Identifying and understanding environmental drivers of deep-sea sponge aggregations is critical for effective protection and management. Current literature suggests an association between internal wave activity, regions of enhanced currents, and the distribution of deep-sea sponge aggregations. It is hypothesised that sponges utilise particulate organic matter associated with internal waves and enhanced currents and therefore aggregate in these regions. Random Forest algorithms – as well as preliminary Generalised Additive Models (GAMs) – were fitted to the presence-pseudo-absence and density datasets to characterise the relationship with each environmental variable. Additionally, residual autocovariate variables were calculated and included in models to account for spatial autocorrelation. All variables were statistically significant when GAMs were fitted to the presence-pseudo-absence datasets, but spatial autocorrelation could not be successfully accounted for. In contrast, a Random Forest model was fitted that also successfully accounted for spatial autocorrelation. No relationships were detected when GAMs were fitted to the density dataset – results confirmed by a poorly performing Random Forest model. Our findings suggest that average temperature variability is an important driver of P. carpenteri distribution and should be incorporated into habitat suitability models. However, available environmental data do not capture key drivers of P. carpenteri density, limiting our ability to model, and therefore identify areas where aggregations are likely to occur. This may reflect a mismatch in the spatial scales in which environmental and biological processes occur. We conclude that further research is required, particularly into feeding strategy and reproduction, to identify missing drivers of P. carpenteri density.



Publication Date


Publication Title

Deep Sea Research Part I: Oceanographic Research Papers



Embargo Period


Organisational Unit

School of Biological and Marine Sciences