Abstract
Anthropogenic global ocean warming is predicted to cause bleaching of many near-sea-surface (NSS) coral reefs, placing increased importance on deeper reef habitats to maintain coral reef biodiversity and ecosystem function. However, the location and spatial extent of many deep reef habitats is poorly known. The question arises: how common are deep reef habitats in comparison with NSS reefs? We used a dataset from the Great Barrier Reef (GBR) to show that only about 39% of available seabed on submerged banks is capped by NSS coral reefs (16 110 km2); the other 61% of bank area (25 600 km2) is submerged at a mean depth of around 27 m and represents potential deep reef habitat that is spatially distributed along the GBR continental shelf in the same latitudinal distribution as NSS reefs. Out of 25 600 km2 of submerged bank area, predictive habitat modelling indicates that more than half (around 14 000 km2) is suitable habitat for coral communities.Harris, P. T., Bridge, T. C. L., Beaman, R. J., Webster, J. M., Nichol, S. L., and Brooke, B. P. 2013. Submerged banks in the Great Barrier Reef, Australia, greatly increase available coral reef habitat. – ICES Journal of Marine Science, 70: 284–293.
Introduction
There has been a well-documented decline in coral reef ecosystems due to natural and anthropogenic causes (Hoegh-Guldberg, 1999; Hughes et al., 2003; Hoegh-Guldberg et al., 2007). Globally, about 19% of coral reefs have already been lost, with a further 35% expected to be lost in the next 40 years (Wilkinson, 2008). Even those reefs not currently under threat are predicted to be affected by climate change impacts, particularly bleaching events due to elevated sea temperature. However, these estimates are largely based on the known extent of reef habitat, which generally only includes near-sea-surface (NSS) coral reefs. Submerged reefs that occur below a depth of around 20 m cannot easily be detected using satellites or aerial photography, even in clear waters. Consequently, their spatial distribution and even their existence are unknown in most reef provinces. For this reason, submerged reefs have been largely neglected in estimates of the available area of coral habitat, despite recent evidence that these areas may be significant (Locker et al., 2010). Understanding the extent of submerged reefs is therefore important because they can support large and diverse coral communities (Bridge et al., 2012) and hence may provide vital refugia for corals and associated species from a range of environmental disturbances (Glynn, 1996; Riegl and Piller, 2003; Bongaertsetal., 2010).
Scientific drilling has demonstrated that most NSS reefs have (relict) limestone foundations, forming a “pedestal” or geomorphic bank upon which modern reef growth occurs (Marshall and Davies, 1984; Hopley et al., 2007). However, not all banks are colonized by NSS coral reefs, and in many cases reef colonies that were established in the late-Pleistocene to early-Holocene were unable to keep pace with post-glacial sea level rise and were subsequently “drowned” (Vecsei, 2003; Abbey and Webster, 2011). Even though their slow vertical growth rate has inhibited them from reaching the sea surface, these submerged reefs do provide habitat for a range of coral reef species, including reef-building corals (Bridge et al., 2011a, b). Now commonly referred to as mesophotic coral ecosystems (MCEs), they are typically found at depths ranging from 30 to 150 m in tropical and subtropical regions, depending on water quality and light penetration (Hinderstein et al., 2010, Kahng et al., 2010). Their potential importance as coral refugia has generated significant interest from both scientists and managers in recent years because observations indicate that deep (>30 m) coral habitats can support diverse coral communities (Puglise et al., 2009; Hinderstein et al., 2010; Bridge and Guinotte, 2012). MCEs have been well documented in the western Atlantic (Armstrong et al., 2006; Garcia-Sais, 2010; Smith et al., 2010; Armstrong and Singh, 2012) and have been reported from the Pacific and Indian Oceans (Kahng and Kelley, 2007; Banks et al., 2008; Bare et al., 2010; Bongaerts et al., 2011; Bridge et al., 2011a, b). In Australia, submerged reefs and associated coral communities have been described in the Gulf of Carpentaria (Harris et al., 2004, 2008), adjacent to Lord Howe Island (Woodroffe et al., 2010), in the Timor Sea off Australia's North West Shelf (Heyward et al., 1997), adjacent to the Ningaloo Reef in Western Australia (Nichol and Brooke, 2011), in the western Coral Sea (Bongaerts et al., 2011), and across the Great Barrier Reef (GBR) continental shelf (Harris and Davies, 1989; Beaman et al., 2008; Bridge et al., 2011a, b).
NSS coral reefs are exposed to a range of natural and anthropogenic threats, particularly the increasing incidence and severity of coral bleaching due to anthropogenic warming of sea temperatures and high light irradiance (Glynn, 1996; Hoegh-Guldberg, 1999; Hughes et al., 2003). In comparison, submerged reefs may be buffered from many of these bleaching events due to the greater depth of the overlying water column and reduced light irradiance (Glynn, 1996; Riegl and Piller, 2003). Importantly, NSS coral reef recovery may be assisted by seed stock supplied from nearby submerged reefs (Harris et al., 2008; Bongaerts et al., 2010, van Oppen et al., 2011). Global ocean warming, combined with a range of other threats (Pandlofi et al., 2003), could potentially result in submerged reefs functioning as refugia for corals and associated species. Therefore, their conservation may be crucial for the persistence of corals and associated species under future climate change impacts and other local stressors.
The Great Barrier Reef (GBR) extends over 15 degrees of latitude and comprises a wide variety of reef systems and different morphotypes (Hopley et al., 2007). The GBR therefore provides an ideal case study to illustrate that the lack of information on deep reef habitats has likely caused a substantial underestimation of the available coral reef habitat in this region. Such a case study, by extension, can be used to highlight an underestimation of the global spatial extent of coral reef habitat. Given that the NSS coral reefs in the GBR are among the best-quantified in the entire Indo-Pacific, we suggest that a significant proportion of global coral reef habitat is currently undocumented, limiting its environmental management as well as our understanding of coral reef biodiversity and also important ecosystem processes, such as connectivity and the effects of climate change on coral reef ecosystems.
Here, we have used a recently developed high-resolution digital elevation model (DEM) for the Great Barrier Reef and Coral Sea (Beaman, 2010) to re-evaluate the geomorphology of the GBR shelf, and address the question of how much surface area of geomorphic banks (potential submerged reef habitat) exists compared to the known area of NSS coral reefs in the GBR. The potential for submerged bank areas to be colonized by corals is assessed based on a test area from the central GBR outer-shelf, where high-resolution (5 m grid) bathymetric data together with observations of mesophotic coral communities are available.
Methods
Mapping geomorphic banks
The geomorphic mapping in this study is based on a GIS analysis of a 100 m bathymetry grid produced by Beaman (2010). The bathymetric data were contoured at 5 m intervals and used to interpret the location of geomorphic bank features, defined as having at least one steep (i.e. greater than ∼2 degrees) slope rising more than 15 m above the level of surrounding seafloor. A “geomorphic bank” is an underwater feature defined by the International Hydrographic Organisation (IHO, 2008) as “isolated (or group of) elevation(s) of the sea floor, over which the depth of water is relatively shallow, but sufficient for safe surface navigation”. Due to their complex, asymmetric morphology (exhibiting both elongate and oval shapes in plan view and both steep and gentle slopes), all banks were digitized by hand (aided using 3D imagery). Bank polygons were created in ArcGIS with the base of slope taken as the outer edge of the bank. Mean bank elevation estimates thus include the bank slopes as well as planar bank-tops. Only banks occurring on the continental shelf of the Great Barrier Reef between the 20 and 200 m isobaths, and between the latitudes of 10 to 25°S were included. GBR_FEATURES.shp, an ArcGIS layer for NSS coral reefs published by the Great Barrier Reef Marine Park Authority (GBRMPA), was used. Land (island) areas were removed, as required, to generate total submerged bank areas. Statistical analyses of depths and surface areas of banks were carried out using ArcGIS. Great Barrier Reef Marine Park zones were downloaded from the GBRMPA web site (http://www.gbrmpa.gov.au/) and used to derive bank areas within the different zones.
Predictive habitat models
In order to estimate the likely occurrence of coral habitat on submerged banks in the GBR, we used Maxent 3.2.19 (Phillips et al., 2004, 2006; Elith et al., 2011) to generate predictive habitat models for the Hydrographers Passage region. Maxent uses maximum entropy techniques to create models of the relative probability of species/community distribution across a study area. It has the advantage of requiring presence-only data, which is beneficial for modelling inaccessible ecosystems, such as deep reefs where occurrence data are sparse, and the lack of reliable absence data renders them unsuitable for traditional modelling methods. Maxent has been used in both terrestrial and marine ecosystems and has been shown to perform favourably relative to other presence-only modelling techniques (Pearson et al., 2007; Elith et al., 2011).
The Hydrographers Passage (19.70°S 150.25°E) site was chosen because its geomorphology and biology have been comparatively well-documented (Harris and Davies, 1989; Beaman et al., 2008; Bridge et al., 2011a; Beaman et al., 2012), and because it is one of very few areas in the GBR where co-located, high-resolution geophysical and ecological data are available. Areas likely to support deep-water coral communities were identified using coral occurrence records derived from optical images taken by autonomous underwater vehicle (AUV) (Williams et al., 2010), and geophysical data on depth, slope, aspect, rugosity, sidescan acoustic backscatter (a surrogate for substratum roughness and type), and geomorphic zone (slope, crest, flat or depression) gridded at 5 ×5 m pixel resolution.
In the model, 70% of occurrence records (n = 100) were used as a training data set, and the remaining 30% used to test model results. The 70/30 split of occurrence records was done randomly using an option available in the Maxent program. The performance of both training and test datasets was evaluated using receiver operating characteristic curves, with the area under the curve (AUC) being a measure of model performance. AUC is a threshold-independent measure of model performance ranging from 0–1. An AUC value of 0.5 represents a model that performs no better than random, whilst 1 is maximally predictive. AUC values in this study were very high for both training (0.984) and test (0.977) data sets, indicating good model performance. Maps of relative habitat suitability were transformed into Boolean maps using a lowest presence threshold technique (Pearson et al., 2007) to identify areas with a high probability of containing coral communities. This approach identifies the lowest probability value associated with an occurrence record, and considers all pixels with equal or higher probability values as being suitable habitat (Pearson et al., 2007). The lowest presence threshold technique therefore provides a conservative estimate of suitable habitat, identifying the minimum predicted area possible whilst maintaining zero omission error in the dataset (Pearson et al., 2007). In this study, the lowest suitability value for any occurrence record was 0.14. Therefore, this value was used as the lowest presence threshold.
Results
Geomorphic banks and NSS coral reefs
A total of 1581 geomorphic bank features were mapped in the GBR (Figure 1), having a mean depth of 27.1 ± 15.8 m and a total surface area of 41709 km2 in the region mapped (Table 1). Within the same study area, NSS coral reefs have been mapped by the GBRMPA using satellite and aerial imagery. Based on our analysis, NSS reefs have a mean depth of 14.9 m and a negatively skewed depth distribution (median depth of 12.0 m). NSS reefs cover an area of 20679 km2, of which 16110 km2 (78%) is located on top of submerged banks mapped in this study. Therefore, 61% of geomorphic bank habitat (25599 km2) is not covered in NSS coral reefs.
Number | Mean depth (m) | Mean depth SD | Mean height (m) | Mean area km2 | Total area km2 | Mean P/A ratio | |
---|---|---|---|---|---|---|---|
All banks incl. NSS reefs | 1 581 | 27.29 | 15.69 | 41.67 | 26.38 | 41 709 | 3.10 |
All NSS reefs | 3 457 | 14.9 | 15.4 | N/A | 5.98 | 20 679 | 8.75 |
Type 1 banks | 1 145 | 26.79 | 11.26 | 43.62 | 20.92 | 23 827 | 2.29 |
Type 2 banks | 251 | 27.23 | 6.29 | 26.29 | 1.98 | 497 | 7.06 |
Type 3 banks | 150 | 58.57 | 16.63 | 36.30 | 8.50 | 1 275 | 2.84 |
All banks with NSS reef area subtracted | 1 546 | 29.94 | 14.66 | 40.09 | 16.56 | 25 599 | 3.10 |
Number | Mean depth (m) | Mean depth SD | Mean height (m) | Mean area km2 | Total area km2 | Mean P/A ratio | |
---|---|---|---|---|---|---|---|
All banks incl. NSS reefs | 1 581 | 27.29 | 15.69 | 41.67 | 26.38 | 41 709 | 3.10 |
All NSS reefs | 3 457 | 14.9 | 15.4 | N/A | 5.98 | 20 679 | 8.75 |
Type 1 banks | 1 145 | 26.79 | 11.26 | 43.62 | 20.92 | 23 827 | 2.29 |
Type 2 banks | 251 | 27.23 | 6.29 | 26.29 | 1.98 | 497 | 7.06 |
Type 3 banks | 150 | 58.57 | 16.63 | 36.30 | 8.50 | 1 275 | 2.84 |
All banks with NSS reef area subtracted | 1 546 | 29.94 | 14.66 | 40.09 | 16.56 | 25 599 | 3.10 |
Banks with reef area subtracted deletes 35 banks that are 100% covered by NSS coral reefs. Types 1, 2 and 3 banks are as defined in the text. The fraction of bank area covered by NSS reef averages 40.2 ± 27.6% over the 1180 banks that support NSS reefs. NSS reef depths estimated using reef polygons available from the GBRMPA, overlain on Beaman's (2010) bathymetric model. Sources of error in the surface area estimates are associated with the pixel size of the bathymetric grid (0.01 km2) plus any (unquantified) human errors associated with digitizing the NSS reefs and banks.
Spatial relationship between banks and NSS reefs
Histograms illustrate that banks and NSS coral reefs exhibit a similar spatial distribution along the length of the GBR. Both banks and NSS reefs are most abundant in the northern GBR between 10° and 12°S, their occurrence reaches a low between 17° and 18°S before increasing again between 20° and 23°S. As noted by Hopley et al., (2007), there is a trend for reef occurrence and mean depth to be greatest in the northern GBR compared with the southern GBR; here we document similar trends for geomorphic banks on the GBR shelf (Figure 1). The relevance to the present study is that MCEs associated with geomorphic banks are spatially concomitant with NSS reefs and are thus proximal to supply seed stock for their re-colonization.
Three types of banks
Out of the 1581 banks mapped, 1180 of them (74.6%) have some portion covered by NSS coral reefs, which are referred to here as Type 1 banks (Table 1; Figure 2a and b). Almost all Type 1 banks are only partially covered by NSS reefs (n = 1,145); a mere 35 banks are completely (100%) covered by NSS coral reefs. The mean depths of the 1145 banks partially covered by NSS coral reefs exhibit a positively skewed, unimodal distribution (Figure 3a). The 401 banks that do not support NSS coral reefs, however, exhibit a bimodal distribution with modal peaks at mean depths of approximately 27 m and 56 m (Figure 3a). The mean and modal depths of banks, combined with co-occurrence or non-occurrence of associated NSS coral reefs, suggests three different types of bank (Figure 3a). Type 2 banks have a mean water depth of 27 m, a similar mean to Type 1 banks, but they are an order of magnitude smaller in surface area and more irregular in shape (larger perimeter/area (P/A) ratio) than the Type 1 banks (Figure 4). Type 3 banks have a mean depth of 56 m but are otherwise geomorphologically similar (similar P/A ratio) to Type 1 banks (Figures 3a and 4). Type 2 banks are common in the northern GBR and are rare in the south, whilst Type 3 banks have the opposite spatial distribution.
Coral cover of submerged banks
Data collected from the test study site of 520 km2 located at Hydrographer's Passage (Figures 1 and 5) were used in this study to investigate coral coverage of submerged banks. Observations of depth, substrate characteristics and occurrence of coral communities were used. In the test site there are nine geomorphic banks covering an area of 71.54 km2 and having a mean coral coverage estimated at 55.4 ± 22.7%.
Discussion
The potential area of MCEs in the Great Barrier Reef is suggested by the area of submerged geomorphic banks not supporting NSS coral reefs, which is equal to an area of 25599 km2. But how much of the 25599 km2 of submerged bank area actually supports living coral communities?
Only a few case studies of MCEs have been published for the GBR (Bridge et al., 2011a, b; 2012). As noted above, 74.6% of banks (1180 out of 1581) support some area of NSS coral reefs. This large proportion of banks colonized by NSS coral reefs is itself compelling evidence that most banks support at least some corals and must therefore be considered as potential coral habitat.
Our results from Hydrographer's Passage indicate the presence of coral communities on Type 1 and 3 banks, covering about 55% ± 23% of bank surface area (Figure 5). Extrapolating this figure to geomorphic banks of the GBR, we estimate that around 14000 ± 6000 km2 of bank area supports mesophotic coral communities. When compared with the surface area of NSS coral reefs (20679 km2) it is apparent that the total area of coral habitat in the GBR is much greater than previously thought. Furthermore, the 20679 km2 of NSS reef surface area includes reefs with sandy lagoons and other habitat types that are not ideal for coral colonization (Roelfsema et al., 2002). Overall, it is evident that the surface area of preferred coral habitat in the GBR is at least ∼50% larger, and may be as much as double the size previously believed to exist.
Coral species diversity versus water depth
If MCEs located on submerged banks are to provide refugia for NSS reefs then the question of coral species occurrence versus depth must be addressed; if communities found in NSS reefs comprise species not found at depth, then deeper MCEs will not provide a source of seed stock to recolonize them. Published research on depth ranges of coral species is sparse and there are few published syntheses. Observations of submerged reefs and banks using both SCUBA and remote imaging methods, such as AUVs, reveal that zooxanthellate coral communities are common features of submerged banks on the GBR outer-shelf to depths of at least 65 m (Bridge et al., 2011a, b; 2012); Bridge et al., (2011a, b) report rich communities of phototrophic megabenthos on even the deepest (Type 3) banks.
The available literature suggests that most coral species are “depth-generalists”, occurring in both shallow and deep water, with most species occurring from very shallow to at least 40 m depth (Carpenter et al., 2008). Furthermore, coral diversity peaks in intermediate depths of 15 to 30 m (Burns, 1985; Huston, 1985; Cornell and Karlson, 2000) and reef fish diversity also peaks at around 30 to 35 m (Cappo et al., 2007, Brockovich et al., 2008). These depth ranges approximately coincide with the mean depth of submerged bank habitat measured in the present study (Figure 3a; Table 1). Taken together, these observations are consistent with the conclusion of Harriott and Banks (2002) that a significant factor controlling coral occurrence is the availability of hard substrate, in this case the antecedent reef habitat provided by submerged banks.
Based on IUCN Red List data presented by Carpenter et al., (2008), 39% of the maximum depths of occurrence of corals is at mesophotic depths of 30 m or greater. These maximum depth values are most likely conservative, because recent studies of mesophotic reefs in the GBR (Bridge et al., 2012, in press) show that many coral species are found to occur at greater depths than those reported by Carpenter et al., (2008). The frequency distribution of maximum depths of occurrence of corals has a mean value of 27.4 ± 17.0 m (n = 675; Figure 3b), a value comparable to the mean depths of Type 1 and 2 banks measured in this study (26.8 and 27.2 m, respectively; Table 1). Therefore, 50% of Type 1 and Type 2 banks in the GBR are in water depths suitable to at least 50% of coral species.
Regional and global significance
Studies from locations such as the Gulf of Carpentaria in northern Australia, where NSS platform and patch coral reefs are absent, indicate that the Gulf contains submerged platform and patch reefs that support MCEs that were only revealed through multibeam sonar mapping combined with towed video and sampling (Harris et al., 2004, 2008). In tropical northern Australia west of Torres Strait, geomorphic banks on the continental shelf are estimated to cover 44290 km2 (Heap and Harris, 2008), much of which is potentially submerged coral reef habitat. Diverse coral communities have been reported on MCEs in other regions (e.g. Bare et al., 2010), and similar submerged banks and reefs throughout the Indo-Pacific are likely to contain coral communities comparable to those in the GBR. For example, in his global analysis of the mean depths of carbonate platforms, Vecsei (2003) noted the modal depth range for all atolls is 20–30 m. Based on the available evidence, it seems likely that submerged reefs and associated MCEs are widespread throughout all of the world's major coral reef provinces and spatially they extend well beyond the known ranges of NSS reefs.
Implications for management
A major rezoning of the Great Barrier Reef World Heritage Area in 2004 resolved to designate a minimum of 20% of each habitat type (referred to as “bioregions”) as no-take protected areas (Fernandes et al., 2005). Although the lack of data on submerged reef habitats meant they were not considered when determining bioregions, approximately 30% of banks currently occur within designated no-take areas and 87.7% of banks that occur within the GBR marine park are protected from bottom trawling (Table 2). This result is consistent with our conclusion (above) that submerged banks and NSS coral reefs exhibit the same latitudinal distribution within the GBR (since NSS reefs were targeted by the GBRMPA for protection). It also suggests that the GBRMPA's approach for design of conservation zones allows for uncertainty and protects future unknown habitats. However, further research is required to determine whether the 30% of submerged reefs and banks that have been protected are those supporting significant areas of MCE habitat, or if they have the greatest chance of surviving the impacts of global ocean warming (or other anthropogenic pressures).
ZONE TYPE | Area (km2) of banks included | Number of banks includeda | Percent of banks by area |
---|---|---|---|
Preservation Zone | 190 | 35 | 0.7 |
Marine National Park Zone | 7 301 | 602 | 28.5 |
Conservation Park Zone | 654 | 42 | 2.6 |
Habitat Protection Zone | 12 983 | 889 | 50.7 |
General Use Zone | 3 157 | 370 | 12.3 |
Banks beyond GBR Marine Park | 1 315 | 166 | 5.1 |
ZONE TYPE | Area (km2) of banks included | Number of banks includeda | Percent of banks by area |
---|---|---|---|
Preservation Zone | 190 | 35 | 0.7 |
Marine National Park Zone | 7 301 | 602 | 28.5 |
Conservation Park Zone | 654 | 42 | 2.6 |
Habitat Protection Zone | 12 983 | 889 | 50.7 |
General Use Zone | 3 157 | 370 | 12.3 |
Banks beyond GBR Marine Park | 1 315 | 166 | 5.1 |
Preservation and Marine National Park and Conservation Park zones provide a “no take” level of protection. Bottom trawling is prohibited in all except the General Use Zone. The study area covers the entire GBR Marine Park, allowing for an assessment of the protection of bank habitat that exists within the park zoning plans. The table shows that although only 0.7 percent of the banks are within “Preservation Zones” that offer the highest level of protection, approximately 87.7% of banks within the GBR marine park (and 82.6% of banks mapped in this study) are protected from bottom trawling. aParts of banks may occur in more than one zone.
Our results show that the spatial extent of coral reef habitat in the GBR Marine Park may be underestimated by as much as 100%, despite the GBR being one of the best-studied coral reef ecosystems on earth. Other reef areas that have received less research effort may also contain a substantially greater amount of reef habitat than is currently assumed. Apart from some immunity from coral bleaching, the water depths of mesophotic coral reefs may protect them from a range of other threats, such as severe tropical storms, the frequency and intensity of which is predicted to increase with climate change (Emanuel, 2005). In the Coral Triangle, the global epicentre of coral reef biodiversity, many reefs have suffered from destructive fishing methods such as dynamite and cyanide fishing. Due to both inaccessibility and poor knowledge of their location, submerged coral reefs (MCEs) seem likely to have escaped many of these pressures. It is therefore critical to identify and effectively manage submerged reef habitats not only on the GBR, but around the world.
Funding
This work was produced with the support of funding from the Australian Government's National Environmental Research Program (NERP) and is a contribution of the NERP Marine Biodiversity Hub. RB acknowledges a Queensland Smart Futures Fellowship for salary support.
Acknowledgements
GBR_Features dataset courtesy of the Great Barrier Reef Marine Park Authority. Thanks to Lachlan Hatch (Geoscience Australia) for assistance with GIS analysis and statistics. PTH, SLN and BPB publish with the permission of the Chief Executive Officer, Geoscience Australia.