16. Discussion

16.1 Limitations and Caveats

This is an Interim bioregionalisation. Whilst it represents a major revision of the CONCOM product, it is the first version of a detailed examination of a portion of the existing marine data resource. The RAP methodology was implemented by concentrating the available search effort to collate fish distribution data from literature records. Much of the verifiable records of institutional collections or even CSIRO Division of Fisheries' extensive collection data remain to be examined.

As with most large-scale biological surveys (Prendergast et al., 1993b), the fish distribution data used for this project contains examples of sampling bias, possibly confounded by collecting effects. Prendergast et al. (1993b) showed that sampling bias could result in the appearance of invalid hotspots, and non-appearance of hotspots where they were warranted. They also found that species richness was a function of sampling levels. We have gone to considerable lengths, including the BioTax '96 Workshop, to vet and refine the high priority species distributions to minimise such effects.

There is a limited amount of information available on off-shelf fish species, and given the time restraints this has not been examined in any detail. While it is thought that those ecosystems proximate to human influence - estuaries, coastal areas, and to a lesser extent shelf ecosystems - are most in need of rapid understanding and protection (Upton, 1992; Ehler and Basta, 1993; Norse, 1995), it is also clear that our ignorance of off-shelf ecosystems is so great that we are incapable of assessing the magnitude of any impacts that human activities may be having.

The general lack of data which handicaps studies of terrestrial and coastal marine biodiversity is crippling in deeper environments. There is currently so little useful and/or reliable biological information regarding continental slope, meso-bathypelagic, and abyssal environments that an attempt at bioregionalisation would be nonsensical. For example, Poore et al. (1994) estimate that just 10% of Australia's slope fauna is described. Studies that have been attempted tend to indicate that there are few detectable patterns of regionalisation of slope fauna, and those which have been shown are related to depth rather than any biogeographical influence (Poore et al., 1994). Lower species diversity is found at greater depths, though this depth gradient appears to be not as severe in southern Australia (Poore et al., 1994). This absence of even baseline biological data for off-shelf environments must be addressed with some urgency.

Finally, we stress that what we have produced with the fish regionalisation is at a provincial scale and so is not directly comparable to the IMCRA regionalisation produced by the States and Territories (MCA, 1995).

In the oceanographic regionalisation, we limited the number of attributes, and used 4 seasonal layers. Time constraints also limited the number of combinations of attributes that could be examined. The seasonal variability in the deep layer may also simply reflect sampling variability rather than a true seasonal signal. Finally, much work has been done with satellite data but here again with the limited timeframe and resources of this project these could not be examined in any detail. The oceanographic regionalisation should be viewed as providing the groups characterising a part of the habitat of the offshore regions. It is not directly comparable to the biological regionalisation either in concept or in area. We do not support any forced matching of the two regionalisation merely from a pattern analysis perspective, and indeed, we warn against misrepresentation of our results by such procedures. Any integration must be soundly based in concept and in the interpretation of the results.

16.2. Summary Overview

The analysis and results of the present bioregionalisation are conceptually very different and unique by comparison to previous schemes. The salient structural features of the present bioregions and their interpretation. are:

1. For the biologically based regionalisation using fish distributions, a provincial scale bioregionalisation was derived for the pelagic and demersal systems separately. The pelagic bioregionalisation comprises 4 bioregions (two provinces and two zootones) of much more extensive spatial scale than the demersal bioregionalisation (17 bioregions, 9 provinces and 8 zootones).
2. The physical regionalisation was conducted for waters 50m or deeper and with a cell size of one-half degree. In contrast to the biological regionalisation, a multivariate classification approach was used with selected high information environmental attributes. These regionalisations provided highly contiguous groups for waters primarily offshore of the shelf. This regionalisation extends well beyond the EEZ in some regions. Remarkable disparities in east-west gradient structures were observed in the oceanographic attribute maps and recognised by the physical classification analysis. These disparities may well be linked to the remarkable zootonal changes noted for the WA region and this is one of the key areas for investigation with the future integration activities.
3. The demersal and pelagic bioregionalisations were derived for the marine system comprising the estuaries, coastal marine and shelf (shelf pelagic, shelf demersal). The timeframe and resources available for this project did not permit an analysis of the slope and deeper ocean biological information (even though good coverage of the slope region fauna was derived for some regions during the project data compilation). Similarly, the offshore oceanographic regionalisation does not provide an adequate regionalisation of the slope environment. Thus, the major spatial gaps in the current regionalisation are in the biological and physical regionalisation for the slope system, and the lack of a biological regionalisation for the areas offshore of the shelf/slope.
4. The pelagic and demersal boundaries extend from the coast to the shelf-edge and reflect the close match of principal disjunctions across the different bands (estuarine, coastal marine, shelf demersal). The data and analyses conducted are capable of delineating subprovincial structuring which exists at the next level down the hierarchy, and in particular, the estuarine system contains a number of subprovincial disjunctions which need to be further examined.
5. A major departure of the current bioregionalisation is in the recognition and demarcation of zootones and their contributory core provincial bioregions. Core provincial bioregions are identified by the presence of provincial species unadulterated by those from other provinces. In contrast zootones are bioregions where species from two or more provincial bioregions mix. Thus, whilst the distribution of species represented in core provincial bioregions may extend well into the neighbouring zootones, it does so in conjunction with species from other provinces. The zootones identified here are not simply "fuzzy" boundaries around a disjunction. On the contrary many of the zootones are at least as extensive and in some cases such as the Great Australian Bight Zootone (GABZ), are more extensive than most of the core provincial bioregions. These zootones must be recognised as unique systems and managed with due recognition of the contribution made from core provinces. Comparisons with such conventional delphic regionalisations as CONCOM are inappropriate as the results generated by this project are novel, focussed on regional interpretations rather than boundary-based, very different in concept and require different conservation interpretations of the derived bioregions.
6. Subprovincial disjunctions across a number of these estuarine, coastal marine, shelf-pelagic and shelf demersal systems exist in zootones. Some of these disjunctions contain large species dissimilarity changes which under a conventional regionalisation system may be marked as primary provincial disjunctions. The view taken in the present project is that these disjunctions represent subprovincial structuring within a zootone. The relative strengths of the zootones and core provincial bioregions do vary and a future task is to identify the relative strengths based on the species distribution information.
7. The biological regionalisation indicates the expected pattern of higher species richness in the tropics relative to the cool temperate bioregions. However, when species are selected based on reliability and information content, the pattern is reversed showing a higher proportion of the temperate species are more reliable and of higher information content. Part of this is attributable to the lower reliability for the northern and north-western tropical species as well as the lower reliability of the north eastern (including the Great Barrier Reef), or the lack of ready access to information. Collation and analysis of tropical information is a high priority for any future extensions and refinement of the current bioregionalisation.
8. The generally higher levels of endemism amongst marine fauna of southern Australia relative to northern areas may be the result of the geological history of the region. Events such as the splitting away of New Zealand 65 - 82 million years ago and the separation of Tasmania from eastern Antarctica 40 million years ago allowed the mixing of eastern faunas with Tethyan faunas from the west (Poore et al., 1994). These conditions are similar to those described by Gilmore (1995) as being important in generating biodiversity.

16.3. Management Implications

1. A number of regionalisations are presented here, all of which have different management implications. One must also be clear about precisely at what level in the hierarchy management actions or outcomes are being sought. Contemporary management has by and large been targeted at the subprovincial and lower scale where management actions at conserving and protecting "habitats" is thought to equate to species conservation and protection. This may well be the case in some situations but the universality of the notion that subprovincial structures can be represented by (or represent) habitats, which in many cases are narrowly interpreted as bottom substrate type, has yet to be proven. This is one of the key objectives of any examination of the subprovincial structuring inherent in the current bioregionalisation project and the largely habitat-based onshore IMCRA regionalisation (MCA, 1995).
2. Accepting that management actions and concerns are at the subprovincial level, the provincial scale regionalisation provides an appropriate national-scale framework for aggregating and assessing the degree of representativeness, protection, bias and threats at a higher level.
3. The provincial scale bioregions and zootones also provide a framework for assessing the connectedness of the lower level bioregions. A key management question is the degree to which actions within a subprovincial, or lower scale, region are likely to impact upon those adjacent to or at a higher level. To a large extent the zootones capture this spatial regional intimacy by recognising the level to which core provincial regions contribute to a particular zootone. The present analysis however needs to be extended to assess the degree of endemicity in provinces and the relative contribution to zootones from core provincial species. Similar statistics will be required of regionalisations at a lower level to indicate the connectedness, or otherwise, with the neighbouring bioregions and to the contribution of that bioregion to the next level up the hierarchy.
4. A principal differentiation between core provinces and zootones is that changes in the conservation and protection status of species in provinces must be treated with greater concern as these are the core source region for the provincial species. Species changes may also appear as contractions or expansions in their distribution to natural or other disturbances (e.g. climate change/variability, fishing). Such alteration may first appear as changes in the endpoints of the species range which may well fall in a zootone or at a disjunction. Monitoring of these endpoints for key indicator species may provide a much more sensitive indicator of change than species abundance or richness change in provincial regions. Indeed, it is precisely because the core region is likely to be the most robust regional sanctuary that it may not be a sensitive indicator of species changes. Conversely, conservation and management actions which affect species measures in the core provinces may dramatically alter distributional ranges in the relevant zootones.
5. The previous discussion highlights the need for different but complimentary management and monitoring actions in provinces and zootones. In particular, the distributional changes in zootones, which in general are more diverse, are likely to be much more dynamic and sensitive to both natural and human actions.

Next Chapter: 17. Conclusions