Integrating Traditional and Close Range Photogrammetric Bathymetric Reconstructions to Enhance Predictions of Fish Abundance and Distribution on the NSW Coast

Abstract

The physical structure of marine habitat is a key determinant of the distribution and abundance of marine biota. Photogrammetry is a new method of obtaining bathymetric reconstructions using overlapping imagery. It is associated with several potential improvements over traditional bathymetric reconstruction methods (e.g., hydroacoustic and optical remote sensing), including finer resolutions, 3D mesh surfaces, and novel metrics of structural complexity. However, the greater cost of photogrammetric data collection requires evaluation of its purported benefits to marine research. This thesis objectively assessed the potential for photogrammetry to improve predictions of marine biota abundance and distribution. Chapter 2 undertook a quantitative review and metanalysis of latest research and the relative performance of metrics. It indicated common metrics, e.g., surface-rugosity, may not always be the best performing. Chapter 3 systematically explored the relationships between metrics derived from common bathymetric reconstructions and reduced a 2,000 predictor dataset to 100 predictors, whilst maximising information captured. Metric relative performance was assessed in Chapter 4. Photogrammetric metrics contributed to 22 / 35 fish species and 10 / 15 trophic-mobility group best performing abundance models and helped explain a third more variability compared to traditional methods. Chapter 5 extrapolated (‘engineered’) broad-scale photogrammetric metrics from traditional metrics to help alleviate the cost of photogrammetry. Using an independent dataset, the variance 26 / 50 fish species distribution models was explained best when engineered photogrammetric metrics were included. These findings help confirm the purported benefits to marine research associated with photogrammetric metrics, which would likely improve predictions of the distribution and abundance of fish, and likely other marine biota, across Australia and worldwide. Engineered metrics would allow greater model performance to be translated to broad-extents required by marine spatial prioritisation, conservation and management. Notably, traditional metrics were important for some fish species and groups, and future studies should seek to combine these methods wherever possible.