1
Habitat Surveys of Saba Bank, Netherlands Antilles:
An Assessment of Benthic Communities and Fish Assemblages Wes Toller
Saba Conservation Foundation Saba, Netherlands Antilles
January 31, 2008 Abstract
Saba Bank is a large and completely submerged carbonate platform in the northeastern Caribbean Sea located approximately 4 km southwest of Saba Island, Netherlands Antilles.
Zonation patterns of reef-like bathymetric features, together with observations of significant shelf edge coral reef development, suggest that Saba Bank is an actively growing coral reef atoll.
Little quantitative data exists to evaluate the composition and distribution of marine benthic communities or fish assemblages of Saba Bank. In the present study, habitat surveys were conducted to investigate the abiotic characteristics, benthic community composition, and fish assemblage structure of habitats from an eastern portion of Saba Bank known as Overall Bank. A random stratified sampling design was developed that utilized remote sensing data for
bathymetry and ocean color superimposed on reef zones. Five sampling strata, which putatively delineated five distinct marine habitat types, were identified along a shelf edge-to-lagoon
gradient. Survey results indicate that the proposed strata correspond to distinct marine habitat types in terms of substrate composition, benthic cover, and dominant macro algae. Significant coral cover was restricted to the outer reef edge in the fore reef habitat (11.5 %) and outer reef flat (2.4 %), declining to near absence in the lagoon habitats towards the bank center. Macro algae dominated benthic cover in all habitats (32.5 – 48.1 % cover) with the composition of dominant algal genera differing among habitats. Gorgonians reached their highest density and greatest average colony height in the fore reef zone. Gorgonian colony height was also
pronounced in softbottom habitats of the lagoon. Fish assemblage structure showed patterns that were concurrent with observed habitat zonation. Highest fish densities were observed in the outer reef flat, fore reef, and inner reef flat zones. Fish abundance and diversity was low in the lagoon zone and lowest over softbottom habitats within the lagoon. The greatest diversity of fishes (average number of species per survey, cumulative number of species) occurred in the fore reef zone and outer reef flat zone. Fish biomass followed the same pattern of distribution, with the greatest weight occurring in the outermost zones and least in the lagoon. Queen conch were most frequently encountered in the softbottom lagoon zone and estimates of average conch density were between 42 and 60 individuals per hectare. Abundance of spiny lobster was not adequately surveyed by the methods employed in this study and recommendations are made for improved field assessment of lobster stocks. Collectively, the results of this study indicate that the benthic communities of Saba Bank follow predictable patterns of distribution, diversity, and abundance across a gradient from shelf edge to lagoon. Recommendations for future research are given.
Introduction
Saba Bank, Netherlands Antilles, is a large and completely submerged island located in the northeastern Caribbean Sea. The Bank is an isolated, flat-topped carbonate platform whose geological origin sparked some controversy among geologists (Vaughan, 1919, Davis, 1926).
Most recent authors consider Saba Bank to be a submerged coral reef atoll (Van der Land, 1977, Meesters et al., 1996). Early studies of Saba Bank focused on geology, geomorphology,
bathymetry and some aspects of biological communities (Macintyre et al., 1975, Van der Land, 1977, Meesters et al., 1996). More recent investigations have placed greater emphasis on investigation of the biological communities found on Saba Bank, including assessments of coral reefs (Klomp and Kooistra, 2003), fisheries resources (Dilrosun, 2000), and inventories of marine biological diversity (e.g. Conservation International, 2006). The present study builds upon previous work by collecting quantitative information on habitat composition and distribution, and associated fish assemblage structure.
The large size of Saba Bank (2,200 km2 above the 200 meter isobath according to Meesters et al., 1996, Figure 1) makes it impractical to assess with visual surveys anything more than a small fraction of Saba Bank’s total area. Discussions among scientists participating in The Saba Bank Project led to the following consensus. Remote sensing data should be used to develop a
stratified sampling design to investigate relationships among reef zones, benthic habitats, and fish assemblages. Survey sites should be selected at random. Assessments should emphasize commercially important finfish and invertebrate stocks from Saba Bank. Where possible, the sampling design should also allow collection of information useful for “ground truthing” the development of Saba Bank benthic habitat maps in the future.
Benthic habitat maps for Saba Bank will likely employ a hierarchical, bottom-up approach. Such hierarchies may be rooted on the geology of the seafloor (e.g. Greene et al., in press) or founded on large-scale patterns reef zonation (e.g. Kendall et al., 2001). Because detailed geological maps are not presently available for Saba Bank, the latter option was selected. Van der Land (1977) provided the first description of Saba Bank reef zonation patterns (see below). His reef zones were selected as the best working hypothesis of the primary determinant for habitat
structure and distribution. Reef zonation was first delineated on maps, and then further evaluated using contemporary remote sensing datasets for bathymetry and ocean color.
Reef Zones and Sampling Strata
Van der Land (1977) considered Saba Bank to be a submerged coral reef atoll and he used coral reef terminology to describe specific reef zones. In the present study, Van der Land’s terms were adopted as they apply to reef zones within the study area at Overall Bank. It is cautioned,
however, that these zones are based upon a submerged (i.e. former) coral reef atoll lying 20-40 m below sea level. Physical and ecological processes of Saba Bank are still poorly known and may be influenced by environmental conditions that do not typify modern (i.e. emergent) coral reef atolls. To give an example, the Saba Bank “lagoon” lacks a barrier reef system altogether, which leaves lagoon sites largely exposed to undiminished wave energy and oceanic currents. Thus, the lagoonal assemblage of Saba Bank might not resemble those from typical protected lagoon environment found elsewhere in the Caribbean.
3 Figure 1. Bathymetric map of Saba Bank.
4
The reef zonation pattern follows a sequence from shelf edge to central bank. At Overall Bank, reef zones occur in the following sequence as one moves from east (windward, open ocean) to west (leeward, towards central Saba Bank): seaward slope, fore reef (with one or more “front reefs”), reef flat, backreef slope (“escarpment”), lagoon, and patch reef (located within the lagoon). These reef zones are illustrated in a typical bathymetric profile across Overall Bank (Figure 2).
-50 -40 -30 -20 -10 0
0 1000 2000 3000 4000 5000 6000 7000 8000
Distance (m)
Depth (m)
Reef Flat Backreef
Slope Lagoon
Patch
Reef Fore
Reef Front Reefs
WEST EAST
Figure 2. Depth profile of Overall Bank study area.
The fore reef zone is a steeply sloping and topologically variable region. Van der Land observed a “front reef” rising from a “reef terrace” at 30-40 m depth. High-resolution bathymetry
confirmed the presence of at least one front reef feature at Overall Bank. To the west (leeward) of the front reef, an area resembling a spur-and-groove reef is found. In the present study, these various reef features are considered elements of a single zone - the fore-reef zone (stratum A).
The seaward slope was not included in the proposed stratification scheme because it lies beyond safe diving limits. For more information on the seaward slope zone of Overall Bank, see
Macintyre et al. (1975) for geological observations and Toller et al. (2008) for observations made by remotely operated vehicle.
Westward (leeward) of the fore reef zone, the reef rises to ~ 15 m depth and forms a wide (>
1000 m) level expanse. Van der Land (1977) identified this area as the reef flat and suggested that it comprised an inner and outer zone distinguished by bathymetry. Examination of recent high-resolution bathymetry data did not differentiate inner and outer reef flat zones within the study area (e.g. Figures 2, 3A). However, LandSat images indicated two distinct color intensities within the reef flat zone (Figure 3B). The outer (eastern, windward) reef flat was light blue in color while inner (western, leeward) reef flat appeared dark green, with the dark band extending continuously across the backreef slope area (Figure 3B). Based on these differences, the outer and inner reef flats were designated as stratum B and C, respectively.
5
Figure 3. Remote sensing data for study area. A. Bathymetry. B. LandSat image. C. Composite.
6
The lagoon zone extends westward (leeward) from the reef flat and backreef slope zones. Van der Land considered the lagoon a single zone, although he distinguished “patch reef’ formations within it. Bathymetry confirmed the presence of patch reef-like features within the lagoon.
However patch reefs were not included as a sampling stratum due to the very small area they occupy. Inspection of LandSat images showed two distinctive color intensities within the lagoon zone. These are shown as light blue and dark green areas in Figure 3B. The spatial distribution of color was independent of patch reefs and other identifiable bathymetric features (Figure 3C).
The light blue areas were hypothesized to be sand bottom habitat within the lagoon (stratum D) and the dark green areas were suspected to be areas of aquatic vegetation within the lagoon (stratum E).
Methods
Study Area
The study area was situated in a central part of eastern Saba Bank known as Overall Bank. This area lies entirely within Saban territorial waters (< 12 nautical miles from shore). Sampling was conducted in a rectangular area (7.3 km long by 5.5 km wide) that encompassed 40.2 km2
(Figure 1). Overall Bank was selected as a study area for a combination of reasons. 1. It lies at the center of Van der Land’s (1977) “Southeastern Reef” - the largest reef system on Saba Bank.
2. It encompasses most of the conspicuous geomorphological reef features known from Saba Bank. 3. Overall Bank reef features are oriented perpendicular to the primary gradients of wind, current, and wave energy (which runs east to west) and it was hypothesized that biological communities would show a response across these same gradients. 4. It is representative of the shallow shelf areas targeted by Saban commercial lobster trap fishermen. 5. Overall Bank is located relatively close to the island of Saba (~ 10 miles), thus enabling frequent visits to field sampling stations. 6. Depths within most of the study area were within safe diving limits, thus facilitating complete surveys using conventional scuba diving.
Remote Sensing Datasets
High-resolution multibeam bathymetry was an essential dataset for developing and refining the stratification scheme. These data come from a 2006 bathymetric survey by the Hydrographic Service of the Royal Netherlands Navy. Multibeam sonar data for Saba Bank and adjacent Small Bank were collected aboard the HNLMS Snellius. Approximately 60% of the Bank’s surface area was surveyed, down to 200 m depth around the entire perimeter of the Bank. Spatial
resolution of bathymetric data was 2 x 2 m and the vertical resolution was < 0.2 m. The resulting bathymetric map of Saba Bank is shown in Figure 1. Note that a central wedge-shaped portion of Saba Bank was not included in the 2006 bathymetric survey due to time limitations. For this area lower resolution (5 x 5 m grid) single-beam sonar data from 1988 and 1996 were used to fill in the map, with the exception of a narrow NE-oriented white band where no digital data were available due to an incorrect prioritization of the Snellius’ survey orders. During the Snellius 2006 hydrographic survey, sidescan sonar data was also collected but it was not available to the author at the time of this study.
Satellite images of ocean color formed another important dataset for formulating the
stratification scheme. A geo-referenced LandSat image from 26-Mar-2002 provided the best
coverage of eastern Saba Bank. Image resolution (pixel size) was ~ 30 x 30 m (900 m ). The LandSat image was imported into GIS and constituent color channels were modulated by eye to maximize visual contrast. After adjusting transparency (~ 60 %), the LandSat image was superimposed on bathymetry datalayers to examine the congruence among bathymetric features and ocean color features. This composite overlay (Figure 3C) served to further refine the stratification scheme for habitat distribution across reef zones.
Delineation of Sampling Strata
To delineate strata for sampling, bathymetric and LandSat images were superimposed in ArcGIS 9.2 (Figure 3C). Initially, the proposed strata were hand-drawn on a hardcopy map as a series of five non-contiguous rectangular boxes within the study area. The boxes were sequentially arranged from the shelf edge, running east to west with the long axis of each box oriented ~ north to south. Later, the five hand-drawn boxes were digitized as polygons using GIS software.
Polygon perimeters were adjusted to avoid areas of transition between reef zones and to exclude areas with depths > 30 m. Polygons were named A, B, C, D, and E, and correspond to fore reef, outer reef flat, inner reef flat, lagoon-softbottom and lagoon-vegetated, respectively. Note that Van der Land’s (1977) reef zones corresponding to backreef slope and patch reef were not specifically delineated in this stratification scheme.
Figure 4. Sampling strata and survey sites.
Survey sites were selected at random positions within each stratum (polygon). Random sites were generated using the “Create Random Points” function in ArcGIS Data Management extension. The location of survey sites is shown in Figure 4 and geographic coordinates are given in Appendix 1. In the field, the position of survey sites was located using a WAAS- enabled GPS receiver (Garmin GPSMAP 76 or GPS 178C). Operational accuracy was < 10 m.
A weighted block with line and buoy was deployed on site to mark the position for underwater surveys.
Field Assessment Methods
Design of field protocols focused on three assessment priorities for data collection: 1) to characterize the habitat in terms of abiotic characteristics (physical substrate, topography), benthic communities, and dominant biota; 2) to quantify the fish assemblage in terms of size, abundance and species composition with particular emphasis on commercially important fish species; and 3) to quantify the size and abundance of queen conch (Strombus gigas) and Caribbean spiny lobster (Panulirus argus). Surveys were designed to be completed by a two- person team in a short dive interval (< 25 minutes). Surveys were conducted at a total of 40 survey sites (Appendix 1), with eight independent surveys completed within each stratum.
Habitat Characterization
At each survey site, the habitat was assessed within a 4 x 25 m belt transect (100 m2). Habitat characterization included the estimation of five physical (abiotic) parameters. Physical substrate was assigned an average percent cover (planar view) into the following categories: hardbottom (consolidated carbonate substrate), rubble (unconsolidated material of < 0.5 m diameter), and sand [Quantifying percent cover of sand posed difficulties at many sites where it was present only as a thin layer (< 1 to 20 cm thickness) on top of hardbottom substrate. The following convention was adopted: if the sand layer was thinner than 0.5 cm, or easily fanned away by hand, it was not included in the score for percent sand cover]. The percent cover of physical substrate always totaled 100%. Note that estimation of physical substrate cover was made independent of any overlying biotic cover (i.e. sponges, corals, algae were considered
“transparent”). Qualitative observations on vertical relief and substrate rugosity were recorded at each survey site. Vertical relief was categorized as low, medium, or high according to reference points of < 0.5 m or > 1 m vertical displacement of substrate as viewed along the transect line.
Similarly, substrate rugosity was categorized into categories of low, medium, or high based on relative degree of substrate involution. Depth and slope were also recorded for each site.
Benthic communities were assessed within the same 100 m2 belt transect. Characterization of benthic communities included estimation of the following parameters. Benthic cover was assigned an average percent cover as estimated from planar view for categories of live coral, sponge, macro algae, and other. When cover was < 1 % it was assigned a value of zero. Note that percent benthic cover was estimated independently of underlying physical substrate. Thus, the percentage of benthic cover often totaled less than 100 % coverage of the substrate (e.g. in sandy areas). The dominant or co-dominant genera of stony corals and macro algae were recorded at each sampling station. Gorgonians were scored using two qualitative categories:
abundance (sparse, medium, dense) and height (low, medium, tall). Criteria for the latter category were < 25 cm (low) and > 1 m (tall).
At each survey site, a set of six digital photos was taken at random points along the transect line.
Photos were taken from planar view ~ 1.8 m above the substrate. Each photo included a measuring bar (1.0 m length marked in 10 cm increments) within the field of view for scale.
Fish Surveys: Belt Transects
Fish surveys were conducted using a belt transect visual survey protocol (Brock, 1954, Bortone et al., 1989). Belt transect dimensions were 4 x 25 m (= 100 m2). A transect tape was affixed to the weighted marker block and the diver swam a random compass bearing while lying a transect tape for a distance of 25 m. Bearing was sometimes modified to accommodate circumstances (e.g. sloping topography, currents or surge). Survey duration was 8-10 minutes. The diver identified, enumerated, and estimated the size of all observed individuals of commercially important fish species (Dilrosun, 2000, Toller and Lundvall, 2008) observed within the belt transect. Only fish passing in front of the diver and within the transect area were counted. The diver estimated the size of each fish (fork length) by reference to a measuring T-bar (Bohnsack and Bannerot, 1986). Fish length was recorded in 5 cm categories. One fish belt transect survey was conducted per survey site. Length data were used to calculate fish biomass using available equations for length-weight relationships (Froese and Pauley, 2007).
Fish Surveys: Roving Diver Surveys
To supplement belt transect assessments of commercially important fishes, a roving diver (RD) survey was also conducted at each survey site. In comparison to belt transects, the RD method enables better enumeration of total species richness (i.e. it yields a more complete fish species list for a given site). As implemented in this study, the RD protocol collected information on species presence or absence, but it did not collect quantitative data for fish abundance or size.
RD surveys were ~ 10 minutes in duration. The diver listed all fish observed while swimming a haphazard circuit within 50-100 m of the survey site marker. Small-bodied demersal species such as gobies, blennies and other cryptic taxa were not included in RD surveys.
Conch and Lobster Surveys
The abundance of queen conch and lobster was quantified within the dimensions of each belt transect (100 m2 per survey). Conch were measured for total length (tip of spire to distal edge of the siphonal canal) to the nearest mm and shell morphology was noted (flared lip or unflared lip).
When more than ten conch were encountered in a single transect, those beyond the tenth were enumerated but not measured. Lobsters were assigned to one of four size classes based upon visual estimation of carapace length (CL): juvenile (< 7 cm), small (7 - 9.5 cm), medium (9.6 - 13 cm), and large (> 13 cm). Following completion of belt transect work, a subsample of conch was collected for measurement of lip thickness and total length. Divers collected the first ten conch found while swimming a haphazard circuit within 50-100 m of the survey site marker.
Lip thickness was measured with plastic calipers to the nearest mm.
Expanded Conch and Lobster Surveys
Both conch and lobster are known to form localized patches of high abundance (i.e. they have clumped distribution patterns). Results from initial surveys indicated that the small size of belt transects (100 m2) relative to the size and distribution of conch and lobster patches yielded a high level of site-to-site variance. In order to increase sampling power and reduce variance, pilot experiments were performed using a belt transect of larger dimensions for conch and lobster.
Expanded conch and lobster survey transects were 50 x 20 m (1000 m2). Two divers surveyed parallel 10 m-wide bands on either side of a 50 m line recording conch and lobster into size categories. A total of 12 such expanded transects were completed at the following survey sites:
A-10, B-3, B-7, C-1, C-6, C-7, C-8, D-1, D-4, D-9, E-11, and E-7. Although this pilot
experiment yielded promising results, the protocol was difficult to implement because it required two additional divers to perform work in parallel with the two divers performing the standardized habitat survey procedures outlined above.
Statistical Analyses
Statistical analyses were performed with Systat Version 9 (SPSS, Inc.) software. Normality tests were used (e.g. Kolmogorov-Smirnov one sample distribution test) to determine if parametric statistical tests were appropriate. When datasets failed to conform to expectations of normality after ln(X) or ln(X+1) transformation, they were analyzed with nonparametric procedures such as Kruskal-Wallis One Way Analysis of Variance. Cluster analyses were performed with Systat software using a data matrix of presence/absence (1/0) data, and dendrograms were constructed using the average linking method based upon Euclidean distance.
Table 1. Abiotic Characteristics
Stratum
A B C D E
Depth (m)
Avg ± StDev 23.6 ± 3.0 13.8 ± 0.4 13.8 ± 0.6 19.4 ± 0.4 20.1 ± 0.7 (Range) (20.7 - 29.9) (13.1 - 14.3) (12.8 - 14.6) (18.6 - 20.0) (19.5 - 21.3) Substrate (%)
Hardbottom
Avg ± StDev 87.4 ± 14.6 56.0 ± 42.0 68.0 ± 29.5 2.5 ± 7.1 44.4 ± 37.6 (Range) (55 - 99) (0 - 100) (5 - 95) (0 - 20) (0 - 95) Rubble
Avg ± StDev 3.5 ± 8.0 39.6 ± 37.1 31.8 ± 29.6 22.6 ± 13.7 50.9 ± 37.6 (Range) (0 - 23) (0 - 85) (5 - 95) (1 - 45) (4 - 100) Sand
Avg ± StDev 9.1 + 8.2 4.4 + 6.8 0.3 + 0.5 74.9 + 17.5 4.8 + 7.1 (Range) (0 - 22) (0 - 15) (0 - 1) (45 - 99) (0 - 20)
Slope (degrees) 5 - 10 < 1 < 1 < 1 < 1
Vertical Relief*
Avg ± StDev 4.3 + 1.0 1.4 + 0.7 1.5 + 0.9 1.1 + 0.4 1.0 + 0.0
(Range) (3 - 5) (1 - 3) (1 - 3) (1 - 2) (1)
Rugosity*
Avg ± StDev 4.5 + 0.9 2.6 + 0.7 1.9 + 0.8 1.0 + 0.0 1.3 + 0.5
(Range) (3 - 5) (1 - 3) (1 - 3) (1) (1 - 2)
* Numeric scores of 1 to 5 (lowest to highest) were assigned to qualitative rank categories for vertical relief and rugosity.
Table 2. Benthic Community Characteristics
Stratum
A B C D E
Benthic Cover (%) Live Coral
Avg ± StDev 11.5 ± 5.8 2.4 ± 1.7 0.4 ± 0.7 0.1 ± 0.4 0.3 ± 0.5
(Range) (5 - 20) (1 - 5) (0 - 2) (0 - 1) (0 - 1)
Sponge
Avg ± StDev 4.2 ± 2.2 2.0 ± 0.5 2.1 ± 1.7 0.4 ± 0.5 1.8 ± 1.5
(Range) (1 - 8) (1 - 3) (0 - 5) (0 - 1) (0 - 5)
Macro Algae
Avg ± StDev 37.5 ± 22.8 46.6 ± 14.6 48.1 ± 21.2 32.5 ± 16.7 43.8 ± 22.8
(Range) (5 - 70) (25 - 65) (10 - 75) (15 - 60) (15 - 80)
Coralline Algae
Avg ± StDev 2.5 ± 4.6 - - - -
(Range) (0 - 10) - - - -
Gorgonian Assemblage*
Density
Avg ± StDev 3.4 ± 0.5 2.4 ± 1.2 1.0 ± 0 1.3 ± 0.7 1.0 ± 0
(Range) (3 - 4) (1 - 4) (1) (1 - 3) (1)
Height
Avg ± StDev 3.1 ± 0.4 1.9 ± 1.0 1.3 ± 0.7 2.9 ± 1.7 1.0 ± 0
(Range) (3 - 4) (1 - 3) (1 - 3) (1 - 5) (1)
Queen Conch (No./100m2)
Avg ± StDev - - - 2.8 + 4.7 N/A
(Range) - - - (0 - 14) (see text)
Dominant Coral Genera† Montastraea (8) Siderastrea (7) none (6) none (7) none (7) Porites (1) Porites (4) Dendrogyra (1) Siderastrea (1) Dichocoenia (1) Diploria (1) Diploria (3) Siderastrea (1) Siderastrea (1) Siderastrea (1) Montastraea (3)
Meandrina (1)
Dominant Algal Genera† Lobophora‡ (8) Sargassum (8) Sargassum (6) Laurencia (5) Lobophora‡ (5) Dictyota (6) Stypopodium (3) Dictyopteris (3) Dictyota (1) Codium (4) Sargassum (1) Dictyopteris (1) Codium (1) Sargassum (2) Dictyota (2)
Halimeda (1) Caulerpa (1) Dictyopteris (1) Eucheuma (1) Sargassum (1) Schizothrix (1)
* Numeric scores of 1 to 5 (lowest to highest) were assigned to qualitative rank categories for gorgonian density and height.
† Dominant or co-dominant genera are listed with number of sites in parentheses.
‡ Two forms of Lobophora variegata differed in distribution: a decumbent form in stratum A and a ruffled form in stratum E.
Results
Habitat Characteristics
The abiotic, physical characteristics of survey sites are summarized by stratum in Table 1 and the benthic community characteristics are presented in Table 2. Collectively, these results indicate that there are consistent differences in biotic and abiotic characteristics of the five strata. In other words, the strata appear to represent valid and distinctive habitat types. However, as illustrated in Figures 5 and 6, the degree of similarity among strata depends to some extent upon which strata are compared and on which habitat characteristics are considered. For example, sand was a significant component of the physical substrate only in stratum D. Although percent cover by sand may distinguish D from the other strata (A, B, C, and E), it is not particularly useful for distinguishing among the remaining four strata. A more detailed multi-factorial approach (e.g. principal components analysis) must be applied to this dataset. Such an analysis may elucidate the most reliable characteristics by which to diagnose habitat type.
Substantial cover by scleractinian corals was only observed in stratum A (average 11.5 %), with sparse cover in stratum B (2.4 %) and minimal or no cover in stratum C, D, and E (Figure 6).
Macro algae was the predominant type of benthic cover in all strata (Figure 6). Average percent cover by macro algae ranged from 32.5 % (stratum D) to 48.1 % (stratum C). Cover by macro algae did not differ significantly among strata (Kruskal-Wallis test statistic = 4.339, p = 0.362).
However, there were clear differences in the dominant genera of algae observed in different strata (Table 2).
0%
25%
50%
75%
100%
A B C D E
Percent Cover
Hardbottom Rubble Sand
Figure 5. Physical substrate characteristics. Average percent cover by substrate category (hardbottom, rubble, sand) is shown for five strata (eight replicates per stratum). Error bars show standard deviation.
Average gorgonian abundance was highest in stratum A while average gorgonian colony height was highest in stratum A and stratum D (Table 2). The large gorgonians observed in stratum D were dominated by Pseudopterogorgia colonies that occurred at low density. A more detailed examination of the distribution and taxonomic composition of Saba Bank gorgonian
communities will be presented elsewhere (Etnoyer et al., in prep.).
0%
25%
50%
75%
100%
A B C D E
Percent Cover
Live Coral Sponge Macro Algae Coralline Algae
Figure 6. Benthic cover. Average percent cover by benthic category (live coral, sponge, macro algae, coralline algae) is shown for five strata (eight replicates per stratum). Error bars show standard deviation.
Stratum A, which corresponds to the fore reef slope zone, was the most distinctive habitat type.
Among the five strata, it was unique for its complex topography and high coral cover (Figure 7).
Figure 7. Stratum A. Habitats of the fore reef slope area were complex reef structures of hardbottom substrate with high vertical relief and rugosity. Coral cover was highest in this zone and was dominated by plating forms such as Montastraea. Gorgonians were abundant. Macro algae were dominated by Lobophora and Dictyota. Photo: site A-12.
Stratum B (Figure 8) and stratum C (Figure 9) comprise the outer and inner reef flat zones of the (former) atoll according to Van der Land’s (1977) zonation scheme. These two strata were similar in terms of abiotic characteristics (Table 1), with stratum B having slightly greater
rugosity. However, benthic community composition differed between B and C in some respects.
Coral coverage was higher in stratum B. Dictyopteris was more frequently observed as the dominant or co-dominant macro algae in stratum C. Dictyopteris is known to form dense algal canopies in some areas of Saba Bank and it is also believed to be seasonally abundant. If dense, seasonal growth of Dictyopteris is characteristic of stratum C but not B, it could explain the basis for color differences in LandSat images that led to differentiation of these two strata.
Figure 8. Stratum B. Habitats of the submerged outer reef flat zone were hardbottom substrates or “pavements” areas. Isolated structures and large rubble fragments provided some vertical relief and rugosity. Coral cover was low in this zone and was dominated by head corals such as Siderastrea siderea and Porites astreoides. Gorgonians were moderately abundant and of medium height.
Sargassum and Stypopodium were the dominant macro algae in this zone were. A queen triggerfish, Balistes vetula, is pictured. Photo: site B-10.
Figure 9. Stratum C. The submerged inner reef flat zone was a low relief, hardbottom substrate or “pavement” area. Rubble fragments provided some vertical relief and rugosity. Coral cover was very low in this zone and
gorgonians were sparse. The dominant genera of macro algae in this zone were Sargassum and Dictyopteris. Photo: site C-3.
Following Van der Land’s classification, stratum D (Figure 10) and stratum E (Figure 11) fall within the lagoon zone of Saba Bank. There were substantial differences in substrate
characteristics (Table 1) between D and E. The preponderance of sand appears to explain why the habitat of stratum D was distinguished as a lighter target in satellite images. Conversely, the habitat of stratum E was characterized by a larger amount of solid substrate and higher
percentage cover by macro algae – either of which may have contributed to creating a darker target in satellite images from the lagoon zone.
Figure 10. Stratum D. The substrate in this part of the former lagoon zone was primarily sand or sand mixed with rubble. Cover by corals and sponges was very low. Gorgonians were sparse although tall colonies of Pseudoterogorgia occurred at some sites. The dominant macro alga was Laurencia. Queen conch were observed with some frequency in stratum D. Photo: site D-9.
Figure 11. Stratum E. The substrate in this part of the former lagoon zone was primarily rubble and hardbottom. Vertical relief was low. Rubble and solution holes provided some rugosity to the seafloor. Corals, sponges, and gorgonians were sparse in this zone. Macro algae were abundant and diverse, Lobophora variegata (ruffled form) and Codium dominated. Photo: E-7.
Fish Assemblages: Belt Transect Surveys
A total of 625 fish were observed in 40 belt transects at Overall Bank [this figure does not include a large school (200 fish) of Bermuda chub, Kyphosus sectator, at site A-1 which was excluded from analyses]. When data from all sites were pooled, the cumulative list includes 34 commercially important fish species that were recorded in belt transects (Table 3). More species occurred in stratum A (24 species) than in stratum B (17 species), C (13 species), D (8 species), or E (10 species).
Fish assemblage structure was compared among strata using metrics of abundance, species richness, and biomass, as determined from belt transect surveys (Figure 12). Average fish density (number of fish per 100 m2) was highest in stratum B (29.9 fish per survey), intermediate in A and C (20.5 and 19.1 fish per survey, respectively) and lowest in D and E (2.6 and 6.0 fish per survey, respectively). Abundance differed significantly among zones (single factor ANOVA, 4 df, F = 14.69, p < 0.001). Tukey tests indicated that stratum A, B, and C each differed
significantly from both D and E (Figure 12A).
Species richness, calculated as the average number of species observed per belt transect, was computed for each stratum (Figure 12B). Average species richness was highest in stratum A and B (7.9 and 8.8 species per survey, respectively), intermediate in C (5.8 species per survey) and lowest in D and E (1.4 and 2.1 species per survey, respectively). Species richness also differed significantly between the five habitat types (single factor ANOVA, 4 df, F = 18.16, p < 0.001).
Tukey tests indicated that stratum A, B, and C each differed significantly from both D and E.
Fish biomass was calculated for each species based upon the number of individuals in each size class. An average value for fish biomass per survey (Kg per 100 m2) was computed by
aggregating the biomass of all species within each stratum. Statistical tests were not applied to this dataset. Nonetheless, computations for fish biomass indicate that, on average, the greater weight of fish occurred in stratum A, B and C (Figure 12C).
It was not possible to apply statistical analyses to species-level questions (e.g. among-zone comparisons of abundance) owing to the relatively small number of observations available for most fishes. Thus, length-frequency analysis for individual species was not performed nor were size datasets tested for differences among zones. Pooled length frequency data is presented in Appendix 4.
0 10 20 30 40 50
A B C D E
No. Fish/100m2
A.
0 2 4 6 8 10 12
A B C D E
No. Species/Survey
B.
0 2 4 6 8 10
A B C D E
Biomass (Kg/100m2 )
C.
Figure 12. Fish assemblage structure as determined from belt transects. Average values are shown for five strata (eight surveys per stratum). Error bars show standard deviation. A. Density. B. Species richness. C. Biomass.
Table 3. Fish Density and Frequency in Belt Transects by Rank Order of Total Abundance
Common Name Density* Freq† Density Freq Density Freq Density Freq Density Freq
ocean surgeon 0.25 13% 8.00 75% 6.13 75% 1.13 38% 1.50 50%
redband parrotfish 2.00 50% 4.13 75% 2.38 50% - - - -
coney 1.50 75% 4.25 100% 1.38 63% 0.13 13% 0.13 13%
princess parrotfish 2.25 63% 3.50 50% - - - - - -
blue tang 1.00 63% 1.75 75% 1.38 50% 0.25 13% 0.13 13%
Bermuda chub‡ 4.38 25% - - - - - - - -
white grunt 0.88 63% 1.00 50% 2.00 75% - - 0.50 13%
doctorfish - - 1.50 50% 2.50 63% - - 0.13 13%
red hind - - 1.00 75% 1.25 63% 0.25 13% 0.75 25%
spotted goatfish 0.25 25% 0.88 50% 0.25 25% - - 1.75 25%
queen triggerfish 0.13 13% 0.88 75% 1.00 63% 0.25 25% 0.13 13%
stoplight parrotfish 1.88 75% 0.25 25% - - - - - -
black durgon 1.63 50% - - - - - - - -
blue runner 1.25 25% - - - - - - - -
great barracuda 0.25 25% 0.50 50% 0.25 13% 0.13 13% - -
rock beauty 0.25 25% 0.88 50% - - - - - -
squirrelfish - - - - 0.38 13% - - 0.50 25%
bar jack - - 0.38 25% - - 0.38 13% - -
graysby 0.63 50% - - - - - - - -
yellowtail snapper - - - - 0.13 13% - - 0.50 25%
longspine squirrelfish - - 0.38 13% - - - - - -
mahogany snapper 0.38 13% - - - - - - - -
redtail parrotfish - - 0.38 25% - - - - - -
striped parrotfish 0.38 25% - - - - - - - -
tomtate 0.38 25% - - - - - - - -
cottonwick - - 0.25 13% - - - - - -
yellow goatfish 0.25 25% - - - - - - - -
black jack 0.13 13% - - - - - - - -
French grunt 0.13 13% - - - - - - - -
nurse shark - - - - 0.13 13% - - - -
queen parrotfish 0.13 13% - - - - - - - -
sand tilefish - - - - - - 0.13 13% - -
smooth trunkfish 0.13 13% - - - - - - - -
Spanish hogfish 0.13 13% - - - - - - - -
* Fish density is reported as the average number of individuals observed/100m2 belt transect (eight replicates per stratum).
† Fish frequency is the percentage of transects where the species was recorded in each stratum (eight replicates per stratum).
‡ A large school of Bermuda chub (200 individuals) observed at site A-1 (stratum A) was exluded from analyses.
Stratum
E
A B C D
Fish Assemblages: Roving Diver Surveys
The Roving Diver (RD) survey was included in order to extend observations of fish diversity to species of no immediate commercial importance. The cumulative species list (pooled data from 40 Overall Bank sites) includes 97 species that were observed in RD surveys (Table 4).
Cumulative species number was greatest in stratum A (72 species) and B (54 species), intermediate in C (46 species) and lowest in D and E (29 and 33 species, respectively).
Species richness of each stratum was compared on the basis of average number of species observed per RD survey – an approach similar to the analysis shown above for belt transect results. Results are shown in Figure 13. The average number of species per survey was greatest in stratum A (28.3), B (28.8) and C (22.4). A lesser number of species were observed per survey in stratum D and E (10.5 and 13.5, respectively). Single factor ANOVA indicated significant differences among strata (4 df, F = 14.81, p < 0.001). Tukey tests indicated that A, B, and C each differed significantly from both D and E.
0 10 20 30 40
A B C D E
No. Species/Survey
Figure 13. Fish species richness as determined from roving diver surveys.
Average species richness (number of fish species per survey) is shown for five strata (eight surveys per stratum). Error bars show standard deviation.
When all sites were pooled, the taxa most frequently observed in RD surveys were the bicolor damselfish (85 %), ocean surgeonfish (80 %), queen triggerfish (70 %), and bluehead (67.5 %).
White grunt, yellowhead wrasse, and blackear wrasse were also frequently observed (each occurring in 65 % of RD surveys). Twenty-one percent of the species were recorded from a single station. Some interesting distribution patterns were found in the RD dataset (Table 4) that are noted briefly here, but must await a more thorough analysis. Some groups showed an
association with shelf edge habitats (stratum A, B, and C) rather than elsewhere (D, E). For example large-bodied parrotfish species such as queen, stoplight, striped, redband and princess parrotfishes, were more common in A, B, and or C and rarely observed in D or E. On the other hand, small-bodied scarids such as the bluelip, greenblotch, and bucktooth parrotfishes, were more evenly distributed among habitats (Table 4).
Cluster analysis was used to investigate the similarity of sites based upon observed fish assemblages. This data matrix was comprised of presence (1) and absence data (0) for 97 fish taxa at 40 survey sites. The resulting dendrogram (Figure 14) indicates that sites within each stratum tend to share similar fish assemblages. In other words, despite a high degree of overlap in species composition among strata, there were consistent within-stratum similarities that tended to form clusters. Distinct fish assemblages were indicated at three of the five strata (A, B, C).
Two strata (D and E) clustered together, but were not differentiated from one another.
0.0 0.1 0.2 0.3 0.4 0.5 0.6 Distances
E1 B6
A3 D4
C8
A6 D11
B12 E7 D9
C1
A10 D1
B3 D10
C11
B10 E11
C6 E6 D3
B1
A8
A1 B2 D7
C12 A12
E4
A7 E3
E2
C3 C7
B7 E5
C4
A5 D5
B4
Figure 14. Cluster analysis of fish assemblages. Data are from roving diver surveys conducted at 40 survey sites. Stratum is indicated by first letter of the site code.
Conch and Lobster
Queen conch were found in seven of 40 belt transect surveys conducted at Overall Bank. This equates to a frequency of occurrence of 17.5 % in surveys (all sites pooled). Six of the conch- positive survey sites fell within stratum D and one within E. Conch were not observed in belt transects in stratum A, B and C. This indicates that conch frequency of occurrence varies
substantially among strata, from 75.0 % (stratum D), to 12.5 % (stratum E), to 0 % (Stratum A, B and C). Average conch density in stratum D, as calculated from belt transect data, was 2.8 conch per 100 m2 (stdev = 4.7). For reasons discussed below, conch density for site E-5 was excluded.
Average conch density for all remaining sites pooled (n = 39 sites) was 0.6 conch per 100 m2 (stdev = 2.3), or 60 conch per hectare. Conch densities were significantly different among habitats (Kruskal-Wallis One Way Analysis of Variance, KW Test Statistic = 26.56, p = 0.000, 4 df).
Conch length-frequency distribution was determined from 97 conch collected in subsamples from Overall Bank survey sites (Figure 15). Average conch shell length was 198.7 mm (± 42.0 mm stdev, range: 89 - 265 mm). Shells of 46 individuals (47.4 %) had flared lips. Shells of the remaining 51 individuals (52.6 %) had no lip flare. For conch with flared lip shells, average lip thickness was 21.9 mm (± 8.4 mm stdev, range: 4 - 37 mm). Linear regression of lip thickness on shell length showed a slightly negative, though non-significant (r2 = 0.086), relationship suggesting that shell length may decrease with age. It was also noted that most conch specimens with thick-lipped shells also had heavy, eroded, and encrusted shells with worn or reduced spines.
0 5 10 15
<11 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 >290
Shell Length (mm)
No. Individuals
Flared Lip No Flare
Figure 15. Length-frequency distribution of queen conch, Strombus gigas, subsampled at Overall Bank survey sites. Conch were assigned to categories of
“flared lip” or “no flare” based upon shell morphology.
Remarkable conch densities were observed at site E-5. Although results from E-5 were treated as outlier data (excluded from analyses), observations from this site deserve a brief account here.
Approximately 100 conch were recorded in the 100 m2 belt transect survey at E-5. Conch were aggregated closely together into a dense “patch”, with individuals abutting against one another on all sides in some areas. The patch was estimated to contain 2,000 individuals, with localized densities as high as 10 to >15 conch per m2. The spatial extent of this patch was limited to an area of approximately 50 x 15 m or ~ 750 m2. Conch of sub-adult size predominated within the patch. Average shell length was 185.2 mm, and 75 % of individuals had shells with no lip flare.
Caribbean spiny lobster, Panulirus argus, was not observed within the belt transect surveys at any of the survey sites. Divers did note the presence of lobster in the vicinity of belt transects at a number of sites. Typically, lobster were observed in crevices, caves, or other shelters either as solitary individuals or in small groups.
Expanded Conch and Lobster Surveys
In expanded transects, conch were found at 7 of the 12 sites surveyed (58.3 %). A total of 50 conch were observed among all sites. Of these, 26% had shells with flared lip, and 74 % had no lip flare. Among the 12 sites, conch occurred at an average density of 4.2 individuals per 1000 m2 (± 7.8 stdev, range: 0 – 27) or 42 conch per hectare. Sample size was too small to make statistical comparisons among strata, however most conch were observed in stratum D (78 % of all conch). Conch were also observed in stratum E (12 %) and C (10 %). No conch were observed in A and B. A single spiny lobster (CL > 13 cm) was observed in the expanded transects.
Table 4. Fish Observed in Roving Diver Surveys
Stratum
Family/Species Common Name A B C D E Total
Ginglymostomatidae
Ginglymostoma cirratum (Bonnaterre, 1788) nurse shark - 2 1 - - 3
Muraenidae
Gymnothorax miliaris (Kaup, 1856) goldentail moray - - 2 - - 2
Gymnothorax moringa (Cuvier, 1829) spotted moray - 2 3 - 1 6
Gymnothorax vicinus (Castelnau, 1855) purplemouth moray - 1 - - - 1
Ophichthidae
Myrichthys breviceps (Richardson, 1848) sharptail eel - 1 - - - 1
Synodontidae
Synodus intermedius (Spix, 1829) sand diver - 1 - - - 1
Holocentridae
Holocentrus adscensionis (Osbeck, 1765) squirrelfish - 2 6 2 8 18
Holocentrus rufus (Walbaum, 1792) longspine squirrelfish 4 7 7 - 1 19
Myripristis jacobus Cuvier, 1829 blackbar soldierfish 2 1 - - - 3
Neoniphon marianus (Cuvier, 1829) longjaw squirrelfish 1 - - - - 1
Aulostomidae
Aulostomus maculatus Valenciennes, 1837 trumpetfish 1 1 1 - - 3
Serranidae
Alphestes afer (Bloch, 1793) mutton hamlet - - 1 - - 1
Cephalopholis cruentata (Lacepède, 1802) graysby 8 - - - - 8
Cephalopholis fulva (Linnaeus, 1758) coney 6 8 8 1 1 24
Epinephelus guttatus (Linnaeus, 1758) red hind 2 8 8 1 5 24
Hypoplectrus chlorurus (Cuvier, 1828) yellowtail hamlet 2 - - - - 2
Hypoplectrus nigricans (Poey, 1852) black hamlet 1 - - - - 1
Hypoplectrus puella (Cuvier, 1828) barred hamlet 5 1 - - - 6
Mycteroperca venenosa (Linnaeus, 1758) yellowfin grouper 2 - - - - 2
Paranthias furcifer (Valenciennes,1828) creolefish 1 - - - - 1
Serranus baldwini (Evermann and Marsh, 1900) lantern bass - - 2 5 5 12
Serranus tabacarius (Cuvier, 1829) tobaccofish 2 - - - 2 4
Serranus tigrinus (Bloch, 1790) harlequin bass 6 5 4 - 1 16
Grammatidae
Gramma loreto Poey, 1868 fairy basslet/royal gramma 1 - - - - 1
Opistognathidae
Opistognathus aurifrons (Jordan and Thompson, 1905) yellowhead jawfish 1 - - 3 1 5
Priacanthidae
Heteropriacanthus cruentatus (Lacepède, 1801) glasseye snapper 2 - - - - 2
Malacanthidae
Malacanthus plumieri (Bloch, 1786) sand tilefish 1 7 6 8 3 25
Carangidae
Caranx bartholomaei Cuvier, 1833 yellow jack 1 - - - - 1
Caranx crysos (Mitchill, 1815) blue runner 2 - - - - 2
Caranx latus Agassiz, 1831 horse-eye jack 1 - - - - 1
Caranx lugubris Poey, 1860 black jack 2 - - - - 2
Caranx ruber (Bloch, 1793) bar jack 3 5 2 1 - 11
Decapterus macarellus (Cuvier, 1833) mackerel scad 2 2 3 2 - 9
Elagatis bipinnulata (Quoy and Gaimard, 1825) rainbow runner 1 - - - - 1
Lutjanidae
Lutjanus apodus (Walbaum, 1792) schoolmaster 3 - - - - 3
Lutjanus mahogoni (Cuvier, 1828) mahogany snapper 2 - - - - 2
Ocyurus chrysurus (Bloch, 1791) yellowtail snapper 3 - 4 1 2 10
