In this section we review the underlying concepts used in past and current regionalisation strategies and details the concepts used in this project.
This project uses fish species composition and richness as an indication of broader general biodiversity of its marine fauna. This is currently the only practical means of obtaining a bioregionalisation of marine Australia since there are no other taxa for which detailed EEZ-wide occurrence data are available. McAllister et al. (1994) used coral reef fish species in their global biodiversity study for similar reasons. Practical necessities aside, is there any scientific justification for the use of fish as an indicator group?
Most studies which have found that biodiversity hotspots are not consistent across taxa have been done in terrestrial environments, and in many cases in areas where human impact has been extreme (e.g. the British Isles in Prendergast et al.'s 1993a study, or North America in the case of McAllister et al., 1994). In all cases, workers were cautious about translating terrestrial findings to the marine environment (Pielou, 1979; Williams and Gaston, 1994). Several recent studies have found that fish biodiversity hotspots can coincide with those for other taxa in tropical environments. McAllister et al. (1994) concluded that fish biodiversity hotspots corresponded with coral hotspots in their global study of shallow tropical marine environments. Harmelin-Vivien (1989) found that there was a strong correlation between species richness of corals and fishes, at least on a broad geographic scale, through 14 Indo-Pacific sites. Yatsu (1995) showed that epipelagic fish distribution patterns in the south Pacific Ocean were basically similar to those of oceanic plankton.
High levels of marine fish biodiversity are often associated with areas having compressed hydrologic and climatic environment gradients, particularly when associated with high habitat diversity, as these conditions enable organisms with widely diverging requirements to coexist (McAllister, 1994; Gilmore, 1995). Under these conditions it is likely that biodiversity hotspots will coincide across taxa, since features which produce fish species richness affect other marine fauna and flora in a similar manner. Such conditions are typical of bioregion boundaries like Shark Bay in WA and the coastline near Sydney, NSW. In these cases fish are probably a valid indicator group.
Gilmore (1995) found that general estuarine biodiversity is reflected in estuarine fish species diversity, and is increased by disturbance-mediated coexistence of species, along with increased spatial heterogeneity of biotic and abiotic habitats. Polysalinity is important in generating habitat heterogeneity (Monaco et al., 1992), which fosters general community biodiversity, including that of the ichthyofauna (Gilmore, 1995). This demonstrates the possibility that under estuarine conditions fish can be considered a valid indicator group.
Monaco et al. (1992) used estuarine fish species assemblages as surrogate habitat indicators in their comparative study of estuaries along the west coast of the USA. Regression analyses of the physical characteristics of study environments identified the topography of the estuary mouth and habitat diversity within the estuary as the most important predictors of fish species richness. However, they found that species richness alone is inadequate for comparing physical characteristics of two estuaries, and suggested that assemblage analysis may provide the means for comparison of habitats.
Consequently, it is possible that fish can act as a valid indicator group, at least under certain habitat conditions. The current project, with its complementary biotic and abiotic approaches, will serve to demonstrate the bioregionalising potential of Australia's fish species.
One commonly advocated approach to countering the problems typically involved in bioregionalisations is to conduct biological, physical and benthic regionalisations as separate and independent activities. The results can then be "merged" using multivariate or pattern analysis techniques. However, the ecological meaning of the final boundaries is generally lost in the subsequent unnatural blend of attributes, which may not be independent and which (in the case of abiotic variables) may not even possess direct ecological meaning. The popularity of the technique is founded mainly in the hope that in regions of high habitat heterogeneity ecological boundaries are sharply defined and forced strongly by abiotic variables. In such cases, independent biological, physical and benthic regionalisations overlap strongly and the resulting boundaries may be ecologically meaningful.
An alternate view, and one favoured by this project, is that ecologically meaningful boundaries are to be more reliably found in the distribution patterns of biological variables. However, it is almost always the case that the distribution and variability of abiotic variables is better known than that of biological variables for a region. Could abiotic variables be used as surrogates for biological variables in investigations of bioregionalisation?
Surrogates are no more than a simple (or simplistic) model of the distribution of biological attributes. An extension of this concept is to use the information from point observations of species and correspondence with abiotic variables to derive distribution models. These models can then be used in conjunction with the better defined distribution of abiotic variables to derive "virtual" distributions of species. These virtual distributions of species contain information from both the biological distribution as well as information of relevance from the abiotic variables. Regionalisations based on these virtual distributions arguably encompass information from the abiotic attributes, and since the distributions themselves are of species, the resulting boundaries are more ecologically meaningful.
Diversity (probably by any reasonable definition) varies with scale so diversity, defined or evaluated at a local scale, will be different to that defined over a regional/provincial scale. Two opposing contemporary paradigms on species diversity dominate current debate on scale effects (Ricklefs and Schluter, 1993): on the one hand, palaeontologists and historical biogeographers emphasise evolutionary changes with the associated phases of species extinction and accumulation as determining current distributions of species diversity. An opposing view is that of the community ecologists who argue that species interactions in the context of small-scale habitats and physical constraints control diversity.
The local ecology theory of species diversity has been unable to explain global comparative patterns of diversity leading to the current emerging view that larger spatial and temporal issues must be addressed. Any assessment must recognise the evolutionary context in which local processes operate. Comparative biogeographical studies are now seen as a way forward in bringing together the disparate views of diversity (Ricklefs and Schluter, 1993).
Community-based ecological models of diversity recognise scale dependence partly by incorporating terms for species immigration and emigration. Thus however a local community is defined, processes external to that community/scale will influence the species composition in that region. This is of special importance in the context of marine systems and in the context of how one manages seemingly isolated MPAs.
Schluter and Ricklefs (1993) identify a number of processes which contribute to observed patterns of diversity. Amongst these are: local ecological interactions (competition, predation), individual movement between habitats (of differing residence times), dispersal between habitats (colonisation as opposed to movement), evolutionary spread of species between neighbouring habitats, allopatric production of species, paleogeographic formation or breakdown of physical barriers and climatic events such as cooling/glaciation. These processes involve a spectrum of space and time scales from local to global. They imply that no matter how provinces are discerned, they cannot be viewed as being absolute in either space or time. The hierarchical conceptual view proposed in this project accommodates the leakiness of such demarcations by recognising the dominant factors and processes operating at various levels in the hierarchy. As such the ecological interpretation of the region depends on the hierarchical level and the associated processes/factors of importance at that level. This of course implies that the same interpretation may not necessarily hold for the same geographic region at a different spatial scale.
In recognition of these considerations, it is appropriate to delimit what this project will attempt to deliver in terms of regionalisations and the context in which other regionalisations can contribute to this process. The scale of interest for this project is that of the province with characteristic spatial scales (generally but not always) of 1000's km at which the characteristic process affecting species diversity is that of species formation (Ricklefs and Schluter, 1993). The regionalisation we will produce recognises the following features:
Provincial Core Regions within which the community assemblage types are consistent.
Zootones in which species characteristic of disparate provincial core regions undergo mixing proceeding from one end of the zootone (which may also be an endpoint for a provincial core region).
Disjunction representing a relatively sharp boundary at which a number of species from a provincial core terminate or start.
Figure 6-1 Boundary delineation illustrating two provincial core regions, denoted "A" and "B" at either ends of a zootone within which species from the provinces mix. The curves are hypothetical representation of species richness. Each provincial core region terminates at a disjunction (these are at the ends of the zootone). Within the zootone an additional disjunction is illustrated; this corresponds to a subprovincial, or mesoscale boundary.
This is conceptually quite a different scheme than existing schemes such as CONCOM (see discussion below) where only sharp provincial boundaries are recognised. Zootonal regions do exist (see section 12) and their non-recognition in past regionalisation is one cause for considerable confusion and non-acceptance.
Attempts over recent decades to produce regionalisations of the worlds marine environments have been based on various scales and have met with varying degrees of success. The following discussion deals with the problems, ideas and solutions presented by a few of these studies.
'Theme regions', based on homogeneity of geology, soils, climate and vegetation, have been described through Canada's terrestrial and coastal marine environments (Davis et al., 1994). Parks Canada used physical and biological themes in a more qualitative (delphic) than quantitative methodology to differentiate 29 natural marine regions. Three themes were selected as components of the physical base case: oceanography, coastal environments, and sea-bottom physiography. Discrepancies were resolved in order to arrive at a physical base-case map. Similarly, a biological base-case was sought within each of the second-order divisions, based on marine mammal, marine bird, marine littoral community, and fish distribution theme maps, and discrepancies were overcome by compromises (Yurick, 1995). To describe subregional distinct landscape features, these regions were further divided into a hierarchical system of districts, units and sub-units (Davis et al., 1994). However, a significant bias arose in the final regionalisation due to the marine regions being terrestrially anchored by imposition. This action was apparently taken partly to assist administration and because coastal physiography was a component to be used in identifying potential park sites (Yurick, 1995).
Davis et al. (1994) used depth and topographic region (inner, middle, or outer) as the principal distinguishing characteristics in defining theme regions through Nova Scotia's shelf ecosystems. Subdivisions of these regions occurred on the basis of temperature, sediments, freshwater influence, and biological processes. They found that physical characteristics of bottom type could adequately describe benthic animal and plant associations, and that while pelagic communities were extremely variable in space and time (Yatsu, 1995), there was some relationship between their presence and local hydrographic mechanisms resulting from bottom physiographic features (O'Boyle et al., 1984; Davis et al., 1994).
An IUCN programme for establishing MPAs through the south Pacific relied firstly on the identification of major biogeographic zones using a biogeographic classification system appropriate for the region (Kenchington and Bleakley, 1994). This biogeographic classification system was based on environmental factors considered important to biological diversity and ecological function and operates regardless of the lack of information regarding species or habitats within the zones (Kenchington and Bleakley, 1994). The authors claim that this technique can be used to determine whether each type of zone contains an adequate area of MPAs, despite the biological information vacuum.
In the Eastern Boundary Current Program (US GLOBEC, 1994), California Current biological provinces were identified from phytoplankton, zooplankton, invertebrate benthos, marine fish, and bird distributions. Of interest to note is the temporal dynamism of provincial boundaries: in that region the effects of El Niño changes in oceanic conditions are rather severe and has led to shifts northwards of species assemblages and oceanic water masses during extreme ENSO events (US GLOBEC, 1994).
In New Zealand, King et al. (1985) produced a 3-tier habitat classification scheme encompassing the entire coastal and nearshore marine realm. This was the first attempt to provide a scientifically based biogeographical framework specifically intended to assist the selection of an ecologically representative system of MPAs. A number of approaches have been used to subdivide New Zealand's marine environment into biogeographic areas or zones, but none of these have been successful in identifying distinct biogeographic regions which were acceptable to the scientific community at large (Walls, 1995).
A primarily biotic information base was used by Walls (1995). Distribution patterns of fish, molluscs, echinoderms, bryozoans, sponges, ascidians, antipatharians, foraminifera, brachiopods and algae were identified. Factors used to differentiate distribution patterns included endemism and species diversity, as well as geological and oceanographic features. It was shown that a systematic approach is required if the New Zealand Department of Conservation is to establish a representative national network of marine reserves. A biogeographic approach was chosen because the marine reserves legislation required that areas be protected for scientific study of marine life, and the scientific community has agreed with this approach (Walls, 1995).
In summary, the international experience shows that regionalisations must be derived from a systematic scientifically-based ecological approach, must consider spatial and temporal variations in boundaries and be designed to address a range of management needs likely to arise from multiple uses of resources. The disparity in the scale and manner in which regions are demarcated reflects a lack of agreement and/or understanding of the hierarchical nature of bioregions.
Next Chapter: 7. Bioregionalising Australia