Abstract

Harris, P. T., Heap, A. D., Anderson, T. J., and Brooke, B. 2009. Comment on: Williams et al. (2009) “Australia's deep-water reserve network: implications of false homogeneity for classifying abiotic surrogates of biodiversity”. ICES Journal of Marine Science, 66: 214–224. – ICES Journal of Marine Science, 66: 2082–2085.

Williams et al. (2009) report on new multibeam sonar bathymetry and underwater video data collected from submarine canyons and seamounts on Australia's southeast continental margin to “investigate the degree to which geomorphic features act as surrogates for benthic megafaunal biodiversity” (p. 214). The authors describe what they view as deficiencies in the design of the Marine Protected Areas (MPAs) in the southeast region of Australia, in which geomorphology information was employed as a surrogate to infer regional-scale patterns of benthic biodiversity. This comment is designed to support and underscore the importance of evaluating MPA designs and the validity of using abiotic surrogates such as geomorphology to infer biodiversity patterns, and also seeks to clarify some of the discrepancies in geomorphic terminologies and approaches used between the original study and the Williams et al. (2009) evaluation. It is our opinion that the MPA design criteria used by the Australian Government are incorrectly reported by Williams et al. (2009). In particular, we emphasize the necessity for consistent terminology and approaches when undertaking comparative analyses of geomorphic features. We show that the MPA selection criteria used by the Australian Government addressed the issues of false homogeneity described by Williams et al. (2009), but that final placement of MPAs was based on additional stakeholder considerations. Finally, we argue that although the Williams et al. (2009) study provides valuable information on biological distributions within seamounts and canyons, the hypothesis that geomorphic features (particularly seamounts and submarine canyons) are surrogates for benthic biodiversity is not tested explicitly by their study.

Inconsistent terminologies and geomorphic approaches

Original classification of geomorphic features

Williams et al. (2009) assert that in Harris et al. (2005), “no clear guidelines were used to differentiate gullies, canyon systems and mature canyons” and that estimates of seamount numbers in the southeast region suffered from “classification uncertainties” (p. 217). It appears from these comments that Williams et al. (2009) may not have clearly understood the procedures and specifications used by Harris et al. (2005; see also a synthesis of this study published by Heap and Harris, 2008) for mapping geomorphic features on Australia's margin. In those studies, geomorphic features were identified using a bathymetric grid of spatial resolution 250 m, with reference to published geological studies. Data sources for the grid included soundings digitized from nautical fair sheets, single-beam echosounder data, and multibeam sonar data. An important dataset used in the mapping of geomorphic features on the southeast margin was derived from a multibeam sonar survey carried out over a large part of the South Tasman Rise, as reported by Exon et al. (1997). A scale of 1:5 000 000 was selected for mapping geomorphic features, together with a minimum length scale of 10 km for the features to be mapped. Geomorphic features <10 km long were either aggregated into larger features (e.g. a field of sandwaves) or were excluded. Drainage maps were generated using ArcGIS and draped over the bathymetry to help identify submarine canyon axes and topographic depressions. The final map was used to differentiate large-scale (>10 km) geomorphic features such as canyons, seamounts, shelf, slope, and abyssal systems across the region (Harris et al., 2005). This scale was constrained by the cell size of the grid (i.e. 40 pixels × 250 m = 10 km) and is consistent with the smoothing carried out for contour mapping which is smoothed over 2250 m. In cases where multibeam sonar data had been collected (e.g. the area mapped by Exon et al., 1997), pinnacle features <2–3 km in size could be resolved, and these were mapped separately. Ultimately, the scale of mapping was a compromise between the resolution of the available data and the national and regional scales required for management decisions to which the dataset would be applied.

Identification and terminology of submarine canyons and seamounts

Submarine canyons

Submarine canyons vary in their size, complexity, and spatial extent (Mitchell, 2008). Several canyons were identified by Harris et al. (2005), including simple blind canyons that do not intrude onto the shelf, and complex, dendritic canyon systems, shelf-incising canyons, and those associated with terrestrial drainage. Submarine canyons smaller than ∼10 km (e.g. gullies) were either aggregated into larger canyon features or excluded. Similarly, the limbs of complex, dendritic canyon systems were grouped as parts of single, large features rather than being reported as discrete separate features. Therefore, although Williams et al. (2009, p. 217) are correct in saying that Harris et al. (2005) did not differentiate between gullies, canyon systems, and mature canyons, we recognized in our earlier work the importance of including a scale when discussing terms like valleys, canyons, and gullies because these features may exist at many different scales. Using this approach, Harris et al. (2005) mapped 131 submarine canyons in the southeast region.

Williams et al. (2009, p. 219) present new 30-kHz multibeam sonar data collected from the upper continental slope in depths of 200–1000 m and identified the presence of an additional 145 new submarine canyons. However, it is unclear whether these additional 145 canyons reflect newly discovered canyons previously unresolved in the earlier bathymetry or whether they are smaller limbs belonging to larger dendritic canyon systems identified by Harris et al. (2005). Williams et al. (2009) do not describe the methods they employed to map the 145 new canyons, and (apart from the small area shown in their Figure 4) they present no information on the location of these canyons relative to those already mapped by Harris et al. (2005). Moreover, the methods they used for identifying these new canyons are not consistent with those applied by Harris et al. (2005), nor were they mapped at a defined spatial scale. Obviously, as more data become available, new features will be found, and existing features will be resolved better. However, where data are presented in the context of a comparative study, information on geomorphic feature size, consistent spatial grid size, and location relative to the original maps are necessary to provide an accurate comparison, particularly where numbers of canyons are being compared between approaches.

Seamounts

Seamounts are defined by geologists and geomorphologists as large isolated features of volcanic origin, characteristically conical in form, >1000 m in relief above the level of the surrounding seafloor (e.g. Kennett, 1982, p. 38). Significantly, this geological definition has been used in global studies of seamount distributions (e.g. Craig and Sandwell, 1988; Kitchingman and Lai, 2004). Submarine features rising <1000 m were defined by Harris et al. (2005) as knolls, banks, or pinnacles based on their depth and shape, and they may have a range of different geological origins and compositions. Based on these definitions, Harris et al. (2005) reported 17 seamounts and 46 pinnacles in the southeast region. Williams et al. (2009) report an additional 102 seamounts, but grouped seamounts, knolls, and pinnacles together, referring to them all as seamounts. Although they present no map showing the location of these seamounts, based on the information available we suggest that none of the 102 additional features reported by Williams et al. (2009) are true (>1000 m in height) seamounts as defined in Harris et al. (2005; see also Heap and Harris, 2008), but are in fact smaller knolls or pinnacles.

Although the distinction between the term seamount and smaller features such as pinnacles is based on an arbitrary height (i.e. 1000 m) and may not reflect clear ecological differences, consistent terminology and definitions are crucial when used in a comparative approach to evaluate the methods and outcomes used during a Marine Protected Area (MPA) planning phase against those of the evaluation study. We also suggest that although the specification of >1000 m elevation in the geological definition of a seamount might be arbitrary, the actual difference in the height of a feature above the surrounding seabed is likely to have important ecological implications. For example, seamounts act as obstacles to oceanic flow, causing local upwelling and eddy formation (Rogers, 1994; Lueck and Mudge, 1997), and lower-relief features such as knolls and pinnacles are unlikely to have the same oceanographic influence. Consequently, grouping seamounts and smaller features together is an example of false homogeneity, in the context of Williams et al. (2009).

Geomorphic features as biodiversity surrogates

Williams et al. (2009, p. 214) aimed to “investigate the degree to which geomorphic features act as surrogates for benthic megafaunal biodiversity”. We do not believe they have actually tested this hypothesis because the benthic megafaunal biodiversity of different geomorphic features was not compared. Instead, Williams et al. (2009) showed heterogeneity of benthic communities within a particular geomorphic feature (e.g. submarine canyons and seamounts) associated with depth. We would like here to clarify the fact that Harris et al. (2005) never implied that within-feature types are homogeneous, just that different large-scale features (>10 km in size; canyons, seamounts, shelf, etc.) support different community types, a point which Williams et al. (2009, p.216) acknowledge (see also Stocks, 2004; Koslow, 2007).

The problem is simply one of data availability over large spatial extents. In the absence of available biological information, geomorphic features were the best available predictor of biodiversity across the southeast region. The approach used by Harris et al. (2005) therefore provided a broad-scale predictor (>10 km resolution), but covered the full spatial extent of the management region (∼1.2 million km2; Harris, 2007). Over such scales, used in the southeast MPA process, this approach is realistic and can be applied practically. Williams et al. (2009) argued that this approach may not capture finer scales of feature complexity and therefore represented false homogeneity. Using geomorphology as a surrogate for benthic communities, however, does not require that the benthic communities within each feature type are identical—simply that the communities associated with a particular geomorphic feature type (e.g. a canyon) are more similar to each other than they are to communities on different geomorphic feature types (e.g. seamounts). We accept that finer-scales of complexity within features are likely to be important predictors of biodiversity, but this does not diminish the value of between-feature geomorphology as a predictor of biodiversity patterns at regional (100s to 1000s of kilometres) spatial scales.

False homogeneity and the evaluation of MPA design criteria

It is important to identify the different scales of variation, different resolution of available data, and the scale at which management decisions need to be made. In general, there are three clear spatial scales in MPA design that are relevant at a regional scale (e.g. the southeast planning region). First is the coarse-scale of the geomorphic feature approach, for which data were available at a nominal length scale of 10 km across the southeast region (Harris et al., 2005). This scale corresponds to regional-scale features such as canyons, seamounts, shelf, slope, and abyssal plains. Second, as Williams et al. (2009) correctly point out, there is finer-scale heterogeneity between these geomorphic features (i.e. not all canyons are the same), which has implications for the representivity of an MPA network. Finally, there is also variability within individual features (e.g. depth-related patterns within a canyon). Williams et al. (2009) combine the latter two sources of variation under a single banner of false homogeneity, which confuses some of the issues. To evaluate the MPA design process for the southeast region, we believe that it is essential to distinguish all three scales with respect to the MPA design criteria.

Williams et al. (2009) recorded the occurrence of rich fauna vs. bare substratum over depths of 150–500 m (standardized into depth-bins of 50 m) within canyons and from the peak to the base (standardized into bins of 100 m) of seamounts. Using these data, they demonstrated that benthic communities vary across depth within the two geomorphic feature types and concluded that communities within a feature type should not be considered as homogeneous, but rather that the 21 geomorphic feature types proposed by Harris et al. (2005) need to be modified by “depth, size, complexity, anthropogenic impact, and other factors” (p. 222). For example, Williams et al. (2009) reported that canyons that intrude onto the shelf (i.e. <200 m) and have exposed rocky substrata along their rims are ecologically different from those that do not have these attributes. Williams et al. (2009; their Figure 3) found that the percentage of rich fauna cover from underwater video analysis peaked at water depths of 200–300 m in submarine canyons, and at 1000–1400 m on the flanks of seamounts. They also observed that this depth range (0–1500 m) is more exposed to the effects of fishing and defined a <1500 m “zone of importance” (p. 215; their Figure 1) where conservation is critical. Williams et al. (2009) conclude that changes should be made to incorporate these biodiversity patterns specifically in future MPA design specifications to improve the conservation outcome.

If finer-scale heterogeneity is to be incorporated into the MPA design, it is essential that these data are available (or at least the variables can be measured) over the entire management region. During the Southeast Australian MPA design process, data were not available to map accurately the finer-scale features at the <10-km resolution identified by Williams et al. (2009; e.g. depicting small gullies from mature canyons) over the entire region. Moreover, it is unlikely that this level of resolution will be available in other Commonwealth regions for MPA designations because of the prohibitive costs of collecting high-resolution physical data over the entire 11.38 million km2 of Australia's marine jurisdiction (excluding the Antarctic Territory). Therefore, it is unclear how fine-scale features can be incorporated into regional-scale MPA designs.

Many of the within-feature complexities reported by Williams et al. (2009) were actually addressed and incorporated into the Southeast Australian MPA design process employed by the Government. For example, depth-related community patterns (Williams et al., 2009; their Figure 3) do not need to be resolved in the mapping phase because the final approach encompassed entire geomorphic features within an MPA (Table 1, Specifications 1 and 2). The approach adopted therefore provides protection for the gamut of communities occurring at all depths within that feature. Similarly, geomorphic features were also distinguished based on human impacts (anthropogenic modifiers; Table 1, Specification 3). For example, where two otherwise equal seamounts were considered, the unmodified, unimpacted seamount is preferred for inclusion in the MPA, so favouring areas that are in a better natural state. Although within-feature variability is undoubtedly an important predictor of community structure, it is unimportant at a regional management level because the design specifications (Table 1) require that the entire feature is included within a single MPA.

Table 1.

Specifications used by Environment Australia (2003) in the design of the southeast region MPAs.

Represent all bioregions and include a sample of all (geomorphic and conservation) features. In general, seek to include a whole feature rather than a fraction of it. 
Wherever possible, include a range of habitats and linked systems across the shelf and extending down the slope—where possible to the abyssal plain and to separated continental blocks. 
Favour areas that are in a highly natural state. For example, if there are areas where human activities are known to have disturbed ecological processes, but where certain known locations are still intact, and include the intact and benthic habitats in MPAs. 
Take account of possible negative and/or positive influences (edge effects) from adjacent human uses, including interactions with other existing conservation measures. 
Select at least two canyons next to each other, and include intervening seafloor. 
Include canyon-rich to canyon-poor transition regions. 
Include entire seamounts, not only part. Where an area includes seamounts on continental block and on abyssal plain, treat these as different; represent each, with some adjacent continental block/abyssal plain. 
As a minimum, include entire cinder cones. Preferably, include at least two cinder cones. 
With consideration of other existing conservation tools, select areas known for high biodiversity or ecologically special areas, which might be identified by: 
   • having high biodiversity inferred indirectly (e.g. seabird and marine mammal feeding areas); 
   • being known significant habitat for a listed species; 
   • being known to contain a refuge for a highly (or over-) exploited species; 
   • being known to contain a nursery, breeding, or spawning site. 
10 Design simple (rather than complex) shapes and reduce fragmentation of areas within each Broad Area of Interest. This can be achieved using straight boundary lines and minimizing the perimeter-to-area ratio. 
Represent all bioregions and include a sample of all (geomorphic and conservation) features. In general, seek to include a whole feature rather than a fraction of it. 
Wherever possible, include a range of habitats and linked systems across the shelf and extending down the slope—where possible to the abyssal plain and to separated continental blocks. 
Favour areas that are in a highly natural state. For example, if there are areas where human activities are known to have disturbed ecological processes, but where certain known locations are still intact, and include the intact and benthic habitats in MPAs. 
Take account of possible negative and/or positive influences (edge effects) from adjacent human uses, including interactions with other existing conservation measures. 
Select at least two canyons next to each other, and include intervening seafloor. 
Include canyon-rich to canyon-poor transition regions. 
Include entire seamounts, not only part. Where an area includes seamounts on continental block and on abyssal plain, treat these as different; represent each, with some adjacent continental block/abyssal plain. 
As a minimum, include entire cinder cones. Preferably, include at least two cinder cones. 
With consideration of other existing conservation tools, select areas known for high biodiversity or ecologically special areas, which might be identified by: 
   • having high biodiversity inferred indirectly (e.g. seabird and marine mammal feeding areas); 
   • being known significant habitat for a listed species; 
   • being known to contain a refuge for a highly (or over-) exploited species; 
   • being known to contain a nursery, breeding, or spawning site. 
10 Design simple (rather than complex) shapes and reduce fragmentation of areas within each Broad Area of Interest. This can be achieved using straight boundary lines and minimizing the perimeter-to-area ratio. 

Williams et al. (2009) demonstrate that the MPAs did not capture important high biodiversity locations such as Horseshoe Canyon, which they viewed as having greater conservation value than other areas within the MPA network. Here, we briefly clarify the principles that were originally applied by the Government to the selection of the Southeast Australian region MPAs. First, an ecosystem-based approach was employed by a team of scientists who identified broad areas of interest that met the ten design criteria (Table 1) and followed the CAR principle (i.e. the MPAs must be Comprehensive, Adequate and Representative; ANZECC, 1999). Second, the final decision phase of selecting and delineating the MPAs was made by the Australian Government based on these design criteria in combination with stakeholder uses and needs (National Oceans Office, 2004). Many of the criticisms of MPA placement identified by Williams et al. (2009), such as the exclusion of Big Horseshoe Canyon, stem from the final decision phase, which was influenced by stakeholder, socio-economic, and political considerations (note also that the upper parts of submarine canyons were excluded in the Murray, Nelson, and Tasman Fracture MPAs, as shown in Figure 2 of Williams et al., 2009). They are not a property of the initial design phase based on known biodiversity patterns (including biodiversity patterns inferred from geomorphology), which indeed met most of these requirements.

Conclusions

The Southeast Australian MPA network was established by the Australian Government in July 2007 based on set design criteria. The study of Williams et al. (2009) is a timely one evaluating the design of MPAs to protect regional biodiversity, and it underscores the need for a multidisciplinary approach to be applied in biodiversity research. This need is also recognized by the Australian Government through its investment in multidisciplinary marine biodiversity research (http://www.marinehub.org), which includes research into biophysical variables as surrogates for patterns of biodiversity. Our conclusions are therefore listed below.

  • Consistent terminologies and approaches are crucial when undertaking comparative studies.

  • Scales of complexity need to be determined specifically with respect to the MPA design criteria. If finer-scale maps are required, they need to be available for the entire management region under consideration at the start of the design process.

  • The specifications used by the Australian Government during the design of the southeast MPAs contained guidelines aimed specifically at avoiding errors of false homogeneity attributable to depth and anthropogenic modifiers. Were the specifications adhered to strictly in every case during the design of the MPAs? Clearly not, because many additional social and economic factors also contributed to the final design.

  • Finally, we agree with Williams et al. (2009) that data availability is a major impediment to obtaining the best possible design for an MPA network. High-resolution data such as presented in their study will guide the use of these data as they become available over larger extents.

Acknowledgements

This paper was prepared with the support of the Australian Government's Commonwealth Environment Research Facilities (CERF) programme and is a contribution of the (CERF) Marine Biodiversity Hub. Thanks are due to Rachel Przelawski and Alix Post (Geoscience Australia), Nic Bax (CERF Biodiversity Hub), and two anonymous reviewers for their helpful suggestions that improved the content of this comment. This paper is published with permission of the Chief Executive Officer, Geoscience Australia.

References

ANZECC
Australian and New Zealand Environment and Conservation Council: Task Force on Marine Protected Areas. Strategic Plan of Action for the National Representative System of Marine Protected Areas: a Guide for Action by Australian Government
1999
Canberra
Environment Australia for ANZECC
pg. 
80
 
Craig
C. H.
Sandwell
D. T.
Global distribution of seamounts from Seasat profiles
Journal of Geophysical Research
 , 
1988
, vol. 
43
 
B9
(pg. 
10408
-
10420
)
Environment Australia
Australia's South-east Marine Region: a User's Guide to Identifying Candidate Areas for a Regional Representative System of Marine Protected Areas
2003
Canberra
Department of Environment
pg. 
92
  
Exon
N. F.
Moore
A. M. G.
Hill
P. J.
Geological framework of the South Tasman Rise, south of Tasmania, and its sedimentary basins
Australian Journal of Earth Science
 , 
1997
, vol. 
44
 (pg. 
561
-
577
)
Harris
P. T.
Todd
B. J.
Greene
G.
Applications of geophysical information to the design of a representative system of marine protected areas in southeastern Australia
Mapping the Seafloor for Habitat Characterisation
 , 
2007
St Johns, NF, Canada
Geological Association of Canada Special Paper 47
(pg. 
449
-
468
)
Harris
P. T.
Heap
A. D.
Passlow
V.
Sbaffi
L.
Fellows
M.
Porter-Smith
R.
Buchanan
C.
, et al.  . 
Geomorphic Features of the Continental Margin of Australia: Report to the National Oceans Office on the Production of a Consistent, High-quality Bathymetric Data Grid and Definition and Description of Geomorphic Units for Part of Australia's Marine Jurisdiction
2005
Canberra
Geoscience Australia
pg. 
147
 
Heap
A.
Harris
P. T.
Geomorphology of the Australian margin and adjacent sea floor
Australian Journal of Earth Science
 , 
2008
, vol. 
55
 (pg. 
555
-
584
)
Kennett
J.
Marine Geology
1982
Englewood Cliffs, NJ
Prentice-Hall
pg. 
813
 
Kitchingman
A.
Lai
S.
Morato
T.
Pauly
D.
Inferences on Potential Seamount Locations from Mid-Resolution Bathymetric Data
FCRR Seamounts: Biodiversity and Fisheries
 , 
2004
Vancouver, BC
Fisheries Centre Research Reports, University of British Columbia
(pg. 
7
-
12
)
Koslow
J. A.
The Silent Deep: the Discovery, Ecology and Conservation of the Deep Sea
2007
Sydney, Australia
University of New South Wales Press
pg. 
270
 
Lueck
R. G.
Mudge
T. D.
Topographically induced mixing around a shallow seamount
Science
 , 
1997
, vol. 
276
 (pg. 
1831
-
1833
)
Mitchell
N. C.
Summary of progress in geomorphologic modelling of continental slope canyons
Geological Society Special Publication
 , 
2008
, vol. 
296
 (pg. 
183
-
194
)
National Oceans Office
South-east Regional Marine Plan: Implementing Australia's Oceans Policy in the South-east Marine Region
2004
Hobart, Tasmania
National Oceans Office
pg. 
111
 
Rogers
A. D.
The biology of seamounts
Advances in Marine Biology
 , 
1994
, vol. 
30
 (pg. 
305
-
350
)
Stocks
K.
Morato
T.
Pauly
D.
Seamount invertebrates: composition and vulnerability to fishing
Seamounts: Biodiversity and Fisheries
 , 
2004
University of British Columbia Fisheries Centre Research Report, 12
(pg. 
17
-
24
78
Williams
A.
Bax
N. J.
Kloser
R. J.
Althaus
F.
Barker
B.
Keith
G.
Australia's deep-water reserve network: implications of false homogeneity for classifying abiotic surrogates of biodiversity
ICES Journal of Marine Science
 , 
2009
, vol. 
66
 (pg. 
214
-
224
)