Content uploaded by William Gladstone
Author content
All content in this area was uploaded by William Gladstone on Jun 27, 2016
Content may be subject to copyright.
1
Title: Application of baited remote underwater video stations to assess benthic coverage in
the Persian Gulf1
Authors: Amir Ghaziloua, Mohammad Reza Shokria,*, William Gladstoneb
Authors' affiliation addresses
a Faculty of Biological Sciences, Shahid Beheshti University, G.C., Evin, Tehran, Islamic
Republic of Iran
b School of the Environment, University of Technology Sydney, PO Box 123, Broadway, NSW
2007, Australia
*Corresponding Author:
Marine Biology Department, Faculty of Biological Sciences, Shahid Beheshti University,
Daneshju Blvd, 1983963113 Tehran-Iran. Tel.: +98-21-2990 2723.
E-mail address: M_Shokri@sbu.ac.ir
1 This is the accepted version of the manuscript that was published in Marine Pollution Bulletin 105:606-612
(2016).
2
Abstract
A Baited remote underwater video station (BRUVS) is generally considered an appropriate
sampling tools for fish. The applicability of BRUVS to determine the substrate coverage were
assessed by comparing stills from BRUVS videos to traditional point intercept transect (PIT)
data to estimate percentage covers(PC) of different benthic substrate categories. Mean PCs of
hard corals, rock, sand, and coral growth forms yielded statistically identical values with the two
survey methods, while PC of motile epibenthic invertebrates were underestimated by BRUVS in
areas of both high and moderate relief. Yet, multivariate analyses revealed that the two methods
yield similar substrate assemblage in an area of moderate relief. Results of our study suggests
that the BRUVS can be effectively used to quantify both presence/absence of a basic set of
benthic habitat characteristics and diversity of coral growth forms on coral reefs in the Persian
Gulf.
Keywords: remote video; point intercept transect; percentage cover; substrate
3
1. Introduction
Coral reef fish are highly associated with their surrounding habitat and actively respond to
changes in habitat structure (Sale, 1991). As such, accurate information on relationships between
habitat structure and reef fish communities seems fundamental to understanding of the effects of
natural/ anthropogenic disturbances on coral reef functioning (Jones and Syms, 1998). The main
attributes of a marine habitat are structural complexity (topography/rugosity), total biological
coverage, and habitat composition (Öhman and Rajasuriya, 1998). Percentage cover (PC) is a
widely used metric of habitat structure that can provide estimates of both biological coverage and
habitat composition (Hill and Wilkinson, 2004), and is widely used an indicator of ecological
change. For example, PC of live corals can be used as an indicator of the health of coral reefs
(Hill and Wilkinson, 2004), changes in microalgae algae cover can be monitored to detect coral–
algal phase shifts (McManus and Polsenberg, 2004), and changes in PC of abiotic/biotic
substrata can be used to predict altered settlement and recruitment in reef fish (Tolimieri, 1995).
Accurate information for PC is usually obtained by diver-based survey methods (e.g. Manta tow,
line intercept transect, point intercept transect (PIT), and timed swimming) (Hill and Wilkinson,
2004). These methods, however, have certain limitations in terms of the diving time and depth,
and the availability of trained divers (Hill and Wilkinson, 2004). Remote systems (e.g., remotely
operated vehicles, autonomous underwater vehicles, drop Video systems) have accordingly been
developed (Mallet and Pelletier, 2014; Singh et al., 2004).
A baited remote underwater video station (BRUVS) is a drop video system which is usually used
for assessing fish populations. Yet, attempts have been made in recent years to make the use of
4
BRUVS in estimating PCs by point sampling of the stills taken off the video from a BRUVS
(Cappo et al., 2011; Dorman et al., 2012). In this case, the BRUVS method can be viewed as a
promising alternative to the diver-based methods for determining PC which suffer from SCUBA
diving limitations. Additionally, it will provide estimates of both cover and fish abundance (and
species richness) in a single observational unit, thereby requiring less labor-intensive operations
in the field. Meanwhile, performance of the BRUVS as a tool for determining PC may be
affected by general problems associated image analysis and/or point sampling procedures, e.g.,
detectability issues in complex/ rugose environments (Leonard and Clark, 1993), and accuracy
issues related to the way(s) that the photos are examined using different point sampling strategies
(Endean et al., 1997). The present study was designed to explore the use of BRUVS still photos
for estimating benthic cover on coral reefs in the Persian Gulf. We looked for the accuracy of the
method by comparing the estimated mean PCs of different substrate categories with the same
measurements taken by performing PIT (a diver-based point sampling survey method) surveys.
We tested two main hypotheses: (i) suitability of the BRUV method for determining benthic
cover is dependent on the habitat rugosity/complexity (expressed as coral relief; (Carpenter et al.,
1981), (ii) reliability of the estimated PC is dependent on the patterns of points overlaid on
BRUVS photos.
2. Materials and methods
2.1. Study area
This study was performed in June 2014 as a part of the Nayband Marine Park monitoring
program that assessed the fish assemblages of Borkouh Bay (27°18.345ʹN, 52°40.389ʹE; an area
of high hard coral relief), eastern Nayband Bay (27°24.153ʹN, 52°35.378ʹE; an area of moderate
5
hard coral relief), and western Nayband Bay (27°28.205ʹN, 52°35.921ʹE; an area of low hard
coral relief). Experimental site were selected according to their live hard coral covers which were
determined in the previous study (Ghazilou et al., unpublished results).
2.2. Experimental procedure and data collection
Two levels of PC measurement were employed:
1- General abiotic/biotic substrate categories including live hard coral (HC), soft coral (SF),
sponge (SP), nutrient indicator algae (NIA: Padina sp.; (Nejatkhah-Manavi et al., 2011)),
recently killed coral (RKC), rock (RC), sand (SD), rubble (RB), and motile epibenthic
invertebrates (MEI).
2- Growth forms of live hard corals including massive, submissive, columnar/digitate, and
encrusting.
A horizontal look-outward system was used for BRUVS deployments. In general, two types of
BRUV systems, a horizontal-BRUVS (HBRUVS) and vertical-BRUVS (VBRUVS), have been
developed for monitoring fish populations but the VBURVS has been shown to be less
appropriate for studying coral reef fish communities (Langlois et al., 2006). The BRUV sampling
apparatus included a GoPro® HERO3 Black Edition HD camera, fixed 0.30 m above the base of
a stainless steel frame, and a plastic bait bag 1.2 m from the camera. All videos were recorded at
depths of 4-6m (a depth range of highest coral cover in the area; (Wilson et al., 2002)). Each cast
recorded 60 min of front-view videos using full HD and wide-angle (170°) casting modes.
Twelve drops (an optimized sample size for estimating reef fish abundances in the area;
Ghazilou et al. unpublished data) were deployed at each study area (Borkouh Bay, eastern
Nayband Bay, or western Nayband Bay) separated by a minimum of 250 m (Dorman et al.,
2012). Snapshots from the recorded videos (one snapshot per video) were captured in the
6
laboratory, imported into Adobe Photoshop CS4, and cropped to produce 10.66 × 4 inch, 300
pixel/inch TIFF images (Fig. 1). The PC of each substrate category or as well as the proportion
of coral growth forms were computed for edited images using Coral Point Count with Excel
extensions (CPCe) software (Kohler and Gill, 2006). Predetermined numbers of random (10, 25,
40, 65, or 80), stratified random (40), or stratified (40) sampling points were overlaid on each
image (Fig. 2). The points were then previewed, and a substrate category was assigned to each
point. Those sampling points which were overlaid on the water column were excluded from the
analyses.
PIT surveys were performed at the same locations. For each PIT survey, a 100 m fiberglass tape
measure (Freemans®- India) was first laid straight on the substratum at the same depth range as
BRUV deployments. Close-up top view photos of the substrate were then taken at 0.5-m
intervals along four intermittent transect segments 20 m in length (modified from PERSGA,
2004). A total of three transect lines were surveyed at each study area. The PC of the substrate
categories and hard coral growth forms were determined by recording the corresponding
categories of substrate that occurred exactly beneath the considered point for each photo.
2.3. Data analysis
The comparative ability of the BRUVS to estimate the benthic coverage was evaluated by
univariate (i.e. PC of different substrate categories, coral growth form richness and diversity
coral growth form diversity) and multivariate (i.e. substrate assemblage) analyses.
2.3.1. PCs of substrate categories
The assumptions of normality and homoscedasticity for the univariate analyses were first
assessed using Kolmogorov-Smirnov and Levene's tests, respectively and arcsine square-root
transformation was applied to achieve normal distributions (Zar, 1998). A one-way analysis of
7
variance (ANOVA) followed by Tukey's HSD test assessed the impact of the survey method on
the estimated PC of each category. A more conservative significance level of P< 0.01 was
considered to account for heteroscedasticity (Zar, 1998).
Pearson's product-moment correlation coefficients were calculated for pair-wise combinations of
BRUV methods to test for the reliability of PCs obtained by applying different point sampling
strategies (Booth et al., 2006). The same approach was used to assess the associations among
substrate components at experimental site.
2.3.2. Coral growth form richness and diversity
Hard coral growth form richness was calculated by counting the number of coral growth forms
that were recorded. Coral growth form diversity was calculated as:
=exp (−�
)
Where is the total number of growth forms in the sample and is proportional coverage of ith
growth form (Wiens and Rotenberry, 1981). Data on richness and diversity were not normal and
normality could not be achieved by transformation. Growth form richness or diversity values
were compared among different methods by means of Kruskal–Wallis tests.
2.3.3. Substrate assemblage
A Bray-Curtis similarity matrix of arcsine square-root transformed data were generated. The
similarity matrices were analysed by a one-factor permutational multivariate analysis of variance
(PERMANOVA) to compare the substrate compositions obtained by the different methods
(Anderson, 2001). Significance of the observed differences was tested at p= 0.05 using 9999
permutaions of residuals under a reduced model. Indices of multivariate dispersion (MVDISP)
were used to check for variability among replicate samples and permutational analysis of
multivariate dispersions (PERMDISP) was performed to determine whether the significant
8
differences were caused by a difference in the position of the treatment groups in multivariate
space and/or differences in their dispersion (Clarke and Warwick, 2001). PERMANOVAs were
followed by post-hoc tests to compare levels of fixed factors. Contribution of each family to the
observed pairwise differences was determined using SIMPER routine in PRIMER V6 software
(Clarke and Gorley, 2006). Non-metric multidimensional scaling (nMDS) on the basis of Bray-
Curtis dissimilarity matrix calculated from arcsine square-root transformed estimates of coverage
ratios was used to visualize variations in substrate assemblage obtained by different PIT or
BRUV methods.
2.3.4. Optimization
The mean levels of precision (standard error as a ratio of the mean) of PC estimates of HC were
calculated for 3–10 replicate BRUV deployments (drops) using bootstrapping with replacement
(modified from Gladstone et al. 2012) to determine the optimum number of drops (sample size)
needed to estimate hard-coral PC. Random simulations (n = 1000) were performed using a batch
of 36 replicates available for each site. Calculated mean precisions were plotted against replicate
size, allowing the determination of the optimum sample size at a precision level of 0.1.
To determine the optimum number of sampling points (sample unit size) for estimating hard-
corals PC, the BRUV data were subjected to a power analysis by statistically comparing (one-
way ANOVA) the minimum change (δ) of substrate coverage that could be detected using 10,
25, 40, 65, or 80 random points (modified from (Lam et al., 2006)).
Optimization of the sampling design was performed to obtain a precise estimation of live hard
coral coverage and other categories were not considered from the analyses. This was due to the
highlighted importance of HC in determining reef status (English et al., 1994)
3. Results
9
3.1. Single category coverage
Only four substrate categories occurred frequently (>50%) in the survey samples of eastern
Nayband Bay, or western Nayband Bay: HC, RC, SD, and MEI. RB only occurred in eastern
Nayband Bay. The other categories were observed very rarely and were thus excluded from the
analyses. Hard corals were predominant at Borkouh Bay, whereas rock constituted the
predominant substrate category at eastern and western Nayband Bay (Fig. 3). PCs of HC, RC,
SD, and RB did not differ significantly (P≥ 0.05) among examined methods (Table 1).
Holothurians and echinoids were the major components of the MEI category. There was a
significant difference in PC of MEI obtained by different methods (Table 1). Pairwise
comparisons showed that the BRUVS method significantly (P<0.05) underestimated the PC of
MEI in areas of high and moderate relief (Fig. 3). The MEI and HC were significantly negatively
correlated in areas of both high (r=-0.88, p=0.00) and moderate relief (r= -0.73, p= 0.01).
The way that the points were overlaid in BRUVS photos had no significant (P≥0.05) effect on
estimated PCs of substrates (Fig. 3). The correlation coefficients between different strategies
were generally strong and significant for all substrate categories except for MEI (Table 2).
Benthic hard coral communities of experimental sites were dominated by digitate corals
contributing 70.3%±4.3 to coral cover at Borkouh Bay, 40.12% ±7.3 at eastern Nayband bay,
and 55.51%±4.44 at western Nayband Bay. Coral growth form richness and diversity did not
differ significantly among BRUV and PIT methods (Table 3)
3.2. Substrate assemblage
The PERMANOVA showed that the estimates of the substrate assemblage of Boukouh Bay,
differed significantly (P<0.05) among the survey methods (Table 4). Follow-up pairwise tests
between methods indicated significant differences between the PIT and BRUV methods (Table
10
4). SIMPER analyses highlighted that the “MEI” category primarily accounted for the observed
differences in substrate assemblage obtained by using PIT and BRUV methods at the Borkouh
Bay (Table 5). Meanwhile, replicate substrate compositions (in terms of percentage cover)
determined by the PIT method (MVDISP=0.62) were less variable than the BRUV (stratified random)
(MVDISP=1.08), BRUV (random) (MVDISP=1.10), or BRUV (stratified) (MVDISP=1.19) and nMDS
ordination plot illustrated some separation between the PIT and BRUV methods in the Borkouh
Bay and eastern Nayband Bay (Fig. 4).
3.3. Optimisation
Increasing the number of points to be sampled in CPCe had a significant effect (p<0.05) on the δ
values for eastern and western Nayband Bay, with significantly higher δs obtained for ≤10
sampling points (Table 6).
The precision of estimates of percentage cover of live hard corals was improved by increasing
the number of replicate BRUV deployments (drops) at examined sites (Fig. 5). Totals of 11, 50,
and 7 replicate deployments were required to achieve the desired precision level (i.e. 0.1) in the
areas of high (Borkouh Bay), moderate (eastern Nayband Bay), and low (western Nayband Bay)
relief, respectively.
4. Discussion
Our results indicated that the analysis of the stills from BRUVS approximates the same estimates
on PCs of HC, RC, SD, or MEI as the PIT surveys in the Persian Gulf. This conclusion,
however, was not robust at different locations across the habitat complexity gradient, suggesting
that the BRUVS may only be effectively applied on relatively low-complexity reefs. Marine
habitats are 3D environments (Costello, 2009), and the ability of a survey method to accurately
and precisely detect the components of the substrata depends on the distributional pattern and
11
size of the specified components and on the topography of the field (Leonard and Clark, 1993).
As such, capturing large and high resolution multi-view images of the substratum is necessary to
accurately detect patchily distributed cryptic components in complex marine environments, e.g.,
coral reefs (Miller and Ambrose, 2000). The lower resolution of digitised large-scale photos,
however, impedes the ability to identify small individuals and cryptic species in 2D planar
photographs (Foster et al., 1991; Leujak and Ormond, 2007; Lirman et al., 2007). In the present
study, a mono-view horizontal video recording system was used for recording videos, resulting
in acquisition of planar 2D still photos. Coral reefs dominated most areas of the 2D photos,
particularly in areas of high or moderate coral relief which made it difficult to recognise
subsurface dwelling sea cucumbers or sea urchins which are generally inhabit subsurface.
Detectability issues have been reported to be a common issue associated with 2D
photography/videography. For example, Foster et al. (1991) found higher estimates of PCs of
sessile animals on point quadrats that photoquadrats. Leujak and Ormond (2007) demonstrated
that that photo-quadrate methods would be less effective in detecting cryptic and/or shaded biota
and suggested the use of filed notes to record presence of cryptic animals. With BRUV video-
photography, the problem may be substantially resolved by analysing the full movie to record the
maximum number of sea urchins/sea cucumbers (MaxN) in a single video frame. The same
approach has been suggested for estimating relative abundances of fish in BRUV surveys (Cappo
et al., 2006). In our study, we did not followed the MaxN approach. Instead, we decided to
analyse the full movie records to obtain the snapshots representing fewest number of fish,
thereby minimizing the shading action of fish on substrates. The captured snapshots were
implicitly used for analyzing substrate coverage (including PC of MEI).
12
The camera viewing angle is another potential negative issue of the BRUV method. We used a
horizontal look-outward system for our study, resulting in production of front view seascape
videophotographs the upper portions of which have been occupied by the water column.
Although, cropping was applied to reduce theses portions, some sampling points were inevitably
overlaid on the water column, and these sampling points were excluded from the analyses. The
effective sampling point density (i.e. number of points overlaid on the actual substrata)
consequently varied among the replicates, which could compromise the accuracy of the
estimated PC. Although not examined in our study, this problem may be partially resolved by
using interactive cropping (fine cropping each image at the water column/substrate interface)
rather than fixed cropping.
Hard coral growth forms have been found as a suitable substitute for coral diversity in areas of
high coral species richness (Hill and Wilkinson, 2004). Information on diversity and richness of
coral growth form can also be used for monitoring changes in reef topography (Hill and
Wilkinson, 2004). Leujak and Ormond (2007) concluded that, compared to the PC of broad
substrate categories, relatively higher levels of resolution should be deployed to get accurate
estimates on PC of coral growth forms. Results of our study indicated that, resolution of the
BRUVS photos were high enough to ensure accurate determination of PC of coral growth forms,
since no magnifications were performed during the process of image analysis. Yet, application of
point sampling-based methods is not recommended for assessing coral growth form diversity due
to low precision (Leujak and Ormond, 2007).
In the present study, analyses of BRUVS photos were performed using differed point sampling
strategies, including random sampling, stratified random sampling, and stratified sampling. We
have compared the accuracy different point sampling strategies and found no significant
13
differences among different treatments. Estimates of PCs of HC, RC, or SD were also highly
correlated among random, stratified random and stratified methods, suggesting high reliability of
different sampling methods. Previous studies employing the use of different sampling scheme for
photo/video analysis revealed somewhat inconsistent results. For example, (Leujak, 2006) found
no differences between random and non-random sampling strategies for video-transect method.
In contrast, Endean et al. (1997) demonstrated that the stratified sampling designs may suffer
autocorrelation particularly for those components that are regularity distributed (e.g. Porites
corals). In the present study, we did not found significant correlation between estimated PCs of
HC obtained by random and stratified/stratified random sampling patterns in an area of low relief
(western Nayband Bay). In this case, our result may highlight the autocorrelation phenomenon
on areas of low coral relief (western Nayband Bay), since coral colonies were more dispersed in
western Nayband Bay.
In general, studies on fish-habitat associations include utilization of different methods for
quantifying fish abundances and habitat structure (e.g. Ahmadia et al., 2012; Bozec et al., 2005).
Baited remote underwater video stations has become a common tool for estimating the
abundance and diversity of fish in marine protected areas, deep sea habitats, and/or
topographically complex marine habitats (Cappo et al., 2006). The method have been found to
provide cost effective alternative to diver-based methods, minimizing issues associated observer
errors and altered behavior of fish in the presence of divers (Watson and Harvey, 2007). Based
on the results of the current study, it can be concluded that the BRUV technique can also be used
to accurately determine the percent coverage of main substrata in marine environments. In this
case BRUVS will simultaneously provide accurate information on both fish abundance and
habitat structure, resulting in less labor intensive operations in the field.
14
Acknowledgments
This research was partially funded by Pars Oil and Gas Company (93-283/pt). Many thanks for
the accommodation provided by Mr. Moazzeni and the boat trips provided by Mr. Khalafi.
References
Ahmadia, G., Pezold, F., Smith, D., 2012. Cryptobenthic fish biodiversity and microhabitat use in healthy
and degraded coral reefs in SE Sulawesi, Indonesia. Mar Biodiv 42, 433-442.
Anderson, M.J., 2001. A new method for non-parametric multivariate analysis of variance. Austral
Ecology 26, 32-46.
Booth, D.T., Cox, S., Berryman, R., 2006. Point Sampling Digital Imagery with ‘Samplepoint’. Environ
Monit Assess 123, 97-108.
Bozec, Y.M., Dolédec, S., Kulbicki, M., 2005. An analysis of fish-habitat associations on disturbed coral
reefs: chaetodontid fishes in New Caledonia. Journal of Fish Biology 66, 966-982.
Cappo, M., Harvey, E., Shortis, M., 2006. Counting and measuring fish with baited video techniques—an
overview in: Lyle, J., Furlani, D., Buxton, C. (Eds.), Australian Society of Fish Biology Conference and
Workshop on Cutting Edge Technologies in Fish and Fisheries Science. Journal of Fish Biology, Hobart,
pp. 101-114.
Cappo, M., Stowar, M., Syms, C., Johansson, C., Cooper, T., Cappo, M., Stowar, M., Syms, C.,
Johansson, C., Cooper, T., 2011. Fish-habitat associations in the region offshore from James price point -
a rapid assessment using baited remote underwater video stations (BRUVS). Journal of the Royal Society
of Western Australia 94, 303-321.
Carpenter, K.E., Miclat, R.I., Albaladejo, V.D., Corpuz, V.T., 1981. The influence of substrate structure
on the local abundance and diversity of Philippine reef fishes., 4th International Coral Reef Symposium,
pp. 497-502.
Clarke, K.R., Gorley, R.N., 2006. PRIMER v6: User Manual/Tutorial. PRIMER-E, Plymouth.
15
Clarke, K.R., Warwick, R.M., 2001. Change in Marine Communities: An Approach to Statistical
Analysis and Interpretation. PRIMER-E, Plymouth.
Costello, M., 2009. Distinguishing marine habitat classification concepts for ecological data management.
Marine Ecology Progress Series 397, 253-268.
Dorman, S.R., Harvey, E.S., Newman, S.J., 2012. Bait Effects in Sampling Coral Reef Fish Assemblages
with Stereo-BRUVs. PLoS ONE 7, e41538.
Endean, R., Cameron, A., Fox, H., Tilbury, R., Gunthorpe, L., 1997. Massive corals are regularly spaced:
pattern in a complex assemblage of corals. Marine Ecology Progress Series 152, 119-130.
English, S., Wilkinson, C., Baker, V., 1994. Survey manual for tropical marine resources. Australian
Institute of Marine Science, Townsville.
Foster, M.S., Harrold, C., Hardin, D.D., 1991. Point vs. photo quadrat estimates of the cover of sessile
marine organisms. Journal of Experimental Marine Biology and Ecology 146, 193-203.
Gladstone, W., Lindfield, S., Coleman, M., Kelaher, B., 2012. Optimisation of baited remote underwater
video sampling designs for estuarine fish assemblages. Journal of Experimental Marine Biology and
Ecology 429, 28-35.
Hill, J., Wilkinson, C., 2004. Methods For Ecological Monitoring Of Coral Reefs : A Resource For
Managers. Australian Institute of Marine Science ( AIMS), Townsville, Australia.
Jones, G.P., Syms, C., 1998. Disturbance, habitat structure and the ecology of fishes on coral reefs.
Australian Journal of Ecology 23, 287-297.
Kohler, K.E., Gill, S.M., 2006. Coral Point Count with Excel extensions (CPCe): A Visual Basic program
for the determination of coral and substrate coverage using random point count methodology. Computers
& Geosciences 32, 1259-1269.
Lam, K., Shin, P.K.S., Bradbeer, R., Randall, D., Ku, K.K.K., Hodgson, P., Cheung, S.G., 2006. A
comparison of video and point intercept transect methods for monitoring subtropical coral communities.
Journal of Experimental Marine Biology and Ecology 333, 115-128.
16
Langlois, T.J., Chabanet, P., Pelletier, D., Harvey, E., 2006. Baited underwater video for assessing reef
fish populations in marine reserves, Secretariat of the South Pacific Community Fisheries Newsletter, pp.
53-56.
Leonard, G.H., Clark, R.P., 1993. Point quadrat versus video transect estimates of the cover of benthic red
algae. Marine Ecology Progress Series 101, 203-208.
Leujak, W., 2006. Monitoring of Coral Communities in South Sinai, Egypt, with Special Reference to
Visitor Impacts. University of London, Millport, p. 517.
Leujak, W., Ormond, R.F.G., 2007. Comparative accuracy and efficiency of six coral community survey
methods. Journal of Experimental Marine Biology and Ecology 351, 168-187.
Lirman, D., Gracias, N., Gintert, B., Gleason, A., Reid, R., Negahdaripour, S., Kramer, P., 2007.
Development and application of a video-mosaic survey technology to document the status of coral reef
communities. Environ Monit Assess 125, 59-73.
Mallet, D., Pelletier, D., 2014. Underwater video techniques for observing coastal marine biodiversity: A
review of sixty years of publications (1952–2012). Fisheries Research 154, 44-62.
McManus, J.W., Polsenberg, J.F., 2004. Coral–algal phase shifts on coral reefs: Ecological and
environmental aspects. Progress in Oceanography 60, 263-279.
Miller, A.W., Ambrose, R.F., 2000. Sampling patchy distributions: comparison of sampling designs in
rocky intertidal habitats. Marine Ecology Progress Series 196, 1-14.
Nejatkhah-Manavi, P., Rafiee, F., Shariat zadeh, S., Seifi, S., 2011. Identification and biomass of
macroalgae in intertidal zone of Bushehr Province. Research in marine science and technology 20, 81-88.
Öhman, M., Rajasuriya, A., 1998. Relationships between habitat structure and fish communities on coral.
Environmental Biology of Fishes 53, 19-31.
PERSGA, 2004. Standard survey methods for key habitats and key species in the Red Sea and Gulf of
Aden, PERSGA Technical series. Regional Organisation for the Conservation of the Environment of the
Red Sea and Gulf of Aden, p. 310.
17
Sale, P.F., 1991. Habitat structure and recruitment in coral reef fishes in: Bell, S.S., McCoy, E.D.,
Mushinsky, H.R. (Eds.), Habitat Structure: The physical arrangement of objects in space. Springer,
Suffolk, Germany, pp. 197-206.
Singh, H., Armstrong, R., Gilbes, F., Eustice, R., Roman, C., Pizarro, O., Torres, J., 2004. Imaging Coral
I: Imaging Coral Habitats with the SeaBED AUV. Subsurface Sensing Technologies and Applications 5,
25-42.
Tolimieri, N., 1995. Effects of microhabitat characteristics on the settlement and recruitment of a coral
reef fish at two spatial scales. Oecologia 102, 52-63.
Watson, D.L., Harvey, E.S., 2007. Behaviour of temperate and sub-tropical reef fishes towards a
stationary SCUBA diver. Marine and Freshwater Behaviour and Physiology 40, 85-103.
Wiens, J.A., Rotenberry, J.T., 1981. Habitat Associations and Community Structure of Birds in
Shrubsteppe Environments. Ecological Monographs 51, 21-41.
Wilson, S., Fatemi, S.M.R., Shokri, M.R., Claereboudt, M., 2002. Status of Coral Reefs of the
Persian/Arabian Gulf and Arabian Sea Region, in: Wilkinson, C.R. (Ed.), Status of coral reefs of the
world. Australian Institute of Marine Science, Townsville, pp. 53-62.
Zar, J.H., 1998. Biostatistical Analysis. Prentice Hall, New Jersey.
18
Table 1. Results of one-way ANOVA testing the effects of survey method on estimated
percentage cover. HC, live hard corals; RC, rock; SD, sand; MEI, motile epibenthic
invertebrates; RB, rubble
Hard coral relief
Substrate category
df
1
df
2
F
p
High
HC
3
44
1.86
0.15
RC
3
44
0.26
0.85
SD
3
44
1.80
0.16
MEI
3
44
48.64
0.00
Moderate
HC
3
44
0.67
0.57
RC
3
44
0.27
0.84
SD
3
44
0.54
0.40
MEI
3
44
4.19
0.01
RB
3
44
0.29
0.83
Low
HC
3
44
0.19
0.90
RC
3
44
0.49
0.61
SD
3
44
1.57
0.21
MEI
3
44
0.60
0.62
19
Table 2. Pearson’s correlation coefficient (r) and levels of significance of correlations between different
strategies of point sampling. HC, live hard corals; RC, rock; SD, sand; MEI, motile epibenthic
invertebrates; R, random; SR, stratified random; S, stratified
Area of high relief
Area of moderate relief
Area of low relief
Substrate category
comparison
r
p
r
p
r
p
HC
R-S
R-SR
SR-S
0.89
0.86
0.90
0.00
0.00
0.00
0.95
0.88
0.96
0.00
0.00
0.00
0.44
0.36
0.70
0.14
0.23
0.01
RC
R-S
R-SR
SR-S
0.94
0.95
0.94
0.00
0.00
0.00
0.92
0.93
0.97
0.00
0.00
0.00
0.50
0.71
0.74
0.09
0.008
0.006
SD
R-S
R-SR
SR-S
0.63
0.78
0.86
0.02
0.002
0.00
0.62
0.76
0.78
0.029
0.004
0.002
0.73
0.90
0.69
0.006
0.00
0.01
MEI
R-S
R-SR
SR-S
0.00
-
-
1.00
-
-
0.56
0.57
0.20
0.055
0.05
0.51
0.14
0.47
0.10
0.65
0.11
0.74
20
Table 3. Results of Kruskal–Wallis tests comparing the effects of survey method on coral growth form
richness and diversity
Hard coral relief
Variable
df
H
P (adjusted for ties)
High
Growth form richness
3
5.76
0.12
Growth form diversity
3
4.58
0.20
Moderate
Growth form richness
3
3.56
0.35
Growth form diversity
3
1.25
0.74
Low
Growth form richness
3
5.34
0.14
Growth form diversity
3
3.05
0.38
21
Table 4. Results of PERMANOVA and follow-up pairwise comparisons on PC estimation data obtained
by different survey methods at high, moderate or low levels of coral relief
Coral relief
Pseudo F
p(perm)
High
3.86
0.00
Moderate
1.76
0.10
Low
1.32
0.23
Pairwise comparisons (the area of high relief)
t- value
P(perm)
PIT Vs. BRUV(random)
2.94
0.002
PIT Vs. BRUV(stratified random)
3.34
0.001
PIT Vs. BRUV(stratified)
2.73
0.002
BRUV(random) Vs. BRUV(stratified random)
0.27
0.92
BRUV(random) Vs. BRUV(stratified)
0.24
0.91
BRUV(stratified random) Vs. BRUV(stratified)
0.55
0.75
Note: multivariate dispersions were not significantly different (p≥0.05) among different groups
22
Table 5. Percent contribution of the top contributor category to the observed differences between survey methods
Top contributor
Average proportion
%δ
δ/SD
Contribution (%)
Group 1
Group 2
PIT Vs. BRUV(random)
OT
6.44
4.01
9.59
2.15
31.42
PIT Vs. BRUV(stratified random)
OT
6.44
4.53
11.09
3.17
34.09
PIT Vs. BRUV(stratified)
OT
6.44
4.49
9.75
2.28
39.99
23
Table 2. Comparison of minimum detectable changes of live hard corals derived from different
number of sampling points used for BRUV analysis.
*Different letters within a row indicate significant differences (P≤0.05).
BRUV-photo (Random)
PIT
10
25
40
65
80
40
Borkouh Bay (an area of high coral relief)
7.1A*
2.0AB
0.9B
0.5B
0.4B
0.3B
Eastern Nayband Bay (an area of moderate coral relief)
10.7A
3.4B
1.6B
0.8B
0.7B
0.2B
Western Nayband bay (an area of low coral relief)
8.8A
1.5B
0.8B
0.6B
0.3B
0.2B
24
Fig 1
25
Fig 2
26
Fig 3
27
Fig 4
28
Fig 5
29
Figure legends
30
Fig. 1. A BRUV video photograph before (left) and after (right) cropping. Note the reduced
proportions of the water column in the cropped image.
Fig. 2. Strategies by which sampling point were overlaid on BRUV stills: (a) random, (b)
stratified random, and (c) stratified.
Fig 3. Mean percentage cover of different substrate categories obtained by BRUV and point PIT
survey methods. HC, live hard corals; RC, rock; SD, sand; MEI, motile epibenthic invertebrates;
RB, rubble; –R, random; -SR, stratified random; -S, stratified. Note: dissimilar uppercase letters
denote significant difference (P< 0.05) between methods. Error bars represent ±1 standard error.
Figure 4. nMDS plots of substrate assemblage driven by different survey methods. High, area of
high relief; Moderate, area of moderate relief; Low, area of low relief
Fig 5. The effect of replicate size on the precision of the estimates of hard-coral coverage
obtained by BRUVs at (a) Borkouh Bay, (b) eastern Nayband Bay, and (c) western Nayband
Bay.
