Remotely operated vehicles as a platform for quantitative visual surveys of demersal fishes and benthic habitat in temperate marine ecosystems

Effective monitoring of marine ecosystems and the redistribution of marine species globally from climate change, overfishing, and other anthropogenic pressures require robust methods for quantifying species, the types of assemblages, and their spatial distributions throughout time. SCUBA diver-based methods that have traditionally been used to monitor these impacts are limited by deep-water (~>35 m) and turbulent environments. This has led to a poor understanding of many remote, offshore ecosystems and marine reserves. Remotely operated vehicles (ROVs) are a novel approach to deep-water surveying that use an unmanned submersible controlled by an operator at the surface to provide direct observations of species and their habitat associations along predetermined transect routes. The exceptional maneuverability of ROVs allows these systems to deploy high-resolution stereo-video along fixed and repeatable transects that can be targeted over a range of habitats and depth gradients. Using an ultra-short baseline (USBL) acoustic positioning system, the location of fish and benthic habitat observed in these surveys can be georeferenced with multibeam echosounder (MBES)-derived seabed terrain attributes (e.g. aspect, slope, rugosity, hardness) for a better understanding of species-habitat relationships.

Recent advancements in technology and the increasing affordability of ROVs have improved the accessibility of this method to researchers. However, as a newly developed method, ROVs still require proper evaluation and standardization before they can reliably be used for monitoring. In this thesis, I address this gap in knowledge by developing effective ROV-based sampling approaches for quantifying fish and benthic habitat in temperate marine environments. I focus on specific sampling design characteristics and challenges associated with using ROVs and MBES-derived seafloor mapping to predict species' distribution patterns.

In Chapter 1, I provide a detailed rationale on the importance of developing deep-water surveying methodology for comprehensive monitoring of the marine environment and how stereo-video-based ROV surveys could fulfill this role. I expand upon this in Chapter 2 with a systematic review of published literature evaluating the suitability of ROVs, and the different sampling strategies and metrics associated with this method for visually surveying fish. Publication trends show clear differences in the application of ROV surveys in different types of studies (e.g. in natural vs. artificial habitat, in marine protected areas, in exploratory surveys, and in method evaluation and comparison surveys). Comparing ROVs with other deep-water surveying methods further illuminated the strengths, limitations, and biases of this method, and provided me with a strong foundation from which to develop effective surveying strategies for demersal fish assemblages.

South-eastern Australia is recognized as a climate change hotspot, which has led to the restructuring of many biological communities in the marine environment. Intensification and expansion of the East Australian Current has facilitated the transport of invasive biota, such as the range-extending long-spined urchin (Centrostephanus rodgersii), from mainland Australia down the east coast of Tasmania. Uncontrolled grazing from C. rodgersii has resulted in large barrens of reef habitat that are devoid of algae, which has substantially impacted local ecosystems. In Chapter 3, I address this pressing management concern by using stereo-video-based ROV surveys to assess the distribution of C. rodgersii and barren habitat cover along a gradient of increasing urchin density reported for three regions surveyed in Tasmania (Governor Island Marine Reserve, Butler's Point, and the Tasman Peninsula). This research trials a spatially balanced sampling approach, called balanced adaptive sampling (BAS), to provide comprehensive surveys of representative habitat for each of the regions. This approach allows urchin density and barren cover to be extrapolated to region-wide estimates for a better understanding of population dynamics and the development of barren habitat within the extended distribution range for this species.

Temperate marine ecosystems host a diverse range of ecologically important fishes, many of which are undergoing anthropogenic impacts that require monitoring. Extractive fishing practices, particularly gillnetting, have led to the depletion of many fish stocks in shallow (<25 m), inshore rocky reef systems in Tasmania. ROV surveys provide a quantitative approach to understanding the distribution of benthic fish and the restructuring of fish assemblages across different habitat features in shallow and deep-water environments. In Chapter 4, I investigate the influence of depth, habitat, and seabed terrain attributes on the abundance and biomass of benthic fishes off the coast of the Tasman Peninsula. Baseline data collected in this study identified important drivers of community distribution patterns that may be used to better inform future research directives and monitoring efforts.

Ecological processes driving distribution patterns of benthic fish can operate at a range of spatial scales, which may not be adequately captured by fine-scale assessments of species-habitat associations. Species' distribution assessments should therefore be undertaken at scales that reflect the habitat utilised by the target species, as well as the scales necessary to predict and manage that species. In Chapter 5, I investigate the influence of seabed terrain attributes calculated over a range of spatial scales (3 m x 3 m to 21 m x 21 m) on stereo-video ROV surveys for three key fisheries-targeted species in Tasmania (Nemadactylus macropterus, Cheilodactylus spectabilis, and Notolabrus tetricus). Considerable differences in accuracy and precision were observed for seabed attributes generated at different scales, highlighting the importance of considering both local and ecosystem-wide processes influencing species distributions.

In Chapter 6, I discuss sampling design aspects for using ROVs to describe and monitor demersal fish across major habitat features and depth gradients associated with typical shelf habitats in temperate Australia. As a novel method that requires further development, I propose numerous possible future directives for improving ROVs as a deep-water surveying tool. I conclude by providing an optimal surveying framework for stereo-video-based ROV surveys of demersal fishes and benthic habitat. This framework provides the foundation for establishing standardised surveying strategies that may guide future research efforts and improve the effectiveness of this method for conserving the marine environment.