Report of Phase 1 of the Great Barrier Reef Seabed Biodiversity Mapping Project

Pitcher R (CSIRO), Venables B (CSIRO), Pantus F (CSIRO), Ellis N (CSIRO), McLeod I (CSIRO), Austin M (CSIRO), Gribble N (DPI), Doherty P (AIMS).

May 2002

The Great Barrier Reef World Heritage Area (GBRWHA) includes the largest multiple-use marine park on the planet – the Great Barrier Reef Marine Park (GBRMP). Under recent Commonwealth legislation (EPBC Act, 1999), major fisheries in the Park (prawn and line) must be assessed and demonstrated to be sustainable. For trawling, the issue is bycatch, despite current requirements to trawl only with bycatch excluding devices. For line fishing, the issue is stock size with major uncertainty about how fish use inter-reefal habitats. Independent of fishing, the managing agency (GBRMPA) is rezoning the marine park and, increasing the amount of no-take areas, through its Representative Areas Program. These initiatives are just some of the reasons why managers might wish to know more about seafloor habitats and biodiversity with the GBRMP.

In 2000, CRC Reef endorsed a proposal to sample the GBRMP seabed in support of sustainable industries and the management of biodiversity. Stakeholder consultation over this proposal raised several issues; primarily about the type of sampling gear to be used in different zones of the Park but also about the adequacy of existing data. In response, the Board of CRC Reef requested a preliminary report into these and other sampling issues.

This report represents contributions from three research providers. QDPI staff provided access to their historical survey data, including videos and biological samples. CSIRO staff from two Divisions reanalysed some of this material and combined it with their own historical data into the effects of trawling. The latter included contemporaneous sampling with a full suite of possible techniques (trawl, dredge, video, acoustics), allowing direct tests of the power of alternative sampling strategies. AIMS staff compared the performance of baited videos with trawls.

Two key results have emerged from this analysis.

First, the comparison of gear types revealed huge differences in performance. For example, a small dredge collected around 1200 taxa (plants and invertebrates) of which more than 500 were also collected by prawn trawling on the same bottom; indicating the potential for impact through incidental catch. Simultaneous deployment of a towed video over the same ground captured just 5% of the biodiversity collected by the dredge, and just 7% of the biodiversity collected by the trawl; indicating the insensitivity of tests based on video. While videos performed better for fish, the low overlap between them and trawls shows that they are complementary rather than alternative sampling devices.

Second, the main reason for the poor performance of video tools is the poor resolution of taxonomic identity. For example, when species were resolved at the level of Class, two thirds of the spatial information was lost compared with patterns established from species level data. One clear reason for this rapid erosion of information content is that related species often have complementary distributions (e.g. an inshore species associated with mud and an offshore relative associated with carbonate sand). When resolved as separate species, the habitat preferences are clear and can be used to predict occurrence elsewhere. When pooled, however, the biophysical relationships are obscured and predictive power is lost.

Most of the information collected by the QDPI surveys was based upon presence and absence of high taxonomic units; mostly identified at Class level. A reduced and weighted set of biophysical variables was used to predict spatial pattern in assemblages across the whole GBRMP based on the QDPI data. Cross-validation, however, revealed that these predictions were highly unreliable. This result was not unexpected (see above) but it does show that the historical surveys do not contain enough information to manage biodiversity within the GBRWHA.

Further work on biophysical relationships from the CSIRO data (resolved at species level) showed that only half of all species have strong relationships useful for prediction based upon surrogates. While generalists may show weak and inconsistent relationships, other reasons for such failure including measuring the wrong proxy, and low or patchy abundance. This is compounded by problems of autocorrelation, where knowledge of the local state is only effective at predicting assemblages within relatively short distances, and uncertainty in many of the environmental descriptors. For about half of the biodiversity, there is no adequate substitute for direct sampling.

The report concludes with a cost-benefit analysis of alternative sampling strategies compared against an unconstrained one. While all parties are agreed that there will be no extraction from highly protected zones (yellow, green & pink), the cost-benefit analysis makes it clear that this would be a small penalty because of the size and distribution of these areas. Conversely, an inability to deploy dredge and/or trawl within General Use B (GUB – dark blue) would seriously undermine the risk assessments required by the fishing industry, because this zone represents a large proportion of the total shelf space.

In summary, the major lessons from this report are:

(a) it is highly desirable to identify species, which requires collecting,
(b) without collections, there can be no estimate of biomass,
(c) without biomass, there can be no quantitative risk assessment of trawl,
(d) without access to GUB, there can be no formal analysis of sustainability.

The analytical and authorship team is to be congratulated for its rigorous approach, which has provided clear answers to these important questions, and for the production of an excellent report.

Peter Doherty
November 2002

Title page, table of contents and summary (Adobe Acrobat file - 269kB)

1. Main aims (Section 1-3, Adobe Acrobat file - 113kB)

2. Major clients and beneficiaries of the resulting information

3. Milestones and deliverables of Phase 1.

4. Information from data already available (Adobe Acrobat file - 123kB)

5. Issues to be considered in future sampling (Adobe Acrobat file -113kB)

6. Evaluation of alternative sampling designs (Adobe Acrobat file - 177kB)

7. Options summary and recommendations (Adobe Acrobat file - 712kB)

Appendix A-1. Information avalable from collation of available data (Adobe Acrobat file - 2.79MB)

Appendix A-2. Analysis of information availabke from QDPI DWS surveys (Adobe Acrobat file - 438kB)

Appendix A-3. Analyses of data available from the Effects of Trawling study (Adobe Acrobat file - 12.4MB)

Appendix B-1. Sample design guidance provided fromthe Effects of Trawling study

Appendix B-2. Comparisons of fish information content of BRUVS and trawl (Appendix B-1 and B-2, Adobe Acrobat file - 1.08MB)

Appendix C. Strategy for stratification and sampling design (Adobe Acrobat file - 536kB)

Appendix D. Information about the proposed sampling and devices (Appendix D and E, Adobe Acrobat file - 106kB)

Appendix E. Outcomes and outputs available from an optimal sampling strategy.