The Marine Environment Classification project aimed to produce classifications of New Zealand's marine environment for resource and conservation management. The statistical process that was used to define the Marine Environment Classification has ensured that the classifications have defined distinctive environmental classes. The classification is hierarchical, enabling the user to delineate environmental variation at different levels of detail and a range of associated spatial scales. Statistical tests determined that the Marine Environment Classification classes are biologically distinctive. The classification provides managers with defensible definitions of environmental and biological pattern. This should provide a useful spatial framework for broad scale environmental and conservation management. The full utility of the classifications will only become clear as the classifications are applied to management issues.
While challenges were encountered all the way through the classification's development, good progress has been made with the analytical assessment of how best to combine, transform and weight candidate classification variables. Fundamental to this was the effort put into clarifying the overall conceptual framework within which we were operating, i.e. one driven by the objective of optimising the measurement of environmental differences in a way that maximises discrimination of biological differences - our choice of Mantel tests to tune variable selection, transformation and weighting stemmed directly from this conceptual starting point. Subsequent use of this test enabled us to substantially increase the correlations between our measures of environmental and biological distance. For example, at the full EEZ scale, environmental distances based on our tuned set of variables had substantially higher correlation with biological distances for the fish (+55%) and chlorophyll a (+50%) datasets than with the initial set of predictors. Smaller gains were made with the benthic dataset and with the geographic subsamples.
There are two points that should be borne in mind when applying the Marine Environment Classification. First, decisions were based on averaging results of tests performed on various biological datasets and based on whole assemblage measures of similarity. Thus, the classification has not been optimised for a specific ecosystem component (e.g. fish communities or individual species) and has sought to provide a general classification that has relevance to a broad range of biological groups. Second, the Marine Environment Classification is based on a particular approach to measuring environmental similarity and method for deriving a structure of classes. Other approaches exist and may have benefits.
The testing and biological characterisation phase (which tested how biologically distinctive the environmentally defined Marine Environment Classification classes were) was also limited by data availability. Testing, and in particular biological characterisation, of the Marine Environment Classification defined classes should be seen as an ongoing process that will continue to occur during the application of the Marine Environment Classification to management issues.
In future, the classification may be improved with new data. In particular, the EEZ classification may benefit from the addition of a freshwater fraction layer. In future, freshwater inputs around the New Zealand coastal region may be able to be described using products derived from remote sensing of ocean colour. Another obvious variable that was omitted from both the Hauraki and EEZ classification is seabed sediment. A point that needs consideration is whether sediment at the same resolution as the other variables (assuming this was available) is necessary. Patterns in seabed sediment may be correlated with bathymetry (depth, shape), tidal currents and swell as well as sources of sediment. It is possible that the Marine Environment Classification classes, particularly at high levels of the classification, already capture broad scale variation in sediment.
One key limitation of both the EEZ and Hauraki Gulf classifications is their discrimination of environmental character in coastal areas. Neither classification includes seabed sediment or substrate (e.g. rocky reefs) as defining variables. Substrates vary at small spatial scales in the coastal area and are a specific cause of habitat heterogeneity. This means that some classes, in particular those that are shallow and coastal, may encompass significantly greater environmental and biological heterogeneity than other classes. Another limitation in coastal areas is the representation of estuaries by the classifications. We do not consider that the classification represents estuaries, even though these features are included in the classification grid. An estuary classification is currently under development (see Hume et al. 2003). This classification system defines estuaries around the New Zealand coastline and could be used to 'mask' the estuarine grid cells out of the Marine Environment Classifications.
At the conclusion of the Marine Environment Classification development project, the steering group was satisfied that the classification provides a useful broad-scale classification of biotic and physical patterns in New Zealand's marine environment. The steering group has supported the Marine Environment Classification as a spatial framework for analysis and management of marine conservation and resource management issues.