Bioregionalisation of Australian waters using brittle stars Echinodermata: Ophiuroidea), a major group of marine benthic invertebrates

Museum Victoria, 2008

Timothy D O'Hara

Brittle Star Australia cover

About the report

To date, large-scale marine bioregional analyses in Australia have been largely based on fish data. However, fish are highly mobile and may not act as a surrogate for seafloor communities. Consequently, to inform future regional marine plans in Australia, a bioregionalisation was undertaken for ophiuroids, a group of benthic marine invertebrates. Ophiuroids are dominant members of the benthic community, occurring in all marine habitats and exhibiting a range of reproductive strategies. Ophiuroid distributional data was accumulated from the collections of all museums in Australia; other museums in New Zealand, Europe, Asia and the United States; and from published historical information.

Ophiuroid distributions were modelled in three depth layers (outer shelf 50-300 m, upper slope 300-750 m and mid slope 750-1500 m) for the seafloor of the Exclusive Economic Zone around the Australian continent. This study used the 'modelling-then-classification' (or interpolate then analyse) approach to mapping multivariate data. Three modelling techniques were attempted: 1) a 'String' analysis, similar to that performed for the Fish Bioregionalisation of Last et al. (2005); 2) Oceanographic envelopes, analogous to terrestrial climatic modelling (eg BIOCLIM), which predicts presence and absence of species according to their known environmental ranges; and 3) Multivariate Adaptive Regression Splines (MARS) that model complex non-linear relationships between environmental variables and presence distribution data.

The six environmental variables that were used for the Oceanographic Envelope and MARS models included: depth, seabed temperature, seabed salinity, sea surface temperature, sea surface productivity, and sea surface current velocity. To eliminate the remaining correlation between depth, seafloor temperature and salinity, the latter two variables were transformed into the residuals from Generalized Linear Models (GLMs) prior to analysis, using depth as a predictor for temperature, and depth and temperature as predictors for salinity. All variables were interpolated to a cell size of 0.02 degrees. The MARS models failed to accurately predict known distributions and were not analysed further. The resulting predictions from the two remaining models were summarized into cells of 1 degree latitude/longitude for multivariate analysis, including cluster analysis, ordinations and plots of Jaccard dissimilarity between neighbouring cells.

The overall pattern is one of continual species turnover around the Australian continent with few definitive biogeographic breaks, just regions of greater or lesser turnover. Two important biogeographic findings are evident from this study. The first is that the overall patterns do not change substantially with depth, within the range analysed (50-1500 m). The same magnitude of faunal transition occurs at the 750-1500 m layer as the 50-300 m layer. There is an almost complete turnover of ophiuroid species on the upper slope between tropical areas and southern Tasmania. The second is that areas of similar biologically-important habitat exist in separate areas off the east and west of the Australian continent. This implies that many species may have discontinuous distributions, which would invalidate the assumptions made under a 'String'-style analysis. This needs testing with further surveys.

A meta-analysis of the six datasets from the Oceanographic Envelope and 'String' analyses resulted in the identification of twelve bioregions around Australia (Figure 4.10). The exact boundaries between these regions differed slightly (1-3 degrees) depending on the technique and depth layer. Nevertheless, there was a remarkable congruity between the various analyses and depth strata within this study, and between this study and the 'String' bioregionalisation based on fish distributions.