gain an optimised Weld of view with visibility of 10-m distance (Harvey and Shortis 1996; Fig. 1a). Each system was deployed by boat and left to Wlm on the sea Xoor for a period of 1 h. Previous research has shown that a Wlming time of at least 36 min is essential for obtaining measures of the majority of Wsh species, and furthermore that the full 60 min is advisable to obtain measures of numerous targeted Wsh species (Watson 2006). A synchronising diode and a bait basket were positioned in front of the cameras (Fig. 1a). Between 6 and 10 stereo BRUV systems were deployed at any one time to increase sampling eYciency. Stereo BRUV systems had 800 grams of pilchards (Sardinops sagax) in a closed plastic-coated wire mesh basket, suspended 1.2 m in front of the two cameras. The pilchards were crushed to maximise dispersal of the Wsh oil and Xesh. Adjacent replicate sites at the Houtman Abrolhos and Ningaloo were separated by at least 250 m to avoid overlap of bait plumes and reduce the likelihood of Wsh moving between sites within the sampling period (e.g. see Cappo et al. 2001).
Mar Biol (2010) 157:1237–1250
1239
A Rope
Camera Camera
Bait bag
Synchronising diode
B
Handholds
Camera
Camera
DOV transects were also conducted in the same Wve shallow habitat types and locations as stereo BRUV deployments. At both Weld sites, stereo DOV transects were collected by two SCUBA divers, one operating the stereo DOV system and the other measuring the distance swum with a chainman cotton counter (bio-degradable cotton). SCUBA divers swam at a slow speed (approx 1 m every 3 s) at a distance approximately 30 cm above the substrate. At the Houtman Abrolhos, six replicate 100 £ 5 m stereo DOV transects were undertaken at each location and at Ningaloo, Wve replicate 50 £ 5 m transects were undertaken in each of the Wve habitats. At both locations, the end of one transect was separated by the start of another by a 15-m gap. The transect width (5 m) was deWned during postprocessing of the stereo-imagery by determining the three-dimensional coordinates (x, y and z) of a Wsh and deciding whether it was inside or outside the transect (see Harvey et al. 2002, 2004). Image analysis
Buoyancy Synchronising diode
Fig. 1 Stereo-video systems for censusing reef Wsh. a baited remote underwater stereo-video system (Stereo BRUV); b diver-operated stereo-video system (Stereo DOV)
At the Houtman Abrolhos, Wve replicate stereo BRUV deployments were made at each of three locations surveyed at each island group (5 £ 3 £ 4 = 60 deployments). At Ningaloo, 30 stereo BRUV deployments were undertaken within the Mandu no-take Sanctuary Zone and 33 within the Osprey no-take Sanctuary with at least 6 replicate stereo BRUV deployments in each of the Wve major habitat types. Diver-operated stereo-video systems (stereo DOV) The stereo DOV technique used in the present study is the same as that used by Harman et al. (2003) and Watson et al. (2005). Similar to stereo BRUV, stereo DOV used two video cameras in underwater housings mounted 0.7 m apart on a base bar and inwardly converged at 8 degrees (Fig. 1b). The stereo DOV system used two Sony TRV 900 video cameras recording on progressive scan. The stereo DOV also had a synchronising diode mounted in front of the cameras and Xoats attached to the base bar to make the system neutrally buoyant (Fig. 1b). At the Houtman Abrolhos, stereo DOV transects were conducted along the same reef slope as the stereo BRUV deployments (GPS located). Stereo DOV transects spanned the reef edge covering a distance that matched the spatial placement of stereo BRUV systems. At Ningaloo, stereo
Stereo BRUV Video images were reviewed, and relative abundance counts were obtained as the maximum number of Wsh belonging to each species present in the Weld of view of the cameras at one time (MaxN; Priede et al. 1994; Cappo et al. 2004). These measurements were made using the Australian Institute of Marine Science (AIMS) purpose-built BRUV database. All Wsh species recorded were assigned to one of two groups: targeted or non-targeted. We deWned targeted Wsh species as those edible species retained when caught by either commercial or recreational Wshermen. The program PhotoMeasure (SeaGIS Pty Ltd) was then used to make length measurements from stereo-video images. To avoid making repeated measurements of the same individuals, measures of length (snout to fork, FL) were made at the time of MaxN. This MaxN ‘time’ is not instantaneous; rather it corresponds to the amount of time all individuals remained in the Weld of view of both cameras. On several occasions, an additional individual was noted outside of the time of MaxN by a large diVerence in size. In these instances, this individual was included in MaxN, and its length measured. To ensure good measurement accuracy and precision, measures of Wsh length were limited to within a maximum distance of 7 m from the cameras. With the cameras used here, measurement accuracy can deteriorate after this distance (Harvey et al. 2002). The software calculates both distance from the cameras and length at the same time, therefore measurements of individuals more than 7 m from the cameras were discarded. During the study period, visibility was excellent at the Houtman Abrolhos and Ningaloo, consistently >8 m.
123
1240
Stereo DOV In the laboratory, pairs of videos from stereo DOV systems were captured and viewed in PhotoMeasure. For each transect, every individual observed on video was identiWed to species level and simultaneously measured (snout to fork length; FL). As described above, the three-dimensional location of each Wsh (distance in front of, and to the left and right of the stereo DOV system) was calculated enabling standardisation of the area surveyed. Individuals further than 7 m in front of, or 2.5 m to the left or right of, the stereo DOV system were not counted or measured. Numbers of individuals of each species were summed for each transect to give measures of abundance. A single species, Chromis westaustralis, was excluded from length analysis for the Houtman Abrolhos, as measurements of these individuals from the stereo DOV footage was diYcult due to their small size, schooling behaviour and the resolution of video imagery. Data analysis The sampling design at the Houtman Abrolhos consisted of three factors: Technique (T, two levels, Wxed: Stereo BRUV, Stereo DOV), Island group (G, four levels, Wxed: Pelsaert, Easter, Wallabi and North) and Location (L, three levels, random, nested in Island group). At Ningaloo, the sampling design consisted of three factors: Technique (T, two levels, Wxed: Stereo BRUV, Stereo DOV), Location (L, two levels, Wxed: Osprey, Mandu) and Habitat (H, Wve levels, Wxed). We used permutational multivariate analysis of variance (PERMANOVA; Anderson 2001) for analyses, because the relative abundances of Wsh were highly skewed and contained many zero counts, i.e. traditional analyses such as MANOVA were not suitable, as they assume normality of errors. All multivariate analyses were conducted using Bray-Curtis dissimilarities on fourth-root transformed relative abundance data, since the data set contained zero values for some species and numbers greater than a hundred for others. PERMANOVA was conducted using the PRIMER-E v6 software (Plymouth Routines in Multivariate Ecological Research; Anderson et al. 2008) on raw MaxN species data and family level data (species MaxN’s summed to family). A canonical analysis of principal coordinates (CAP; Anderson and Robinson 2003; Anderson and Willis 2003) was also conducted to examine whether each technique sampled diVerent assemblages of Wsh. Individual species likely to be responsible for any observed diVerences between stereo BRUV and stereo DOV techniques were determined by examining Pearson correlations of species counts with
123
Mar Biol (2010) 157:1237–1250
canonical axes. A correlation of |r| > 0.25 was used as an arbitrary cut-oV to display potential relationships between individual species and the canonical axes. Kolmogorov–Smirnov tests were conducted on length measurements to determine whether the size frequency of assemblages diVered when surveyed using stereo BRUV or stereo DOV. DiVerences in the size structure of Wsh species surveyed using stereo BRUV and stereo DOV are also illustrated graphically.
Results Assemblage description Houtman Abrolhos Islands Stereo DOV recorded a greater number of individuals than stereo BRUV but fewer species and families (Table 1). A greater proportion of the species recorded by stereo BRUV are considered targeted by Wshers than those recorded by stereo DOV (Table 1). In contrast, stereo DOV recorded an assemblage comprising a greater proportion of non-targeted species than stereo BRUV. Fifty-six species were only recorded by stereo BRUV (50% of all species viewed). The majority of these species were either targeted by Wshers, predatory and/or cryptic in nature (e.g. Lethrinidae spp, Muraenidae spp). Eighteen species were only recorded by stereo DOV (25% of all species viewed) with most of these species being small-bodied Labrids or Pomacentrids. Families which comprised the greatest number of species were similar for both techniques, with the most common family being Labridae followed by Chaetodontidae and Pomacentridae (Table 1). Ningaloo Marine Park Within the two no-take Sanctuary Zones sampled within the Ningaloo Marine Park, stereo BRUV recorded 4,000 fewer individuals than stereo DOV but 60% more species and 30% more families (Table 1). Stereo DOV recorded an assemblage comprising a greater proportion of non-targeted species than stereo BRUV, while a greater proportion of the species recorded by stereo BRUV are considered to be targeted by Wshers (Table 1). Eighty-nine species were recorded only on stereo BRUV (Table 1), while 18 species were seen only on stereo DOV. Many of the species unique to stereo BRUV were predatory and/or cryptic (e.g. Carangidae spp, Muraenidae spp), while those unique to stereo DOV were all non-targeted species (e.g. Pomacentridae spp). The Wve families that each comprised the greatest number of species were similar for both techniques at Ningaloo reef (Table 1).
Mar Biol (2010) 157:1237–1250
1241
Table 1 Description of the Wsh assemblage recorded by stereo BRUV and stereo DOV and the Houtman Abrolhos and at Ningaloo Reef Houtman Abrolhos Islands Stereo BRUV
Ningaloo reef
Stereo DOV
Total
Stereo BRUV
Stereo DOV
Total
Total # individuals
5,617
6,923
12,540
5,368
9,757
15,125
Total # species
110
73
128
183
112
201
Total # families
30
27
38
39
27
41
% Targeted
28
18
24
23
18
21
% Non-targeted
72
81
76
77
82
79
# Unique to technique
56
18
89
18
Top Wve families (# spp)
Labridae (20)
Labridae (15)
Labridae (22)
Labridae (29)
Labridae (26)
Chaetodontidae (10)
Chaetodontidae (10)
Chaetodontidae (12)
Pomacentridae (23)
Pomacentridae (14)
Pomacentridae (23)
Pomacentridae (10)
Pomacentridae (10)
Pomacentridae (12)
Acanthuridae (13)
Chaetodontidae (12)
Chaetodontidae (14)
Serranidae (9)
Scaridae (5)
Serranidae (9)
Chaetodontidae (12)
Scaridae (9)
Acanthuridae (13)
Scaridae (8)
Serranidae (4)
Scaridae (8)
Serranidae (11)
Acanthuridae (5)
Scaridae (12)
Table 2 Houtman Abrolhos Islands: PERMANOVA based on Bray Curtis dissimilarities of fourth-root transformed relative abundance data for 130 Wsh species (raw MaxN data) and of 38 families (MaxN’s Source
Labridae (35)
df
summed to family level) in response to factors technique (T), Island group (G), location (L) and their interactions
Species-level data
Family-level data
MS
Pseudo-F
P(perm)
MS
Pseudo-F
P(perm) stereo DOV Labracinus lineatus
¡0.60
0.65 § 0.10
50
0
0
N
Gymnothorax woodwardi
¡0.58
0.52 § 0.08
43
0
0
N
Pagrus auratus
¡0.57
1.18 § 0.27
42
0
0
Y
Lethrinus miniatus
¡0.52
0.50 § 0.10
35
0
0
Y
Coris auricularis
¡0.44
6.80 § 1.20
68
1.30 § 0.40
Lethrinus sp
¡0.46
0.92 § 0.23
32
0
Lethrinus nebulosus
¡0.45
0.83 § 0.19
33
0.03 § 0.02
3
Y
Lethrinus atkinsoni
¡0.43
0.82 § 0.23
30
0
0
Y
32
N
0
Y
Scarus ghobban
¡0.40
0.25 § 0.07
22
0
0
N
Plectropomus leopardus
¡0.39
1.27 § 0.17
63
0.49 § 0.11
29
Y
Chlorurus microrhinos
¡0.38
0.37 § 0.09
25
0.01 § 0.01
1
N
Choerodon rubescens
¡0.38
1.52 § 0.14
82
0.90 § 0.15
49
Y
¡0.35
0.68 § 0.27
23
0
0
Y
Chaetodon lunula
¡0.33
0.53 § 0.18
20
0.04 § 0.04
1
N
Pseudocaranx sp
¡0.32
2.92 § 1.64
15
0.08 § 0.08
1
Y
Parma mccullochi
¡0.30
0.23 § 0.08
18
0.03 § 0.02
3
N
Bodianus bilunulatus
¡0.29
0.17 § 0.06
13
0
0
N
Seriola hippos
¡0.29
0.38 § 0.15
18
0.07 § 0.05
3
Y
Scombridae spp (mackeral)
Epinephelides armatus
¡0.29
0.12 § 0.05
10
0
0
Y
Apogon angustatus
¡0.29
0.20 § 0.08
12
0
0
N
Austrolabrus maculatus
¡0.28
0.10 § 0.04
10
0
0
N
Chaetodon assarius
¡0.28
0.38 § 0.12
22
0.11 § 0.06
6
N
Epinephelus rivulatus Gymnothorax Xavimarginatus
¡0.27
0.23 § 0.06
20
0.07 § 0.04
6
Y
¡0.25
0.07 § 0.03
7
0
0
N
Chaetodon plebeius
0.26
0.50 § 0.14
23
1.24 § 0.20
43
N
Dascyllus reticulates
0.28
0.17 § 0.11
5
1.44 § 0.47
25
N
0
0.18 § 0.07
Stereo DOV > Stereo BRUV
Chaetodon trifascialis
0.34
0
Chromis westaustralis
0.44
47.90 § 19.70
35
73.40 § 11.80
9
N
81
N
Listed are the mean relative abundance per drop/transect (§SE), and the per cent of drops/transects each species were recorded at. Those species considered targeted by commercial and recreational Wshers are indicated
Scombridae (0.36), Apogonidae (0.31) and Monacanthidae (0.28). There were large diVerences in the abundances of individuals within each family depending upon the technique used (Fig. 2). The majority of the assemblage surveyed using stereo DOV comprised individuals in the families Pomacentridae, Scaridae and Labridae. While these families were also common to stereo BRUV deployments, the spread of abundances across families surveyed was far greater (Fig. 2). Ningaloo Marine Park Stereo BRUV and stereo DOV recorded diVerent assemblages of Wsh within the Ningaloo Marine Park (Table 4). Again this diVerence was evident for both species and
123
family data and across all levels of the model. Given the signiWcant all-ways interaction, we considered the reef locations Mandu and Osprey separately for further analyses. Examination of the signiWcant Technique £ Habitat interaction for species and family level data (Table 4), separately for Osprey and Mandu, showed that stereo BRUV and stereo DOV techniques surveyed diVerent assemblages at each habitat (all 20 P < 0.05; Fig. 3a, b). This distinction in Wsh assemblages across diVerent habitat types is illustrated by clear separation of points in Fig. 3a, b. The interaction term appears caused by variability in the magnitude of diVerence between the techniques in each habitat. At Mandu, over tabulate coral habitat, 29 Wsh species were more abundant on stereo BRUV than on stereo DOV
1243
50
Stereo BRUV Stereo DOV
40 30 20 10
Sphyraenidae
Sparidae
Serranidae
Scaridae
Scombridae
Pomacentridae
Muraenidae
Other Families
Mullidae
Lethrinidae
Labridae
Kyphosidae
Chaetodontidae
Carangidae
0 Caesionidae
Proportion of the fish assemblage (%)
Mar Biol (2010) 157:1237–1250
Family
Fig. 2 Total relative abundances of species summed to family level and expressed as a per cent of the total population surveyed using stereo DOV and stereo BRUV. Twenty-three families not included, each made up stereo DOV vs. 9 spp stereo DOV > stereo BRUV). Stereo DOV surveyed many more species in greater relative abundance than stereo BRUV over branching coral (35 spp vs. 22) and algal pavement habitats (19 vs. 2), while both techniques recorded a similar number of species in greater relative abundance over reef Xat habitats. While very diVerent assemblages were recorded by each technique in each habitat (Fig. 3a), many species were consistently associated with stereo BRUV or stereo DOV, regardless of habitat (CAP |r| > 0.25). Consistently more abundant on stereo BRUV than stereo DOV were Abalistes aculeatus, Carangoides fulvoguttatus, Carcharhinus melanopterus, Chaetodon lineolatus, C. lunula, Table 4 Ningaloo Marine Park: PERMANOVA based on Bray Curtis dissimilarities of fourth-root transformed relative abundance data for 201 Wsh species (raw MaxN data) and of 41 families (MaxN’s summed Source
df
Epinephelus fasciatus, E. rivulatus, Gymnothorax javanicus, Lethrinus atkinsoni, L. nebulosus, Scarus ghobban, S. rubroviolaceus and Siganus fuscenscens. Conversely more abundant on stereo DOV than stereo BRUV were Abudefduf sexfasciatus, Acanthurus triostegus, Chaetodon citronellus, C. plebeius, Chlorurus sordidus, Ctenochaetus striatus, Dascyllus aruanus, D. reticulatus, Neoglyphidodon melas, Pomacentrus coelestis, P. moluccensis, Scarus frenatus, S. schlegeli, Thalassoma lutescens and Zebrasoma scopas. A large proportion of the Wsh assemblage surveyed by stereo DOV was from the Scaridae and Pomacentridae families (Fig. 4a). In contrast, a broader suite of families were sampled by stereo BRUV (Fig. 4a). At Osprey, greater species richness was found with stereo DOV than on stereo BRUV over algal pavement (25 spp stereo DOV > stereo BRUV vs. 11 spp stereo BRUV > stereo DOV), tabulate coral (40 spp vs. 30) and lagoonal bommie (39 spp vs. 26 spp) habitats. Species richness was greater on stereo BRUV than on stereo DOV in reef pass (45 spp stereo BRUV > stereo DOV vs. 26 spp stereo DOV > stereo BRUV) and branching coral habitats (11 spp vs. 7). Very diVerent assemblages were recorded by each technique across all habitat types (Fig. 3b), but regardless of habitat, species which were more abundant on stereo BRUV than stereo DOV transects (CAP |r| > 0.25) included Acanthurus aculeatus, Carangoides fulvoguttatus, Carcharhinus amblyrhynchos, Echeneis naucrates, Gymnothorax Xavimarginatus, G. javanicus, Kyphosus biggibus, Lethrinus atkinsoni, Parupeneus cyclostomus, P. indicus, Scarus rubroviolaceus and Triaenodon obesus. Consistently more abundant on stereo DOV than stereo BRUV were Abudefduf sexfasciatus, Acanthurus grammoptilus, Chaetodon citronellus, C. trifacialis, Chlorurus sordidus, Coris auricularis, Ctenochaetus striatus, Dascyllus reticulatus, to family level) in response to factors technique (T), location (L), habitat (H) and their interactions
Species-level data
Family-level data
MS
Pseudo-F
P(perm)
MS
Pseudo-F
P(perm)
T
1
29,277.0
16.32
