Dynamic Distributions of Coastal Zooplanktivorous Fishes

Home , Barber perch, Great Southern Reef, Herring cale, Kelpfish, Latridae, Little weed whiting

Dynamic distributions of coastal zooplanktivorous fishes

Matthew Michael Holland

A thesis submitted in fulfilment of the requirements for a degree of

Doctor of Philosophy

School of Biological, Earth and Environmental Sciences

Faculty of Science

University of New South Wales, Australia

November 2020

4/20/2021 GRIS  Welcome to the Research Alumni Portal, Matthew Holland!

You will be able to download the finalised version of all thesis submissions that were processed in GRIS here.

Please ensure to include the completed declaration (from the Declarations tab), your completed Inclusion of Publications Statement (from the Inclusion of Publications Statement tab) in the final version of your thesis that you submit to the Library.

Information on how to submit the final copies of your thesis to the Library is available in the completion email sent to you by the GRS.

Thesis submission for the degree of Doctor of Philosophy

Thesis Title and Abstract Declarations Inclusion of Publications Statement

Corrected Thesis and Responses

Thesis Title

Dynamic distributions of coastal zooplanktivorous fishes

Thesis Abstract

Zooplanktivorous fishes are an essential trophic link transferring planktonic production to coastal ecosystems. Reef-associated or pelagic, their fast growth and high abundance are also crucial to supporting fisheries. I examined environmental drivers of their distribution across three levels of scale. Analysis of a decade of citizen science data off eastern Australia revealed that the proportion of community biomass for zooplanktivorous fishes peaked around the transition from sub-tropical to temperate latitudes, while the proportion of herbivores declined. This transition was attributed to high sub-tropical benthic productivity and low temperate planktonic productivity in winter. As temperatures declined in autumn, zooplanktivores migrated away from reefs, likely seeking warmer waters offshore.

This was supported in a related bioacoustics survey around Montague Island, southern NSW, which showed the coast-to-shelf distribution of pelagic zooplanktivores in early spring was closely correlated with temperature, with schools seeking warmer waters of the East Australian Current at the shelf edge. Some zooplanktivores may select warm water to improve physiological performance and may also avoid shallow coastal areas, where vertically constrained distribution increases predation risk.

To examine spatiotemporal dynamics of zooplanktivores around coastal reefs a non-scientific portable multibeam echosounder was deployed. As calibration of such instruments is impracticable, I assessed a novel metric of school thickness, delineating the spatial boundaries of schools rather than quantifying backscatter.

These methods were applied to study school distribution around natural and artificial reefs, and variability attributed to current exposure and time of day. Schools regularly distributed upstream of structure, as fish competed for access to un-grazed prey. Schools blanketed the seafloor at night and rose high in the water column during the day to feed, particularly around high vertical https://gris.unsw.edu.au/alumni/ 1/2 4/20/2021 GRIS relief artificial structures. These observations can inform the design of artificial reefs to include high vertical relief to facilitate foraging in the water column. These findings also agree with observations of dense schools associated with offshore oil and gas infrastructure and provide support for leave-in-place decommissioning. These distribution patterns have important implications for fisheries and for the predators which depend on regular and predictable access to prey.

These methods were applied to study school distribution around natural and artificial reefs, and variability attributed to current exposure and time of day. Schools regularly distributed upstream of structure, as fish competed for access to un-grazed prey. Schools blanketed the seafloor at night and rose high in the water column during the day to feed, particularly around high vertical relief artificial structures. These observations can inform the design of artificial reefs to include high vertical relief to facilitate foraging in the water column. These findings also agree with observations of dense schools associated with offshore oil and gas infrastructure and provide support for leave-in-place decommissioning. These distribution patterns have important implications for fisheries and for the predators which depend on regular and predictable access to prey.

https://gris.unsw.edu.au/alumni/ 2/2 4/20/2021 GRIS  Welcome to the Research Alumni Portal, Matthew Holland!

You will be able to download the finalised version of all thesis submissions that were processed in GRIS here.

Information on how to submit the final copies of your thesis to the Library is available in the completion email sent to you by the GRS.

Thesis submission for the degree of Doctor of Philosophy

Thesis Title and Abstract Declarations Inclusion of Publications Statement

Corrected Thesis and Responses

ORIGINALITY STATEMENT

 I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.

 I hereby grant the University of New South Wales or its agents a non-exclusive licence to archive and to make available (including to members of the public) my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known. I acknowledge that I retain all intellectual property rights which subsist in my thesis or dissertation, such as copyright and patent rights, subject to applicable law. I also retain the right to use all or part of my thesis or dissertation in future works (such as articles or books).

For any substantial portions of copyright material used in this thesis, written permission for use has been obtained, or the copyright material is removed from the final public version of the thesis.

AUTHENTICITY STATEMENT

 I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis.

https://gris.unsw.edu.au/alumni/ 1/2 4/20/2021 GRIS  Welcome to the Research Alumni Portal, Matthew Holland!

You will be able to download the finalised version of all thesis submissions that were processed in GRIS here.

Information on how to submit the final copies of your thesis to the Library is available in the completion email sent to you by the GRS.

Thesis submission for the degree of Doctor of Philosophy

Thesis Title and Abstract Declarations Inclusion of Publications Statement

Corrected Thesis and Responses

UNSW is supportive of candidates publishing their research results during their candidature as detailed in the UNSW Thesis Examination Procedure.

Publications can be used in the candidate's thesis in lieu of a Chapter provided:

The candidate contributed greater than 50% of the content in the publication and are the "primary author", i.e. they were responsible primarily for the planning, execution and preparation of the work for publication. The candidate has obtained approval to include the publication in their thesis in lieu of a Chapter from their Supervisor and Postgraduate Coordinator. The publication is not subject to any obligations or contractual agreements with a third party that would constrain its inclusion in the thesis.

 The candidate has declared that some of the work described in their thesis has been published and has been documented in the relevant Chapters with acknowledgement. A short statement on where this work appears in the thesis and how this work is acknowledged within chapter/s:

https://gris.unsw.edu.au/alumni/ 1/2 4/20/2021 GRIS

My thesis contains three data chapters that have been published (or in press) in peer-reviewed journals, while a fourth data chapter is currently in review. The paper 'Latitudinal patterns in trophic structure of temperate reef-associated fishes and predicted consequences of climate change' comprises Chapter 2 and is published in Fish and Fisheries. The paper 'Pelagic forage fish distribution in a dynamic shelf ecosystem – Thermal demands and zooplankton prey distribution' comprises Chapter 3 and is published in Estuarine, Coastal and Shelf Science. The paper 'Characterising the 3D distribution of schooling reef fish with a portable multibeam echosounder' comprises Chapter 4 and is in press in Limnology and Oceanography: Methods. The paper 'Fine spatial and diel dynamics of zooplanktivorous fish on temperate rocky and artificial reefs' comprises Chapter 5 and is currently in review in Marine Ecology Progress Series (MEPS). Acknowledgments of the work contributed by other authors of these papers are made in my Acknowledgments section, at the end of my General Introduction, and at the start of Chapters 2, 3 and 4.

Candidate's Declaration

I declare that I have complied with the Thesis Examination Procedure.

https://gris.unsw.edu.au/alumni/ 2/2

Abstract

This was supported in a related bioacoustics survey around Montague Island, southern

NSW, which showed the coast-to-shelf distribution of pelagic zooplanktivores in early spring was closely correlated with temperature, with schools seeking warmer waters of the

East Australian Current at the shelf edge. Some zooplanktivores may select warm water to improve physiological performance and may also avoid shallow coastal areas, where vertically constrained distribution increases predation risk.

To examine spatiotemporal dynamics of zooplanktivores around coastal reefs a non- scientific portable multibeam echosounder was deployed. As calibration of such instruments is impracticable, I assessed a novel metric of school thickness, delineating the spatial boundaries of schools rather than quantifying backscatter.

These methods were applied to study school distribution around natural and artificial reefs, and variability attributed to current exposure and time of day. Schools regularly distributed

upstream of structure, as fish competed for access to un-grazed prey. Schools blanketed the seafloor at night and rose high in the water column during the day to feed, particularly around high vertical relief artificial structures. These observations can inform the design of artificial reefs to include high vertical relief to facilitate foraging in the water column. These findings also agree with observations of dense schools associated with offshore oil and gas infrastructure and provide support for leave-in-place decommissioning. These distribution patterns have important implications for fisheries and for the predators which depend on regular and predictable access to prey.

Acknowledgements

I would like to thank my dedicated team of supervisors Iain Suthers, Jason Everett, James

Smith, Alistair Becker and Martin Cox for their guidance and support throughout my candidature. Thanks to the honours and graduate students in the FAMER Lab and the

Centre for Marine Science and Innovation at UNSW who have helped immensely with problem solving through casual work chats around our desks. A further thanks to my collaborators Martina Doblin and Adriana Vergés, for all their input and support on two of my manuscripts.

I am very grateful to Iain for the opportunity to learn the fundamentals of fisheries acoustics through an ICES training course in Copenhagen, as well as two research cruises aboard the RV Investigator which taught me fundamental skills in marine science and formed Chapter 3 of this thesis. Another big thanks to Sven Gastauer from the Australian

Antarctic Division for his guidance in geostatistics and acoustic analysis at the start of my candidature, and to Ben Maslen and Zhixin Liu from UNSW Stats Central for their guidance with numerical modelling. My attendance at the Australian Marine Science

Association annual conference in Adelaide in 2018 was supported by the Graduate Research

School at UNSW.

The Sydney Institute of Marine Science provided invaluable financial support for boat work and thanks must go to their staff, particularly Hayden Schilling, Nigel Coombes and

Andrew Niccum, for their skippering services and allowing me use of their facilities as well as the ability to retrofit their vessels for coastal fieldwork.

Thanks to Captain John Paton of Bravo Fishing Charters and the dedicated volunteers who braved long cold nights and seasickness to document the nocturnal distribution of forage

iii

fish around natural and artificial reefs. Hopefully, our encounters with the odd humpback, dolphin or mako shark helped the time go by.

A big thanks to the staff of Taylor Marine, particularly Brendan Bourke, for technical assistance with the WASSP multibeam. The teams at Echoview in Hobart and WASSP in

Auckland were especially supportive, creating and testing software to allow me to process

WASSP data in Echoview.

Thanks Mom and Dad for all your understanding and support throughout this long process, and to Milly for all your patience and encouragement to keep me motivated.

This work was funded in part by an Australian Research Council Linkage Grant

(LP160100162) and The Sydney Institute of Marine Science. I was personally supported by an Australian government Research Training Program Scholarship.

I would like to dedicate this thesis to my parents, Peter and Dorothy Holland, for always encouraging me to follow my passions in life. Without their love and support none of this would have been possible.

I hope that the research I have conducted in this thesis can support a better understanding of the importance of planktivory in western boundary current ecosystems and inform advancements in the design of artificial reefs.

Table of contents

Abstract ...... i

Acknowledgements ...... iii

List of tables ...... ix

List of figures ...... x

1. General introduction ...... 1 1.1. Pelagic subsidies to temperate coastal ecosystems ...... 1 1.2. Broader spatiotemporal scales and tropicalisation ...... 5 1.3. Contribution of zooplanktivorous fishes to temperate coastal ecosystems ...... 9 1.4. What distribution can reveal about behaviour ...... 13 1.5. Enhancing fisheries with artificial reefs ...... 16 1.6. Modern methods for studying fish distribution ...... 18 1.7. Thesis structure and aims ...... 21

2. Latitudinal patterns in trophic structure of temperate reef-associated fishes and predicted consequences of climate change ...... 25 2.1. Abstract ...... 25 2.2. Introduction ...... 26 2.3. Methods ...... 30 2.3.1. Study region ...... 30 2.3.2. Biomass of fish trophic groups ...... 31 2.3.3. Trophic classification ...... 35 2.3.4. Characterising latitudinal trends ...... 35 2.3.5. Explanatory variables ...... 37 2.3.6. Model prediction ...... 40 2.4. Results ...... 41 2.4.1. Broad latitudinal trends ...... 41 2.4.2. Multivariate latitudinal trends ...... 44 2.4.3. Fish Biomass GAMMs - spatiotemporal effects ...... 46 2.4.4. Fish Biomass GAMMs – environmental effects ...... 49 2.4.5. Predicted Fish Biomass from the GAMM ...... 50 2.5. Discussion ...... 51 2.5.1. Latitudinal gradients in trophic structure ...... 52 2.5.2. Disappearing fishes – Winter declines in rocky reef associated fish biomass ...... 53 2.5.3. Dynamic temporal distribution - A means of managing predation? ...... 54

2.5.4. Climate-change risks for temperate rocky reef communities ...... 55 2.5.5. Conclusion ...... 57

3. Pelagic forage fish distribution in a dynamic shelf ecosystem – Thermal demands and zooplankton prey distribution ...... 60 3.1. Abstract ...... 60 3.2. Introduction ...... 61 3.3. Methods ...... 64 3.3.1. Field surveys ...... 64 3.3.2. Acoustic data collection and processing ...... 66 3.3.3. Comparing fish distribution between surveys ...... 69 3.3.4. Measuring the distribution of zooplankton biomass ...... 70 3.3.5. Bathymetry as a driver of fish and zooplankton vertical distribution ...... 71 3.3.6. Measuring in situ oceanographic variables ...... 72 3.3.7. Establishing long-term oceanographic and fishery context ...... 72 3.3.8. Fish and zooplankton surface distribution ...... 73 3.4. Results ...... 76 3.4.1. Survey and long-term oceanographic conditions ...... 76 3.4.2. Comparing fish distribution between surveys ...... 79 3.4.3. Bathymetry as a driver of fish and zooplankton vertical distribution ...... 80 3.4.4. Simulating fish and zooplankton surface distribution ...... 82 3.4.5. Comparing fish and zooplankton predicted surface distribution ...... 86 3.5. Discussion ...... 88 3.5.1. Consistency in coast-to-shelf distribution (Hypothesis 1) ...... 88 3.5.2. Bathymetry as a driver of forage fish distribution (Hypothesis 2) ...... 89 3.5.3. Temperature as a driver of forage fish distribution (Hypothesis 3) ...... 90 3.5.4. Spatial mismatch between predators and prey (Hypothesis 4) ...... 92

4. Characterising the 3D distribution of schooling reef fish with a portable multibeam echosounder ...... 95 4.1. Abstract ...... 95 4.2. Introduction ...... 96 4.3. Methods ...... 101 4.3.1. Site description ...... 101 4.3.2. Species composition from remote video ...... 101 4.3.3. Multibeam acoustic data collection ...... 103 4.3.4. Multibeam acoustic data processing ...... 103 4.3.5. Spatial subsetting ...... 106 4.3.6. Deriving school thickness ...... 108 4.3.7. Simulating school distribution around an artificial reef ...... 110

4.3.8. Comparison of school thickness with calibrated acoustics ...... 111 4.4. Results ...... 113 4.4.1. Comparison of school thickness with calibrated acoustics ...... 113 4.4.2. Simulating school distribution around an artificial reef ...... 116 4.5. Discussion ...... 118

5. Fine spatial and diel dynamics of zooplanktivorous fish on temperate rocky and artificial reefs ...... 123 5.1. Abstract ...... 123 5.2. Introduction ...... 124 5.3. Methods ...... 127 5.3.1. Site selection ...... 127 5.3.2. Data collection ...... 133 5.3.3. Water column acoustic data processing ...... 135 5.3.4. Raster image processing ...... 138 5.3.5. Data analysis ...... 139 5.4. Results ...... 148 5.4.1. School composition and diversity ...... 148 5.4.2. School distribution ...... 151 5.4.3. Diel effects on school characteristics ...... 155 5.5. Discussion ...... 162 5.5.1. Foraging behaviour drives spatial distribution around benthic structure ...... 163 5.5.2. Accessing the water column ...... 165 5.5.3. The effects of current velocity ...... 166 5.5.4. Nocturnal behaviour ...... 168 5.5.5. Conclusion ...... 169

6. General discussion ...... 170 6.1. Trophic composition along western boundary currents globally ...... 174 6.2. Coast-to-shelf gradients on continental scales ...... 177 6.3. Decommissioning offshore infrastructure ...... 180 6.4. Designing artificial reefs to capture pelagic subsidies ...... 182 6.5. Concluding remarks ...... 185

7. Literature cited ...... 186

8. Appendices ...... 222 8.2. Supplementary material for Chapter 2 ...... 222 8.2.1. Supplementary methods ...... 222 8.2.2. Supplementary tables ...... 225

vii

8.2.3. Supplementary figures ...... 302 8.3. Supplementary material for Chapter 3 ...... 311 8.3.1. Supplementary methods ...... 311 8.3.2. Supplementary tables ...... 313 8.3.3. Supplementary figures ...... 316 8.4. Supplementary material for Chapter 4 ...... 322 8.4.1. Supplementary tables ...... 322 8.4.2. Supplementary figures ...... 323 8.5. Supplementary material for Chapter 5 ...... 332 8.5.1. Supplementary methods ...... 332 8.5.2. Supplementary tables ...... 334 8.5.3. Supplementary figures ...... 340

viii

List of tables

Table 2.1 The final GAMMs generated through testing every possible combination of variables in the full model and selecting the model with lowest AIC score...... 47

Table 3.1 Summary of survey characteristics and instruments on both surveys. In this case, ✓ indicates that data was available for the corresponding dataset, survey year combination, while - indicates the data was not available...... 67

Table 3.2 The three sets of models selected via the forward-stepwise AIC-based selection process for NASC in 2016 (NASC.2016), NASC in 2017 (NASC.2017) and zooplankton biomass in 2017...... 83

Table 4.1 Glossary of terms ...... 98

Table 5.1 Site names, locations and descriptions, and the aims they were used to address.130

Table 5.2 Results for the binomial GLMs describing the distribution of schooling fish around artificial and natural reefs...... 152

Table 6.1 Key behavioural findings and corresponding design implications for artificial reefs...... 183

List of figures

Figure 1.1 The Australian ‘Great Southern Reef’ (red line), an 8000 km strip of interconnected habitat, adapted from Bennett et al. (2016), Figure 1...... 1

Figure 1.2 Graphical abstract illustrating the three major energy inputs to a temperate reef (kelp forest) ecosystem, from Zuercher and Galloway (2019), Figure 1.Zuercher and Galloway (2019), Figure 1...... 3

Figure 1.4 Simulated global distribution of pelagic forage fish biomass adapted from Petrik et al. (2019), Figure 7a...... 7

Figure 1.5 The provision of pelagic subsidies to a tropical coral reef, from Morais and Bellwood (2019), Figure 4a...... 10

Figure 1.6 Average relative biomass of all observed fish trophic groups as a proportion of total fish biomass for: three temperate rocky reef sites, an average of 14 sites from the Reef Life Survey, and an average of all sites, from Truong et al. (2017), Figure 3...... 12

Figure 1.7 Percentage of functional studies in the journal ‘Coral Reefs’ by trophic group, relative to the number of species (a) and abundance (b) in each group at the Great Barrier Reef, from Bellwood et al. (2019), Figure 3...... 13

Figure 1.8 Schematic displaying the three different types of acoustic data that can be obtained simultaneously from a multibeam echosounder, including bathymetric measurement (a), seafloor backscatter (b) and water column measurement (c), from Colbo et al. (2014), Figure 1...... 21

Figure 2.1 Variation in the proportional representation of herbivores and zooplanktivores (sum = 1) from select studies examining reef fish trophic structure along environmental gradients ...... 29

Figure 2.2 Chart of the southeast coast Australia, with black tick marks representing the 567 Reef Life Survey (RLS) sites...... 32

Figure 2.4 Mean monthly total fish biomass represented by the height of each bar...... 43

Figure 2.6 2D tensor contour plots showing fitted GAMM relationships with contours of trophic group biomass (in g m-2) ...... 49

Figure 2.7 Predicted per-survey total fish biomass trajectories calculated by month and latitudinal bin using the total fish biomass GAMM...... 51

Figure 3.1 Graphical abstract outlining the hypotheses under investigation in this study, indicating possible forage fish distribution relative to horizontal gradients in zooplankton and phytoplankton, bathymetry and temperature and vertical gradients of predation pressure from the surface and the seafloor...... 63

Figure 3.2 Chart of ship tracks (solid white line) of acoustic surveys in 2016 (a) and (c), and 2017 (b) and (d)...... 65

Figure 3.3 Mean along shelf current for southeast Australia along the 200 m isobath (a) with red dashed line indicating the latitude of Montague Island...... 77

Figure 3.4 Temperature-salinity plots generated from the ship’s underway data displaying NASC values from fish as point colour on a logarithmic scale for each of the two surveys. . 78

Figure 3.6 Vertical distribution of mean zooplankton biomass (a), mean fish NASC (b) and mean temperature (c) for three along-shelf transects with mean bathymetry (62, 100 and 118 m bottom depth) from the 2017 survey...... 81

Figure 3.7 Model structure and predictor responses for the most parsimonious models for fish NASC in 2016 (a) and 2017 (b) and with the additional predictor for zooplankton biomass (c) which was not selected in the most parsimonious model, and the model for zooplankton biomass from the 2017 survey (d)...... 85

Figure 3.8 Model prediction results for pelagic forage fish in 2016 (a) and 2017 (b), and for zooplankton biomass (c), generated using satellite data and bathymetry maps...... 87

Figure 4.1 Diagram of the reef module layout with the approximate position of the remote camera assembly, anchored to be floating above the seafloor (not to scale)...... 102

Figure 4.2 Flow chart outlining the method for processing acoustic data from both split- beam and MBES to produce the ‘school thickness’ variable...... 104

Figure 4.3 One-metre resolution digital elevation model for a natural reef site generated from data collected with the multibeam echocounder...... 105

Figure 4.4 Raw 3D spatially referenced samples from one survey of the artificial reef, coloured by mean volume backscatter (Sv)...... 106

xii

Figure 4.8 GAMM prediction for the mean distribution of school thickness around the artificial reef. Black squares indicate the locations of the concrete reef modules...... 116

Figure 5.2 Gridded school thickness for four survey dates at the SS Annie Miller (AM – top row) and Sydney Offshore Artificial Reef (OAR – bottom row)...... 137

Figure 5.4 Ordination plot output for the two-variable generalised linear latent variable model of relative fish abundance...... 150

Figure 5.8 Index of aggregation (unitless) for each individual site survey and line of best fit for artificial and natural reefs, separately...... 155

xiii

volume of an individual school (c), and the mean number of schools per hectare of area surveyed (d)...... 156

Figure 5.10 Generalised least squares model effects plots displaying mean values (symbols) and 95% confidence intervals for the interaction of reef site (N5 and JDN) and diel period (night: blue and day: red) with variables describing school vertical and horizontal distribution, including mean school height above the seafloor (a) school thickness, or the sum of each 1 m depth interval containing fish (b)...... 158

Figure 5.11 Generalised least squares model effects displaying mean values (symbols) and 95% confidence intervals for the interaction of reef site (N5 and JDN) and diel period (night: blue and day: red) with variables describing school aggregation, including mean school perimeter to area ratio (a) area to volume ratio (b) and horizontal distance from the reef centre (c)...... 159

Figure 5.12 Comparison of the average prey (from gut contents analysis of three fish species, Schilling et al. unpublished data) and day/night zooplankton net capture size distributions. Error bars represent standard error of the mean...... 160

Figure 6.1 The main distributional patterns documented in each of the four studies of this thesis, including the gradient in reef community trophic composition across 16 degrees of latitude (Chapter 2), the model-predicted coast-to-shelf gradient in zooplanktivore density around Montague Island (Chapter 3), the distribution of school thickness around an artificial reef field near Sydney (Chapter 4) and differences in the distribution of zooplanktivore schools between night and day at artificial and natural reefs (Chapter 5). . 172

Figure 6.2 Global distribution of the 3537 Reef Life Survey sites surveyed by divers between 2008 and 2020, from Edgar et al. (2020), Figure 2...... 177

Figure 6.3 Zooplankton biomass (biovolume converted to mg m-3 based on the density of seawater) distributions from transects conducted with a Laser Optical Plankton Counter (LOPC) on an undulating towed body at four locations across the Australian continental shelf, from Schilling et al. (In review), Figure 3.Temperature (°C) isotherms indicated in black...... 178

xiv

Figure 6.4 The distribution of oil and gas wells drilled in Australian waters (a), from Evans et al. (2017), and three different methods for reefing retired platforms, including tow-and- place, topple-in-place and partial removal, from Bull and Love (2019)...... 182

Figure 6.5 Conceptual diagram of an optimised plankton-supported artificial reef. The vector representing current direction extends out of the page...... 184

1. General introduction

1.1. Pelagic subsidies to temperate coastal ecosystems

Temperate rocky reef ecosystems support coastal populations globally through commercial and extractive operations and tourism, in addition to the ecosystem services and intrinsic benefits they provide, which are naturally more difficult to quantify (Bennett et al., 2016;

Blamey & Bolton, 2018). While the plight of coral reefs, such as the Great Barrier Reef, has received significant media attention and research funding, temperate rocky reef ecosystems are arguably more valuable to coastal communities. In Australia, temperate rocky reefs and their productive kelp forests have recently been recognised as the ‘Great Southern Reef’

(Figure 1.1), a single interconnected entity spanning five states, adjacent to ~70% of

Australia’s population (Bennett et al., 2016). This recognition has greatly contributed to raising the profile of this important coastal ecosystem amongst the scientific community, government and the general public (National Marine Science Committee, 2015; Coleman &

Wernberg, 2017; Layton et al., 2020).

Figure 1.1 The Australian ‘Great Southern Reef’ (red line), an 8000 km strip of interconnected habitat, adapted from Bennett et al. (2016), Figure 1.

Like temperate reefs in other parts of the world, the Great Southern Reef ecosystem is characterised by dense stands of canopy-forming macroalgae, namely: Sargassum linearifolium (Coleman & Wernberg, 2017), Phyllospora comosa (Coleman & Kelaher, 2009) and especially the dominant habitat-forming species Ecklonia radiata (Wernberg et al.,

2019). These macroalgae provide essential habitat and contribute to the reef food web through herbivory and the production of detritus (Krumhansl & Scheibling, 2012; Bennett et al., 2016). Rates of carbon production from macroalgae on temperate reefs are among the highest of any ecosystem globally, with some locations exceeding 1000 g C m-2 yr-1 (Mann,

1973; Cebrian, 1999), greater even than the most productive wheat fields (Bennett et al.,

2016). Only a small proportion of this production (<20%) is consumed locally by herbivores.

Most energy in the world’s oceans is produced by benthic autotrophs (e.g. macroalgae, turfs, seagrasses, coral with symbiotic zooxanthellae etc.) and planktonic autotrophs (e.g. phytoplankton and photosynthetic bacteria), with the exception of hydrothermal vent ecosystems (Jannasch & Mottl, 1985) and some terrestrial inputs. Benthic primary productivity can be very high in temperate reef systems, mainly in the form of macroalgal growth (Cebrian, 1999), however, most local production (~80%) is converted to detritus and exported to neighbouring marine and terrestrial habitats (Krumhansl & Scheibling, 2012).

Benthic production may directly support a large portion of fish community biomass on some highly disturbed temperate reefs (Bennett et al., 2015), but the role of herbivores is more commonly occupied by invertebrates within temperate systems (Meekan & Choat, 1997). It is now generally accepted that pelagic subsidies may be equally, if not more important than local benthic production across temperate (Bulman et al., 2001; Truong et al., 2017) and even some tropical reefs (Morais & Bellwood, 2019; Skinner et al., 2021).

In some systems the energy provided by pelagic subsidies can exceed that provided by benthic primary production in fuelling benthic invertebrate communities and thus indirectly

supporting benthivorous fish (Docmac et al., 2017; Skinner et al., 2021). Reefs are supplied by currents through the pelagic pathway, with holoplankton produced over areas of ocean surface many times greater than the reef footprint, and with meroplankton produced upstream over the continental shelf (Zuercher & Galloway, 2019). In such systems coastal currents act like conveyor belts, delivering planktonic energy to fishes and invertebrates on reefs. In regions where currents are reliable, some reef ecosystems have evolved to become dependent on this steady supply of planktonic energy and propagules (Figure 1.2) (Zuercher

& Galloway, 2019).

At tropical coral reefs, the intense competition for resources has facilitated the speciation of fish into a diverse array of trophic niches (Tornabene et al., 2015; Siqueira et al., 2020).

Many species have evolved highly derived physiology to allow them to live off a wide variety of low-quality benthic algal sources (Floeter et al., 2004; Bellwood et al., 2014). A

large proportion of the energy entering tropical coral reefs is derived from direct consumption of algae by herbivorous fishes (Longo et al., 2019). At temperate rocky reefs, this level of specialised herbivory in fishes has not occurred to the same degree. In fact, very few temperate reef fishes have evolved the derived physiology required to live directly off benthic primary production alone (Floeter et al., 2004). Consequently, despite the high benthic productivity on temperate rocky reefs, much of this production must transfer through intermediate trophic steps before it becomes available to fish (Floeter et al., 2004).

Thus, it is the contribution of pelagic subsidies to temperate rocky reefs that allows them to support much greater fish biomass than if they were dependent solely on local benthic production (Truong et al., 2017).

Reef ecosystems benefit from this conveyor belt of phytoplankton and zooplankton through organisms which can capture and retain planktonic energy in place (Erez, 1990). This role is predominantly played by sessile filter feeders (Gili & Coma, 1998; Genin et al., 2009;

Monismith et al., 2010; Wyatt et al., 2013) and by zooplanktivorous fishes (Pinnegar et al.,

2007), which typically form large schools and dominate fish biomass and abundance around temperate rocky reefs (Floeter et al., 2004; Champion et al., 2015; Truong et al., 2017).

These fish are particularly abundant around topographically complex habitats featuring high vertical relief, due to the provision of additional refuges for avoiding predation and currents (Hixon & Beets, 1993; Johansen et al., 2008). Thus, topographically complex reefs receive much of their energy from pelagic subsidies (Figure 1.3). In addition to the essential ecosystem role these fish play in concentrating energy from zooplankton into a form that is accessible to higher predators, they also enrich the benthic substrate through their faeces

(Robertson, 1982; Pinnegar & Polunin, 2006), providing nutrients which facilitate benthic primary production and offer a source of energy for communities of benthic invertebrates

(Reeds et al., 2018).

Figure 1.3 Graphical abstract detailing the greater relative contribution of pelagic subsidies to reef food webs in reef habitats featuring high topography (complexity), from Morais and Bellwood (2019), Graphical Abstract. Notice that although pelagic subsidies are greater in the high topography habitat, the contribution of local benthic production is unaffected.

1.2. Broader spatiotemporal scales and tropicalisation

Tropical seas typically have lower planktonic primary productivity, and greater water column stratification than temperate systems (Brainerd & Gregg, 1997). As the shallow mixed layer mixes infrequently with deeper waters rich in nutrients, productivity is often low, but mostly consistent throughout the year. Despite this, tropical coral reefs still receive a large proportion of their energy from planktonic sources (Floeter et al., 2004; Morais &

Bellwood, 2019). Conversely, temperate coastal ecosystems often have higher pelagic primary productivity, and net productivity tends to increase with latitude (Everett et al.,

2014). This productivity is not distributed evenly throughout the year due to the more

seasonal nature of temperate environments (Longhurst, 2010). As a result of the colder winter temperatures experienced by temperate marine environments, the mixed layer increases in depth throughout the winter and stratification is reduced (Kara et al., 2000).

This facilitates enrichment of surface waters with nutrients and trace elements that are normally in limited supply, however, increased turbulence limits the time planktonic organisms spend within the photic zone and reduced temperatures impose physiological constraints on productivity (Edwards & Richardson, 2004; Petrou et al., 2016).

With the arrival of spring and the warming of surface waters, increased stratification leads to a shallowing of this mixed layer, allowing phytoplankton to spend more time in the photic zone and use this abundance of nutrients, leading to the algal spring bloom

(Rumyantseva et al., 2019). The massive abundance of phytoplankton produced during this period often far exceeds what can be consumed by zooplankton and higher trophic levels, and much of this production inevitably settles on the seafloor, enriching the benthos

(Parrish et al., 2009) and fuelling demersal trophic pathways (van Denderen et al., 2018).

Similar strong bentho-pelagic coupling is regularly documented in association with strong upwelling zones, such as northern Chile, where 98% of the energy consumed by benthivorous fish has a planktonic origin (Docmac et al., 2017). Around such upwelling zones, namely the southern Benguela, settlement of this excess production can be so high that it can generate an anoxic zone at the seafloor (Pitcher et al., 2014). Wind-driven upwelling is also a common occurrence along the east coast of Australia, particularly in summer when strong north-easterly winds drive deep water to the surface through Ekman transport (Roughan & Middleton, 2002).

Despite the greater net primary productivity in high latitudes, the constraints of cold temperatures on fish physiology, combined with seasonal paucities in winter (Harris et al.,

1987), may make it difficult for planktivorous organisms, including fishes, to persist in large

abundances. The exceptions of course are krill, which fulfil the niche typically occupied by zooplanktivorous fishes in polar and sub-polar regions (Pikitch et al., 2014), as they are able to undergo a period of starvation under sea ice throughout the long winter (Hagen et al.,

1996; Schaafsma et al., 2016). However, at lower latitudes the amplitude of seasonal cycles is much less and although net primary productivity is lower, it is generally more consistent.

This difference is likely a major contributor to the persistence and success of planktivorous fishes across tropical and temperate regions globally (Figure 1.4).

Figure 1.4 Simulated global distribution of pelagic forage fish biomass adapted from Petrik et al. (2019), Figure 7a. This simulation was generated using a spatially explicit mechanistic model of three fish functional types (forage, large pelagic, demersal), which the authors have named the FishErIes Size and functional TYpe model, or ‘FEISTY’. This model was forced with physical and plankton food web dynamics data from GFDL’s ESM2.6 high-resolution earth system model (Delworth et al., 2012). Forage fish biomass was greatest around coastal upwelling areas and oligotrophic gyres where secondary production was too low to support large populations of predators. Forage fish biomass was also generally lower in shelf seas, where they were vulnerable to both demersal and pelagic predators.Along the fringes of continents within these temperate regions, ocean gyres generate boundary currents that effectively supply planktonic energy to coastal ecosystems. Eastern boundary currents, including the California, Humboldt, Benguela and Canary Currents, carry cold water from polar regions towards the equator (Carr, 2001). These currents are associated with some of the world’s most significant upwelling zones, which support exceptionally high primary productivity (Philander & Yoon, 1982) and the world’s largest fisheries for planktivorous fishes (Pikitch et al., 2014).

Along the opposite sides of the ocean basins, western boundary currents are high velocity currents that carry warm, nutrient-poor water from equatorial regions towards the poles

(Wu et al., 2012). These include the Gulf Stream, the Brazil, Agulhas, Kuroshio, and the

East Australian Current. While these currents are generally less productive, they are instrumental in dispersing the larvae of fishes and invertebrates along continental margins and drive sporadic upwelling, delivering planktonic energy which fuels coastal food webs.

This is particularly the case around separation zones (Roughan & Middleton, 2002) and around capes (Aguiar et al., 2014), where these warm currents diverge eastward from continental margins, generating large upwelling zones downstream. They also facilitate the expansion of tropical species into temperate regions as climate change continues to increase ocean temperatures (Vergés et al., 2019).

In recent decades, temperate reefs near the warm edge of their distribution are becoming

‘tropicalised’, as many tropical species respond to warming by shifting their distribution towards cooler regions (Vergés et al., 2014a). Tropicalisation processes are particularly prominent along western boundary currents, which are intensifying in both velocity and temperature as a direct result of climate change (Wu et al., 2012). Within these current systems, the eggs and larvae of tropical marine organisms are transported to higher latitudes, where some individuals are able to persist and overwinter due to warming temperatures, and in some cases establish local populations (Johnson et al., 2011; Vergés et al., 2019).

This intrusion of tropical species can profoundly alter species interactions like herbivory, leading to overgrazing and regime shifts as canopy-forming kelp declines (Vergés et al.,

2016). Southeastern Australia is a climate change hotspot which is already experiencing substantial impacts driven by the East Australian Current (Sunday et al., 2015), with many species observed to be shifting their distributions poleward (Ling et al., 2009; Last et al.,

2011). In the coming years, tropicalisation will continue to alter temperate reef ecosystems from historic baselines, and in some cases lead to the creation of ‘no-analogue’ communities, unlike anything currently found in today’s oceans (Williams & Jackson, 2007).

1.3. Contribution of zooplanktivorous fishes to temperate coastal ecosystems

At more local scales, the current-driven delivery of zooplankton to reef systems drives patterns in the distribution of zooplanktivorous fishes (Hamner et al., 1988; Kingsford &

MacDiarmid, 1988; Morais & Bellwood, 2019), as they attempt to maximise their individual feeding rates while maintaining proximity to the refuge afforded by the reef itself

(Forrester, 1991; Motro et al., 2005). At coral reefs these zooplanktivorous fishes assemble along a gradient from the reef crest to the reef flat (Figure 1.5) and gradually strip the water of zooplankton along a predation gradient, forming what Hamner et al. (1988) famously referred to as the ‘Wall of Mouths’. Similar behaviour has been observed at temperate reefs, where eastern hulafish (Trachinops taeniatus) have been found to associate with locations containing higher densities of zooplankton (Gregson & Booth, 2005). At temperate reefs, schools of zooplanktivores may cause local depletions in zooplankton density (Kingsford & MacDiarmid, 1988; Gregson & Booth, 2005). Due to large differences in the geomorphology of tropical coral reefs versus temperate rocky reefs, these depletions tend to be more localised in nature and currents generally reconverge downstream of rocky reefs (Kingsford & MacDiarmid, 1988), rather than emptying into enclosed lagoons (Gal,

1993).

Figure 1.5 The provision of pelagic subsidies to a tropical coral reef, from Morais and Bellwood (2019), Figure 4a. Zooplanktivorous fishes assemble along the forereef zone and gradually strip water currents of zooplankton as they pass over the reef, boosting reef productivity.

Zooplanktivorous fishes tend to be either ‘reef-associated’ (e.g. Pomacentridae, Plesiopidae,

Pempherididae, Apogonidae), which obligately associate with and do not generally stray far from reef structure (Glasby & Kingsford, 1994; Champion et al., 2015), or ‘pelagic’, often referred to as ‘forage fish’ (e.g. Engraulidae, Clupeidae, Carangidae, Scombridae), which often live as large, dynamic schools in open water and form some of the most productive fisheries globally, often in association with major upwelling zones along eastern boundary current systems (Pikitch et al., 2014).

While forage fish exhibit pelagic tendency, some species (notably in the family Carangidae) form temporary associations with reefs (Malcolm et al., 2007; Scott et al., 2015). Owing to their mobility, these zooplanktivores are some of the first fishes to locate and colonise newly created benthic habitat, such as artificial reefs (Folpp et al., 2011; Lowry et al., 2014). While these fishes are not permanent members of the reef fish community, they maintain high densities around benthic structure when present (Champion et al., 2015; Becker et al., 2017)

and contribute significantly to reef food webs through providing food for predators and enriching sediments with their faeces (Robertson, 1982; Pinnegar & Polunin, 2006).

Zooplanktivorous fishes contribute energy to marine ecosystems through three main pathways: direct consumption by their predators (e.g. fish, seabirds, marine mammals, squid), extraction by humans (e.g. animal feed, fertiliser or direct consumption), and their contribution to predatory species which are extracted by humans (Pikitch et al., 2014). They also supplement benthic productivity indirectly through their waste (Robertson, 1982;

Pinnegar & Polunin, 2006). Per annum, these fish contribute an estimated US $16.9 billion to fisheries globally, accounting for ~20% of global catch value (Pikitch et al., 2014). Yet, despite the low trophic level they occupy, fisheries for zooplanktivores are vulnerable to overfishing and can experience stock collapse (Essington et al., 2015). Many marine ecosystems exhibit ‘wasp waist’ structure, with the zooplanktivore trophic level occupied by only a few species, despite greater species richness in the adjacent trophic levels (Bakun,

2006). In such systems, zooplanktivore stock collapse can have devastating consequences for predators. Both the breeding success of seabirds (Cury et al., 2011) and the abundance of predatory fish (Kaplan et al., 2013) have shown dramatic declines when forage fish abundance falls below certain thresholds as a result of overfishing. The link between overfishing and stock collapse is still hotly debated (Hilborn et al., 2018; Pikitch et al.,

2018), as there is evidence that in some cases bottom-up processes may play a more important role than human impacts (Hilborn et al., 2017).

The work of Truong et al. (2017) recognised the contribution of reef-associated and pelagic zooplanktivores that visit reefs in large abundances. They synthesised citizen science underwater visual census data with their own field studies to determine the relative biomass of twelve functional groups across the Sydney region (Figure 1.6). They combined this relative biomass data with dietary information to conduct a food web analysis, allowing

them to calculate the relative contribution of benthic and planktonic energy sources to temperate rocky reefs. They found that zooplanktivores dominated reef fish communities, contributing on average 41% of total fish biomass. Their food web analysis also revealed that, on average, 56% of total fish biomass was supported by plankton. One major objective for this thesis is to expand upon their work and to understand more about the latitudinal distribution of these fishes and how their distribution varies across a much larger spatial scale along the East Australian Current.

It is well established that herbivorous fishes abound at tropical coral reefs (Longo et al.,

2019; Vergés et al., 2019), but it has not yet been established how a transition from

herbivores to zooplanktivores might occur along a gradient from tropical to temperate latitudes. This may be at least partially attributable to the greater difficulty associated with monitoring zooplanktivores in situ (Bellwood et al., 2019), as they are typically small, highly abundant and dynamic, and constantly shift positions in the water column. Because of this, they are more difficult to observe with underwater visual census and remote underwater video methods. This discrepancy has resulted in disproportionately few studies of reef associated zooplanktivores relative to their high diversity and abundance on reefs

(Figure 1.7). As climate change continues to impact temperate marine ecosystems, through warming water temperatures and the poleward expansion of tropical species altering habitats in their wake (Vergés et al., 2019), it is increasingly important to understand present day distributions and how they will likely shift in the future (Hobday, 2011).

1.4. What distribution can reveal about behaviour

The causes of variation in the local abundance and distribution of organisms is a fundamental concern for terrestrial (Andrewartha & Birch, 1954) and marine ecology

(Hedgpeth, 1957). Emergent patterns in the distribution of organisms at multiple spatial and temporal scales can reveal underlying aspects of the evolutionary and physiological

requirements which ultimately drive their behaviour (Bertrand et al., 2014). When multiple coexisting species share similar requirements, their species identity can often be ignored and they can be categorised as a functional group (Villéger et al., 2017). Examining marine ecosystems through the lens of functional ecology allows us to resolve a mechanistic understanding of physiological processes and requirements, and reduces the emphasis on discrete characteristics of individual organisms (Rosenfeld, 2002).

Patterns in the distribution of individuals sharing common traits and requirements reveal adaptations which have likely spread across species through their impacts on evolutionary fitness (Hamner, 1995). For example, a relatively narrow range of thermal tolerance is a common trait across endothermic organisms, including fish (Brett, 1956). Thermal range can be determined through laboratory experiments, but also in situ through examining the distribution of fish relative to the distribution of water temperatures they occupy (Ferguson,

1958). This forms the foundation of species distribution modelling (Brodie et al., 2017;

Phillips et al., 2017). Combining fisheries catch rates with satellite remote sensing reveals that many commercially important fish species show strong affinities for specific temperature and chlorophyll regimes (Klemas, 2013). Similarly, examining fish distributions in associations with static environmental variables such as bathymetry, benthic cover and substrate type can reveal habitat requirements at the level of species or functional groups, although benthic habitat mapping is more typically used to predict simplistic metrics such as species richness (Diaz et al., 2004; Pittman et al., 2007). Additionally, frequent co- occurrences of fish with other organisms can reveal details of community composition, and bottom-up and top-down controls on trophic groups (Madin et al., 2020). Thus, investigating such patterns in the distribution of fishes at multiple spatial and temporal scales in association with other organisms and environmental variables can reveal fundamental ecological aspects of both group and individual behaviour.

A good example of group behaviour is the tendency for conspecific groups of fish to form high density local aggregations, otherwise known as shoaling. Shoaling is a form of group living, a behaviour which has emerged many times through convergent evolution from an ancestral tendency for solitary living (Hamner, 1995). The primary benefits include reduced individual risk from predation, improved ability to locate food, and a reduction in energy expenditure (Pitcher, 1986; Marras et al., 2015). The term ‘schooling’ refers to a specific type of shoaling behaviour, when individual fish swim in the same direction in a coordinated manner (Shaw, 1962). The significant degree of group cohesion and information transfer

(Marras et al., 2012) occurring within some fish schools suggests they can be viewed as distinct biological entities, rather than simply as collections of individual organisms (Vabø

& Nøttestad, 1997). In such cases of high information transfer, the ideal free distribution

(IFD) theory provides a foundation for predicting the distribution of animals that forage in groups, rather than as individuals (Tregenza, 1995).

Within IFD models, foragers have a choice of multiple prey patches with varying levels of quality (e.g. prey density). In this case, foragers can freely move between patches at no energetic cost and have perfect knowledge of the quality and location of patches

(Matsumura et al., 2010). They also share prey equally among members foraging within the same patch. In such cases, stable distribution can only occur if the individual rate of prey consumption is even across all patches, otherwise there will be a tendency for individuals to converge on a higher quality patch until equilibrium in prey allocation is achieved

(Stephens, 2008). This theory, based on the economics of foraging, holds up in some cases in nature (Stephens, 2008), although it is also often the case that foragers overuse low quality patches and underuse high quality patches due to a lack of knowledge and an actual cost associated with travel among sites (Kennedy & Gray, 1993; Tregenza, 1995). Still, such an

idealised framework provides a foundation for what drives the distribution of organisms that forage in groups, including schooling forage fish.

The distribution of groups of animals and the distribution of individuals within groups can reveal separate drivers of animal behaviour. We can draw conclusions related to physiological requirements and habitat selection by examining the distribution of schools over large geographic areas and we can infer factors driving the behaviours of individual fish competing within schools by studying features of schools themselves. A detailed understanding of the spatial distribution of organisms is essential for identifying priority areas for conservation (Malcolm et al., 2016), and is also vital to informing effective artificial reef designs for enhancing fisheries (Taylor et al., 2017).

1.5. Enhancing fisheries with artificial reefs

Fisheries enhancement is a useful approach for managing the productivity of heavily exploited fisheries. Enhancement in this case refers to human interventions that can result in improvements to stock recruitment and productivity, through restocking, fishery closures, marine protected areas, and habitat restoration (Taylor et al., 2017). The design and deployment of artificial reefs (Becker et al., 2018) and re-use of decommissioned offshore oil and gas infrastructure are also common practices carried out often with the intention to enhance fisheries (Jørgensen et al., 2002; Claisse et al., 2014). To understand the effectiveness of artificial reef projects it is vital for managers to monitor the successional processes of colonisation by marine organisms and the development of the local fish and invertebrate community (Baine, 2001). To judge the effectiveness of a design, managers must also understand how various species use the space around a structure and how the structure affects fish behaviour (Becker et al., 2019).

The earliest recorded artificial reefs were deployed in Japan over 300 years ago simply as a means of attracting fish, thus making them easier to harvest (Thierry, 1988). More recently there has been substantial research into optimising artificial reef design and the ‘attraction versus production’ debate (Bohnsack & Sutherland, 1985; Pickering & Whitmarsh, 1997;

Smith et al., 2016). This ongoing debate focuses on whether artificial reefs actually enhance the production within a system, or whether they simply aggregate fish from adjacent natural habitats, thus making them easier to harvest and potentially overexploit (Pickering

& Whitmarsh, 1997). Although there is still no scientific consensus on this debate, it is now generally accepted that the design (Sherman et al., 2002) and location (Strelcheck et al.,

2005) of artificial reefs both play an important role. It may be possible to optimise both design and location to increase the influence of pelagic subsidies in sustaining local secondary production at artificial reefs.

Artificial reef research has revealed promising results around the potential for artificial reefs to enhance local production and improve fisheries efficiency and sustainability (Komyakova et al., 2019). However, owing to the many complexities of artificial reef design, including size, dimensions, materials, region and location, well-structured comparative studies are rare (Becker et al., 2018). Further, long-term monitoring over the time frames necessary to document successional processes (>10 years; Becker et al., 2018) is expensive and results are not widely applicable to other projects due to common issues of pseudoreplication (Bortone,

2006). Thus, most reefs are typically only monitored for short periods of time after deployment, and often no baseline data is gathered prior to deployment (Becker et al., 2018).

Consequently, the long-term impact and effectiveness of many artificial reef projects goes unknown. To help address this issue and encourage longer-term monitoring post- deployment it may be beneficial to develop faster, more efficient methods for the regular monitoring of fish distribution and community composition.

1.6. Modern methods for studying fish distribution

There are several methods commonly used to determine the makeup of fish communities and monitor fish distribution in the vicinity of natural and artificial benthic structures.

These methods typically involve the use of gillnets (Santos & Monteiro, 1997; Creque et al.,

2006), optical sensors (Becker et al., 2019) or acoustic sensors (Sala et al., 2007). The most commonly used methods, including underwater visual census (UVC) and diver operated video (DOV) (Lowry et al., 2012; Lowry et al., 2014; Davis & Smith, 2017), remote underwater video (RUV) (Champion et al., 2015; Scott et al., 2015; Becker et al., 2019) and remotely operated vehicles (ROV) (Ajemian et al., 2015a; Ajemian et al., 2015b; McLean et al., 2017), rely on optical sensors which must be operated in close proximity to subjects to be effective. These methods have been shown to affect fish behaviour and distribution, which can lead to biased results (Holmes et al., 2013; Schramm et al., 2020). These methods are all reliant on water clarity and light availability, making them difficult to apply in turbid or dark conditions without introducing artificial light into the environment (McLean et al.,

2019; Sheehan et al., 2020), which can affect fish behaviour (Becker et al., 2013). They also generally require manual analysis of many hours of video footage, although modern developments in machine learning have the potential to improve efficiency (Marini et al.,

2018; Sheehan et al., 2020).

Alternatively, light-independent methods, such as acoustic telemetry, are increasingly used to monitor variation in the distribution of fish around benthic structures with high spatiotemporal resolution (Jørgensen et al., 2002; Lowry et al., 2017). Unlike visual methods, acoustic telemetry is as effective at night as it is during the day. However, due to the high cost of transmitters and installation of receiver arrays, this method is generally only useful for fishes with high site fidelity and sufficient body size to facilitate viability after implantation (Reese Robillard et al., 2015). Fisheries acoustics methods employing

echosounders offer an alternative which can be operated from a vessel, at a greater distance from subjects, with less potential to introduce bias by affecting fish distribution and behaviour. Although vessel avoidance is an issue associated with fisheries acoustics (Vabø et al., 2002), acoustic surveys are generally more efficient and require minimal running costs, apart from the initial outlay on instrumentation.

Echosounders function by using a transceiver to generate electrical signals, which are sent to a submerged transducer which transmits a ping of acoustic energy, or sound, through the water column (MacLennan & Simmonds, 2013). When this ping encounters an object in its path which possesses a density differing from the surrounding water, such as a fish or the seafloor, some acoustic energy is reflected back to the transducer. The transducer detects the returned ping and converts it back to an electrical signal which is sent to the transceiver. The transceiver then calculates the time elapsed between signal transmission and receipt to determine the range to the reflective object, based on the speed of sound through water (MacLennan & Simmonds, 2013). When operated from a moving vessel with

GPS, individual pings can be stitched together to form a profile of the water column along the vessel track.

One of the main drawbacks of acoustic methods is that it can be difficult to determine which species are present (Korneliussen et al., 2009). This must be ascertained independently through local knowledge, fishery catch data, video methods, or through supplemental fishing operations (Hwang et al., 2004). With species-level information and length- frequency distributions, traditional single-beam or split-beam transceivers can provide accurate estimates of fish distribution when parallel transects are conducted over large geographic areas of pelagic ocean (Bailey & Simmonds, 1990). Narrow beam-widths limit the horizontal resolution provided by these instruments, so traditional echosounders are not well-suited for detecting fine-scale three-dimensional patterns in fish school characteristics

and distribution around benthic structures (Paramo et al., 2007). The superior horizontal resolution provided by multibeam echosounders makes them better-suited for measuring such fine-scale variation (Gerlotto et al., 1999; Colbo et al., 2014).

Multibeam echosounders were first developed for seafloor mapping in the 1960s by the

United States Navy. This technology remained classified until the 1970s when a lower- resolution version was released for civil use, and by the 1980s there was strong uptake of this revolutionary technology among oceanographic research vessels (Theberge & Cherkis,

2013). With the completion of the Global Positioning System in the 1990s, the spatial reckoning of multibeam echosounders was dramatically improved and recent advances in computer processing power, data storage and visualisation have further increased their utility (Theberge & Cherkis, 2013). Over time multibeam echosounders have become more affordable and today there are a range of portable models which are marketed towards high- end recreational boat owners, commercial fishers, and researchers operating smaller, low- cost trailer vessels for applications in coastal areas and shallow water.

Unlike traditional echosounders, multibeam echosounders use a technique known as beamforming, which allows for the directional partitioning of the transmitted and received acoustic energy through the constructive and destructive interference of sound waves (Jung et al., 2018). They can be used to provide near complete coverage of the seafloor and the adjacent epi-benthic zone, something no other instrument or technique can currently achieve (Aziz et al., 2018). Further, they have the added advantage that bathymetry and seafloor backscatter can be detected concurrent with measurements of fish in the water column (Figure 1.8). This seafloor data can be used for habitat mapping and direct comparison of seafloor features with the midwater targets that co-occur (Mackinson et al.,

2004).

Due to the various drawbacks associated with existing methods, there is clearly a need to develop better methods for monitoring the distribution of shoaling fishes around reefs and artificial structures. The recent availability of affordable multibeam echosounders provides an opportunity to develop such methods. Multibeam echosounders are particularly suited to studying the distribution of multi-species schools of reef fish associated with benthic structures. In such cases, target strength can vary widely across species within a multispecies school (Bakhtiar et al., 2020), requiring a single midwater remote underwater video deployment to determine the dominant species present (Hwang et al., 2004).

1.7. Thesis structure and aims

Specifically, I aimed to:

1. Quantify regional scale patterns in the distribution of reef-associated

zooplanktivores and other fish trophic groups across southeastern Australia;

2. Investigate the effects of environmental and biological drivers on the distribution of

pelagic zooplanktivorous fishes within a highly dynamic continental shelf ecosystem;

3. Develop new methods for applying modern consumer-grade multibeam

echosounders to study patterns in the distribution of schooling reef fish;

4. Apply the above multibeam echosounder methods to study variation in the

distribution of shoaling reef fish at natural and artificial reefs in relation to current

flow and between night and day.

These aims will contribute towards a better understanding of the relative contributions of planktonic subsidies, and more specifically zooplanktivorous fishes, to a temperate western boundary current ecosystem. This thesis will also contribute towards informing the deployment of future artificial reefs and will provide more efficient methods to facilitate their monitoring with increased temporal resolution and duration. Following the overall aim for this thesis, I had two broad expectations spanning regional and local scales. First, I expected the regional distribution of coastal zooplanktivores would be driven by the sustained influence of pelagic subsidies and their ability to support long-term population viability; Second, locally across reef habitats, the distribution of zooplanktivores would be driven by the collective behaviour of fish attempting to maximise their individual fitness.

In Chapter 2, I analysed a high-resolution dataset of citizen science UVC surveys to uncover trends in trophic composition and seasonality of reef fishes across ~15 degrees of latitude along the coast of southeastern Australia. I used findings from this study to predict how modern-day patterns in trophic structure could shift as ocean temperatures continue to rise.

The manuscript for this chapter has been published in Fish and Fisheries and is cited in this thesis as Holland et al. (2020b):

Holland, M. M., Smith, J. A., Everett, J. D., Vergés, A., & Suthers, I. M. (2020b). Latitudinal patterns in trophic structure of temperate reef‐associated fishes and predicted consequences of climate change. Fish and Fisheries, 21(6), 1092-1108. https://doi.org/10.1111/faf.12488

In Chapter 3, I used data collected in situ by a state-of-the-art blue-water research ship to document spatial patterns in the distribution of zooplankton and pelagic zooplanktivores around an island in a dynamic shelf ecosystem. I supplemented this analysis with long-term satellite data and fishing records to highlight a particularly dynamic ecosystem that is subject to regularly alternating influence of the East Australian Current and the Tasman

Sea. I expected the distribution of zooplanktivore density to follow IFD theory, with higher densities of fish associated with areas containing high zooplankton density. The manuscript for this chapter has been published in Estuarine, Coastal and Shelf Science and is cited in this thesis as Holland et al. (2020a):

Holland, M. M., Everett, J. D., Cox, M. J., Doblin, M. A., & Suthers, I. M. (2020a). Pelagic forage fish distribution in a dynamic shelf ecosystem – thermal demands and zooplankton prey distribution. Estuarine, Coastal and Shelf Science, 107074. https://doi.org/10.1016/j.ecss.2020.107074

In Chapter Error! Reference source not found., I developed new methods for applying m odern consumer-grade multibeam echosounders to study the distribution of schooling fish around reefs. This method is designed to be more robust to noise, seafloor misclassification and multi-species schools. Rather than relying on traditional methods, which integrate the acoustic energy contained within volumes of water, I quantified the vertical height of fish schools over a horizontal grid. I then applied this method to predict the distribution of schooling fish around an artificial reef. The manuscript for this chapter has been accepted

(16 March 2021) and is in press in Limnology and Oceanography: Methods:

Holland, M. M., Becker, A, Smith, J. A, Everett, J. D., & Suthers, I. M. (In press). Characterising the 3D distribution of schooling reef fish with a portable multibeam echosounder. Limnology & Oceanography: Methods. Accepted 16 March 2021.

In Chapter 5, I applied the methods developed in Chapter 4 to measure variation in the distribution of schooling fish around natural and artificial reefs in response to current direction. I also measured differences in the distribution of schooling fish between natural and artificial reefs, and between night and day. These observed differences in school distribution and school characteristics were then applied to inform improved designs for artificial reefs in the study region. This chapter is the largest because two studies were combined due to fieldwork restrictions in the early months of the COVID-19 pandemic.

Chapter 6 is used to synthesise the findings of the four data chapters (Chapters 2 to 5) into their broad implications, along with presenting potential directions for future research.

The four data chapters have been written as independent manuscripts to prioritise publication in peer-reviewed journals. As such, there will be some repetition among chapters. Literature cited has been assembled as a single list at the end of the document to avoid duplication (Chapter 7). Similarly, supplementary material for each chapter has been combined into an appendices section at the end of this thesis (Chapter 8). The electronic version of this thesis contains a clickable table of contents, clickable internal links and DOIs to facilitate navigation.

2. Latitudinal patterns in trophic structure of temperate reef-associated fishes and predicted consequences of climate change

The manuscript for this chapter has been published in Fish and Fisheries and is cited in this thesis as Holland et al. (2020b). The reference is:

Holland, M. M., Smith, J. A., Everett, J. D., Vergés, A., & Suthers, I. M. (2020). Latitudinal patterns in trophic structure of temperate reef‐associated fishes and predicted consequences of climate change. Fish and Fisheries, 21(6), 1092-1108. https://doi.org/10.1111/faf.12488

2.1. Abstract

Some dramatic consequences of climate change are caused by shifting species interactions and associated changes to trophic structure and energy flow. In coastal ecosystems, the relative abundance of feeding guilds indicates dominant energy sources sustaining food webs. Here, we use a space-for-time substitution to investigate potential climate change impacts on trophic structure and energy flow in reef fish communities. We investigated latitudinal and seasonal patterns in the biomass distribution of five trophic groups across subtropical to temperate latitudes (29 to 44°S) in eastern Australia. Along western boundary currents, temperatures are increasing up to three times faster than the global average, making them ideal for studying climate change impacts. Using 10 years of Reef

Life Survey data, we investigated potential determinants of fish biomass and community composition with generalised additive mixed models. Biomass decreased towards higher latitudes, from 220 g m-2 in the subtropics to 13 g m-2 in the south. Dominant trophic group also changed latitudinally, with herbivores and omnivores dominating lower latitudes

(~30°S), zooplanktivores at mid-latitudes (~35°S) and benthic invertivores at higher latitudes (~40°S). Biomass varied seasonally, with lower latitudes experiencing a 3.2-fold increase between spring and autumn, while variation at higher latitudes was 1.9-fold. We

found strong evidence that factors linked to latitude and seasonality are important determinants in the distribution of fish trophic structure. As climate-driven species redistributions accelerate in the 21st century, expected poleward shifts in trophic structure include overall increases in fish biomass linked to enhanced herbivory at mid-latitudes and increased planktivory at higher latitudes.

Keywords: citizen science, planktivory, reef fish, Reef Life Survey, rocky reefs, trophic composition

2.2. Introduction

Climate change has impacted all ecosystems on Earth, despite an average warming of only

~1C so far (Scheffers et al., 2016; Pecl et al., 2017). One of the most widely documented impacts of warming is the global redistribution of species (Parmesan, 2006; Poloczanska et al., 2013). To stay within their preferred thermal ranges, many species are moving towards the poles, to greater altitudes on land, and into deeper waters in the ocean (Chen et al.,

2011; Pinsky et al., 2013). This redistribution of species is leading to new biological interactions between previously separated species, disrupting trophic structures and altering food webs (Scheffers et al., 2016).

Species redistributions are happening particularly fast in the ocean, with marine species shifting, on average, at least four times faster than on land (Poloczanska et al., 2013). These rapid shifts have already led to profound disruptions, especially on shallow temperate reefs, which are highly productive (some exceed 1000 g C m-2 yr-1 (Mann, 1973; Cebrian, 1999)) and have major economic and intrinsic benefits to their adjacent towns and cities (Bennett et al., 2016; Blamey & Bolton, 2018).

In coastal marine environments, the relative importance of herbivory and planktivory provides important information about dominant energy pathways (Truong et al., 2017;

Morais & Bellwood, 2019). Fish communities in temperate reefs are mostly underpinned by planktonic energy sources, with local benthic primary production being a minor contributor to local food webs (Truong et al., 2017). In these systems, seasonal fluxes of planktonic primary productivity can enrich sediments and fuel benthic food webs indirectly as excess organic material sinks to the seafloor (Heip, 1995; Wassmann, 1997; Parrish et al., 2009).

However, species redistributions due to climate change have increased the dominance of tropical species on temperate reefs, fundamentally altering species interactions and increasing the relative importance of herbivory (Vergés et al., 2014a; Vergés et al., 2016). It has been proposed that these climate-mediated shifts in energy flow from low to high herbivory can have important ecosystem function implications as a higher proportion of primary production is incorporated into higher trophic levels (Vergés et al., 2019). Such consequences of tropicalisation, however, remain untested.

‘Space-for-time’ is a widely used approach to predict future trajectories of ecological systems based on present-day patterns, which is underpinned by the assumption that drivers of spatial gradients are similar to drivers of temporal changes (Blois et al., 2013; Elmendorf et al., 2015). These approaches often use latitudinal gradients to predict the effects of warming, as temperature is a key environmental variable that varies predictably with latitude (Wogan

& Wang, 2018). Here, we examine latitudinal patterns in the trophic structure of fish communities and how reef-fish functional biomass is spatially distributed to infer potential future changes in energy flow and fish biomass.

Previous studies measuring latitudinal patterns in reef associated fish communities have tended to concentrate on specific trophic processes such as herbivory (Meekan & Choat,

1997; Floeter et al., 2005) and predation (Barnes et al., 2010; Freestone et al., 2011), while only a few have examined patterns in overall trophic structure (Floeter et al., 2004; Longo et al., 2019). Although zooplanktivores are the most speciose fish groups on many reefs

(Bellwood et al., 2019; Morais & Bellwood, 2019) and can make up over 40% of fish biomass on temperate rocky reefs (Truong et al., 2017), we know little about large-scale latitudinal patterns of planktivory across the tropical to temperate interface of nearshore reefs. The distinct lack of studies may be partially attributed to the greater difficulty associated with observing zooplanktivores in situ (Bellwood et al., 2019).

Here, we use eastern Australia’s temperate rocky reefs as a model system to uncover large- scale latitudinal patterns in trophic structure and total fish biomass, in a region strongly influenced by the East Australian Current (EAC) (Suthers et al., 2011). Regions influenced by western boundary currents like the EAC, i.e. eastern Japan (Kuroshio Current), eastern

USA (Gulf Stream), northern Brazil (Brazil Current), and southeastern Africa (Agulhas

Current), are climate change hotspots where waters are warming faster than the global average (Wu et al., 2012). Western boundary currents facilitate the poleward dispersion of warm water species (Vergés et al., 2014a) such as tropical herbivores (Figure 2.1), making them ideal sentinel ecosystems to understand and predict the impacts of climate change and associated poleward shifts in species distributions.

The dispersal of tropical fishes into temperate waters is not limited to western boundary currents and is occurring rapidly in other regions as well, leading to novel ‘tropicalisation gradients’. For example, tropical herbivores have dispersed into the Mediterranean Sea from the Suez Canal and have established local populations, leading to a longitudinal gradient in herbivory (Vergés et al., 2014b). The Leeuwin Current, an exceptional boundary current which flows poleward along the west coast of Australia, has also experienced recent tropicalisation of fish communities in association with marine heat waves (Wernberg et al.,

2013), affecting temperate ecosystems where planktivory may normally dominate.

Figure 2.1 Variation in the proportional representation of herbivores and zooplanktivores (sum = 1) from select studies examining reef fish trophic structure along environmental gradients using three metrics: relative fish abundance (square symbols, Holmes et al., 2013), relative fish biomass (circle symbols, Sala et al., 2012; Truong et al., 2017) and relative feeding pressure (triangle symbols, Longo et al., 2019). Results from this study are presented on the far right. Major tropicalisation gradients associated with boundary currents (solid red) and with invasion processes derived from the building of the Suez Canal (dashed red) are also indicated (methods in 8.2.1 Supplementary methods). Bubbles to the right of the map depict a conceptual model of temperate reef fish trophic group dominance and biomass distribution. The lower (most equatorial) latitude sites are dominated more by herbivores and omnivores and have high total biomass, low phytoplankton biomass and low zooplankton abundance. Mid-latitudes are dominated by zooplanktivores with high-moderate total biomass and productivity and the greatest zooplankton abundance, and high (more polar) latitudes are dominated by benthic invertivores with generally lower total biomass, while zooplankton abundance is low and much of phytoplankton productivity settles on the seafloor as a result of seasonal blooms.

We quantified how the biomass and structure of rocky reef associated fish trophic groups varies over a latitudinal gradient across 1800 km and 16 degrees of latitude. Although latitude itself cannot be a causal factor shaping trophic structure, it indicates covarying environmental variables (such as temperature) that may be important determinants. We accounted for regional and local variables that may influence trophic structure and overall fish biomass by evaluating chlorophyll a, zooplankton abundance, and human population density, as well as non-dynamic variables such as depth. As temperature also shifts predictably with season and because seasonality is likely to be more influential at higher latitudes, we also quantified seasonal patterns in fish trophic structure and biomass.

We hypothesised that with increasing latitude trophic composition would shift away from herbivore dominance in favour of trophic pathways that derive energy from plankton and benthic invertebrates (Figure 2.1). We predicted that these changes in trophic structure would be linked to lower overall fish biomass in higher latitudes, as a lower proportion of local benthic primary production is consumed (Vergés et al., 2019) and because of greater inter-seasonal variability in plankton production. At a regional level, we hypothesised that the amount of energy available as plankton would influence trophic group dominance, with the biomass of zooplanktivores linked positively to the abundance of zooplankton and/or phytoplankton. At a local level, we expected declines in the biomass of some trophic groups with high human population density (due to human impacts such as fishing) and we expected trophic group biomass to be linked to the depth of sites sampled (due to influences of depth on light availability and exposure to pelagic subsidies). Finally, we discuss how climate-mediated poleward shifts of the observed distributions may alter the trophic ecology of temperate rocky reef communities during this century.

2.3. Methods

2.3.1. Study region

Along the southeast coast of Australia, shallow rocky reefs range from urchin barrens to sponge gardens to outcrops dominated by dense stands of canopy-forming macroalgae

(Bennett et al., 2016), particularly the laminarian kelp Ecklonia radiata (Bennett et al., 2016;

Wernberg et al., 2019). Collectively, these reefs combine to form a single entity spanning thousands of kilometres and connected by processes of the EAC (Roughan & Middleton,

2004; Suthers et al., 2011).

Water temperatures here can range from 10 to 27 °C between the highest and lowest latitudes, respectively, generally peaking in March (early autumn) and reaching their

minimum in September (early spring). Current-driven localised upwellings are prevalent, particularly in spring and summer, facilitated by coastal winds (Roughan & Middleton,

2002, 2004). Upwelling events tend to be smaller and more episodic at lower latitudes and much larger at higher latitudes with the onset of spring (Everett et al., 2014). This spring upwelling is the driver of overall greater mean primary productivity at higher latitudes, despite the very low productivity of these waters in winter (Everett et al., 2014).

2.3.2. Biomass of fish trophic groups

Fish count data were sourced from the Reef Life Survey (RLS)

(https://reeflifesurvey.com/survey-data/, Accessed: 22/11/2017), a global dataset of systematic aquatic biodiversity surveys conducted by trained recreational divers in collaboration with experienced marine ecologists (Edgar & Stuart-Smith, 2014). The RLS follows a rigorous survey methodology globally, ensuring sites can be compared across large geographic and temporal extents. In this analysis, we combine RLS data with environmental and biological records across the same date range.

Survey data were collected from 3032 RLS surveys (between 01/2008 – 10/2017) across

567 sites (Figure 2.2). All sites were south of ~29°S, across 16 degrees of latitude, spanning the subtropical to temperate east coast of Australia, within an area commonly referred to as the ‘Great Southern Reef’ (Bennett et al., 2016). Each survey consists of a single 50 m transect along a constant isobath. After laying the measuring tape along the transect, divers spend a few minutes preparing equipment (~5 minutes) providing an acclimation period for diver effects on fish to subside (Dickens et al., 2011). Subsequently, two divers swim the length of the tape and record the identity, abundance and length-class of all fishes observed within 5 m either side of the transect. Our analysis of these surveys was limited to non- cryptic teleost fishes, and to taxa observed in at least 5% of surveys. Fish biomass for each

survey was calculated by RLS from fish counts, using observed fish total length and species- specific length-weight relationship variables available in FishBase (Froese & Pauly, 2009).

Figure 2.2 Chart of the southeast coast Australia, with black tick marks representing the 567 Reef Life Survey (RLS) sites.

For the purposes of our analysis, it was essential that fish detectability remained consistent across latitude. The effects of divers on fish detectability have been studied in tropical

(MacNeil et al., 2008; Dickens et al., 2011) and temperate systems (Edgar et al., 2004;

Watson et al., 2005; Watson & Harvey, 2007; Holmes et al., 2013). Although diver effects have been recorded for most families studied (Dickens et al., 2011), the fish most susceptible to variation in detectability generally include larger predators, chondrichthyans and cryptobenthic species (Brandl et al., 2019), which we have excluded from our analyses to avoid potential bias. Although large predators can dominate fish biomass at remote isolated reefs (Stevenson et al., 2007), their contribution to total fish biomass is often much less at reefs exposed to fishing and human disturbance (Valdivia et al., 2017), which are both prevalent factors along the heavily populated coastline of southeast Australia. Further, the primary reason for limiting our study to latitudes greater than 29 degrees south was to restrict our focus to only macroalgae-dominated rocky reefs and avoid the increased complexity of coral-dominated tropical sites. Therefore, we do not believe a latitudinal bias in detectability exists in the analysed data.

There are certainly sources of bias associated with underwater visual census that can affect the estimation of fish lengths, abundances and ultimately biomass (Edgar et al., 2004;

Harvey et al., 2004). However, it is important to note that due to rigorous consistency in survey methods across the RLS program, these biases should be consistent across locations and times. Our analyses were only concerned with comparisons of relative biomass and as such, absolute biomass values were not necessary for survey results to be comparable.

To evaluate comparability of surveys across latitude and seasons, we compared site characteristics across our study domain. We tested whether survey depth, swell exposure or visibility, differed by season or four-degree latitudinal bin, using two-factor ANOVA

(season and latitude bin) and Tukey’s post hoc tests in R (R Core Team, 2020). Differences

in visibility would be of particular concern, as they strongly influence fish detectability, particularly for pelagic species that inhabit the water column around reefs (Figueroa-Pico et al., 2020). Swell exposure was determined using the ‘dist2Line’ function from the R package

‘geosphere’ (Hijmans et al., 2020) to locate the nearest point to each site on a polygon shapefile of the Australian coast (Whiteway, 2009) and subsequently calculate the bearing between each point with its corresponding site. We then extracted mean wave direction per month by site coordinates from the Australian Wave Energy Atlas (Hemer et al., 2018) and rescaled their alignment with site aspect from 0 to 1, similar to Turnbull et al. (2018).

Visibility data were diver estimates (in m) and were only available for 1820 out of the 3032 surveys. Although some pairwise differences in depth (max: 3.26 m), swell exposure (max:

0.07) and visibility (max: 3.48 m) were statistically significant, they were all very small and unlikely to be ecologically meaningful (Supplementary Table 8.2.1, Supplementary Figure

8.2.1).

Photo quadrats, which contained 36 categories for benthic cover, were only available for 202 out of 567 sites. They were used to calculate a matrix of proportionate benthic cover using the R package ‘vegan’ (Oksanen et al., 2019). These results were then compared visually with non-metric multidimensional scaling (NMDS) and bar charts (Supplementary Figure

8.2.2, Supplementary Figure 8.2.3), and statistically using the ‘adonis’ function (Oksanen et al., 2019), which computes a two-factor permutational MANOVA, and post-hoc comparisons of centroid distance were made using the ‘betadisper’ function. As sites spanned ~16 ° of latitude, differences in benthic composition across latitude were to be expected (max: 0.18), while comparatively minimal variation across seasons was observed between winter and spring (max: 0.08) (Supplementary Table 8.2.1).

To further evaluate comparability of sites and surveys, we tabulated the number of surveys for each season and year by site (Supplementary Table 8.2.2, Supplementary Table 8.2.3,

respectively) and by one-degree latitudinal bands (Supplementary Figure 8.2.4,

Supplementary Figure 8.2.5, respectively). Besides apparent under-sampling between 38 to

39°S, and at latitudes higher than 37°S in winter and spring, there was no distinct seasonal bias to temporal survey distribution across the nearly ten-year period of survey data. This lack of winter and spring data precludes any seasonal conclusions for high latitudes. Most latitudes had regular survey coverage over the ten-year period, except for between 38 to

42°S, coinciding with the Bass Strait.

2.3.3. Trophic classification

A species list of 163 fish species (all taxa observed in ≥5% of surveys) and trophic group classifications was generated, with the use of data from FishBase (Froese & Pauly, 2009),

Fishes of Australia (Bray & Gomon, 2018) and existing classifications made by Truong et al.

(2017), using a similar methodology to Soler et al. (2015) which classified fishes into trophic groups for a global analysis of RLS data (Supplementary Table 8.2.4). Taxa were allocated to one of five categories: zooplanktivores (feed primarily on zooplankton); herbivores (feed primarily on algae and/or seagrass); omnivores (have algae and/or seagrass as a minor component of their diet); benthic invertivores (carnivorous species that feed on zoobenthos but generally not fishes); and piscivores (carnivorous species that feed primarily on other fishes). This species list was then used to classify fishes in the biomass database, and biomass totals were summed among taxa to calculate the biomass in each trophic group for each survey.

2.3.4. Characterising latitudinal trends

To examine generally how trophic composition varied across latitude, we analysed multiple metrics: total fish biomass, trophic group biomass, and taxa-level biomass. We conducted linear regression to test whether total fish biomass broadly varied with latitude. The study

region was then divided into 1-degree latitudinal bins and mean observed total fish biomass per survey was calculated for each bin. Similarly, the mean proportion of biomass for each of the five fish trophic groups was calculated for each latitudinal bin which were used to construct bar plots of biomass and trophic composition by latitude. The total number of surveys used in calculations was also tabulated and displayed with each bar plot for validation of spatiotemporal coverage.

To assess whether any seasonal variation in fish biomass was driven by growth or migration, we plotted length frequency distributions for each combination of season and four-degree latitudinal bin and calculated the median value for each distribution. Minimal variation in median length across seasons within latitude bins should indicate an effect of migration, rather than growth, on the total biomass of fish observed.

For a multivariate analysis of taxa-level biomass, Bray-Curtis dissimilarity was calculated from a matrix of 4th root transformed mean biomass per site for the 163 fish species, and visualised using nonmetric multidimensional scaling (NMDS) with the R package ‘vegan’

(Oksanen et al., 2019) in R (R Core Team, 2020). For a multivariate analysis of trophic group biomass, each site’s mean per-survey biomass of each trophic group was calculated, and these means were then summed for each site and used to calculate the proportionate biomass contribution for each of the five trophic groups. These proportions were used to calculate Bray-Curtis dissimilarity and visualised using NMDS. To determine which species and trophic groups were important in driving differences in biomass across latitude, trophic group loadings were generated using the ‘envfit’ function in the ‘vegan’ R package and overlaid on the trophic group proportion NMDS plot as vectors.

2.3.5. Explanatory variables

2.3.5.1. Zooplankton abundance Site-level data for zooplankton abundance were not available, thus we evaluated the influence of zooplankton abundance at the regional level only. The Integrated Marine

Observing System (IMOS) maintains a network of National Reference Stations (NRS) at strategic locations around Australia. Regular zooplankton biomass surveys are conducted approximately bimonthly at NRS sites and data was selected to span the same date range

(09/2008 – 08/2017) as RLS data (01/2008 – 10/2017). Three reference sites (North

Stradbroke Island: 27°S, Port Hacking: 34°S and Maria Island: 42°S) represent the northern, the middle and the southern extent of the RLS survey sites, respectively. Since zooplankton biomass and abundance are regularly measured simultaneously at these locations, these point location data were used to measure the relationship between zooplankton abundance and biomass with linear regression, producing the following equation (p < 0.001, Adjusted r2 = 0.31):

Biomass = e0.518∗lnAbundance − 1.57 (2.1)

This relationship between zooplankton abundance and biomass suggested that continuous zooplankton abundance data from the IMOS Continuous Plankton Recorder (CPR) could also be used in our models. This assumes that the CPR over the continental shelf is representative of zooplankton abundance at our spatial scale. We extracted zooplankton abundance from CPR surveys between 03/2008 to 03/2017 and summed the abundance across each standardised 10 nautical mile section of CPR mesh (~19 km) for each voyage.

We excluded all small zooplankton with a geometric mean size <0.6 mm diameter

(Champion et al., 2015), to determine the mean zooplankton abundance for 0.5-degree

latitudinal bins (~56 km each) for each of the four seasons. These bins were used for matching with RLS surveys which were conducted in the same season and latitudinal bin.

2.3.5.2. Human population density Human population density along the east coast was calculated in ArcGIS (ESRI, 2011) from

Australian Bureau of Statistics Australian Population Grid 2016 (Australian Bureau of

Statistics, 2017). A circular buffer with a 50 km radius was created for each RLS site in the analysis as in Bennett et al. (2016). The total number of people living within each 50 km radius was then calculated and applied to each site as a predictive variable in the full model.

This variable is intended to represent potential human impacts such as fishing or urbanisation near a site.

2.3.5.3. SST and Chlorophyll

Chlorophyll a concentration was used as a measure of phytoplankton biomass. Sea surface temperature (SST) and chlorophyll a concentration (Chl; using OC3 algorithm) were derived from Level-3 MODIS satellite data and were obtained from the IMOS Data Portal

(http://imos.aodn.org.au/imos/) at daily 1 km resolution. The extent of the satellite data was limited to a 10 × 10 pixel grid (~100 km2) centred on each RLS site and bounded by the coast in order to maximise the data retrieval for each reef and minimise cloud interference.

Seasonal means of SST and Chl (e.g. Spring, 2010) were calculated for each site for use in the modelling.

2.3.5.4. Trophic group biomass modelling To quantify whether latitudinal patterns in fish community composition were related to environmental drivers, generalised additive mixed models (GAMMs) were created for

‘trophic group biomass’ and ‘trophic group biomass proportion’ for four of the trophic groups (zooplanktivores, herbivores, omnivores and benthic invertivores) and total fish biomass. This was done using the R packages ‘GAMM4’ (Wood & Scheipl, 2020). We did

not model reef piscivores because they were absent from 38% of surveys and only contributed 4.8% to mean total fish biomass. Generally, underwater visual census is not an effective method for surveying piscivores and many of the larger species are shy of divers, particularly in areas where spearfishing is practiced (Kulbicki, 1998; Ward-Paige et al.,

2010; Lindfield et al., 2014; Gray et al., 2016; Goetze et al., 2017).

The explanatory variables included in the full GAMM were ‘site latitude’, ‘month’, ‘mean zooplankton abundance’, ‘mean chlorophyll concentration’, ‘human population density’, plus

‘site’ and ‘year’ as random intercept factors. Site and year were included as random factors to account for residual spatial and temporal dependency in the response variable. Population density was included as a potentially explanatory variable because many factors that impact temperate reefs (e.g. fishing, pollution and urbanisation) are correlated with human density

(Stallings, 2009; Brewer et al., 2013) and there is an uneven distribution of people along the latitudinal coastline. Residual deviance was tested for spatial autocorrelation by applying

‘Moran’s I’ function from the R package ‘ape’ (Paradis & Schliep, 2019) against an inverse distance matrix of projected site coordinates. Spatial and temporal patterns were modelled with a two-dimensional variable (tensor product) of site latitude and month (a cyclical variable). Sea surface temperature was examined but not included due to collinearity with site latitude (r = 0.66, p < 0.001). The percentage of surveys that recorded zero biomass for a trophic group ranged from 0.1% for benthic invertivores to 9.9% for omnivores, thus the

Tweedie distribution was selected as a family capable of modelling continuous non-negative data containing zeros (Foster & Bravington, 2013).

To model trophic group biomass proportion, the same response variable was used (each trophic group’s biomass) but the model included the total observed fish biomass from each survey as a log-linear offset term. Including this offset standardised trophic group biomass to total biomass i.e. a model of ‘trophic group biomass proportion’. There were thus nine

response variables: the absolute and proportionate biomass observed for each of four trophic groups (zooplanktivores, herbivores, omnivores, benthic invertivores), and total fish biomass. The full model calculated for each response variable was (in script notation; offset was only included for biomass proportion models):

Response = t2(SiteLat, Month) + s(Depth) + s(ln(MeanAbundZoo)) (2.2) + s(ln(Chl)) + s(ln(Population)) + (1|Site) + (1|Year) + offset(ln(TotalBiomass))

Where ‘s’ indicates a penalised regression spline type smoother, and ‘t2’ is a tensor product smooth. The Tweedie distribution parameter is specified when using the ‘GAMM4’ package, and this was found by fitting a GAM without random effects in the ‘mgcv’ R package (Wood, 2020). Residual plots and Q-Q plots were evaluated to ensure model assumptions were sufficiently met.

A model selection process was done to identify the most parsimonious model for each response. For each response variable, we applied the ‘dredge’ function from R package

‘MuMIn’ (Barton, 2020) to test every possible combination of variables. The model with the lowest Akaike information criterion (AIC) was selected as the best model.

Goodness of fit of the models was assessed using the percentage of explained deviance calculated without including the random effects. Response plots showing the relationship between response and covariate (holding all other covariates as constant) were used for visual interpretation of the effects for each variable.

2.3.6. Model prediction

Predictions of the GAMMs of total fish biomass were used to visualise total fish biomass over a standard year for four latitudinal bins within the study area. Predicted biomass was calculated at each month, for four evenly spaced latitude bins, using mean values for depth,

zooplankton abundance, and population density within each bin. In some cases (n = 3) zooplankton data were unavailable, so the mean zooplankton abundance for the latitude bin was used and month was disregarded.

2.4. Results

2.4.1. Broad latitudinal trends

Mean total fish biomass (± SE) decreased with latitude (slope = 19.4 g m-2 per degree latitude, adj r2 = 0.16, p < 0.001) and ranged from 220 ± 88 g m-2 in the north (29.5°S) to 13

± 1 g m-2 in the south (43.5°S) (Figure 2.3a). Similarly, proportional composition by trophic group for the same 1-degree latitudinal bins (Figure 2.3b) indicated a shift in proportional dominance of trophic groups across latitude, with herbivores and omnivores dominating the lower latitudes, zooplanktivores dominating mid-latitudes and benthic invertivores dominating the higher latitudes.

Figure 2.3 Mean biomass and biomass proportions across one-degree latitudinal bins, with (a) representing the mean observed biomass for all fish across fifteen one-degree bins from north to south. Error bars represent the standard error of the mean and (b) representing the mean proportionate biomass for each of the five trophic groups across the same bins. Numbers to the right of bars in (a) represent the number of unique surveys used to calculate both means and proportions.

There was also a seasonal component to these latitudinal trends (Figure 2.4). In the northern low latitudes (29 to 33°S), there was a peak in biomass driven by omnivores,

herbivores and zooplanktivores around late-autumn (282 ± 37 g m-2 for May; Figure 2.4a).

Further south, (33 to 37°S) followed a similar trend, with observed biomass also peaking in late-autumn (211 ± 32 g m-2 for May; Figure 2.4b) and a minimum at the end of spring (57

± 7 g m-2 for November). At 37 to 41°S, the maximum mean biomass was observed in mid- winter (281 ± 88 g m-2 for July, based on only four surveys). At the most southern latitudes

(41 to 45°S), there was an order of magnitude lower fish biomass, although there were too few winter surveys to confirm seasonality for this region (Figure 2.4d). Some individual sites were only sporadically sampled across seasons. Therefore, to provide further validation to seasonal patterns, we examined a subset of five sites within a region with very regular survey coverage and these seasonal patterns persisted (Supplementary Figure 8.2.6). Length frequency distributions indicated minimal variation in median length across seasons, suggesting that seasonal variation was likely more influenced by patterns in migration, rather than growth (Supplementary Figure 8.2.7).

Figure 2.4 Mean monthly total fish biomass represented by the height of each bar. Stacked bars represent the proportionate contribution of each trophic group to the monthly mean total fish biomass. Error bars represent standard error of the mean for total fish biomass. Numbers above each error bar indicate the number of unique surveys used to calculate totals and proportions for each corresponding bar. Months where no data was available for a latitude bin are therefore represented with ‘0’.

2.4.2. Multivariate latitudinal trends

As expected, there was clear latitudinal structure in the fish assemblage at the taxa level

(Figure 2.5a) and based on proportional composition by trophic group (Figure 2.5b), with latitude associated most clearly with axis NMDS1. Sites occur along a gradient of samples dominated by zooplanktivores aligned with NMDS1, as shown by the trophic group loading vectors for the trophic proportion NMDS (Figure 2.5c), and predominantly at mid-latitudes

(zoo: r2 = 0.97, p < 0.001; with r2 representing the squared correlation coefficient of each trophic group with their respective vector, and p representing the proportional rank of the statistic observed among those evaluated through 1000 permutations). Those sites dominated by benthic invertivores (ben.inv: r2 = 0.97, p < 0.001) aligned with NMDS1 in the higher latitudes, while herbivores and omnivores aligned with NMDS2 and contributed to community composition especially in the northern lower latitudes (herb: r2=0.46, p <

0.001; omni: r2 = 0.37, p < 0.001). The effect of piscivores was much weaker as they were never observed reliably in high abundance (pisc: r2 = 0.10, p < 0.001). These distinct differences in trophic compositions with latitude are also evident in bar plots of the proportional contributions to biomass across trophic groups and latitudinal bins (Figure

2.3b).

Figure 2.5 Nonmetric multidimensional scaling (NMDS) plots displaying Bray-Curtis dissimilarity calculated across a matrix of mean observed biomass for individual taxa at RLS sites (a) and proportional mean biomass in each trophic group for each site (b and c). For (a) and (b), points are coloured by latitude and for (c), points are coloured by the mean proportional biomass of planktivorous fishes for each site. Point sizes are scaled to the mean total fish biomass for each site. For (c), overlaid vectors indicate trophic group loadings for the corresponding groups: zoo = zooplanktivores, herb = herbivores, omni = omnivores, ben.inv = benthic invertivores, pisc = piscivores.

2.4.3. Fish Biomass GAMMs - spatiotemporal effects

The model selection process resulted in the selection of five biomass models (Table 2.1) and in almost all cases most of the variation in biomass was described by the tensor product smoother of Month and Site Latitude (Supplementary Table 8.2.5). Moran’s I statistics ranged between 0.03 to 0.14 across all models and thus residual spatial autocorrelation was not important. The results of calculating relative variable importance (RVI; 0-1) for each model and taking the mean of RVI scores across both sets of models suggests that the

Month-Latitude tensor is the most influential explanatory variable, followed by Depth and

Population density for the biomass and biomass proportion models, respectively

(Supplementary Table 8.2.6).

Table 2.1 The final GAMMs generated through testing every possible combination of variables in the full model and selecting the model with lowest AIC score. The AIC of the full model is given as AIC points greater than the selected parsimonious model. Variable codes are as follows: SiteLat = site latitude, Month = survey month, Depth = mean site depth, MeanAbundZoo = mean zooplankton abundance corresponding with survey season and latitude, Chl = chlorophyll concentration corresponding with survey season and latitude, Population = number of people living with a 50 km radius of a site). The Biomass models model the per-survey biomass of each of the listed trophic groups, while the Biomass Proportion models model the biomass of each trophic group relative to the total fish biomass observed for each survey.

Dependent Model Parsimonious model terms Random effects Deviance Parsimonious Parsimonious Full model variable type (bold = signiﬁcant eﬀect) explained model p model AIC ΔAIC (%)

Zooplanktivores Biomass SiteLat,Month + Depth + SiteCode + Year 26.9 <0.001 57796 0 ln(MeanAbundZoo) + ln(Chl) + ln(Population)

Herbivores Biomass SiteLat,Month + Depth + SiteCode + Year 30.1 <0.001 19808.1 18 ln(MeanAbundZoo) + ln(Chl)

Omnivores Biomass SiteLat,Month + Depth + SiteCode + Year 44.5 <0.001 14828.6 5.5 ln(Population)

Benthic Biomass SiteLat,Month + Depth + SiteCode + Year 15.8 <0.001 21324 6.8 invertivores ln(Population)

Total fish Biomass SiteLat,Month + Depth + SiteCode + Year 25.7 <0.001 63118.8 11.5 ln(MeanAbundZoo) + ln(Population)

Dependent Model Parsimonious model terms Random effects Deviance Parsimonious Parsimonious Full model variable type (bold = signiﬁcant eﬀect) explained model p model AIC ΔAIC (%)

Zooplanktivores Biomass SiteLat,Month + Depth + SiteCode + Year 26.8 <0.001 54706.3 0 proportion ln(MeanAbundZoo) + ln(Chl) + ln(Population) + offset(TotalBiomass)

Herbivores Biomass SiteLat,Month + Depth + SiteCode + Year 22.6 <0.001 18455.8 0 proportion ln(MeanAbundZoo) + ln(Chl) + ln(Population) + offset(TotalBiomass)

Omnivores Biomass SiteLat,Month + Depth + SiteCode + Year 15.1 <0.001 14130.1 7.5 proportion ln(Population) + offset(TotalBiomass)

Benthic Biomass SiteLat,Month + SiteCode + Year 16 <0.001 20592.3 7.3 invertivores proportion ln(MeanAbundZoo) + ln(Population) + offset(TotalBiomass)

There was clear spatiotemporal variation in fish biomass, shown in the contour plots of

Latitude by Month tensor splines (Figure 2.6). Tensor product smoothers treated missing high latitude winter survey data as zero, and were flexible at integrating real data from adjacent summer and mid-latitude surveys. Total fish biomass is generally higher at lower latitudes and in winter months (Figure 2.6e), which is a summation of the surfaces for the other trophic groups. The most variation among trophic groups was shown for zooplanktivores, which generally dominated total fish biomass (Figure 2.6a).

Figure 2.6 2D tensor contour plots showing fitted GAMM relationships with contours of trophic group biomass (in g m-2) by Month (cyclical) and Site Latitude for: (a) Zooplanktivores, (b) Herbivores, (c) Omnivores, (d) Benthic invertivores, (e) Total fish biomass. The second row of contour plots (f, g, h and i) represents tensors from the corresponding biomass proportion models. Note the two different colour scales for the two rows of plots.

2.4.4. Fish Biomass GAMMs – environmental effects

At the regional level, one consistent finding was the minimal influence of phytoplankton density (chlorophyll a) on fish biomass or proportionate biomass across all groups, including zooplanktivores (Supplementary Figure 8.2.8, Supplementary Figure 8.2.9). Total fish biomass and zooplanktivore biomass and proportional biomass showed a positive relationship with zooplankton abundance, although this had less influence than the month-

latitude tensor or site depth. The biomass proportion of benthic invertivores declined with increasing zooplankton abundance.

At the local level, the biomass of trophic groups was generally unimodal with site depth.

The peak in herbivore biomass occurred at a shallower depth than for all other trophic groups, while omnivores peaked at deepest depths (Supplementary Figure 8.2.8b).

Zooplanktivores and benthic invertivores showed low biomass at shallow sites and increasing variability with increasing site depth. As depth increased, the proportion of zooplanktivores and herbivores generally decreased, with a corresponding increase in omnivores (Supplementary Figure 8.2.9) (Parsons et al., 2016). Responses for human population should be interpreted with caution due for the few surveyed sites with low covariate values, however a general theme across omnivores, benthic invertivores and total fish was a decline in biomass at high levels of human population density.

Across all models, the month-latitude tensors explained most of the variability in the data, and primary productivity (chlorophyll a) the least (Supplementary Table 8.2.6). The biomass proportion models showed a similar set of relationships (Supplementary Figure

8.2.9).

2.4.5. Predicted Fish Biomass from the GAMM

There was a wide range of variability in the seasonal fluctuations of fish biomass observed at rocky reefs, with the greatest variability observed at the most northern latitudes of our study region (Figure 2.7). At this northern latitude bin (29 to 33°S), the mean predicted biomass more than tripled from 66 to 214 g m-2 (a 3.2-fold increase) between November and

May. This region does contain an area of offshore islands within a marine reserve, but these patterns persisted when marine park sites were excluded. By comparison, predicted biomass at mid to high latitudes increased only ~1.8-fold between spring and autumn, although the

highest latitudes have few winter surveys. Overall, predicted fish biomass was lowest in spring and generally peaked in autumn (Figure 2.7). Model prediction results are also reinforced by the monthly bar plots of mean observed biomass for each latitude bin (Figure

2.4).

Figure 2.7 Predicted per-survey total fish biomass trajectories calculated by month and latitudinal bin using the total fish biomass GAMM. Confidence intervals for each latitudinal bin represent two times the standard error.

2.5. Discussion

Along our ‘space-for-time’ latitudinal gradient we revealed some remarkable changes in fish biomass and trophic structure. Overall fish biomass declined over an order of magnitude from north to south, along with a dramatic shift in the trophic composition of fish assemblages associated with temperate rocky reefs. The relative biomass of zooplanktivores and benthic invertivores were the strongest drivers of assemblage composition across this

range, with zooplanktivores dominating mid-latitudes and benthic invertivores dominating high latitudes. These observed differences in trophic structure over 16 degrees of latitude

(Figure 2.1) reveal potential changes to the trophic structure of fish assemblages due to a warming ocean and species range shifts.

2.5.1. Latitudinal gradients in trophic structure

Herbivores and omnivores made up the greatest proportion and the greatest total biomass in the lower latitude sites. Similar patterns were also described in the western Atlantic

(Floeter et al., 2005), where herbivore abundance (Longo et al., 2014), plant-herbivore interactions (Longo et al., 2019), and herbivore biomass decrease towards higher latitudes, across both coral (Floeter et al., 2004) and rocky reef environments (Floeter et al., 2004;

Morais et al., 2017).

We also observed a clear dominance of planktivory (> 40% of the fish biomass) in the mid- latitudes between 31 and 37°S, confirming and extending earlier local studies (e.g.

Kingsford & MacDiarmid, 1988; Parsons et al., 2016; Truong et al., 2017) despite the fact that planktonic primary production increases at higher latitudes (Everett et al., 2014). This is linked to the dominance of benthic invertivores at higher latitude sites where the contribution of benthic and detrital energy pathway is particularly important, based on infaunal (Gee, 1989) and epiphytic invertebrates (Poore & Steinberg, 1999), such as polychaetes, gastropods, bivalves, decapods, amphipods and echinoderms (Morton et al.,

2008). Higher latitudes are characterised by spring plankton blooms and very low winter zooplankton abundance (Harris et al., 1987), and such seasonality could not support large year round populations of zooplanktivores. Spring blooms enrich sediments as excess organic material sinks to the seafloor (Wassmann, 1997) and are associated with subsequent increased biomass and abundance of zoobenthic organisms (Heip, 1995; Parrish et al., 2009).

This benthic enrichment coupled with high inter-seasonal variability in zooplankton abundance may favour benthic food webs and thus account for the dominance of benthic invertivores over zooplanktivores at higher latitudes.

2.5.2. Disappearing fishes – Winter declines in rocky reef associated fish biomass

The strong seasonality in fish biomass was driven mostly by zooplanktivores but the reasons for these fluctuations were not clear. There were also very few observations of large piscivores, to account for the large reduction in fish biomass through winter and spring.

Given that these seasonal increases in total fish biomass in the low to mid latitudes were not accompanied by corresponding increases in median fish length, it is likely that these patterns were more a result of migration, or of reduced activity (torpor or hibernation) causing a reduction in detectability (Speers-Roesch et al., 2018), rather than growth.

A plausible explanation is that during late winter, zooplanktivores, which are dominated by yellowtail scad (Trachurus novaezelandiae), may migrate to access the East Australian

Current, which is warmer offshore in winter (Suthers et al., 2011). These fish typically spawn in spring and summer along the outer shelf, which may explain the low zooplanktivore biomass observed during summer across mid-latitudes (Horn, 1993; Stewart

& Ferrell, 2001). Similar behaviour of offshore seasonal migration in response to declining bottom temperature was documented for Japanese horse mackerel (Trachurus japonicus) in the East China Sea (Sassa et al., 2009). Although these pelagic zooplanktivores are not permanent reef residents, they have an important role in reef food webs through enriching detrital pathways with their faeces, and as prey (Robertson, 1982; Pinnegar & Polunin,

2006; Morais & Bellwood, 2019). In contrast to other regions, the higher latitudes were characterised by very low overall biomass of zooplanktivores although seasonal conclusions are not possible due to the lack of winter survey data.

The peak in biomass of herbivorous fishes at low latitudes in autumn was driven primarily by large schools of the tropical surgeonfish, Prionurus microlepidotus, which have been expanding their range southward as ocean temperatures continue to rise (Vergés et al.,

2016). This peak in abundance coincides with the time of year when water temperatures are typically at their warmest. As these fish are one of the greatest contributors to macroalgae decline (Vergés et al., 2016), they may be undertaking seasonal southward migrations to feed during the warmer months and returning northward or succumbing to mortality as temperatures cool below their thermal limits in winter.

2.5.3. Dynamic temporal distribution - A means of managing predation?

Seasonal variation in mid-latitude sites was driven by variation in the biomass of zooplanktivores, as other groups fluctuated only slightly and followed no clear pattern.

Such patterns were driven largely by Trachurus novaezelandiae, which can form large schools and regularly travel between the coast and deeper waters. These fish make large seasonal contributions to the trophic ecology of shallow temperate reefs, particularly for these mid- latitude sites (Deegan, 1993). For example, they make up the bulk of diet for piscivorous predators (Scharf et al., 2003) and the large scale fluctuations in their biomass are almost certain to influence the suitability of reef habitats for piscivores as prey encounter rates are closely linked to prey capture rates (Breck, 1993). Therefore it is possible that the seasonality of zooplanktivores may suppress the proliferation of their predators (Durant et al., 2007).

In pelagic environments, zooplanktivores such as sardines and anchovies form large aggregations which are generally separated by large distances, and predators must invest significant time and energy in their search for the next aggregation (Sims et al., 2006).

Therefore, while pelagic zooplanktivores manage predation through remaining spatially patchy, reef zooplanktivores may achieve a similar outcome through temporal patchiness.

2.5.4. Climate-change risks for temperate rocky reef communities

South-eastern Australia is a climate change hotspot which is already experiencing substantial impacts (Sunday et al., 2015), with many species shifting their distribution poleward (Champion et al., 2018; Champion et al., 2019). Temperate reefs near the warm edge of their distribution are becoming ‘tropicalised’, as many tropical species respond to warming by shifting their distribution towards these cooler, higher latitude regions (Vergés et al., 2014a). The latitudinal and seasonal patterns on trophic community composition and total fish biomass we have documented may also shift poleward with the cumulative range shifts of taxa. These shifting patterns will likely result in increased biomass of herbivorous and omnivorous fishes in high to mid-latitudes. In some regions this has already led to overgrazing of seaweed forests and caused regime shifts as canopy-forming kelp declines towards low-biomass turf dominated habitats (Bennett et al., 2015; Vergés et al., 2016;

Filbee-Dexter & Wernberg, 2018).

Expanded range and increased abundance of herbivorous fishes under climate change has been linked to macroalgae losses in several parts of the world, including Japan, Australia and the Mediterranean (Sala et al., 2012; Vergés et al., 2014a). However, there is evidence that existing macroalgae can also inhibit the spread of tropical fish species (Beck et al.,

2017). Thus, preventing further losses of macroalgae may be vital in slowing the poleward expansion of these tropical herbivores. Their influence may not be entirely negative however, as some range expanding herbivorous fishes, such as rabbitfishes in the

Mediterranean, are already being exploited as new fisheries (El-Haweet, 2001). Further, as grazing by herbivores prevents the proliferation of macroalgae, their poleward expansion

may facilitate a parallel expansion in habitat-forming corals (Cheal et al., 2013; Booth &

Sear, 2018) which can shift poleward at rates of up to 14 km year-1 (Yamano et al., 2011).

Climate-mediated changes in habitat composition and shifts in the distribution of foundation species such as kelp and corals can further influence and accelerate changes in fish community composition, as individual fish species lose or gain specific habitats used for settlement and recruitment, as refuge or as food (Vergés et al., 2019).

A poleward shift in the patterns observed could also lead to an increase in the influence of zooplanktivores in the Bass Strait and Tasmania. However, this space-for-time implication is dependent on specific changes in oceanography, as high biomass of zooplanktivores appears to be tightly linked to regions of regular upwelling, such as the separation zones of boundary currents (Bakun et al., 2015). As the East Australian Current strengthens

(Ridgway, 2007) it is predicted that its separation zone could shift 100 km poleward by 2060

(Oliver & Holbrook, 2014), shifting patterns of nutrient enrichment to fish distribution along with it (Bakun et al., 2015).

Shifting oceanographic conditions, alongside milder winter water temperatures and more consistent planktonic primary production throughout the year could facilitate subsequent increased abundances of planktivorous reef associated fish at higher latitudes. In this case, increased direct consumption of planktonic primary production would reduce the proportion of organic matter reaching the benthos, shifting trophic composition away from currently dominant benthic invertivores. This increased availability of smaller body size, highly abundant zooplanktivores would increase overall fish biomass at high latitude reefs and provide improved feeding opportunities for marine mammals, seabirds and piscivores

(Kaschner et al., 2006; Smith et al., 2011). This could be a net benefit for temperate reef ecosystems, however trophic interactions are complex, and it is difficult to speculate on the full range of potential impacts. Clearly, further examination into how poleward shifts in

such patterns would influence local conditions under multiple climate change scenarios would be useful for understanding potential future impacts for temperate rocky reefs.

Ecosystem modelling could prove invaluable in this context.

2.5.5. Conclusion

It is well established that species’ distributions have begun shifting poleward due to climate change and that these redistributions will have significant implications for human systems

(Pecl et al., 2017). Our work contributes critically by producing highly informed mechanistically-based predicted changes in the latitudinal patterns of trophic structure and their potential consequences. The latitudinal patterns in trophic structure observed off eastern Australia in this study are relevant for other poleward-flowing boundary currents systems (Figure 2.1) (Vergés et al., 2014a). For example, reef fish communities in the western Atlantic, which are also influenced by poleward-flowing boundary currents, shift from being primarily dependent on low-energy food sources, such as algae and seagrass, to reliance on higher-energy foods, such as plankton and invertebrates, along the transition from tropical to temperate zones (Floeter et al., 2004). Herbivores that dominate near the tropics may also be expected to expand their distribution in both northern and southern poleward directions in the western Atlantic (Longo et al., 2019). Similarly, in southeastern

Japan the intensification of the Kuroshio boundary current has already been linked to the poleward expansion of tropical herbivorous fish (Kumagai et al., 2018). The observed patterns may also be relevant in the Mediterranean, where the arrival of tropical species from the Red Sea is initially facilitated by the Suez Canal and where the expansion of herbivorous fish is occurring longitudinally in a western direction as this basin continues to warm (Azzurro et al., 2017).

Many questions surrounding the tropicalisation of rocky reefs remain against a backdrop of urbanisation and invasive species on temperate coasts. Our results suggest that overall herbivorous and planktivorous fish biomass at these reefs will increase, which could lead to subsequent losses of macroalgae in lower latitudes and a shift towards more direct planktonic trophic pathways at higher latitudes. More generally, the large latitudinal differences in the trophic composition and ecology of fishes observed in our study highlight the future of valuable, temperate rocky reefs under climate change.

The observed patterns in total fish biomass do not match latitudinal gradients in planktonic primary productivity, which are greatest in higher latitudes (Everett et al., 2014), suggesting that other drivers such as temperature are important in determining total fish biomass and trophic structure. However, there was a positive relationship between zooplankton abundance and total fish and zooplanktivore biomass, indicating that low inter- seasonal variability in zooplankton may be a vital factor driving the structure of reef associated fish assemblages. It could also be that the on-reef abundance of zooplankton is influenced by fine-scale processes not apparent in our estimate of zooplankton abundance, and studies that sample coastal zooplankton abundance at a finer spatial resolution are necessary to quantify the dynamic importance of zooplankton biomass to reef associated fish assemblages. Documenting patterns in the spatiotemporal variability of plankton communities in nearshore waters represents a pressing area for future research. Regardless of the root cause driving these patterns, it is likely that the relative contribution of planktonic subsidies to both tropical (Morais & Bellwood, 2019) and temperate (Truong et al., 2017) reefs has been greatly underestimated.

Due to the dependence of numerous predators on zooplanktivores, it is also important to consider the root causes of these seasonal migrations. These seasonal abundances and paucities of fish biomass would undoubtedly impact the behaviour of mobile piscivorous

predators and make large contributions to reef energy flow (Robertson, 1982; Pinnegar &

Polunin, 2006), however we do not at this stage understand what is causing them. Our along-shore analysis indicates the need for more research into these cross-shelf seasonal movements and what drives them.

Finally, it is important to acknowledge that climate-driven redistributions of species will continue to lead to novel species interactions, changes in foundation species and the emergence of ‘no-analogue’ communities, i.e. communities that are compositionally unlike any found today (Williams & Jackson, 2007). Therefore, while we can learn much about trophic structure and energy pathways from contemporary distributions, we should also expect and be prepared for early detection of ecological ‘surprises’ in coming decades

(Lindenmayer et al., 2010).

3. Pelagic forage fish distribution in a dynamic shelf ecosystem – Thermal demands and zooplankton prey distribution

The manuscript for this chapter has been published in Estuarine, Coastal and Shelf Science and is cited in this thesis as Holland et al. (2020a). The reference is:

3.1. Abstract

The fine-scale distribution of pelagic forage fish is shaped by competing factors as fish optimise foraging while avoiding predation. We investigated the distribution of forage fish in surface waters of a dynamic coastal environment during two spring seasons to examine their distribution in relation to environmental variables. Using a multi-frequency echosounder and a towed Laser Optical Plankton Counter (LOPC), we investigated the effects of bathymetry, temperature, chlorophyll a concentration and zooplankton biomass on forage fish density. Relationships between fish density and these variables were consistent between surveys, despite large differences in total acoustic energy attributed to fish. Fish density showed a strong positive relationship with bathymetry and water temperature, and no relationship with surface zooplankton biomass density or chlorophyll a. This mismatch between fish and zooplankton may be caused by differences in the way fish perceive the distribution of prey versus temperature and predators in shallow coastal waters. Seeking out warmer temperatures along the shelf break may also improve fish physiological performance when cooler spring temperatures are below their thermal optimum.

Understanding the distribution of coastal forage fish may contribute to interpreting nearshore movements of their predators.

Keywords: East Australian Current, fisheries acoustics, forage fish, Montague Island, predator avoidance, thermal tolerance

3.2. Introduction

Continental shelf ecosystems form some of the world’s most productive waters, often due to the upwelling of nutrient-rich bottom water caused by boundary currents, particularly in areas with a narrow continental shelf (Lucas et al., 2011). These productive waters not only support the majority of fisheries landings (Watson et al., 2004; Pauly et al., 2005), but also marine predators through the planktonic food web, from phytoplankton to zooplankton to pelagic zooplanktivorous fish, commonly referred to as small pelagic fish or forage fish

(Pikitch et al., 2014). As a group, forage fish (e.g. herring, sardines, anchovies, mackerel) are some of the most abundant fishes in the world’s oceans and are an essential link in the provision of energy to higher trophic levels (Bakun, 2006; Pikitch et al., 2012).

Over broad geographic scales we know that the distribution of forage fish is highly variable

(Holland et al., 2020b), but generally should peak in biomass in temperate regions (Sala et al., 2012; Holmes et al., 2013; Longo et al., 2019; Holland et al., 2020b). However, we do not currently know about the finer scale patterns of their distribution in coastal environments.

There is evidence to suggest they are constrained by bathymetry (Maravelias, 1999), temperature (Sato et al., 2018) and the distribution of their zooplankton prey (Ayón et al.,

2008). In offshore environments, forage fish distributions are driven by the dynamics of oceanographic and ecological conditions of temperature, fronts and prey density (McInnes et al., 2017; Sato et al., 2018). In contrast, shallow coastal environments constrain the vertical distribution of forage fish around static spatial features (e.g. reefs) and therefore consistent patterns in distribution may emerge (Maravelias, 1999).

Of particular importance for fish distribution is water temperature. The metabolic rates of fish impose a physiological requirement to seek out water masses within their thermal niche

(Pribyl et al., 2016). Oceanographic fronts can present temperature boundaries to fish distribution and may cause a mismatch between forage fish and their prey (Sato et al., 2018).

These thermal barriers to distribution may be particularly prevalent in early spring, when water temperatures are generally at their annual minimum and fish actively migrate to remain on the warmer, offshore side of these fronts (Sato et al., 2018). In comparison, the distribution of zooplankton prey of forage fish is driven by passive processes of accumulation (Genin, 2004; Aarflot et al., 2019) and the availability of phytoplankton

(Benoit-Bird & McManus, 2012). In coastal environments, these accumulations are generally caused by a combination of currents and topographic constraints on vertical migration (Genin, 2004). Shallow topographies, such as in nearshore environments, present a barrier to the pre-dawn descent of zooplankton, resulting in their accumulation in surface waters after they are advected onto the shelf at night (Aarflot et al., 2019). Aarflot et al.

(2019) proposed that these ‘topographically constrained’ zooplankton distributions likely have an important influence on forage fish.

For forage fish in coastal environments, the ability to locate the densest zooplankton aggregations would result in optimal foraging under ideal free distribution (IFD) theory

(Tregenza, 1995; Matsumura et al., 2010). However, fish only acquire limited information from their immediate surroundings and therefore some resources distributed along large- scale environmental gradients, such as temperature, should be easier for fish to navigate than more patchily distributed resources such as prey.

To explore the effects of these potentially competing drivers of thermoregulation and access to high-density prey aggregations on the distribution of forage fish, we examined the coastal waters off southeastern Australia, a part of the Great Southern Reef (Bennett et al.,

2016). This area hosts a combined purse seine and midwater trawl fishery (Marton &

Steven, 2019) for yellowtail scad (Trachurus novaezelandiae), jack mackerel (Trachurus

declivis), blue mackerel (Scomber australasicus) and redbait (Emmelichthys nitidus), with the greatest catch occurring in early summer (Stewart & Ferrell, 2001). The inshore area of this region, in the vicinity of Montague Island (36.251°S, 150.227°E), is home to year round populations of predators that feed on pelagic forage fish, including Australian and New

Zealand fur seals (Arctocephalus pusillus doriferus and Arctocephalus forsteri, respectively)

(Shaughnessy et al., 2001) and little penguins (Eudyptula minor) (Carroll et al., 2016). The narrow continental shelf here is only ~20 km wide, enabling large scale eddies formed by the East Australia Current (EAC) to encroach onto the shelf. Unlike areas to the north, the

EAC eddy field (Everett et al., 2012) facilitates alternating influences of EAC and Tasman

Sea water, making this location particularly dynamic. We expected both static (bathymetry) and dynamic (temperature, chlorophyll a, zooplankton density) drivers would influence the distribution of forage fish in coastal areas in relation to IFD theory (Figure 3.1).

More specifically, we hypothesised that (Hypothesis 1) the magnitude and the direction of associations would be consistent when sampling in the same season in successive years;

(Hypothesis 2) bathymetry would drive consistent distributions of forage fish between surveys; (Hypothesis 3) dynamic variation in the distribution of forage fish would be linked to temperature; and (Hypothesis 4) the distribution of forage fish would follow bottom-up drivers, roughly matching the distribution of zooplankton in line with IFD theory. Finally, we aimed to contextualise our surveys with long-term data on the dynamics of the East

Australian Current.

3.3. Methods

3.3.1. Field surveys

Data was collected on two oceanographic voyages aboard the Australian research vessel, RV

Investigator on 10 September 2016 and 17 September 2017, hereafter referred to as the 2016 and 2017 surveys. A series of six near-parallel ~15 km along-shelf acoustic transects (215 km total for 2016 and 145 km for 2017) were conducted at an average speed of 4.5 m s-1 both north and south of Montague Island, New South Wales (36.25° S, 150.22° E), which were planned to roughly follow the 30, 100 and 130 m isobaths (Figure 3.2). For safety in nearshore waters, both surveys were conducted primarily during daylight. By total duration, 8% of the 2016 survey was conducted before sunrise and 18% after sunset.

Similarly, 4% of the 2017 survey was conducted before sunrise and the survey ended at sunset. For both surveys transects were oriented parallel to the shoreline due to the risks of navigating a large (94 m) vessel with towed gear nearshore.

Figure 3.2 Chart of ship tracks (solid white line) of acoustic surveys in 2016 (a) and (c), and 2017 (b) and (d). Montague Island is pictured at 36.25° S. Vectors extending from the ship track (a) and from two fixed mooring locations (b) represent the current speed and direction between 20-30 m depth over a one-hour period, as measured by acoustic doppler current profilers (ADCP) (see Results section for current velocity measurements). The dashed white line represents the 200 m isobath. Background data for each plot represents 1 km resolution MODIS (Integrated Marine Observing System, 2018) sea surface temperature (a) and (b) and chlorophyll a (c) and (d) averaged over a period extending six-days prior to each survey and smoothed using a focal mean around each pixel. Yellow and navy dots on (c) and (d) indicate sunrise and sunset, respectfully. Red dot on inset map (centre) included for approximate location. Note that each panel has an independent colour scale.

3.3.2. Acoustic data collection and processing

Acoustic data was collected concurrently using the RV Investigator’s built-in Simrad EK60

(Kongsberg Maritime AS, Horten, Norway) multi-frequency echo sounder which pinged simultaneously at 18, 38, 70, 120, 200 and 333 kHz. Transducers were mounted ~6.4 m below the waterline on a lowered drop-keel. During the 2017 survey, the 120 kHz transducer was switched off due to interference with another sounder (data not shown;

Table 3.1). The ship’s acoustic doppler current profiler (ADCP) was active during the 2016 survey, however it was malfunctioning during the 2017 survey, so fixed mooring data was used instead to determine current bearing and velocity. This fixed mooring data was unavailable for the 2016 survey (Table 3.1).

Acoustic data were processed using Echoview v8 (Echoview Software Pty Ltd, Hobart,

Australia). Acoustic data were divided into a grid of cells 30 pings long by 10 m deep and integrated (McKelvey & Wilson, 2006). Given a mean ship speed of 4.5 m s-1, this resulted in grid cells approximately 150 m long. All subsequent processing was conducted with R v3.5.0 (R Core Team, 2020). Details regarding echosounder parameters and processing steps undertaken in Echoview are explained in Supplementary 8.3.1.1 Acoustic processing methods.

2016 2017

Date 10 Sept 2016 17 Sept 2017

Local Time Start 04:51 05:25

Local Time Finish 20:39 17:52

EK60 18, 38, 70, 200, 333 kHz ✓ ✓

EK60 120 kHz ✓ -

Underway logging of temperature, salinity, ✓ ✓ chlorophyll a fluorescence

ADCP ✓ -

Oceanographic mooring - ✓

Triaxus with LOPC - ✓

Acoustic echoes arising from fish were identified using the relative frequency response of

Korneliussen and Ona (2002) and isolated from the 38 kHz data. The 38 kHz fish echoes were then transformed into the linear domain (nautical area scattering coefficient (NASC) values) for further analysis. Multi-frequency data were used to derive relative frequency response signatures for samples collected in situ. We determined the frequency response of each grid cell relative to the response for the 38 kHz band. Relative frequency response, r(f), is defined as follows:

Sv(f) (3.1) r(f) = Sv38 kHz

Where Sv is the volume backscattering coefficient of the grid cell, f is the acoustic frequency being queried and Sv38 kHz is the Sv of the 38 kHz band. Korneliussen and Ona (2002) made their comparisons using the following frequencies: 18, 38, 70, 120 and 200 kHz. As we did not have 120 kHz data for the 2017 survey (Table 3.1), we omitted this band from the analysis to ensure an identical classification method across the two surveys. Volume backscatter values were subject to set of logical criteria outlined in Korneliussen and Ona

(2002) to classify cells based on whether or not they contained schooling swim-bladdered fish. Cells were classified as containing fish if their relative backscatter decreased with increasing frequency, or in other terms:

Sv18 kHz Sv70 kHz Sv200 kHz (3.2) > 1 > > Sv 38 kHz Sv 38 kHz Sv 38 kHz

For validation, we integrated sections of echograms which were visually identified to be likely fish schools and examined the resulting frequency response to inform our classification (Supplementary Figure 8.3.1). This trend of decreasing relative backscatter with increasing acoustic frequency is consistent with targets that are relatively strong scatterers, with gas-filled swim bladders (Miyanohana et al., 1990; Hwang et al., 2015).

From fisheries logbook data for the months of August to October in our study region,

Trachurus declivis (jack mackerel) and Scomber australasicus (blue mackerel) contributed on average 74% and 19%, respectively, of the total catch by weight in Commonwealth Small

Pelagic Fishery for 2016 to 2019 (n = 67), inclusive of bycatch and discards (Supplementary

Figure 8.3.2). Thus, it was likely that the large pelagic schools of fish detected in both surveys predominately belonged to these two species. The relative frequency response employed in our classification was consistent with the acoustic scattering characteristics of

Japanese horse mackerel (Trachurus japonicus), a species closely related to T. declivis, across a range of fork lengths from 134 to 190 mm (Hwang et al., 2015), and for S. australasicus

between 216 and 360 mm (Miyanohana et al., 1990). Fisheries landing records from the

2016/17 season in our study region recorded the fork length range of these species to be

144 to 306 mm for T. declivis and 173 to 351 mm for S. australasicus (Ward & Grammer,

2018). While we could confidently classify cells as either ‘fish’ or ‘not fish’ with this approach, we were unable to discriminate between the two species.

Once grid cells were classified, measurements of the nautical area scattering coefficient

(NASC), in units of m2 nmi-2 (nmi = nautical mile; MacLennan et al., 2002), for the 38 kHz band were summed across each vertical section of water column for each cell classified as

‘fish’ down to a water depth of 30 m. We also retained a copy of the data in gridded format for our analysis of bathymetry as a driver of fish and zooplankton vertical distribution. Cells within this range that were not classified as ‘fish’ were assigned a value of zero. NASC (sA) is represented here by the following equation from MacLennan et al. (2002):

2 sA = 4π(1852) sa (3.3)

2 -2 Where sa represents the area backscattering coefficient (m m ) of a volume of water. This method provided a proxy for the relative density of fish in the surface waters along the ship track which could then be compared to the Laser Optical Plankton Counter (LOPC) and underway measurements of environmental variables. The resulting classifications, using an identical method for both surveys, resulted in consistent mean relative frequency responses for fish classifications across both surveys (Supplementary Figure 8.3.3).

3.3.3. Comparing fish distribution between surveys

To compare the distribution of surface forage fish between surveys (Hypothesis 1), the region was divided into 12 polygons around the island to test for spatial differences in mean

NASC using one-way ANOVA. The number of replicates used to calculate mean values for each polygon was also determined (Supplementary Table 8.3.1). Polygons were generated

by extending the coastline into three 0.065° longitudinal bands divided into four 0.1125° latitudinal blocks, centred at Montague Island at 36.25° S. These polygons were also bounded to the east by the 200 m isobath to exclude data collected off the continental shelf.

The area of each polygon was approximately 70 km2, encompassing a total area of 840 km2.

To compare the raw acoustic classifications between surveys with finer detail (Hypothesis

1), sample coordinates for both years were projected as UTM Zone 56 South so that a specific distance radius could be set for point matching between both surveys. Coordinates for the 2017 survey were matched to 2016 survey points using a 500 m search radius around each point and nearest neighbour search with the ‘nn2’ function in the R package

‘RANN’ (Arya et al., 2019). Points were only matched to one nearest neighbour and if there were no corresponding points within the search radius, the point was excluded.

3.3.4. Measuring the distribution of zooplankton biomass

During the 2017 survey only, a Laser Optical Plankton Counter (LOPC) mounted on a towed-body Triaxus (MacArtney A/S, Esbjerg, Denmark) was towed at 4.5 m s-1 ~250 m behind the vessel at a depth of 20 m for the transects south of Montague Island. For the second half of this survey (the transects north of the island) the Triaxus was undulated (15-

80 m depth) to examine vertical structure (Supplementary Figure 8.3.4). To ensure the data collected north and south of the island were comparable, we only extracted readings for the northern part of the survey between 15 to 25 m water depth.

Data from the LOPC from the 2017 survey were divided (binned) into size categories based on the equivalent spherical diameter (ESD) of measured particles and was then converted to biomass density (mg m-3) assuming the density of water and the volume of water sampled

(Suthers et al., 2006). The particles able to be accurately recorded by the LOPC ranged in size from 100 to 10,000 μm ESD. A previously unpublished analysis of gut-contents of some

common forage fish in the region determined that the lower size limit of the prey they consumed was approximately 250 μm. Therefore, all particles smaller than 250 μm ESD

(~1% of total biomass) were excluded from our analyses.

Although it may have been possible to use the ship’s echosounder to derive zooplankton biomass for the 2016 survey, such methods introduce significant uncertainty (Stanton et al.,

1994). To maintain consistency and to avoid over-complicating the analysis zooplankton density was derived exclusively from the more reliable towed LOPC in 2017.

3.3.5. Bathymetry as a driver of fish and zooplankton vertical distribution

The vertical distribution of zooplankton biomass was examined across three levels of bottom depth (three isobaths) to determine whether the horizontal distribution of zooplankton at the surface was constrained by topography (Aarflot et al., 2019). A similar approach was also taken with fish NASC as well as temperature, by plotting their relationship with water depth for the same three isobaths (as a steeper decline in biomass with depth offshore versus along the coast would be evidence of such an influence). These isobaths corresponded to the three transects conducted north of Montague Island in 2017, where the towed body Triaxus was undulated through the water column and thus LOPC and temperature data were collected over a range of depths (15 to 80 m). These relationships were compared statistically using a GLM with gamma distribution (link = log) and analysis of covariance (ANCOVA), with an interaction of depth in the water column and a fixed factor for three levels of ‘isobath’ as covariates. Comparisons were conducted using Tukey’s post-hoc tests.

3.3.6. Measuring in situ oceanographic variables

In situ temperature, salinity, chlorophyll a (Chl-a) concentration and bathymetry were examined as potentially important environmental variables affecting the distribution of fish and zooplankton. For inferring the distribution of fish, these variables were generated from the ship’s underway instruments which continuously sampled seawater from the 5 m depth drop keel. Bathymetry values were obtained from the ship’s EK60 echosounder.

Temperature and salinity were gathered by the ship’s thermosalinograph and matched to acoustic data by 1-minute sampling intervals. Chl-a concentration was calculated from the ship’s underway fluorometer using linear regression. The details of this approach are described in Supplementary 8.3.1.2 Chlorophyll a methods.

For zooplankton distribution models, the towed-body Triaxus was mounted with a Sea-Bird

SBE911 CTD, which recorded salinity, temperature, and depth, and a calibrated Sea-Bird

ECO-Triplet FLBBCD2K, which was used to derive Chl-a measurements (Sea-Bird

Electronics, Bellevue, USA). All data from these instruments were integrated into 2-second intervals. Bathymetry measurements obtained from the ship’s EK60 echosounder were matched to the Triaxus data, based on the distance between the ship and the Triaxus. This variable distance offset was derived from trigonometry, using the length of cable out and the depth sensor on the Triaxus.

3.3.7. Establishing long-term oceanographic and fishery context

Montague Island is located south of the EAC separation zone (Cetina‐Heredia et al., 2014) and is characterised by dynamic eddies (Everett et al., 2012). To contextualise the high temporal variability in oceanographic conditions for the study region, we examined MODIS

Level 3 satellite altimetry, sea surface temperature and chlorophyll a (Chl-a) concentration data from IMOS (Integrated Marine Observing System, 2018) for the date range September

2011 to February 2020. Using grid cells along the continental shelf edge only, we calculated monthly mean velocity vectors relative to the coastline angle. We also calculated temperature anomaly relative to climatological monthly mean from this date range, and the along-shelf current velocity within the southeast Australian coast (27.8 to 37.8 °S). We constructed a linear model with monthly mean values as replicates to determine whether variability in water temperature was driven by current direction at this location.

Additionally, we examined monthly mean Chl-a concentration by temperature with a linear model to examine the influence of colder Tasman Sea water on phytoplankton biomass.

To establish the distribution of water temperatures in which Trachurus declivis and Scomber australasicus were caught in the Commonwealth Small Pelagic Fishery, we extracted satellite sea surface temperature for the date and location of midwater trawls across a one-degree latitude band centred on Montague Island, occurring between August and October. We then compared the temperature in which T. declivis and S. australasicus were caught with the mean regional temperature between the 100 and 200 m isobaths, as fishing operations were concentrated around the 150 m isobath (149 ± 2.7 m). We also calculated the mean

(weighted by catch per unit effort (CPUE) in kg hour-1) temperature difference between fishing operations and regional mean temperature and conducted one-sample t-tests to determine whether this difference significantly differed from zero. This was intended to establish whether successful fishing operations were associated with temperature anomalies.

3.3.8. Fish and zooplankton surface distribution

3.3.8.1. Fish and zooplankton distribution along vessel track To examine the effects of bathymetry on the distribution of fish during the ship surveys

(Hypothesis 2), two independent generalised linear models (GLMs) were created using surface NASC (summed NASC for cells classified as fish, <30 m deep) as the response

variable for each of the two surveys. Both models used a single predictor, ‘bathymetry’. A gamma distribution was used due to the zero-inflation of the response and strong right- skewness of the error distribution. In order to eliminate zeros to meet the requirements of a gamma distribution, a value equivalent to half the minimum non-zero value in the response was added to all response values (Johnson et al., 1970). Residual and response plots were scrutinised to ensure the GLMs did not violate model assumptions. The initial model structure (identical for both surveys) was (in script notation):

NASC ~ bathymetry (3.4)

Two full models were then created using the additional predictor variables of modelled Chl- a and underway water temperature (Hypothesis 3). Salinity data was available but not included as a predictor due to strong collinearity (correlation coefficient: >0.95 in 2016) with temperature. Both models were then subject to forward-stepwise Akaike information criterion (AIC) based selection. The full model structure (identical for both surveys) was as follows (in script notation):

NASC ~ bathymetry + temperature + log10(Chl-푎) (3.5)

The resulting 2017 model was then combined with an additional predictor of zooplankton biomass to examine its effect on fish distribution, in line with IFD theory (Hypothesis 4).

AIC was used to assess this model. This step was only performed on the 2017 data as the

LOPC was not deployed during the 2016 survey (Table 3.1).

An additional zooplankton biomass model was generated for the 2017 survey to predict the distribution of zooplankton in the region, using the following initial predictors and was subject to the same forward-stepwise variable selection process as described above:

Zooplankton Biomass ~ log10(Chl-푎) + bathymetry + temperature (3.6)

3.3.8.2. Model inference and predicting continuous surface distribution from satellite data

All fish and zooplankton models were assessed for the presence of serial autocorrelation using the autocorrelation function and predictor test statistics were updated to incorporate heteroskedasticity and autocorrelation consistent (HAC) estimates using the ‘sandwich’ package (Zeileis, 2006). Thus, our adjusted inference takes autocorrelation into account and reduces the possibility of Type I errors.

Model predictions for NASC were generated from the final parsimonious NASC models for each survey. To generate continuous two-dimensional predictions across our study area, we used gridded satellite and bathymetry data. Predictions were conducted using 1 km resolution MODIS satellite sea surface temperature and Chl-a averaged over a six day period prior to each survey to eliminate cloud cover (Integrated Marine Observing System,

2018). MODIS data were smoothed using a focal function which calculates a mean value for the 24 pixels surrounding each pixel. For continuous bathymetry, 9 arc second bathymetry data from Geoscience Australia was used (Whiteway, 2009). The same method was applied to predict the distribution of zooplankton at the time of the 2017 survey.

3.3.8.3. Comparing fish and zooplankton predicted surface distribution To examine differences among two-dimensional continuous predictions for NASC among surveys, model responses were scaled from 0 to 1 by dividing by the maximum predicted value for each survey. The absolute difference between the 2016 and 2017 values were then calculated and plotted spatially. Similarly, predictions from the zooplankton distribution model were scaled to have a mean of 0 and a standard deviation of 1 along with the predictions of fish NASC for 2017 to examine differences in the distribution of fish and zooplankton. The difference between these two distributions was calculated and constrained

to fall between -1 and 1, with negative values indicating a greater proportional density of zooplankton versus fish and positive values indicating the opposite.

3.4. Results

3.4.1. Survey and long-term oceanographic conditions

The two surveys happened to sample two different oceanographic regimes (Figure 3.2), characterised by a southward current in 2016, and a northward current in 2017. In 2016, a large warm core eddy of the East Australian Current (EAC) pushed currents southwards at a mean velocity of 0.6 m s-1 and bearing of 189° (as measured by the ship’s 150 kHz ADCP), with mean water temperature (± SE) of 18.02 (± 0.02)°C and salinity (in PSU) of 35.62 (±

0.003). In contrast, during the 2017 survey, a large cold core cyclonic eddy east of the island generated a NNE flow past the study-site, with a mean velocity of 0.5 m s-1 and bearing of

25° (calculated from the 120 m depth mooring averaged over the duration of the survey) and a mean temperature (± SE) of 15.81 (± 0.02)°C and salinity of 35.64 ± (0.001).

During the 2016 survey a frontal edge between water masses was apparent from the southern extension of the EAC pushing warmer water along the edge of the shelf, whereas in 2017 there was a decreasing gradient in temperature from offshore to onshore and temperatures were less variable throughout the region, although colder overall (Figure 3.2).

Ship-derived Chl-a (± SE) averaged 1.41 (± 0.02 mg m-3) in 2016 and 1.96 (± 0.04 mg m-3) in 2017 across the domain surveyed. Chl-a was significantly higher in 2017 (coeff. = 0.52, p

< 0.0001). In both years, Chl-a concentration increased in the cooler coastal side of the temperature front (Figure 3.2). In the 2017 survey, Chl-a distribution was patchier throughout the survey region in comparison to 2016 (2016: CV = 53.2, 2017: CV = 83.1).

To place these two surveys in context, the long-term (8 year) oceanographic data for the shelf break indicated that Montague Island experiences net zero along-shelf current velocity, with an overall mean (± SE) of only 0.02 ± 0.01 m s-1 (Figure 3.3a). This differed greatly from the coastline to the north, which experienced strong net southward flow of up to 0.47 m s-1. Despite net zero velocity, along-shelf currents around Montague Island reversed net directionality on average 3.6 times per year and monthly mean current ranged from -0.36 to 0.30 m s-1 (Figure 3.3b). In agreement with ADCP data, current data from satellite altimetry indicated that the 2016 survey was conducted at a time with weak southward current (-0.04 ± 0.02 m s-1) and the 2017 survey was conducted during moderate northward current (0.18 ± 0.02 m s-1).

Figure 3.3 Mean along shelf current for southeast Australia along the 200 m isobath (a) with red dashed line indicating the latitude of Montague Island. For this location, monthly aggregated time- series for current direction (b) temperature anomaly (c) and Chl-a concentration (d) along the shelf edge are displayed. For (b), blue shading indicates net northward flow and red shading for net southward flow. For (c), blue shading indicates negative temperature anomaly and red shading for positive temperature anomaly. Grey shading indicates the standard error of the mean for all panels. Black dashed lines indicate the point of zero flow (a, b) or temperature anomaly (c). Blue dashed lines (b, c, d) correspond with the dates of the two surveys.

Long-term sea surface temperature anomaly (Figure 3.3c) indicated a moderate negative correlation with along-shelf current velocity (coeff. = -3.06, n = 102, R2 = 0.21, p < 0.001), indicating that variability in temperature was affected by the alternating influence of EAC and Tasman Sea water. Long-term temperature data indicated that the 2016 survey was conducted during a period of positive temperature anomaly (1.1 ± 0.2 °C), while the 2017 survey was conducted at a time of negative temperature anomaly (-0.6 ± 0.1 °C).

Temperature-salinity plots produced from underway data provide further evidence for a separation in water mass signatures between the two surveys (Figure 3.4).

Figure 3.4 Temperature-salinity plots generated from the ship’s underway data displaying NASC values from fish as point colour on a logarithmic scale for each of the two surveys.

This long-term satellite data indicated that the log10 of Chl-a concentration (Figure 3.3d) was negatively correlated with water temperature (coeff. = -0.11, n = 102, R2 = 0.43, p <

0.001), with the annual peak in Chl-a concentration generally occurring in early spring, when water temperatures were at their annual minimum. The annual cycle of Chl-a

concentration suggested that both surveys were conducted during this spring bloom, with a mean concentration (± SE) of 0.44 ± 0.09 mg m-3 in September 2016 and 0.84 ± 0.11 mg m-

3 in September 2017.

The distribution of fishing operations in the region for the months of August to October (n

= 67) was concentrated around 17.7 °C for trawls that caught Trachurus declivis and 17.6 °C for trawls that caught Scomber australasicus (Supplementary Figure 8.3.5) while the background mean temperature was 17.2 °C. For both species, 90% of catch was caught in water above 17.1 °C. The mean temperature anomaly from the background (weighted by

CPUE) was +0.24 °C for T. declivis (p < 0.001) and +0.31 °C for S. australasicus (p < 0.001).

This temperature difference, although small, was consistently positive for both targeted species.

3.4.2. Comparing fish distribution between surveys

In both years, NASC (nautical area scattering coefficient) was greatest in water with relatively high salinity and temperature (Figure 3.4). There was greater overall NASC

(mean ± SE) recorded in the 2017 survey (360 ± 12 m2 nmi-2) than in the 2016 survey (39 ±

1 m2 nmi-2, f1,3898 = 1286, p < 0.001). In 2016 mean NASC was greatest to the southeast of

Montague Island (Figure 3.5a) in the region along the shelf break (polygon S2.3: 136 ± 10.7 m2 nmi-2; f11,2483 = 81, p < 0.001) while the lowest NASC was recorded south of the island along the coast (polygon S1.1: 0.1 ± 0.06 m2 nmi-2). There was no significant difference among the four polygons adjacent to the coast (f3,658 = 1.4, p = 0.246). Comparing complete longitudinal bands (each band of four latitudinally-divided polygons), there was a statistically-significant gradient of increasing NASC from the coast to the shelf break (coast:

3.4 ± 1.1, mid: 38.6 ± 1.4, shelf: 70.9 ± 4.2 m2 nmi-2; f2,2492 = 162, p < 0.001). In 2017 the greatest mean NASC was also recorded to the southeast of the island (Figure 3.5b, polygon

S2.2: 894 ± 43 m2 nmi-2; f11,1393 = 89, p < 0.001). Similar to the 2016 survey, the lowest values were recorded along the coast (polygon S1.1: 89 ± 12 m2 nmi-2) and there was a similar coast to shelf gradient (coast: 112.7 ± 8.6, mid: 472.5 ± 23.3, shelf: 522.5 ± 17.9 m2 nmi-2; f2,1402 = 148, p < 0.001).

Figure 3.5 Mean NASC (nautical area scattering coefficient) values for each of twelve polygons of equal interval latitude and longitudinal spacing from the coast bounded in the central latitude by Montague Island for the 2016 survey (a) and the 2017 survey (b). Polygons are labelled so they can be referenced in text. Note both panels use a different colour scale and polygons are approximately 70 km2.

Using nearest neighbour matching (Supplementary Figure 8.3.6) we were able to geographically pair 79% of NASC measurements from 2017 with measurements from 2016

(n = 1107). The log10(NASC) of both surveys were significantly and positively correlated (r

= 0.40, t1105 = 14, p < 0.001). This indicates a moderate degree of correlation among spatially co-located measurements between the two surveys, despite the different oceanographic conditions and overall greater NASC in 2017.

3.4.3. Bathymetry as a driver of fish and zooplankton vertical distribution

The mean bottom depths of each of the three transects north of Montague Island were 62,

100 and 118 m. Zooplankton biomass (± SE) at the surface was 4143 ± 344 mg m-3 for the

transect nearest to the coast and 461 ± 90 mg m-3 for the transect nearest to the shelf break

(Figure 3.6a). There was a 3.8-fold increase in biomass between the surface and 50 m depth for the shallowest transect in comparison to a 13.5-fold increase in biomass across the same depth range for the deepest transect. The decline in zooplankton biomass with depth was less steep along the coast than it was offshore (GLM, 62 m isobath: coeff. = 1.034; 118 m isobath: coeff. = 1.059; p < 0.0001).

In surface waters (<30 m from the surface), NASC from fish was greater offshore than in coastal waters. The peak in NASC, with respect to water depth, became increasingly shallow with increasing distance from the coast (Figure 3.6b). The greatest NASC from fish for the most coastal transect (± SE) occurred at 40 m depth (173 ± 75 m2 nmi2), while for the furthest offshore transect the peak in NASC occurred at 20 m depth (143 ± 26 m2 nmi2).

Although surface temperatures (as measured by the towed-body) were similar across all three bottom depths (15.6, 15.8, 15.5°C from shallow to deep, respectively), there was a much faster decline in temperature with depth offshore, ranging between 14.9 to 15.5°C, while temperature only ranged between 15.5 to 15.6°C for the most coastal transect.

3.4.4. Simulating fish and zooplankton surface distribution

The initial NASC linear models with bathymetry as the only predictor met all assumptions of normality (2016: n = 2841, 2017: n = 1572). Coefficient and intercept estimates (± SE) indicated an elevated and more consistent overall NASC in 2017 (coeff. = 0.03 ± 0.001, int.

= 2.67 ± 0.12) compared with 2016 (coeff. = 0.06 ± 0.003, int. = -2.93 ± 0.27) across the range of bathymetry surveyed.

Following this, we generated NASC models which included temperature and Chl-a as additional covariates (Table 3.2). In both cases bathymetry and temperature were selected to explain further variation in the forward-stepwise variable selection process, resulting in the two final models with identical structure described (Figure 3.7a,b).

Table 3.2 The three sets of models selected via the forward-stepwise AIC-based selection process for NASC in 2016 (NASC.2016), NASC in 2017 (NASC.2017) and zooplankton biomass in 2017. In this case, ΔAIC represents the difference in AIC between the most parsimonious model and the model specified. Columns named after variables correspond to coefficient (Bath: bathymetry, Temp: temperature, chl: chlorophyll-a concentration, zoo: zooplankton density) and intercept (Int) estimates and p values are indicated for each predictor variable where appropriate.

Model formula ΔAIC Int Int p Bath Bath p Temp Temp p log10(chl) log10(chl) p log10(zoo) log10(zoo) p

NASC.2016 ~ 0.00 -12.81 <0.001 0.05 <0.001 0.56 <0.001 - - - - bathymetry + temperature

NASC.2016 ~ 104.93 -2.93 <0.001 0.06 <0.001 ------bathymetry

NASC.2016 ~ 1 1054.36 3.77 <0.001 ------

NASC.2017 ~ 0.00 -6.94 <0.001 0.02 <0.001 0.72 <0.001 - - -0.24 0.079 bathymetry + temperature + log10(zooplankton)

NASC.2017 ~ 0.55 -2.28 0.011 0.03 <0.001 0.34 <0.001 - - - - bathymetry + temperature

NASC.2017 ~ 39.61 2.67 <0.001 0.03 <0.001 ------bathymetry

NASC.2017 ~ 1 363.57 5.89 <0.001 ------

Model formula ΔAIC Int Int p Bath Bath p Temp Temp p log10(chl) log10(chl) p log10(zoo) log10(zoo) p

Zooplankton 0.00 3.14 0.008 -0.01 <0.001 0.42 <0.001 2.20 <0.001 - - Biomass ~ log10(chlorophyll) + bathymetry + temperature

Zooplankton 33.42 9.04 <0.001 -0.01 0.001 - - 3.01 <0.001 - - Biomass ~ log10(chlorophyll) + bathymetry

Zooplankton 46.90 8.56 <0.001 - - - - 3.36 <0.001 - - Biomass ~ log10(chlorophyll)

Zooplankton 144.36 8.64 <0.001 ------Biomass ~ 1

We subsequently included zooplankton biomass as an additional predictor within the 2017 model (the only year for which we have zooplankton biomass) but this did not explain any additional variability in NASC (Table 3.2, Figure 3.7c), as the inclusion of this additional predictor only improved AIC by 0.55. The zooplankton biomass model retained all three of its initial predictors in the variable selection process and Chl-a explained the greatest amount of variation in the response, followed by bathymetry and temperature respectively

(Figure 3.7d).

Examination of the autocorrelation function indicated that some degree of serial autocorrelation was present in all models, at scales ranging from ~ 2 km for the zooplankton biomass models to ~ 4 km for the NASC models from both years. Our adjusted inference (Supplementary Table 8.3.2) took this into account by using HAC estimates to adjust the test statistic accordingly for all model predictors. All predictors which demonstrated statistical significance (p < 0.05) in the original inference remained significant after HAC estimates were included, except for the effect of Chl-a on zooplankton biomass.

3.4.5. Comparing fish and zooplankton predicted surface distribution

Spatial model predictions of NASC were generated for each of the two fish models (Figure

3.7a,b) using bathymetry and sea surface temperature as predictors (Figure 3.8a,b). There were some differences in the distribution of NASC from fish across the two predictions, even though a static variable (bathymetry) predicted most of the variability in both models.

Overall, predicted NASC (± SE) in 2017 (430.7 ± 11.0 m2 nmi-2) was higher than 2016

(100.5 ± 5.8 m2 nmi-2). In 2016 there was a much steeper decrease in NASC between the shelf break and the coast. Standardised proportional differences in the distribution of NASC between years (Figure 3.8d) indicated that overall variation was low (mean = 15%), while

80% of the study area showed less than 25% variation. The greatest variability in NASC

between surveys occurred along the edge of the continental shelf, driven by variation in water temperature.

The additional prediction (Figure 3.8c) generated from the zooplankton biomass model for

2017 (Figure 3.7d), using MODIS satellite Chl-a, sea surface temperature and bathymetry as predictors displayed an almost opposite distribution to NASC in the same survey (Figure

3.8b), as evidenced by the plot of the difference in standardised NASC and zooplankton biomass (Figure 3.8e). The proportional difference plot indicated greatest zooplankton biomass along the coast and greatest NASC from fish along the shelf break.

Figure 3.8 Model prediction results for pelagic forage fish in 2016 (a) and 2017 (b), and for zooplankton biomass (c), generated using satellite data and bathymetry maps. Absolute proportional differences in the standardised distribution of fish between the 2016 (a) and 2017 (b) surveys is represented by (d). Proportional differences in the simultaneous distribution of zooplankton (c) and fish at the time of the 2017 survey (b) is represented by (e), with negative values indicating greater proportional density of zooplankton and positive values indicating greater proportional presence of NASC from fish (relative to the entire study area).

3.5. Discussion

Our research reveals new insights into the spatial dynamics of forage fish and their prey in temperate coastal waters that form a key foraging area for iconic marine fauna (Carroll et al., 2017). We found moderate inter-annual consistency in the spatial distribution of forage fish (r = 0.40), despite a reversal in current direction which reduced the influence of the

EAC in favour of colder, more productive Tasman Sea water in 2017. Such reversals in current direction are common for this coastal region (Figure 3.3b), driving alternate influences of cold and warm water masses.

We found that springtime distribution of water temperature played a more important role than zooplankton density, both vertically and horizontally, in structuring the daytime distribution of forage fish (Figure 3.1). Overall, our findings indicate that forage fish distribution was not guided by ideal free distribution (IFD) theory. In this case, it was apparent that the balance of needs for forage fish was skewed towards prioritising thermal requirements over locating the highest densities of prey. Furthermore, we found that the shelf break was an important habitat for forage fish in spring. Despite the paucity of zooplankton available in surface waters along the shelf break, there may have been adequate zooplankton available for forage fish at dawn or dusk as these trophic groups converge during diel vertical migration.

3.5.1. Consistency in coast-to-shelf distribution (Hypothesis 1)

While other studies have found evidence for consistent seasonal distribution of pelagic forage fish at large spatial scales (100s of kms; Maravelias, 1999; Grémillet et al., 2008), it was unclear whether the distribution of forage fish in surface waters would demonstrate consistency at the scale of our study (10s of kms). We found evidence to suggest a degree of relative similarity in fish distribution from the coast to the shelf break between annual

surveys, despite variations in temperature and the large differences in current velocity and direction as well as Chl-a. Differences in the distribution of water masses between surveys were a potential driver for the smaller interannual variation we observed in fish distribution between surveys, and this was reflected by our model predictions that showed only 15% mean variation between surveys.

3.5.2. Bathymetry as a driver of forage fish distribution (Hypothesis 2)

A strong driver of this consistent distribution was a positive correlation with bathymetry, as the greatest densities of forage fish in surface waters of the continental shelf were located along the shelf break and associated with warmer offshore water in both 2016 and 2017.

Along the outer shelf, high densities of fish were observed near the surface in both surveys and density declined with depth. Surface waters along the coast showed an opposite pattern, with low fish density at the surface and higher density near the seafloor. The horizontal gradient in fish density was steepest in the 2016 survey. Other studies in the North Sea have documented similar findings, with the greatest densities of Atlantic herring (Clupea harengus) associated with deeper areas of the continental shelf near the shelf break, and with specific substrate types (Maravelias, 1999). It seems likely that this spatial pattern is related to other biological processes not measured in this study that may be associated indirectly with depth. For example, blue mackerel (Scomber australasicus), which we know to be prevalent in the region (Marton & Steven, 2019), congregates at lower latitudes along the shelf break to spawn (Neira & Keane, 2008). This strong association with the shelf break is not just limited to times when fish are spawning, it may also be preferred in foraging periods (Maravelias, 1999). The positive relationship with bathymetry and the distribution of pelagic forage fish could also be influenced by an inshore-offshore gradient in perceived predation risk in shallow coastal waters.

Little penguins nesting on Montague Island tend to feed in the top 20 m of the water column on the inshore side of the island during daytime foraging trips (Carroll et al., 2016;

Carroll et al., 2017). There is also evidence that fur seals forage in water depths of 60-80 m

(Arnould & Kirkwood, 2007). Similar to zooplankton, pelagic fish are topographically constrained within shallow coastal environments (Aarflot et al., 2019), which puts them in closer proximity to both benthic (Woodland & Secor, 2013) and surface predators such as seabirds (Maxwell & Morgan, 2013) (Figure 3.1). Thus, shallower regions of the shelf in the vicinity of the island present a high predation risk for forage fish. Other studies have demonstrated that forage fish will avoid areas with relatively high predation risk, even if this avoidance restricts them to lower-density food resources (Werner et al., 1983;

Holbrook & Schmitt, 1988; Logerwell & Hargreaves, 1996). Thus, in our study region forage fish may have been restricting their distribution to be further offshore, where they are less susceptible to predators. In this case, it may not have been bathymetry specifically that was driving the distribution of forage fish, but rather an attraction to the outer shelf that was driving the association with bathymetry.

3.5.3. Temperature as a driver of forage fish distribution (Hypothesis 3)

While there are many variables that co-vary between the coast and the outer shelf, temperature was one of the more likely variables to structure the distribution of forage fish.

Temperature increased from inshore to offshore in both surveys, however there was enough latitudinal variation in the dynamic distribution of warm water masses to separate the effects of bathymetry and temperature. Thus, the inter-survey variability in fish distribution in the outer shelf region is partially attributed to the redistribution of warm water masses.

It is likely that this positive relationship with temperature is unimodal over a greater range of temperatures, given that our surveys were conducted in early spring (Castillo et al., 1996;

Carroll et al., 2016). Zooplanktivorous fish dominate the biomass observed around rocky reefs in this region, however their biomass generally peaks here in early autumn when water temperatures are several degrees warmer (> 20°C) (Holland et al., 2020b).

The primary catch of the local fishery is dominated by Trachurus novaezelandiae, Trachurus declivis and Scomber australasicus (Marton & Steven, 2019). To determine the optimum temperature for each species, we calculated the average of the 10th and 90th preferable temperature percentiles from AquaMaps (Kaschner et al., 2008) in a method adapted from

Serpetti et al. (2017). Calculated optimum temperatures were: 18.7°C for T. novaezelandiae,

17.2°C for T. declivis and 20.7°C for S. australasicus (Kaschner et al., 2008; Serpetti et al.,

2017). The range of temperatures across our study area during the 2016 survey were 16.7 to

18.8°C while during the 2017 survey was 14.6 to 16.7°C, which were well-below the optimum temperatures calculated for these three species. Only in 2016 were temperatures recorded which exceed the optimum temperatures for T. novaezelandiae and T. declivis, and these temperatures were only recorded along parts of the shelf break.

The combined fishery for Trachurus novaezelandiae and Scomber australasicus is highly linked to temperature, as these fish largely disappear from our study region in winter when water temperatures drop below 13°C. More northern regions of the fishery operate year-round

(Stewart & Ferrell, 2001). Commercial fishers have reported that these two species do not return to our study region in large numbers until water temperatures reach 17°C in summer, between November and January (Stewart & Ferrell, 2001). While our surveys may have been conducted too early to be representative of the more robust distributional patterns exploited by fishers, given the substantial strengthening of the East Australian

Current that has occurred since Stewart and Ferrell (2001) was published (Wu et al., 2012;

Yang et al., 2016), we expect these fish are entering southern coastal waters earlier. Similar

patterns have been documented for other temperature-linked species within the same region

(Champion et al., 2019).

As spring temperatures during our surveys were also likely below the thermal optimum for most forage fish species that inhabit the area, it is likely that fish were actively seeking out warmer water through behavioural thermoregulation (Magnuson et al., 1979).

Nevertheless, across the entire study area, there was higher fish density in the 2017 survey

(9-fold increase in mean NASC), even though mean water temperature was 2°C cooler than

2016.

3.5.4. Spatial mismatch between predators and prey (Hypothesis 4)

The hypothesis that forage fish were guided by ideal free distribution (IFD) theory was not validated. There was a mismatch in the distribution of pelagic forage fish and zooplankton, with the greatest densities of fish found offshore, while the greatest densities of zooplankton were found close to the coast (Figure 3.8e). The size-distribution of particles recorded by the LOPC spanned the full detection-range of the instrument (100 to 10,000 μm ESD), so we were confident that we captured a representative distribution of likely zooplankton prey.

The model variable selection process indicated that Chl-a was the most influential variable that correlated with the density of zooplankton. Several studies have concluded that zooplankton density is driven by bottom-up processes, so their distribution may be driven by phytoplankton abundance (Canfield & Watkins, 1984; Strömberg et al., 2009). The results of inference and our examination of zooplankton distribution with depth over three levels of bathymetry suggests that topographic constraints may be a more important driver in our study area (Genin, 2004; Aarflot et al., 2019). We observed a much steeper decline in zooplankton density with depth from the surface along the shelf break than near the coast

(Figure 3.6a). This pattern is typical of a topographic blockage, when zooplankton are

advected onto shallow shelf waters at night and their pre-dawn descent is blocked by shallow bathymetry (Genin, 2004).

In practice, studies attempting to match the distribution of predators to that of their prey seldom find true synchronisation (McInnes et al., 2017; Sato et al., 2018), with physical drivers often discussed as the reason for mismatch. There are large overall differences in the way water temperature and zooplankton are spatially distributed, which could explain the mismatch between forage fish and zooplankton. Fish migrate through a dynamic seascape to optimise their physiological performance. A major assumption of IFD theory is that foragers have perfect knowledge of the quality and distribution of resource patches within their environment (Tregenza, 1995; Matsumura et al., 2010). Variation in temperature is more predictable than zooplankton patches and generally occurs gradually and over a coarser spatial scale than zooplankton distribution. If temperature is an ecological resource

(Magnuson et al., 1979), it may be easier for fish to navigate to optimal thermal conditions than it is for them to locate optimal foraging conditions. Despite this horizontal mismatch, the near universality of diel vertical migration (Lampert, 1989) suggests that any predator- prey mismatch would temporarily dissolve as trophic groups converge at dawn and dusk.

These depths are relatively shallow for benefits of diel vertical migration to transpire, and there is typically not a large effect on diel zooplankton biomass in near coastal waters.

Night-time feeding may also be important, as some Trachurus species have been observed to feed throughout the night (Emmett & Krutzikowsky, 2008).

Sato et al. (2018) found that in the Northern California Current System forage fish distribution shifted offshore to remain in warmer water as summer upwelling intensified and the upwelling front moved cold water further offshore. As the EAC is strengthening with climate change (Ridgway, 2007), topographically induced upwelling intensity in our study area is also expected to increase in the coming decades. This could cause upwelling

fronts to expand, shifting cold water further offshore of the shelf break and likely shifting forage fish distribution along with them. This may impact the colonies of little penguins and fur seals which nest and pup on Montague Island and are limited in the distance they can travel on daily foraging trips (Carroll et al., 2017).

Our analysis of two surveys showed the distribution of pelagic forage fish across our study region showed consistent relationships with bathymetry and temperature. Overall, our findings suggest that in periods when temperatures are sub-optimal and the distributions of prey and warm water do not align, forage fish may prioritise thermoregulation over optimised foraging. In the coldest months when temperatures are well below optimal it is possible that even minor increases in body temperature may provide substantial benefits for improved somatic growth (Neuheimer & Taggart, 2007; Neuheimer et al., 2011). In seasons when temperatures are closer to the thermal optima for specific taxa (Holland et al., 2020b), this balance may shift to prioritise improved foraging.

To resolve these relationships, we suggest day and night surveys in other seasons to explore the horizontal and vertical relationships among zooplankton, forage fish and water temperature. The oceanographic bidirectionality in current flow at the poleward end of a western boundary current is intriguing, especially around an important foraging ground for seals, whales and penguins. With water temperatures along western boundary currents rising at rates two to three times the global average (Wu et al., 2012), it is becoming increasingly important that we understand what motivates the behaviour of these highly mobile and temperature-dependent species, so that we can forecast their future distribution.

4. Characterising the 3D distribution of schooling reef fish with a portable multibeam echosounder

The manuscript for this chapter has been accepted and is in press in Limnology and

Oceanography: Methods. The reference is:

4.1. Abstract

Multi-species schools of small planktivorous fishes are important constituents of reefs and coastal infrastructure with high vertical relief, however, determining the extent and distribution of these schools is challenging. Here we describe a novel use of a low-cost portable multibeam echosounder from a small vessel, which can concurrently measure detailed bathymetry and the distribution of mid-water targets with high spatial accuracy, regardless of light availability or water clarity. Fish abundance and biomass are not easily quantified by multibeam echosounders, so we developed a new metric for delineating the gridded horizontal distribution of school thickness, and assessed the metric’s efficacy by examining its correlation with mean volume backscattering strength derived from a calibrated 38 kHz split-beam echosounder (r = 0.73). We measured the distribution of fish school thickness around clusters of large concrete modules of an artificial reef using a multibeam echosounder, complemented with underwater video to aid species identification.

The mean distribution of school thickness was mapped around the reef field with generalised additive mixed models (GAMMs). Model spatial predictions indicated schools aggregated around module clusters, rather than individual modules. Dynamic schools of fish in relatively shallow coastal waters (~30 m) can be surveyed over 400,000 m2 at 3 m s-1 in

just 60 minutes. Portable multibeam echosounders are an accessible and valuable addition to quantifying the dynamic distributions of coastal fishes.

Keywords: artificial reefs, Atypicthys, echosounder, multibeam, reef fish, spatial distribution, Trachurus, WASSP, zooplanktivores

4.2. Introduction

Littoral marine ecosystems are characterised by diverse and abundant fish communities, particularly around reefs and areas of high structural complexity (Hunter & Sayer, 2009;

Graham & Nash, 2013; Parsons et al., 2016; Holland et al., 2020b). Fish display a diversity of habitat associations in these environments, with varying use of the water column and benthic environment. Measuring the fine-scale distribution of fishes is valuable for a better understanding of coastal ecosystems, because distribution can indicate essential habitats, trophic processes such as predation and plankton depletion, and the contribution of natural and artificial habitats to fish production and fishery enhancement (Hamner et al., 1988;

Claisse et al., 2014; Champion et al., 2015; Smith et al., 2016). Some of the most abundant structure-associated fishes in the littoral zone are zooplanktivores (Truong et al., 2017;

Morais & Bellwood, 2019), but their fine-scale habitat use is difficult to assess due to their often broad use of the surrounding water column to school and forage (Bellwood et al.,

2019).

Several studies have simulated the horizontal distribution of fish around artificial structures by saturating the reef field with stationary underwater video (Scott et al., 2015; Smith et al.,

2017; Becker et al., 2019), while others have investigated vertical distribution with underwater visual census (Rilov & Benayahu, 1998) and camera-mounted remotely operated vehicles (McLean et al., 2019), but measuring three-dimensional distribution remains a challenge. While these visual techniques do provide valuable species-composition data (dos

Santos et al., 2010; Scott et al., 2015; Smith et al., 2017; Becker et al., 2019), they are not

well suited for studying the distribution of schooling fishes like zooplanktivores, which can aggregate over large volumes not easily surveyed in one dimension. Shoaling fish are also notoriously difficult to enumerate accurately based on visual observations (Pais & Cabral,

2017). Under these circumstances, fisheries acoustics techniques have often proved valuable.

Acoustic methods have been extensively applied to monitoring the regional distribution of small schooling fishes – typically forage fish such as herring and sardines – which often form large single-species schools in low-diversity pelagic environments (O'Donnell et al.,

2019; Kuriyama et al., 2020). Echosounders have also been used to measure fish distributions around structures such as shipwrecks (Paxton et al., 2019), artificial reefs (Lee,

2013) and oil and gas extraction infrastructure (Soldal et al., 2002; Punzo et al., 2015).

Most modern scientific echosounders operate using split-beam transducers (Table 4.1), with a narrow beam (typically <15°, Taylor & Ebert, 2012) to resolve a detailed volume-adjusted two-dimensional profile of the water column in the along-track direction (Rudstam et al.,

1999). Backscatter values (Table 4.1) obtained from these instruments can be calibrated using metal spheres of precise diameter and acoustic properties (Foote & MacLennan, 1984).

These instruments have high spatial resolution along the vessel track, however, the fine- scale spatial distribution of fish and bathymetry in the across-track direction cannot be resolved. Thus, school shape, size and volume are difficult to determine as it is unknown whether the vessel intersected the centre or edge of a school (Kieser et al., 1993). Similarly, the distribution of schools relative to fine scale variations in bathymetry is only detectible along the vessel track. The precise ranging capabilities and large swath coverage of multibeam echosounders makes them better suited to approaching this problem.

Table 4.1 Glossary of terms

Term Definition

Fisheries acoustics A field of research which uses active acoustic instruments (echosounders) to measure the distribution of marine organisms with reflected sound waves (MacLennan & Simmonds, 2013)

Echosounder An instrument which generates and detects reflected sound waves to measure physical and/or biological underwater objects, including the seafloor, fish, plankton, aquatic vegetation, and the separation of water masses

Split-beam echosounder Instrument containing three or more beams, which facilitates positioning of targets within the beam pattern. This transducer can compensate for the variation in echo returns that can be affected by whether a fish is detected at the edge or centre of the beam pattern

Multibeam echosounder Instrument containing potentially hundreds of beams, which are (MBES) divided using a technique known as beamforming, which allows for the directional partitioning of the transmitted and received acoustic energy through the constructive and destructive interference of sound waves

Frequency (kHz) The number of cycles (in thousand cycles per second) that an acoustic wave repeats itself (e.g. 38 kHz is a standard frequency for detecting swim-bladdered fish with split-beam transducers)

Calibration Correcting the drift in sensitivity experienced over time by acoustic instruments by placing a metal sphere with precise acoustic properties within the beam and adjusting the receiver parameters to register a known value

Mean volume A measure of the acoustic energy reflected by targets, averaged backscattering strength over a finite volume (MVBS in dB re 1 m-1)

Multibeam echosounders (MBES) operate with much larger swath width (typically 120-

150°, Colbo et al., 2014). Unlike traditional echosounders, MBES use a technique known as beamforming, which allows for the directional partitioning of the transmitted and received acoustic energy through the constructive and destructive interference of sound waves (Jung et al., 2018). This enables the across-track dimension to be divided up into hundreds of separate beams, facilitating the resolution of spatial features in the direction perpendicular

to the vessel track. In this way, MBES record data in both the along-track and across-track directions, providing a means for generating detailed three-dimensional output.

Recently a range of more affordable portable MBES have been introduced, which are marketed towards high-end recreational boat owners, commercial fishers, and researchers operating smaller, shared-use trailer vessels for applications in shallow water and coastal areas. In relatively shallow depths (20-100 m), these modern MBES can very efficiently survey large areas of seafloor and the lower water column. For example, at 30 m depth a

120° swath would be over 100 m wide at the seafloor and a vessel travelling 3 m s-1 could survey 1 km2 of seafloor in just 60 minutes. The high number of narrow beams and integrated real-time compensation for pitch, heave, roll and tidal height allow MBES to measure the georeferenced position of targets, with horizontal and vertical accuracies generally less than 5 cm at ranges under 15 m (Dix et al., 2012).

MBES generate significant data, in the order of gigabytes per minute, presenting challenges for researchers in terms of processing and storage (Colbo et al., 2014). Due to the immense quantity of data generated, the ability of MBES to retain water column data, in addition to seafloor data, has only been possible since the late 1990s (Mayer et al., 2002). Many modern

MBES still do not allow for logging of water column data (Colbo et al., 2014). MBES data often require considerable aggregation to be used practically for statistical modelling approaches with traditional computer hardware (Colbo et al., 2014). Another drawback is the technical difficulty of calibration (Foote et al., 2005) which typically requires gain adjustment parameters to be measured for each beam independently (Trenkel et al., 2008).

For an instrument with potentially hundreds of beams, this is impracticable for many users and as a result many consumer-grade (rather than scientific-grade) MBES cannot be user- calibrated. While most MBES are calibrated initially by the manufacturer, they have a tendency to drift in sensitivity over time (Roche et al., 2018). A major consequence is that

backscatter values generated from consumer-grade MBES cannot be reliably used to derive fish abundance or biomass.

The comparatively low price-tag (relative to scientific-grade echosounders), portability and user-friendly interface associated with many of the newer consumer-grade MBES may appeal to marine ecologists who may not have the technical skills of acousticians. They can provide quantitative data pertaining to the spatial distribution and behaviour of schooling fish, for such applications as: monitoring the successional establishment of schooling fish at newly deployed artificial structures (Leitao et al., 2008; Folpp et al., 2011; Lowry et al.,

2014); examining the distribution and behaviour of schooling fish at night (Rieucau et al.,

2015; Becker et al., 2016); and, recording the distribution of schools concurrent with benthic habitat classification (Monk et al., 2010; Monk et al., 2012). In short, MBES offer superior technology for measuring the three-dimensional distribution of schooling reef-associated and near-shore fishes. However, due to the complexities associated with deriving estimates of abundance or biomass from MBES, it remains challenging to quantify the fine-scale distribution of schools in these environments.

Here we demonstrate a simple and computationally efficient method for defining the distribution of fish schools from MBES water column data in relatively shallow coastal areas. We apply a similar approach to that used to generate bathymetric surfaces from

MBES point clouds. We extract three-dimensional georeferenced samples and project them onto a two-dimensional grid by calculating a mean value for each grid cell. This results in a spatial variable representing the thickness of a school for any given point in two dimensions, which could be used as an index of distribution for schooling fish. A key attribute of this ‘school thickness’ approach is that it can used to map the distribution of multi-species fish schools in shallow water from small vessels within proximity of the seafloor. This method is robust to spikes attributed to noise and can be used on instruments

100

that cannot be user-calibrated, so long as the GPS antenna, motion adjustments and range estimation are accurate. The overall goal of this article is to present a convenient methodology for marine ecologists to use MBES to map the three-dimensional distribution of schooling fish in relation to shallow coastal features. We have two main aims: (1) to validate our approach by comparing two methods for calculating ‘school thickness’ with backscatter from a calibrated split-beam echosounder; and, (2) demonstrate a practical application of the ‘school thickness’ method using a portable MBES and generalised additive mixed models (GAMMs) to map the distribution of schooling fish around an artificial reef.

4.3. Methods

4.3.1. Site description

For both the split-beam validation and the MBES practical application we surveyed a set of artificial reef modules, 900 m off the coast of Sydney, Australia (34.0943° S 151.1776° E).

The reefs were deployed in 30 m water over an area of bare sand in 2017 to enhance recreational fisheries. Each concrete module weighs 25 tonnes and stands 5 m high, with a footprint of 4  4 m. Modules are deployed in five clusters of 3-4 modules, with one module in each cluster supporting a 4 m steel tower. Artificial reefs such as these typically host isolated, compact and discrete epi-benthic schools (Becker et al., 2019) which facilitate the resolution of individual schools with MBES.

4.3.2. Species composition from remote video

We used remote video to identify the fish species present in each survey. MBES are only suitable for identifying species of schooling fish when there is some prior knowledge or in situ validation of taxa-specific school characteristics (Guillard et al., 2011; Innangi et al.,

2016). Therefore, to identify the dominant species present, we deployed a remote camera

101

assembly with two opposite-facing GoPro Hero 4 cameras (GoPro Inc., San Mateo, United

States) at the centroid of the reef module clusters ~2.5 m above the seafloor for 30 minutes

(Figure 4.1). We generated a relative proportion of abundance by subsampling each video with five random two-minute intervals (Basford et al., 2016) and calculating the mean of the proportionate abundance, from MaxN, for each taxon across surveys. In this case, MaxN was the maximum abundance of each species visible within a single frame within each sampling interval. Since the remote video camera was deployed at the centroid of the reef complex, the bulk of schools at greater distances would not have been observed, however, camera deployments were only intended to complement MBES and enable species identification.

Figure 4.1 Diagram of the reef module layout with the approximate position of the remote camera assembly, anchored to be floating above the seafloor (not to scale).

102

4.3.3. Multibeam acoustic data collection

We conducted five MBES surveys at the Sydney artificial reef approximately 3 hours after sunrise (14 August, 3 September, 26 September, 31 October, and 8 November 2019) using a

WASSP WMB-1320Fi portable 160 kHz multibeam echosounder (WASSP Limited,

Auckland, New Zealand). The transducer was side-mounted on a custom-built aluminium pole assembly, which was stabilised by steel cables under high tension and suspended 1.5 m below the surface. This MBES has a chirp range of 130 to 190 kHz and a maximum range resolution of 2 cm. The MBES was calibrated by the manufacturer to be within 3 dB of the nominal model at 160 kHz. Acquisition range was set to 50 m, with transmission power of

36 W and variable pulse duration and frequency (~17.5 Hz). Acoustic energy of 160 kHz travels through seawater with a wavelength of <1 cm, which is substantially smaller than any individual fish we wished to measure. The transducer assembly was fitted with an integrated Hemisphere Vector V103 Smart Antenna (Hemisphere GNSS, Scottsdale, United

States) which adjusts for pitch, heave and roll in real-time and has a positioning accuracy of

~2 cm under ideal conditions. The recording software also compensated for tidal height in real-time. The seafloor of artificial reef field was completely ensonified with six evenly spaced parallel transects. Each subsequent transect was designed to have 50% overlap with the previous transect at the seafloor.

4.3.4. Multibeam acoustic data processing

All acoustic data were processed using Echoview v9.0 (Echoview Software Pty Ltd, Hobart,

Australia), although other hydroacoustic analysis software would also be suitable (e.g. QPS

Qimera). Associated Echoview files and R-scripts for a working example are available from: https://github.com/hollam2/WASSP_methods.

103

After data were imported into Echoview, the first processing step involved isolating fish schools from noise and bottom backscatter (Figure 4.2). First we excluded some of the outer beams, due to their tendency to produce erroneous measurements (Step M1), caused by their oblique angle of intersection with the seafloor (Colbo et al., 2014). We then conducted an XYZ convolution to blur the acoustic data (Step M2) (Kang, 2011) and generated a bathymetry surface from the same data, using Echoview’s built-in algorithm (Step M3).

Figure 4.2 Flow chart outlining the method for processing acoustic data from both split-beam and MBES to produce the ‘school thickness’ variable. Procedures which are carried out in Echoview are represented in blue, while procedures in R are represented in pink. Dashed grey lines indicate the incorporation of unaltered data from a previous step to be extracted with a Boolean mask. The major parts of the process are categorised and labelled with the three grey boxes in background. Process steps are numbered in sequence and labelled according to whether they pertain to split-beam (prefix: S) or multibeam (prefix: M).

This surface was exported (Step M4) and modified in R v3.6.3 (R Core Team, 2020) to incorporate the three-dimensional outlines of artificial structures (Step M5), such as the concrete artificial reef modules, as these hollow objects are not readily detected by the algorithm for bathymetry detection. This additional step was unnecessary for natural reef sites as the MBES generally performed well at detecting natural seafloor variation (Figure

4.3). The bathymetry surface was then offset by 1 m to provide a small buffer for manually rendered structures and seafloor detection anomalies (Ona & Mitson, 1996) and was

104

subsequently imported back into Echoview (Step M6) so that seafloor and surface data could be masked out (Step M7). It is important to note that due to the high level of accuracy required for this bathymetry surface, it should ideally be generated from this concurrent

MBES data, rather than relying on archival data, as the interaction of currents and benthic organisms with artificial reef structures can gradually alter seafloor morphology (Tassetti et al., 2015) and reef structures may shift and deteriorate over time (Sala et al., 2007).

Figure 4.3 One-metre resolution digital elevation model for a natural reef site generated from data collected with the multibeam echocounder. Note the z-axis has a 3x exaggeration to highlight variations in bathymetry.

After the masking step, we calculated the median value for every three pings in sequence to achieve minor smoothing of the original data (Step M8) and finally applied an additional

XYZ convolution with a top-hat transformation (Dougherty, 1992) to facilitate feature extraction (Step M9). These data were then subset with a user-specified threshold value, in our case -65 dB (Brandt, 1996; Loures & Pompeu, 2015), to create a three-dimensional

Boolean mask, which was then applied to the original unaltered data to exclude volume most likely to be noise and extract samples most likely to contain schools (Step M10). The combination of these processing steps essentially detects and refines the major transitions in mid-water acoustic energy across space that define the surfaces around schools.

105

4.3.5. Spatial subsetting

Extracted georeferenced samples were exported from Echoview as three column XYZ data

(Step M11). All subsequent analysis was conducted using R v3.6.3 (R Core Team, 2020).

Samples and bathymetry were projected into a UTM coordinate system (Step M12) and plotted in three dimensions and visually inspected (Step M13) using the ‘RGL’ package

(Adler & Murdoch, 2020). This package can be used to generate detailed interactive three- dimensional visualisations which can be exported to html and viewed in a web browser

(Figure 4.4).

Upon plotting, any visually apparent noise spikes (sometimes caused by bubbles) which were not suitably removed by the previous steps were then excluded by manually defining the three-dimensional extent of rectangular prisms to exclude erroneous samples. This step was rarely required as the previous processing steps were generally effective at isolating schools from noise. In addition, any erroneous below-bottom samples were also excluded during this step by using the bathymetry surface generated in step M5.

Figure 4.4 Raw 3D spatially referenced samples from one survey of the artificial reef, coloured by mean volume backscatter (Sv). Bathymetry surface is also shown and has been modified to accommodate the shapes of artificial reef modules, in red.

106

For all types of echosounders, comparing targets at different depths is complicated by the fan shape of the acoustic beam. Based on the shape of the beam, with the transducer deployed facing directly downward, it is generally more effective at detecting benthic targets than those in the upper water column, where the beam is much narrower. This is particularly evident when using MBES due to their considerably large swath width.

Additionally, the characteristic tendency for MBES to produce arc-shaped noise artefacts

(across-track), caused by interference from sidelobes, reduces the usability of data from the outer beams near to the seafloor (Clarke, 2006). However, as the wide swath angle used by

MBES generates a large ensonified volume, it is possible to eliminate some data to reduce this bias and interference while retaining enough data to still take advantage of the wider swath.

To alleviate this disparity in detectability, samples were cropped to effectively modify the swath so that detectability remained more consistent throughout (Step M14). Further, this step virtually eliminated spurious sidelobe seafloor detections, which tend to be stochastic due to the scattering effects of beams intersecting seafloor objects at oblique angles (Colbo et al., 2014). Data were initially subset to exclude samples occurring shallower than a user- specified depth, effectively cropping the beam pattern. The width of the swath at this specified depth was then calculated using trigonometry by using the swath width (120° in our case) and range from the transducer face (10 m in our case). A polygon was then generated, consisting of the union of circular buffers calculated from the GPS-detected coordinates along the vessel track, with radii equivalent to the swath widths calculated in the previous step. This polygon was used to crop the edges of the swath. This process cropped the top and sides of a triangular prism to transform it into an internal rectangular prism with varying maximum depth and more consistent detectability throughout its

107

vertical and horizontal extents (Figure 4.5). Even with this cropping, a vessel travelling 3 m s-1 could survey 400,000 m2 of seafloor and adjacent water column in one hour.

Figure 4.5 Swath (light grey) conversion to rectangular prism (blue and outlined in red), through setting a minimum depth threshold and limiting the extent of acoustic data in the across-track direction. This effectively excludes most of the swath likely to be affected by sidelobe artefacts (yellow) in the outer beams where they intersect the seafloor. Note that beams are included to aid visualisation and do not correspond to the number of beams used in this study.

4.3.6. Deriving school thickness

To derive a metric for school thickness (a measure of a fish school’s vertical extent), two methods were explored and compared using the same dataset. Both methods involved initially binning data into 1 m x and y horizontal intervals, creating a topographic grid

(Step M15).

The first method, referred to hereafter as the ‘difference method’, involved calculating the minimum and maximum depth of all samples with Sv ≥ -65 dB occurring within each cell of the grid and calculating the difference between these two values to derive a measure of school thickness (Step M16). A sensitivity analysis was conducted to determine this threshold using a subset of the split-beam data. Increasing this threshold to be above -55 dB substantially reduced mean school thickness, as this threshold exceeded the level of backscatter typically exhibited by small schooling fish targets (Supplementary Figure 8.4.1).

108

One drawback of this method is the potential to produce artificially inflated values when a lone fish or school higher in the water column overlaps with a school lower in the water column. This method presumes that the empty water between the highest and lowest occurring samples contains fish and would thus create artificially inflated values for such cases of overlap. Furthermore, as schools increase in size, individual fish are less aware of the overall distribution of their school and must base their movements and behaviour on that of their more immediate neighbours (Paramo et al., 2007). As a result, large schools tend to exhibit more heterogeneous density throughout their volume, and are composed of a complex arrangement of high density volumes or nuclei, and empty or low density volumes, also known as vacuoles (Guillard et al., 2011).

We tested a second method to exclude these empty volumes, referred to hereafter as the

‘sum method’. This involved an additional step, which binned samples into depth layers of 1 m thickness (Step M16). For each grid cell, the number of layers containing at least one target detection were summed. If a grid cell contained fish in every layer, then the calculated school thickness derived using the sum method and the difference method would be equivalent. The sum method should be more robust to the presence of vacuoles (Fréon et al., 1992; Paramo et al., 2007; Guillard et al., 2011), spatially overlapping schools or lone targets overlying schools. The use of 1 m resolution layers to derive school thickness from the sum method is arbitrary and does not affect whether grid cells are classified as containing fish, however, the choice of vertical resolution does affect the precision of the school thickness measurement. This was verified by conducting a sensitivity analysis on a subset of data with four levels of vertical resolution (0.1, 0.5, 1, 2 m; Supplementary Figure

8.4.2). It is possible to modify the vertical resolution for other applications depending on the range resolution of the MBES, however, for the sake of simplicity we maintained the same resolution in all three dimensions.

109

4.3.7. Simulating school distribution around an artificial reef

In order to quantify the distribution of school thickness around the artificial reef, school thickness from each MBES survey was processed into a raster grid, on a transect by transect basis, with the previously described sum method using the ‘rasterize’ function from the ‘raster’ package (Hijmans, 2020) in R. The combined extent of grids was cropped to fit a

300 m square positioned over the reef centroid. Grids were aggregated to 5 m resolution to improve the computational efficiency of model fitting. Aggregation was conducted by calculating the mean value for school thickness for each grid cell within each transect (n =

9034).

To simulate the mean spatial distribution of school thickness around the reef we created a

GAMM that used orthogonal coordinates as predictors (Equation 4.1):

thickness ~ s(x, y) + (1 | date) (4.1)

Where the variable s(x,y) is a two-dimensional product smoother of grid cell coordinates to resolve the spatial distribution of school thickness in both horizontal directions (x and y).

We also included survey date as a random factor to account for day-to-day variation. We assessed several values for the k-parameter for the smooth term (5, 10, 25, 50, 75, 100;

Supplementary Figure 8.4.3), however, we chose to use k=50 to resolve spatial patterns in the data without overfitting. Several studies have found that fish abundance drops off rapidly after 30 to 50 m (dos Santos et al., 2010; Scott et al., 2015; Smith et al., 2017; Becker et al., 2019). To determine the spatial extent to limit our model, we created a binomial

GAMM with logit-link to predict probability of target occurrence, with a smoother of

‘distance from the nearest module’ as the only predictor and date as a random effect. Model predictions indicated probability of target occurrence levelled off at ~20% at distances beyond 70 m (Supplementary Figure 8.4.4). Thus, it was appropriate to restrict the spatial

110

extent to 70 m from the nearest module to model school thickness. We chose a Tweedie distribution with log-link to model continuous positive values with zero inflation. We refer to this model hereafter as our ‘school thickness distribution model’. The advantage of this model structure is that it creates a flexible surface, which can be compared to the distribution of spatial features (such as artificial reefs) to examine their influence on the distribution of school thickness. School thickness is visualised in two dimensions but represents a depth-aggregated metric of fish abundance.

4.3.8. Comparison of school thickness with calibrated acoustics

To validate our school thickness variable, we re-analysed three acoustic surveys of the artificial reef site using a calibrated split-beam echosounder. This allowed traditional split- beam backscatter measurements (mean volume backscatter or MVBS) to be compared with our two variations of calculating school thickness. Because split-beam echosounders only collect data along-track (two dimensional), the derivation of school thickness for split-beam data was conducted at the level of 1 m intervals as replicates, whereas the resolution of

MBES data was at the level of 1  1 m grid cells.

The three surveys were conducted from a 6 m vessel travelling at 2.5 m s-1 with a Simrad

EK80 echosounder and a side-mounted 38/200 kHz Combi C transducer (Kongsberg

Maritime AS, Horten, Norway). Transmitted pulse length was set to 0.512 ms with a transmission power of 500 W and sampling frequency of 15.6 kHz. Prior to each survey, the echosounder was calibrated using standardised methods with a 38.1 mm tungsten carbide sphere (Foote & MacLennan, 1984). Surveys were conducted on 3 December 2018, 1

October 2019 and 25 February 2020, imaging 68 distinct fish schools in total. The difference method and sum method were examined on the 38 kHz wideband (34 to 45 kHz)

111

data obtained from the EK80 so that we could examine correlation between school thickness and MVBS.

A slightly different approach was necessary to process this two-dimensional split-beam data to extract georeferenced samples (Figure 4.2). Data were cleaned to remove background noise, impulse noise and transient noise prior to echo integration following the methods of

De Robertis and Higginbottom (2007) and Ryan et al. (2015) through built-in functions in

Echoview (Step S1). A bottom detection line was generated using Echoview’s built-in algorithm and was manually corrected for errors (Step S2). This line was offset by 1 m to exclude any potential bottom backscatter (Ona & Mitson, 1996). All data below this line were excluded, as well as all data above a surface line at 6 m depth (Step S3). Schools detection was then performed (Step S4) using the SHAPES algorithm (Fernandes, 2009) implemented within the schools detection tool to generate polygon boundaries around distinct schools using a set of schools detection parameters (Supplementary Table 8.4.1) and a detection threshold of -65 dB (Brandt, 1996; Loures & Pompeu, 2015). For each isolated school, georeferenced samples along with mean volume backscatter for each 1 m interval along the school length were isolated (Step S5) and exported to R (Step S6) where data were visually inspected by plotting in two dimensions.

The two-dimensional data from both school thickness methods were projected to UTM

(Step S7) and binned into 1 m intervals (Step S8) so that school thickness could be calculated as above (Step S9). School thickness was assessed at two scales, which included binning into 1 m intervals and aggregating binned data at the level of individuals schools, by calculating mean school thickness and mean volume backscatter, after conversion to the linear domain. Correlation tests were conducted for each of the two scales for each method.

We then conducted paired t-tests on the correlation coefficients using the ‘paired.r’ function from the ‘psych’ package (Revelle, 2020) to determine which method demonstrated the best

112

correlation with MVBS. We also investigated whether measurements of backscatter might be affected by depth, with high intensity targets shadowing targets below them in the water column. Individual georeferenced targets were used as replicates in a linear model with the following structure (in script notation):

Sv ~ depth ∗ school_id (4.2)

Where Sv represents backscatter, depth represents the depth of the target in the water column, and school_id is a factor to group targets belonging to the same school.

As the sum method proved a more comparable estimate to traditional backscatter (see

Results section), it was then further examined to determine which characteristics of schools led to better correlation between school thickness and backscatter. These characteristics included school profile area, perimeter-to-area ratio and mean backscatter, which we compared with the correlation coefficients for each of the schools to determine which affected the reliability of our spatial variable.

4.4. Results

4.4.1. Comparison of school thickness with calibrated acoustics

4.4.1.1. Comparison of the difference and sum methods Based on the EK80 split-beam dataset, the difference method applied at the scale of 1 m intervals along the vessel track demonstrated a moderate correlation with mean volume backscatter (r = 0.56, p < 0.0001; Figure 4.6a), however, correlation was higher for the sum method at this fine scale (r = 0.73, p < 0.0001; Figure 4.6b).

When data were aggregated to the level of individual schools, mean volume backscatter ranged from -62.5 to -42.1 dB. There was higher correlation with school thickness at the

113

level of individual schools than at the level of 1 m intervals for both the difference method (r

= 0.69, p < 0.0001) and the sum method (r = 0.81, p < 0.0001; Figure 4.6). Paired z-tests indicated that these differences were not significant (difference: z = 1.21, p = 0.23, sum: z =

1.03, p = 0.30).

Figure 4.6 Correlation of mean volume backscatter (dB re 1 m-1) with mean school thickness (split- beam data), with school thickness calculated using the difference method (a) or the sum method (b). Black points indicate samples at the scale of 1 m horizontal intervals and red points indicate the mean values for each of the two variables aggregated at the level of individual schools. Correlation coefficient (R) and level of statistical significance are indicated in the top left of each plot, with corresponding colours to the plotted data. Note that in (b), x-axis positions are jittered to aid visual interpretation.

Paired t-tests of the difference between the two dependent correlations indicate that the sum method had higher correlation with backscatter than the difference method at the level of 1 m samples (df = 3093, t = -17.23, p < 0.0001) and at the level of individual schools (df =

65, t = -2.46, p = 0.02). Due to its higher correlation with backscatter at both scales, the sum method was selected as a more appropriate method for all further analysis.

The effect of depth on backscatter was small, but significant (coeff. = 1.76, SE = 0.88, p =

0.045), indicating there was a slight decrease in backscatter with depth within a given school, however, this difference only ranged from -52.5 to -55 across the range of depths where targets were detected.

114

4.4.1.2. Performance with variation in school shape Based on the sum method, correlation coefficients of school thickness with volume backscatter (at the scale of 1 m intervals) were compared to two aspects of school dimensionality ─ school size (area of the profile) and diffuseness (perimeter to area ratio of the profile) ─ to determine whether these characteristics affected the reliability of our method (Supplementary Figure 8.4.5, Supplementary Figure 8.4.6). The total area contained within school profiles was positively correlated with these correlation coefficients for individual schools (r = 0.35, p < 0.01; Figure 4.7a), suggesting that larger schools are more easily resolved by our method.

Figure 4.7 Plots representing the correlation coefficients for school thickness and backscatter at the level of individual schools generated from the sum method, with (a) school profile area; (b) school profile perimeter:area ratio, and (c) mean volume backscatter. Correlation coefficient (R) and level of statistical significance are indicated in the top right of each plot. A linear regression line (blue) and 95% confidence interval (grey) are also displayed.

Perimeter to area ratio negatively correlated with correlation coefficients (r = -0.44, p <

0.001; Figure 4.7b) suggesting more dispersed and complex school shapes are more difficult to resolve or quantify with our method. Larger areas tend to have lower perimeter to area ratios, so perimeter to area ratio is probably a more appropriate metric for comparing schools in this case. Mean backscatter was positively correlated with the correlation coefficients (r = 0.56, p < 0.001), indicating that schools with greater density or larger fish are better resolved by our school thickness method (Figure 4.7c).

115

These findings suggest our method should perform best for larger and denser schools with compact boundaries and an approximately elliptical shape, as ellipses exhibit smaller perimeter to area ratio than other shapes with comparative dimensions.

4.4.2. Simulating school distribution around an artificial reef

Our school thickness distribution model explained 18.8% of residual deviance in the absence of random effects (Figure 4.8) and a moderate degree of standard error throughout (mean

SE = 1.8 m) (Supplementary Figure 8.4.7). Visual assessment of the semivariogram of residuals indicates that residual spatial autocorrelation was not present in this model

(Supplementary Figure 8.4.8), which primarily exhibited a pure nugget effect (Smith et al.,

2007).

Figure 4.8 GAMM prediction for the mean distribution of school thickness around the artificial reef. Black squares indicate the locations of the concrete reef modules.

116

Mean predicted school thickness ranged from 3.0, within 5 m of the nearest reef module, to

0.8, between 65 and 70 m from the nearest reef module. Our GAMM indicated that when summed, 84% of mean school thickness occurred within 50 m of the nearest module. This finding agrees with other studies of the distribution of fish around artificial reefs in the same region (Scott et al., 2015; Smith et al., 2017; Becker et al., 2019).

Overall, our model had limited predictive power (deviance explained: 18.8%) relative to that generated from drop cameras in a similar study (deviance explained: 40.1% to 66.6%, Becker et al., 2019). This relatively low explained deviance is probably due to large variation in the distribution of fish schools among days, which can be evaluated by exploring each survey’s data (Supplementary Figure 8.4.9). The high proportions of school thickness occurring within 50 m of modules suggests that we likely captured most of the fish schools associated with the reef field. The lower performance than Becker et al. (2019) may also be due to differences in model structure, as Becker et al. (2019) explained relative fish abundance as a function of bearing to, and distance from, the nearest reef (rather than a function of geographic location).

The processed survey outputs and fitted GAMM indicate that the greatest peaks in school thickness may not occur in the immediate vicinity of reef modules. Rather, school thickness was generally concentrated to the northeast of both the northern and the eastern module cluster. Since the three-dimensional module outlines were excluded from the acoustic data before samples were exported (Figure 4.4), it is unlikely that peaks were driven by the modules themselves. The fact that peaks in school thickness were offset with the same relative distance and bearing from module clusters, rather than centred around module clusters, could be ecologically meaningful. The artificial reef field was designed with this specific module placement so that the provision of fish habitat would be greater than the sum of the individual modules. These observations are important because they highlight an

117

association between schooling fish and benthic structure that would not have been well- resolved by alternate survey methods.

Remote video camera footage from each MBES sampling occasion showed the mid-water artificial reef fish community to be dominated by yellowtail scad (Trachurus novaezelandiae,

71% of total community abundance) and Australian mado (Atypichthys strigatus, 28%). Three other schooling taxa were also observed, silver sweep (Scorpis lineolata), one-spot puller

(Chromis hypsilepis) and silver trevally (Pseudocaranx georgianus), but these contributed less than 1% of observed total abundance.

4.5. Discussion

Rapid spatial sampling and estimation of fish distribution across shallow coastal reefs is of critical importance as we increase pressure on our coastal environments (McClanahan et al.,

2014; Caldwell et al., 2016). Here we show that a consumer-grade MBES was effective in simulating the distribution of schooling fish around an artificial reef and was able to delineate ecologically meaningful peaks and boundaries in the distribution of fish schools around the reef field. The general patterns measured were similar to those predicted by models produced from drop-cameras (Smith et al., 2017; Becker et al., 2019), a more labour- intensive method that poorly integrates vertical information. Our approach using a relatively low-cost and portable MBES demonstrates that these instruments can provide useful quantitative information on fish distribution, despite their inability to be easily calibrated. Overall, our results highlight the potential value to coastal ecologists of simple, spatial metrics produced from the new generation of consumer-grade MBES. Although these spatial metrics and MBES are by no means new (Reid & Simmonds, 1993; Theberge &

Cherkis, 2013), the application of non-scientific MBES to marine spatial ecology has been limited. Extraction of three-dimensional targets based on a threshold dB value and

118

subsequently projecting their vertical distribution onto a grid represents a novel approach for working with this large, uncalibrated, and often noisy data to address quantitative ecological questions. While the three-dimensional visualisations generated by MBES can be useful for interpretation, depth-aggregation into a two-dimensional grid of school thickness facilitates spatial calculations and statistical approaches that would otherwise be computationally impractical.

Applying acoustic techniques to monitor fine-scale distribution of multi-species assemblages around benthic structure, such as reefs, presents significant challenges (Boswell et al., 2020), as it is common for multiple species to forage together in the water column (Lukoschek &

McCormick, 2000). This school heterogeneity significantly complicates traditional fisheries acoustic approaches (Boswell et al., 2020). Although there exists high variability in translating school thickness to actual fish abundance (Zwolinski et al., 2009), the high correlation of this variable with mean volume backscatter from the split-beam echosounder suggests that school thickness can be a useful representation for the relative distribution of schools. As a measure of relative distribution, it is not that dissimilar to more common methods for deriving relative abundance such as MaxN or MeanCount from remote video count data (Schobernd et al., 2014; Campbell et al., 2015), from which absolute abundance is unknown but expected to scale relatively.

Due to the spatially continuous nature and high resolution associated with MBES data, it is possible that models of the mean fish distribution generated using our method will also have lower predictive power than those produced from point data, such as underwater video

(Saveliev et al., 2007; Li & Heap, 2011). However, achieving a similar outcome using underwater video methods is substantially more laborious. Becker et al. (2019) predicted the distribution of fish around an artificial reef by saturating a similarly designed reef field with

40 drop camera deployments of 2 minutes duration each over 24 days, resulting in 960

119

deployments and 32 hours of footage. By comparison, we conducted only 5 MBES surveys of 40 minutes duration at our site, which took 3.33 hours in total. Like Becker et al. (2019), we also observed a sharp decline in fish density with increasing distance from the nearest module, as has been documented in other studies of artificial reefs (Boswell et al., 2010;

Champion et al., 2015). This pattern, known as the thigmotaxic response, or the tendency for animals to move towards structure rather than bare habitat, is a dominant process structuring the distribution of fishes (Brickhill et al., 2005; Sheehan et al., 2020), even when it acts to their detriment (Hallier & Gaertner, 2008). In the context of artificial reefs, this process is the basis for the enduring ‘fish attraction vs production’ debate (Bohnsack, 1989;

Pickering & Whitmarsh, 1997; Smith et al., 2015).

Despite the strong attraction of fish to the reef structure, fish schools will relocate among surveys, and even redistribute over the course of a single survey. Due to the non- instantaneous nature of data acquisition, this redistribution is an important consideration for acoustic surveys, as it is for underwater visual census and video methods (Schobernd et al., 2014). At the finest scale, this may lead to aliasing of individual targets (Colbo et al.,

2014), although our grid-based averaging approach should reduce this effect. At a coarser scale, redistribution could cause some schools to be surveyed multiple times and some to remain unsurveyed. This potential source of error should be diminished in the vicinity of large reefs, as many schooling fish maintain close proximity to structure (Szedlmayer &

Schroepfer, 2005; Lowry et al., 2017) and can demonstrate high site fidelity, even within multi-reef arrays (Taylor et al., 2018). Nevertheless, a robust sampling design would evaluate variation in the abundance of fish schools and their distribution to ensure sufficient surveys are done to estimate metrics of interest (such as the mean distribution).

In addition to the school thickness variable we derived, there are additional gridded spatial variables pertaining to school distribution which would be straightforward to generate with

120

only minor modifications to our rasterization methods. The simultaneous high-resolution measurement of water column and seafloor data produced by MBES makes them particularly well suited for studying the spatial distribution of fish relative to their habitat.

A good example of this is ‘height above bottom’, or simply the vertical distance between the fish in a grid cell and the seafloor (Weber et al., 2009). This variable can indicate how strongly schooling fish associate with the seafloor (Holzman et al., 2007). It would be similarly easy to apply these methods to studying the depth of fish relative to the water surface. This could be useful in studying vessel avoidance (Vabø et al., 2002) or the response of schooling fish to the threat of surface predators, such as seabirds (Fauchald, 2009).

We demonstrated our method for processing and summarising MBES data by mapping the depth-aggregated distribution of fish around a near-shore artificial reef, but this approach could be readily adapted to other applications concerning the distribution of schooling fish around benthic structures, such as sewage outfalls (Grigg, 1994) and subsea pipelines associated with the oil and gas industry (Bond et al., 2018). These artificial habitats present linear sources of structure (rather than point sources, as in our case) which may be used by schooling fish in a similar way to artificial reefs (Love & York, 2005; McLean et al., 2017).

Typical approaches to monitoring pipelines have involved the use of manned submersibles

(Love & York, 2005), remote operated vehicles (ROV) (McLean et al., 2017), and baited remote underwater video (BRUV) (Bond et al., 2018), which provide species-level abundance data and, with stereo methods, can be used to estimate fish biomass. However, for more routine monitoring of fish distribution around these structures, MBES and summary metrics such as school thickness are likely to be more efficient. An additional benefit is that MBES do not affect fish behaviour in the way that manned-submersible, ROV and BRUV methods can (Schramm et al., 2020). MBES can also be deployed on autonomous underwater vehicles (AUVs) and kept at a sufficient distance from targets to survey with

121

high replication in deep water (Dunlop et al., 2018). Although echosounders cannot match the species-level detail provided by underwater video, the superior survey efficiency provided by MBES could be used to increase the frequency of surveys for applications where high temporal resolution is necessary.

122

5. Fine spatial and diel dynamics of zooplanktivorous fish on temperate rocky and artificial reefs

5.1. Abstract

In rocky reef ecosystems, plankton is an important component of the food web. Ocean currents act like conveyor belts, supplying plankton to resident and reef-associated planktivorous fishes, and thus subsidising local production. Measuring the fine-scale distribution of these schooling fishes provides insight into their habitat use and how they balance risk and reward while foraging. Maintaining their proximity to benthic structure provides refuge from predation but may also limit foraging opportunities. We used a portable multibeam echosounder to survey schooling fish at five natural and three artificial reefs, during day and night, and under varying current conditions. We isolated midwater acoustic targets and used binomial generalised linear models to explain the distribution of schools as a function of current exposure, distance from structure and seafloor complexity.

We also isolated individual schools and used generalised least squares to measure how school characteristics differed between night and day, using spatial metrics of school area, perimeter length and height above the seafloor. Our results revealed that the occurrence of schools was almost twice as likely upstream versus downstream of artificial reefs, although distance to structure was the main driver. School occurrence was also more likely on artificial versus natural reefs. Schools at artificial reefs exhibited greater volume and areal coverage at night and rose higher in the water column while aggregating around the reef during the day. These findings suggest that artificial and natural reefs featuring enhanced vertical relief and direct exposure to the prevailing current are preferred habitats for planktivorous fish.

123

Keywords: artificial reefs, Atypicthys, echosounder, multibeam, reef fish, spatial distribution, Trachurus, WASSP, zooplanktivores

5.2. Introduction

Reef ecosystems are supported by a combination of benthic primary productivity and pelagic subsidies in the form of plankton delivered by ocean currents (Morais & Bellwood,

2019). In some temperate reef ecosystems, plankton supports the bulk of the fish community and over half of fish biomass (Truong et al., 2017; Holland et al., 2020b). While herbivores within these systems receive their energy from local benthic production, the overall contribution of planktonic production to reef food webs is often much greater because it is generated over a much larger spatial area (Docmac et al., 2017; Udy et al., 2019; Zuercher &

Galloway, 2019). Ocean currents deliver a large supply of plankton to coastal reefs, with supply influenced by multiple processes including upwelling and productivity, current speed, and upstream consumption. This planktonic subsidy can support large abundances of schooling and reef-associated planktivorous fish, especially when reefs exist as pockets of structural complexity in otherwise relatively featureless sandy environments (Morais &

Bellwood, 2019; Zuercher & Galloway, 2019).

Due to their high abundance and frequency as prey, planktivorous fish are a key pathway by which planktonic energy supports reef food webs (Bakun, 2006; Pikitch et al., 2014). To reduce their individual risk of predation, these fish form schools or aggregations and often maintain their proximity to structure while feeding on drifting zooplankton (Motro et al.,

2005; Yahel et al., 2005b; Paxton et al., 2019). Proximity to structure provides access to physical refuges where fish can shelter to escape or avoid predators. An indirect benefit of aggregation behaviour results when schools associate with the same object (Sandin &

Pacala, 2005), providing increased vigilance, dilution of individual risk and increased predator confusion (Morgan & Godin, 1985). However, individual fish must also compete

124

with their neighbours for access to this dispersed food source and there may be a benefit to spreading out and reducing their density, with the trade-off of increased distance to reef structure and consequently increased predation risk (Motro et al., 2005).

Zooplankton are delivered to reefs by currents, where they are encountered by reef- associated planktivorous fish, which graze on them as they pass over the reef. This process, often referred to as the ‘Wall of Mouths’ (Hamner et al., 1988), has been shown to deplete zooplankton density across tropical coral reefs (Kiflawi & Genin, 1997; Holzman et al., 2005;

Motro et al., 2005; Yahel et al., 2005b) by up to 45% h-1 (Gal, 1993). Tropical coral reefs tend to exhibit common geomorphology, from open ocean to fore reef to reef crest to reef flat. Gradients in zooplankton density are possible here because currents bearing pelagic zooplankton must first pass from the open ocean over the reef crest before reaching the reef flat. However, temperate rocky reefs exhibit very different geomorphology, with no such analogous barrier or bottleneck, making such large-scale gradients in zooplankton density unlikely. While such large-scale zooplankton depletion may be challenging to observe along rocky reef coastlines, localised depletion of zooplankton density has been observed across individual temperate rocky reefs (Kingsford & MacDiarmid, 1988; Paxton et al., 2019).

As zooplanktivore feeding rates can be highly correlated with zooplankton density (Kiflawi

& Genin, 1997), localised depletion may be an important process structuring the small-scale distribution of fish schools. An obvious solution to improve individual foraging success within a fish school is to disperse, both vertically and horizontally. However, with the limited vertical relief of many temperate rocky reefs, spreading out vertically requires fish to feed higher in the water column, thus increasing their distance from the safety afforded by physical structure (Motro et al., 2005). Similarly, in the case of isolated reefs such as patch reefs and artificial structures, spreading out horizontally also increases distance from reef structure.

125

Designed artificial reefs are increasingly being deployed for fisheries enhancement and habitat restoration purposes in coastal marine environments, and are incorporating design features intended to improve recruitment and sustain large schools of associated zooplanktivores (Sherman et al., 2002). These reefs are usually deployed over relatively bare, soft sediment environments to create hard structure where none previously existed.

Reef location is important for fish recruitment (Komyakova et al., 2019), as location influences current flow and plankton supply. Reef size and design also influence fish abundance, species composition, and patterns of space use, with effective artificial reef designs generally incorporating internal spaces (Sherman et al., 2002) and structurally complex features intended to exaggerate the limited vertical relief found at natural reefs

(Rilov & Benayahu, 2002; Davis & Smith, 2017). Artificial reef modules are often arranged in such a way that multiple modules interact to form a ‘reef field’, greatly increasing the horizontal footprint of the reef (Becker et al., 2019). These features facilitate enhanced attraction and recruitment by offering access to a greater proportion of the water column and a wider horizontal footprint for fish that associate with hard structure.

The large vertical relief of many artificial reefs is intended to extend habitat into the water column, allowing schooling fish to feed while maintaining proximity to structure, and thus facilitate the conversion of additional planktonic energy into the production of fish biomass.

Both designed and incidental artificial reefs (e.g. shipwrecks, oil rigs) that contain tall vertical features are consistently found to support the greatest abundance and biomass of fish (Rilov & Benayahu, 2000; Claisse et al., 2014). However, only a few studies have examined how schooling fish actually use this enhanced vertical relief (Sala et al., 2007;

Champion et al., 2015; Becker et al., 2019).

The goal of this study was to evaluate schooling behaviour during both day and night using multibeam echosounders (MBES) around rocky reef ecosystems. We used artificial reefs as

126

model systems to document how fish aggregate around isolated high vertical relief structures and included nearby natural reefs to determine whether aggregation behaviours measured at artificial reefs could also be observed across less abrupt natural topography.

We used MBES to measure variation in fish school distribution and characteristics and used remote underwater video to determine fish composition. We present our observations of fish school distribution to improve our understanding of fish behaviour and better understand how artificial reefs alter habitat use and contribute to fish production.

We conducted two separate surveys. The first examined differences in school distribution relative to the location of hard structure and with variation in both bathymetry and current velocity, and the second survey measured variation in school characteristics, including school size and shape, number, dimensionality and position (both horizontal and vertical), at night and during the day. Our specific aims were to: (1) examine school position relative to fine-scale variation in benthic hard structure and how it varies in response to current direction; (2) determine whether school aggregation behaviour varies in response to current velocity; (3) quantify differences in the dominance of schooling fish at an artificial and a natural reef; (4) measure differences in the distribution and schooling behaviours of fish between night and day at artificial and natural reefs; and, (5) quantify the amount of zooplankton supplying an artificial reef and examine the effects of zooplankton supply on school aggregation.

5.3. Methods

5.3.1. Site selection

We conducted all fieldwork off the coast of Sydney, Australia (33.87° S, 151.21° E), where

41% (and up to a maximum of 71%) of fish biomass around rocky reefs consists of schooling zooplanktivores (Truong et al., 2017). In total, eight sites were selected (Figure 5.1)

127

incorporating the two separate surveys. The sites consisted of five natural reef locations, a shipwreck and two designed artificial reefs. The two designed artificial reefs were deployed by the NSW Department of Primary Industries (NSW DPI) to increase fishing opportunities for recreational anglers, with one constructed from steel (Scott et al., 2015) and the other consisting of a field of concrete modules (Becker et al., 2019).

Of these eight sites, six were selected to examine patterns in the distribution of schooling zooplanktivores relative to artificial reef structures and specific benthic habitat variables

(Aims 1, 2; Table 5.1). Two of these six sites were artificial structures. The additional four natural reef sites were selected from analysis of bathymetry data collected by New South

Wales Office of Environment and Heritage (NSW OEH) and encompassed a range of slopes and water depths around a central point, typically the point of greatest relief relative to the surrounding seafloor. This point was used as the centroid of each surveyed area. Due to the shelf-like morphology of reefs in this region, the four natural reef sites covered only a portion of much larger contiguous reefs. Also owing to morphology, areas surveyed consisted of roughly half rocky reef, dropping away to relatively flat, sandy seafloor.

One artificial reef and one natural reef site were selected to quantify differences in the characteristics of fish schools and fish assemblage at a natural and an artificial reef at night and during day (Aims 3, 4, 5; Table 5.1). These two relatively isolated sites were selected for safety reasons due to the requirements of working at night. The natural reef selected for comparison was the only site in the vicinity of the artificial reef with a similar range of vertical relief.

128

Figure 5.1 The locations of the three artificial ( ) and five natural ( ) reef sites used to study fish school distribution ( ) and characteristics ( ) and their general location in Australia ( ). Shaded panels with contours indicate bathymetry and structures of the eight corresponding site locations. The outlines of artificial reef structures (panels: OAR, AM and JDN) are also indicated, with OAR indicating the position of concrete anchor blocks ( ) and the steel reef structure ( ), AM indicating the position of the shipwreck lying on its side, and JDN indicating the position of concrete modules with ( ) and without towers ( ). Contours on main chart indicate a 25 m change in bathymetry while contours on panels indicate a 2 m change.

129

Table 5.1 Site names, locations and descriptions, and the aims they were used to address.

Site Reef Coordinates Survey Aims Name and year Survey dates Site description Data type of deployment

OAR Artificial 33.8466° S School 1, 2 Sydney Offshore 23/10/2018 Singular steel lattice Multibeam acoustics 151.2998° E distribution Artificial Reef, 06/11/2018 structure with two 2011 19/11/2018 towers Remote underwater 21/01/2019 video Structure height: 4 m Drogue coordinates Total height: 12 m

Mean depth: 36 m

AM Artificial 33.8666° S School 1, 2 SS Annie M 23/10/2018 48 m long historic Multibeam acoustics 151.2984° E distribution Miller, 1929 06/11/2018 shipwreck 19/11/2018 Remote underwater 21/01/2019 Structure height: 5 m video

Mean depth: 42 m Drogue coordinates

N1 Natural 33.8492° S School 1, 2 - 23/10/2018 Mean depth: 26 m Multibeam acoustics 151.2942° E distribution 06/11/2018 19/11/2018 Min depth: 20 m Remote underwater 21/01/2019 video Max depth: 32.5 m Drogue coordinates

130

Site Reef Coordinates Survey Aims Name and year Survey dates Site description Data type of deployment

N2 Natural 33.9410° S School 1, 2 - 23/10/2018 Mean depth: 34 m Multibeam acoustics 151.2751° E distribution 06/11/2018 19/11/2018 Min depth: 30.5 m Remote underwater 05/12/2018 video Max depth: 38 m Drogue coordinates

N3 Natural 33.9588° S School 1, 2 - 23/10/2018 Mean depth: 27.5 m Multibeam acoustics 151.2685° E distribution 06/11/2018 19/11/2018 Min depth: 18.5 m Remote underwater 05/12/2018 video Max depth: 34.5 m Drogue coordinates

N4 Natural 33.9728° S School 1, 2 - 23/10/2018 Mean depth: 33 m Multibeam acoustics 151.2681° E distribution 06/11/2018 19/11/2018 Min depth: 28 m Remote underwater 05/12/2018 video Max depth: 39 m Drogue coordinates

131

Site Reef Coordinates Survey Aims Name and year Survey dates Site description Data type of deployment

JDN Artificial 34.0943° S Diel effects on 3, 4, 5 John Dunphy 14/08/2019 Five clusters of pre- Multibeam acoustics 151.1776° E school Reef, North 03/09/2019 formed concrete characteristics cluster, 2017 26/09/2019 modules. One tower Remote underwater 31/10/2019 per cluster video 08/11/2019 Total height: 9 m Vessel coordinates

Structure height: 5 m Zooplankton net samples

N5 Artificial 34.0756° S Diel effects on 3, 4, 5 - 14/08/2019 Mean depth: 25.5 m Multibeam acoustics 151.1824° E school 03/09/2019 characteristics 26/09/2019 Min depth: 20.5 m Remote underwater 31/10/2019 video 08/11/2019 Max depth: 27.5 m Vessel coordinates

Zooplankton net samples

132

5.3.2. Data collection

5.3.2.1. School distribution (Aims 1, 2) School distribution surveys (six sites; Table 5.1) were conducted from a 6.5 metre vessel.

Five surveys were conducted between 23 October 2018 and 21 January 2019 (Table 5.1). A

WASSP WMB-1320Fi portable 160 kHz multi-beam echosounder (WASSP Limited,

Auckland, New Zealand) was used to record tide-corrected bathymetry and water column targets simultaneously. The transducer was side-mounted to a portable purpose-build aluminium pole with an integrated Hemisphere Vector V103 Smart Antenna (Hemisphere

GNSS, Scottsdale, United States) which adjusted for pitch, heave and roll in real-time.

Digital tidal charts for the local area were programmed into the WASSP software to adjust for tidal height in real time. For each survey, eight parallel transects were conducted at ~2.5 m s-1, initially aligned into the prevailing swell to ensure the vessel remained on a straight course (Supplementary Figure 8.5.1a). The central transect was repeated twice to ensure that the artificial reef structure, or point of greatest prominence at the four natural reefs, was ensonified in at least one of the two passes. If both central transects had suitable coverage of the reef structure, one was randomly discarded.

To characterise current velocity and bearing, a remote camera drogue was deployed upon the initial pass of each site. This drogue consisted of a 40 cm diameter spherical polystyrene float attached to a variable length of 1 cm thick nylon line with a weighted steel frame (20 ⨯

30 cm, 2 kg, Supplementary Figure 8.5.2). The frame supported two opposite-facing GoPro

Hero 4 cameras (GoPro Inc., San Mateo, United States) which were used to determine the species present during each survey. While it was certainly possible for this drogue to be affected by wind or vertically sheared currents (Lumpkin et al., 2017), surveys were conducted on days with negligible wind to improve the quality of acoustic data, so any deviance from the true current bearing was expected to be minor.

133

This video footage was observed in its entirety and species observations were recorded.

Since many video deployments did not observe fish schools, despite schools being detected by the echosounder in all surveys, this footage was only suitable for determining species presence. The deployment duration and coordinates of deployment and retrieval for the drifting camera assembly were used to calculate current velocity and direction based on the haversine formula. To characterise water mass characteristics, temperature, and salinity at the surface, and at 20 m depth, were recorded using a calibrated Hydrolab multi-meter

(OTT Hydromet, Loveland, United States).

5.3.2.2. Diel effects on school characteristics (Aims 3, 4, 5) Five school characteristics surveys (two sites; Table 5.1) were conducted from a 14 m wooden vessel between 3 September and 8 November 2019. The same side-mounted MBES was used to record bathymetry and water column targets, at a vessel speed of ~2.5 m s-1. To ensure a similar total area of seafloor was surveyed at each of the two sites, and given that swath width at the seafloor is dependent on water depth, five evenly spaced parallel along- shelf transects (along lines of constant longitude) were conducted at the artificial reef site

(Supplementary Figure 8.5.1b), and nine transects were conducted at the shallower natural reef site. At both sites, the central transect was surveyed twice. If both central transects had suitable coverage of the reef structure, one was randomly discarded. Each transects was 300 m long and centred at the middle of each reef (Figure 5.1). Each site was surveyed before sunrise, and again after sunrise on the same day. The timing of surveys was scheduled to allow a one-hour buffer either side of sunrise to avoid the morning crepuscular period

(Yahel et al., 2005a; Myers et al., 2016) and subsequent surveys of the same reef were separated by over four hours.

To collect midwater zooplankton, following each acoustic survey two replicate horizontal tows were conducted using a 40 cm diameter circular net with a 200 μm mesh size. The net

134

was towed by 30 m of rope with the rudder locked to cause the vessel to travel in a circular pattern (~50 m radius) over the centre of each reef for a duration of 5 minutes per tow at a speed of 1.25 m s-1. The net had a mechanical flowmeter (General Oceanics, Miami, USA) to calculate the volume of water filtered. Each plankton sample was preserved in a 5% formaldehyde solution for post-processing in laboratory.

At the end of each daytime acoustic survey, a remote camera assembly was deployed to identify the fish species assemblage at the centre of each reef. This equipment was identical to that used to map fish school distributions, however, the opposite-facing cameras were anchored with a lead weight and the length of rope was varied so that the camera would hover in a fixed location ~2.5 m above the seafloor, similar to Sheehan et al. (2020), to maximise the chances of recording fish schools. This assembly was deployed for 30 minutes, during which time temperature and salinity were recorded at 20 m depth and at the surface.

Current bearing and direction were recorded based on drift of the vessel’s GPS position while engines were in neutral. The effect of wind was considered minimal, as the vessel had a 3 m deep keel, and all surveys took place on days with negligible wind to improve the quality of acoustic data.

5.3.3. Water column acoustic data processing

For both school distribution and school characteristics datasets, raw acoustic data were imported into Echoview v10.0 (Echoview Software Pty Ltd, Hobart, Australia) and processed to isolate fish targets from noise and bottom backscatter (Chapter Error! R eference source not found.). We used the bathymetry detection tool in Echoview, along with the known geographic locations of reef structures and their dimensions, to construct bathymetry surfaces which were inclusive of artificial structures. These updated bathymetry surfaces were used to subset midwater acoustic data so that echoes from reef structures and

135

seafloor could be completely excluded. We then calculated median values across every three pings, blurred the data with an XYZ convolution, and applied a threshold of -65 dB to create a 3D Boolean mask to extract the original unaltered data. Georeferenced samples were then exported from Echoview and all further analysis was conducted in R v3.6.3 (R

Core Team, 2020).

Unlike many split-beam echosounders the WASSP cannot be user-calibrated, eliminating the ability to determine fish abundance based on backscattering data; we therefore focused on variation in school thickness as our measurable proxy for abundance (Chapter Error! R eference source not found.). Georeferenced midwater samples were converted to 1 m resolution horizontal rasters using the ‘rasterize’ function from the ‘raster’ package

(Hijmans, 2020) in R. Three raster layers were created for each transect, to represent the minimum and maximum depth of target detections, as well as the ‘school thickness’ for each grid cell. We calculated school thickness by binning samples into vertical depth layers of 1 m thickness, and for each horizontal grid cell, counting the number of layers containing at least one target detection. The rationale behind this approach is that it should be more robust to the presence of vacuoles (Fréon et al., 1992; Paramo et al., 2007; Guillard et al.,

2011), spatially overlapping schools or lone targets overlapping with schools. Because raster resolution was 1 m, this school thickness variable could essentially be considered aggregated school volume for each 1 m2 of seafloor.

For transects undertaken to map school distribution, gridded values were aggregated to 5 m resolution by calculating mean values for each grid cell, to improve computational efficiency of modelling (Figure 5.2). Transects from the same survey were combined into a single raster layer by calculating the mean value of overlapping cells. The extent of this raster was then cropped to a 200 m square grid centred on the centroid of each reef.

Transect data for school characteristics retained the 1 m resolution so image analysis

136

techniques could be used to define fine-scale school boundaries. Transects were processed individually, rather than being combined into single raster layer. These transects were subset to conform to a 300 m square grid centred on the centroid of each reef.

Figure 5.2 Gridded school thickness for four survey dates at the SS Annie Miller (AM – top row) and Sydney Offshore Artificial Reef (OAR – bottom row). Vectors extending from the origin of each raster represent the current over a 5-minute period. Reef structures are displayed as black polygons for each panel.

Since our analyses only examined the geographically referenced thickness of schools, we must consider the possibility that there will be variability among surveys, and possibly among transects, in abundance and total fish density contained within measured levels of school thickness. For the purposes of these analyses, we do not make assumptions on the density or abundance of fish observed. Rather, in this case we are primarily concerned with the spatial distribution of fish schools relative to current, the location of fixed structures and changing bathymetry. Although school thickness in this case is likely a poor index of absolute fish abundance, especially among surveys, school thickness is probably better interpreted as relative abundance, especially within surveys (Misund et al., 1995).

137

5.3.4. Raster image processing

5.3.4.1. Image analysis - School distribution (Aims 1, 2) For the study of school distribution (Aims 1, 2; Table 5.1), raw point clouds generated by the WASSP software’s bathymetry detection algorithm were projected into UTM Zone 56 and rasterised over a 200 m square grid with 5 m resolution. Ordinary kriging was used to interpolate over any small holes. The ‘terrain’ function in the ‘raster’ package (Hijmans,

2020) was then used to calculate seafloor aspect and roughness for natural reef sites (Wilson et al., 2007). Roughness was calculated as the difference between the minimum and maximum depths of each cell and an unweighted kernel of its eight neighbouring cells. This measurement assigns greater values to seafloor with steep slope and to seafloor with high complexity, even if the overall slope remains minimal (Wilson et al., 2007).

To represent the degree of exposure to prevailing currents for the artificial reef sites, a continuous raster surface was generated, where cells directly upstream from the reef were assigned a value of 1 and cells directly downstream of the reef were assigned a value of 0

(Supplementary Figure 8.5.3a). We refer to this indicator variable hereafter as ‘relative bearing’. Similarly, for natural reef sites, seafloor aspects fully exposed to the prevailing current were assigned a value of 1, and aspects fully sheltered from the current were assigned value of 0. We refer to this indicator variable as ‘relative aspect’ (Supplementary

Figure 8.5.3b). In both cases, remaining bearings were interpolated between these two values, similar to the derivation of wave exposure in Turnbull et al. (2018). These ‘relative bearing’ and ‘relative aspect’ variables were calculated for each survey to account for differences in current direction among sites and survey dates.

138

5.3.4.2. Image analysis – Identifying schools (Aim 4) For the study of school characteristics, all rasterised transects were processed individually using image analysis techniques and kernel functions to isolate individual aggregations or schools, following the methods of Reid and Simmonds (1993), which outlined image analysis techniques for isolating fish schools from vertical acoustic profiles (explained in 8.5.1

Supplementary methods).

5.3.5. Data analysis

While most of our analysis focused on investigating differences between natural and artificial reefs, it is important to note that differences in the physical variables we measured

(diel behaviour, position in the water column, etc.) may be attributable to fundamental differences between natural and artificial reefs, or they may be due to differences in species assemblages. In this case differences between reef types cover both possibilities and the data at hand does not allow these two alternatives to be resolved.

5.3.5.1. Modelling - School distribution Gridded school thickness from all surveys was combined with gridded bathymetric variables so that school volume could be examined relative to seafloor characteristics. The artificial reefs were fitted separately to natural reefs because they were structurally very different, consisting of high-relief isolated structures surrounded by relatively featureless seafloor. There were naturally many grid cells in a typical survey which did not contain fish, thus data were highly zero-inflated. Because of the zero-inflation of school thickness, and its uncertain association with fish abundance, we modelled the binomial probability (0 to 1) of fish school presence at reef sites using generalised linear models (GLMs).

139

5.3.5.1.1. School distribution at artificial reefs (Aim 1)

The GLM of fish schools around artificial reefs included the following explanatory variables: ‘distance from reef’, ‘relative bearing’, and ‘reef site’ as a fixed effect to account for a difference in school occurrence between the two artificial reefs. Data were collected over 5 survey dates, but we chose to pool all surveys for model fitting, with the rationalisation that general fish distribution patterns are those measured across multiple days, and the assumption that all days contributed equally to the modelled patterns. We compared this full model with a restricted ‘site only’ model, using the Akaike Information Criterion (AIC), to evaluate the contribution of the spatial variables to explained information. We calculated area under the curve (AUC) of the full model to evaluate its goodness-of-fit (Elith et al.,

2006). Finally, we generated a semivariogram from Pearson residuals to visually assess whether spatial autocorrelation was present. This model satisfied all assumptions regarding the normal distribution of residuals and quantile-quantile plots. The semivariogram generated from the final model (Supplementary Figure 8.5.4a) suggested no change in semivariance with increases in spatial separation, indicating that spatial autocorrelation was not a concern.

5.3.5.1.2. School distribution at natural reefs (Aim 1)

The natural reef GLM was similar to the artificial reef GLM, but because these natural reefs did not contain a single point of high relief from which to measure ‘distance from reef’, we included the explanatory variables ‘roughness’ and ‘relative aspect’. In this case, ‘roughness’ was used to examine whether fish school distribution was affected by changes in bathymetry and ‘relative aspect’ was used to determine whether fish school distribution was affected by the relative exposure of seafloor to the prevailing current. This model was subject to the same model validation process as the artificial reef GLM. Like the artificial reef model, the semivariogram generated from the final model (Supplementary Figure 8.5.4b) suggested no

140

change in semivariance with increases in spatial separation, indicating that spatial autocorrelation was not a concern.

5.3.5.1.3. Index of aggregation at artificial and natural reefs (Aim 2)

To measure the effects of current velocity on the aggregation behaviour of schooling fish, we calculated an index of aggregation for each site survey. This index of aggregation (Iagg)

(Bez & Rivoirard, 2001) was calculated to describe the spatial patchiness of fish school distribution at each reef site (Benoit‐Bird et al., 2019). This index is calculated by dividing the survey area into ‘blocks’. We divided the survey area (200 ⨯ 200 m) into 40 m square blocks (5 ⨯ 5 = 25 blocks total) so the centre of the reef would not be split across multiple blocks, and to ensure the total survey area was divisible by block size (1600 m2 each). In this case, we used school thickness as a measure of abundance to calculate the index of aggregation for artificial and natural reefs, generating a single value per survey (Equation

5.1):

N 2 N N (5.1) ti ti Iagg = (∑ ( ) ⁄∑ ( ))⁄∑ ti Si Si i=1 i=1 i=1

Where ti represents the sum of school thickness contained within a block (in m), Si represents the area contained within a block (in m2) and N represents the total number of blocks contained within a survey. This index is a scaleless measure of aggregation which does not require defined boundaries (Bez & Rivoirard, 2001). This measure is essentially an index of crowding and is normalised by the total abundance recorded at the site. We then compared this index with current velocity using correlation tests. Essentially, Iagg will be low when school thickness is spread over a larger number of blocks and high when school thickness is spread over a small number of blocks.

141

5.3.5.2. Modelling - Diel effects on school characteristics 5.3.5.2.1. Diel effects on school size and abundance (Aim 4)

To examine differences in the size and abundance of schools between sites and between night and day, we used the extracted fish aggregations relative to the total seafloor area surveyed (after cropping) to calculate the number of distinct schools per unit area, and the average area and volume contained within a single school. In cases where school width exceeded swath width, the average area of schools would be underestimated, however, it would still provide a relative area for comparison across transects. Mean per transect values were used as the response variables in three generalised least squares models, with the following structure (in script notation):

gls(Response ~ diel ∗ reef + date, correlation (5.2) = corGaus(formula = ~ mean_x + mean_y | survey), method = ′ML′)

Where ‘Response’ is either schools per unit area, area per school or volume per school, ‘diel’ is a factor with two levels for diel period (night or day), ‘reef’ is a factor with two levels (N5 or JDN), ‘date’ is a factor intended to account for day-to-day variability (5 levels) consistent across diel period and sites. Date was included as a fixed, rather than a random, effect because this model type does not allow for specifying random effects in addition to spatial correlation structure. In this case, a Gaussian spatial correlation structure was included to account for possible spatial autocorrelation of residual values. Values for the correlation structure (‘mean_x’ and ‘mean_y’) were calculated as the centroids for each transect for each survey. Model residuals were assessed for normality and fitted models were compared to identical models with ‘date’ as the only predictor (lacking the interaction term ‘diel * reef’) using analysis of variance (ANOVA) and through comparison of Akaike information criterion (AIC).

142

5.3.5.2.2. Diel effects on school vertical position (Aim 4)

To test for differences in the vertical distribution of fish schools, extracted fish aggregations from the image analysis process were used to calculate mean values per transect for school thickness and height above bottom. Height above bottom was calculated as the difference between seafloor depth and the mean of the deepest and shallowest target. Both values were calculated with a resolution of 1 m2 in both horizontal dimensions and the per transect mean was calculated for each variable. Two generalised least squares models, representing school thickness and height above bottom, were generated with identical model structure and validation as in the previous example (Equation 5.2).

5.3.5.2.3. Diel effects on school aggregation (Aim 4)

Extracted fish aggregations from the image analysis process were used to calculate the perimeter-area and the area-volume ratios for each distinct school. Whereas the metrics of

‘area per school’ and ‘volume per school’ used in the first model indicate school size, these metrics indicate the diffuseness of schools. The mean per transect values for perimeter-area and area-volume ratios were used as response variables in two generalised least squares models, as above (Equation 5.2). In cases where school width exceeded swath width (17% of all cases), the straight sides of the cropped swath were classified as school perimeters.

Similarly, we examined the horizontal position of schools relative to the centre of each reef.

In this case, smaller mean distance should be indicative of stronger aggregation around the reef. We calculated the distance from each school centroid to the reef centre. In the case of the artificial reef, we considered the reef centre to be the centroid of the reef module clusters, and in the case of the natural reef the centre was the peak of bathymetry. We generated a generalised least squares model with identical structure to the previous examples (Equation 5.2), but with the spatial autocorrelation structure represented by the school centroid coordinates.

143

5.3.5.3. Underwater video analysis 5.3.5.3.1. Video processing (Aim 3)

The video footage was analysed by subsampling each thirty-minute camera deployment by five intervals of two minutes each. Time intervals were selected by randomly sampling five timestamps in R from the duration of each video and assigning these as the midpoint of each interval, as in Basford et al. (2016). To reduce issues introduced by temporal autocorrelation, each sample had a gap of at least two minutes from adjacent samples. Video samples were viewed in VLC (VideoLAN, Paris, France) and for each sample the maximum number of individuals within a species that could be observed within a single frame (MaxN) was recorded. In cases where large schools were observed, screenshots of video were imported into ImageJ (National Institute of Health, Washington D.C., USA) and fish were counted using the ‘multi-point’ counting tool. Since footage from the two opposite-facing cameras could essentially be considered part of the same recording, the maximum abundance for each species observed across both cameras was recorded as the MaxN for each time interval (Becker et al., 2019).

Video MaxN data was converted to a relative abundance matrix for the five schooling species which were recorded over the course of the study. Species recorded on only one occasion and non-schooling species (e.g. blue-spotted flathead (Platycephalus caeruleopunctatus), red morwong (Cheilodactylus fuscus), Port Jackson shark (Heterodontus portusjacksoni)) were excluded from analysis. Relative abundance data were then used in a multivariate analysis to examine differences in the fish community assemblage between the natural and the artificial reef. Further univariate analysis was subsequently conducted to resolve potential drivers of differences and similarities.

144

5.3.5.3.2. Multivariate differences in community composition (Aim 3)

For the multivariate analysis, the relative abundance matrix was used to generate a generalised linear latent variable model (GLLVM) using the R package ‘gllvm’ (Niku et al.,

2020). A Poisson family distribution was used to model this count data, including an observation-level (i.e. row) factor to control for differences in abundance among surveys.

The model used the following formula (in script notation):

gllvm(y = Abundance, X = environmental_variables, num. lv = 2, family (5.3) = poisson, row. eff = TRUE, formula = ~ date + reef)

Where ‘Abundance’ is the matrix of species relative abundances (n = 50 observations),

‘environmental_variables’ is a matrix of environmental variables, containing factors for

‘date’ and ‘reef’. In this case, ‘date’ was included as a fixed factor, since this model type did not allow for this type of random effect, with one level per survey date. The variable ‘reef’ was coded as a factor with two levels, one for each of the two reefs surveyed. Ordination plots were generated from this model but with the ‘reef’ term removed and covariate coefficient plots were generated with the ‘date’ term removed to visualise general patterns.

5.3.5.3.3. Univariate differences in biodiversity (Aim 3)

For the univariate analysis, this relative abundance matrix was used to calculate total fish abundance (by summing the MaxN for each species observed), species richness and the

Shannon diversity index for each video sample. These three variables were used as the response variable for three generalised least squares (GLS) models with the following structure (in script notation):

gls(Response ~ reef + date, correlation (5.4) = corCAR1(form = ~time |survey), method = ′ML′)

Where ‘Response’ (n = 50 observations) is either the abundance, species richness or

Shannon diversity index of samples. In this instance, we defined a continuous AR1

145

correlation structure (corCAR1), with ‘time’ as the video time stamp and ‘survey’ as a grouping factor for each video deployment (each set of five samples).

For validation, each multivariate and univariate model was assessed for normality of residuals and was compared to an identical model with ‘date’ as the only predictor

(excluding the ‘reef’ term) using analysis of variance (ANOVA) and through examination of

ΔAIC.

5.3.5.4. Zooplankton density and flux 5.3.5.4.1. Zooplankton sample processing (Aim 5)

Zooplankton samples were rinsed of formaldehyde and subsampled using a 0.5 litre Folsom plankton splitter. Samples were split between 1/2 to 1/8 of their original volume, based on a visual assessment of sample density. Samples were processed through a lab-based Laser

Optical Particle Counter (LOPC) (Rolls Royce Canada Ltd, Peterborough, Canada) using a header tank, water pump and proprietary software from the manufacturer (Moore &

Suthers, 2006). Post-processing of the binary LOPC output files was performed using

MATLAB (MathWorks, Natick, United States). Equivalent spherical diameter (ESD) of measured particles was used to calculate volume contained within each particle, assuming

ESD to be the longest dimension of an oblate ellipsoid with a 3:1 ratio. These volumes were converted to biomass density (mg m-3) using the density of water, the volume of water filtered in each sample and the number of times each sample was split (Suthers et al., 2006).

The particles able to be accurately recorded ranged in size from 346 to 30,000 μm ESD. All particles measured in this size range were included in our analyses of biomass density.

5.3.5.4.2. Zooplankton density and flux analysis (Aim 5)

To confirm whether zooplankton samples were representative of the prey resident zooplanktivores were feeding on and to alleviate concerns regarding net avoidance, we compared the size distribution of net samples with size distribution obtained from a local

146

diet analysis (Schilling et al. unpublished data). Average size distributions for day and night zooplankton samples were generated to be compared to similar size distributions obtained from the gut contents analysis for three of the most abundant species of zooplanktivore in the region.

To determine the rate of zooplankton delivery to the artificial reef (JDN), we calculated the mean current bearing across all surveys and estimated the profile area of the reef field perpendicular to this bearing using 95% confidence intervals for the mean height above bottom of fish targets as the vertical dimension. We were then able to calculate the rate of zooplankton delivery to the artificial reef (in g s-1), by multiplying the biomass density (in g m-3) by the current velocity (in m s-1) to derive the flux of zooplankton biomass (in g m-2 s-1) and multiplying by the profile area of the reef. Because the natural reef site contained contiguous habitat not entirely covered by our surveys, it was not possible to delineate site boundaries to calculate rates of zooplankton delivery to this site.

We also examined differences in zooplankton biomass and abundance (from lab-processed zooplankton samples) between night and day at natural and artificial reef sites as potential evidence of localised depletion by zooplanktivores. We expected abundance and biomass at night to be higher than during the day and to be similar between the two sites, as zooplanktivores are visually hunting predators and are often unable to feed effectively at night (Yahel et al., 2005a). Furthermore, we expected higher biomass and abundance at the natural reef site during the day due to localised depletion from the larger abundances of schooling fish at the artificial reef site. While additional zooplankton sampling in the absence of fish may have been useful in identifying whether night versus day differences were a direct result of depletion by fish, this was not possible due to the timing constraints of pre-dawn surveys.

147

Finally, we calculated the index of aggregation for schooling fish (Equation 5.1), using the volume contained in isolated schools as the abundance variable and a block size of 10 m. We used this index as the response variable in two GLS models to examine possible effects of zooplankton biomass density and biomass density flux on fish aggregation behaviour. These models were evaluated in an identical manner to previously described GLS models

(Equation 5.2).

5.4. Results

5.4.1. School composition and diversity

From the video footage obtained by the drift camera assembly used in the evaluation of school distribution, only two fish species (Trachurus novaezelandiae and Pseudocaranx georgianus) were observed at artificial reefs and these fish were only observed during one survey, despite visibility being >10 m on most deployments. At the natural reef sites, six species of schooling fish were observed (Trachurus novaezelandiae, Pseudocaranx georgianus,

Chromis hypsilepis, Scorpis lineolata, Atypicthys strigatus and Seriola lalandi). Overall, these drifting midwater video cameras had a low rate of success in observing fish schools. Fish schools were detected by the MBES during all surveys; however, they were only captured on drifting video in 33% of surveys.

Greater success was achieved using the benthic deployed camera when evaluating school characteristics, where five species were detected (Trachurus novaezelandiae, Pseudocaranx georgianus, Chromis hypsilepis, Scorpis lineolata and Atypicthys strigatus). All species except P. georgianus were zooplanktivores (Froese & Pauly, 2009). The covariate coefficient plot indicates that T. novaezelandiae was just as likely to be found at either type of reef, whereas all other taxa were more likely to be found at N5 (Figure 5.3). The GLLVM was found to outperform the null model when the fixed term for reef site was included as a predictor

148

(ΔAIC = 65.1, p < 0.001). This indicates that the artificial reef and the natural reef contained statistically distinct fish community composition.

Figure 5.3 Covariate coefficient plot generated from the Generalised Linear Latent Variable Model (GLLVM), indicating mean values (red dots) and confidence intervals for the effect of reef site on MaxN of the five schooling species detected. The plot indicates that T. novaezelandiae was just as likely to be observed on either type of reef, while the four other species were more likely to be observed at the natural reef (N5). Some symbols do not display confidence intervals due to symbol size. Images obtained from efishalbum.com.

149

The ordination shows the clear difference in composition, especially along LV1 axis (Figure

5.4). There was also some clustering of survey date, indicating structure at the weekly- monthly scale. Coefficient plots and confidence intervals for each species plotted against the factor for reef site indicate that these differences were driven by highly abundant T. novaezelandiae at the artificial reef and by S. lineolata, C. hypsilepis and P. georgianus at the natural reef. A. strigatus was just as likely to be found at either reef, as their confidence interval spanned zero.

Figure 5.4 Ordination plot output for the two-variable generalised linear latent variable model of relative fish abundance. Points are coloured by reef site and the ellipses represent 95% confidence intervals for each of the two reef sites. Symbols represent the survey dates when video footage was collected.

The artificial reef had greater total fish abundance than the natural reef (Figure 5.5a) (coeff.

= 104, p = 0.019) (Supplementary Table 8.5.1). However, the artificial reef site also hosted

150

lower species richness (Figure 5.5b) (coeff. = -1.44, p < 0.001) and lower Shannon diversity

(Figure 5.5c) (coeff. = -0.49, p < 0.001).

Figure 5.5 Generalised least squares model effects plots displaying mean values (symbols) and 95% confidence intervals for the interaction of reef site with total fish abundance (a), species richness (b), and Shannon diversity index (c) based on video data. Raw data values for each factor have been jittered to aid visualisation. Asterisks above each panel indicate significance level (*: p < 0.05, **: p < 0.01, ***: p < 0.001).

5.4.2. School distribution

5.4.2.1. Model results 5.4.2.1.1. School distribution at artificial reefs

For the GLM examining fish school presence at the artificial reef, the full model had a much lower AIC than a model including only site, indicating that distance and bearing to reef were important (Table 5.2).

151

Table 5.2 Results for the binomial GLMs describing the distribution of schooling fish around artificial and natural reefs. Model terms are as follows: presence – grid cell with fish detected (1) or no fish detected (0), reef_site – a factor for the individual reef sites, cell_dist – cell distance (in m) from the reef centre, rel_bear – bearing of cell from the reef centroid relative to prevailing current, rough – seafloor roughness (m), rel_asp – relative exposure of bathymetry aspect to the prevailing current. Columns are as follows: R df – residual degrees of freedom, N-R dev – difference between null and residual deviance, ΔAIC null – ΔAIC between the full model and an intercept-only model, ΔAIC reef site – ΔAIC between the full model and a model containing only the reef site term, AUC – area under the curve.

GLM formula R df N-R dev ΔAIC null ΔAIC reef site AUC Predictor Estimate SE z value Pr(>|z|)

presence ~ reef_site + 12796 1854 1848 1839 0.74 intercept 0.658 0.074 8.936 <0.0001 **** cell_dist + rel_bear reef_site OAR -0.163 0.045 -3.628 <0.001 ***

cell_dist -0.032 0.001 -38.050 <0.0001 ****

rel_bear 1.002 0.079 12.734 <0.0001 ****

presence ~ reef_site + 25443 1353 1343 403 0.69 intercept -2.063 0.056 -36.730 <0.0001 **** rough + rel_asp reef_site N2 -0.162 0.046 -3.510 <0.001 ***

reef_site N3 -1.595 0.063 -25.450 <0.0001 ****

reef_site N4 -0.932 0.053 -17.580 <0.0001 ****

rough 0.322 0.018 18.270 <0.0001 ****

rel_asp 0.551 0.068 8.074 <0.0001 ****

152

Effects plots from the artificial reef model indicate that there was a slightly greater probability of fish schools occurring in any 5 ⨯ 5 m cell (Figure 5.6a) at AM versus OAR

(Table 5.2; p < 0.001). The probability of school occurrence declined with increasing distance from the reef (Figure 5.6b). Although there was a bearing signal (with schools almost twice as likely upstream as downstream; Figure 5.6c), it was predominantly distance to structure that determined the occurrence of fish schools.

Figure 5.6 Partial effects from the GLM examining the probability of occurrence of schools at the artificial reefs, showing the fixed effect of the two reef sites (a), ‘distance’ from the reef (b) and ‘relative bearing’ from the reef (c; 0 = downstream, 1 = upstream). Error bars in (a) and ribbon in (b) and (c) represent 95% confidence intervals. Rug plots for the x-axes indicate the distribution of the raw data used to create the model.

5.4.2.1.2. School distribution at natural reefs

For the GLM examining fish school presence at the natural reefs, AIC declined with removal of the roughness term (ΔAIC = -315) and also with removal of the relative aspect term (ΔAIC = -251). This outcome suggests that all terms contained in this model were important.

For this natural reef model, effects plots indicated variability in the probability of school occurrence amongst reef sites (Figure 5.7a), with the greatest probability at N1 and the

153

least at N3. The probability of school occurrence was positively correlated with seafloor roughness (Figure 5.7b). The response also indicated positive correlation with relative aspect, suggesting that patches of seafloor more exposed to current had a greater probability of school occurrence, however, the magnitude of the effect was less important than the effect of roughness, and less strong than observed on artificial reefs.

Figure 5.7 Partial effects from the GLM examining the probability of occurrence of schools as a binomial family GLM, with plots for the fixed effect of the four reef sites (a), seafloor ‘roughness’ (b) and ‘relative aspect’ (c; 0 = minimum exposure, 1 = maximum exposure). Error bars in (a) and ribbon in (b) and (c) represent 95% confidence intervals. Rug plots for the x-axes indicate the distribution of the raw data used to create the model.

5.4.2.2. Effect of current velocity on index of aggregation For natural reef sites there was positive correlation between current velocity and index of aggregation across surveys, however, this correlation was not statistically significant for artificial reef sites (Figure 5.8) (natural: df = 14, r = 0.44, p = 0.091; artificial: df = 6, r =

0.69, p = 0.056). The proportion of surveyed area containing schools was also negatively correlated with current velocity (df = 6, r = -0.74, p = 0.035). This correlation was not present at natural reef sites (df = 14, r = -0.14, p = 0.593). However, given the low number of replicates, more data is needed to properly test the relationship between current velocity and aggregation.

154

Figure 5.8 Index of aggregation (unitless) for each individual site survey and line of best fit for artificial and natural reefs, separately.

5.4.3. Diel effects on school characteristics

5.4.3.1. Diel effects on school size and abundance Fish school attributes measured with the MBES were examined using GLS models to test for differences in spatial characteristics between sites and between night and day

(Supplementary Table 8.5.2). The GLS model for mean area covered by a single contiguous school (m2) (Figure 5.9a) indicated a significant interaction of diel period and reef site, such that schools associated with the artificial reef had a smaller mean area during the day (coeff.

= -1615.1, p = 0.008).

155

Figure 5.9 Generalised least squares model effects displaying mean values (symbols) and 95% confidence intervals for the interaction of reef site (N5 and JDN) and diel period (night: blue and day: red) with variables describing school size and abundance, including mean area of an individual school (a), the proportion of surveyed area containing schools (b), the mean volume of an individual school (c), and the mean number of schools per hectare of area surveyed (d). Significant fixed effects (Site, Diel) or variable interactions (Site:Diel) are indicated above each panel, along with significance level (*: p < 0.05, **: p < 0.01, ***: p < 0.001). Raw data are included for each combination of factors. Note, all models also contain a fixed effect for survey date to account for day to day variability (not shown).

156

There was a significant interaction of diel period and reef site (Figure 5.9b), such that the total area covered by schools decreased between night and day at the artificial reef site only

(coeff. = -0.159, p < 0.001). Main effects indicate a greater proportion of the area around the artificial reef contained fish schools when compared to the natural reef site (coeff. = 0.301, p

< 0.001), but there was no overall difference between night and day in terms of the area covered by schools (coeff. = 0.019, p = 0.459).

Mean school volume (m3) (Figure 5.9c) did not differ between night and day (coeff. =

1097.8, p = 0.669), however schools associated with the artificial reef had greater volume overall (coeff. = 14424.6, p < 0.001). There was also no significant interaction (coeff. = -

2978.9, p = 0.461).

There was no difference in the total number of individual contiguous schools (schools per hectare) (Figure 5.9d) between the natural and the artificial reef sites (coeff. = -1.12, p =

0.103). There was a reduced number of individual schools between night and day across both sites (coeff. = -1.82, p = 0.002). This effect did not differ between the two sites as there was no evidence of an interaction (coeff. = 0.42, p = 0.647).

5.4.3.2. Diel effects on school vertical position Mean school height (m) was generally higher during the day, but this was statistically clear only at the artificial reef (Figure 5.10a) (coeff. = 2.57, p < 0.001). The consistency in the direction at both sites also resulted in a significant main effect for diel period (coeff. = 1.36, p < 0.001). The overall height of schools above the bottom did not differ between the natural and the artificial reef (coeff. = 0.27, p = 0.598).

157

There was a significant interaction between diel period and site for the mean vertical thickness of fish schools (m) (Figure 5.10b), indicating that the increase in school thickness between night and day was much greater at the artificial reef than at the natural reef (coeff.

= 3.52, p < 0.001).

5.4.3.3. Diel effects on school aggregation The mean perimeter to area ratio of individual fish schools (Figure 5.11a), which measures the two-dimensional ‘diffuseness’ of a school, was lower during the day compared to night

(coeff. = -0.20, p = 0.041). The perimeter to area ratio of schools at the artificial reef site was less than at the natural reef site (coeff. = -0.35, p = 0.002) and there was no interaction of reef site and time of day (coeff. = 0.09, p = 0.535).

158

The area to volume ratio (Figure 5.11b), which measures the three-dimensional diffuseness of a school, indicated a significant interaction between site and diel period, such that at the artificial reef site this ratio was reduced between night and day (coeff. = -0.10, p = 0.010).

The significant interaction term indicated that schools at the artificial reef had a more compact three-dimensional structure after sunrise at the artificial reef site. This ratio did not differ between the natural and the artificial reef site (coeff. = -0.02, p = 0.417) or generally between night and day (coeff. = -0.01, p = 0.714).

The model for distance from reef centre (Figure 5.11c) indicated a significant effect of reef site, with schools at the artificial reef likely to be closer to the centre of the reef (coeff. = -

15.21, p = 0.018), however there was no significant difference between night and day distribution (coeff. = -6.24, p = 0.292).

159

5.4.3.4. Current and zooplankton density The size distributions of zooplankton samples showed strong overlap with the size distributions of prey obtained from gut contents analysis (Figure 5.12). Our zooplankton net samples collected a greater proportion of zooplankton in the larger size bins than was present in the gut contents of three of the most abundant zooplanktivores species (Schilling et al. unpublished data), suggesting that we adequately captured the range of zooplankters being captured as prey.

160

There was no difference in zooplankton abundance between night and day during our sampling period (± SE) at either N5 (coeff. = 342, p = 0.79) or JDN (coeff. = 659, p = 0.68)

(N5: night = 3745 ± 916, day = 4087 ± 765, JDN: night = 4510 ± 1582, day = 5169 ± 1004 ind. m-3). No significant difference was found in zooplankton biomass between night and day

(± SE) at either N5 (coeff. = 16, p = 0.89) or JDN (coeff. = 231, p = 0.37) (N5: night = 317 ±

56, day = 333 ± 57, JDN: night = 396 ± 121, day = 628 ± 130 mg m-3). While this lack of difference is surprising, considering the universality of diel vertical migration, it may be attributable to the relatively low sample size and natural patchiness of zooplankton distribution. Diel vertical migration is a dynamic behaviour in response to predation, both perceived and actual. A lack of diel vertical migration could be due to patchiness in predation.

Daytime zooplankton biomass density (coeff. = 0.0002, p = 0.97, ΔAIC = -3.95) and biomass flux (coeff. = -0.006, p = 0.81, ΔAIC = -3.59) were not found to affect the index of aggregation for fish schools at either of the JDN or N5 reefs. Furthermore, the zooplankton flux across the reef (± SE) at JDN (1788 ± 317 ind. s-1 m-2) was several times the rate of depletion observed across a reef in a comparable study (Genin et al., 2016), so it is unlikely that

JDN was limited by zooplankton availability.

Based on the mean current velocity and bearing (± SE) (0.33 ± 0.02 m s-1 at 173 ± 10°), and the mean biomass density recorded at JDN (day: 628 ± 130, night: 396 ± 212 mg m-3), we calculated the horizontal flux of zooplankton at JDN (day: 207 ± 43, night: 131 ± 40 mg m-2 s-

1). Based on a rectangular cross-section of the reef profile, calculated using the width of the reef in the dimension perpendicular to the current (136 m) and height above bottom for 95% of samples (day: 4.2 m, night: 3.4 m), net zooplankton delivery to JDN would have been 118 ± 25 g s-1 during the day (or 425 ± 90 kg h-1) and 61 ± 18 g s-1 at night (or 220 ± 65 kg h-1).

Divided evenly over the entire reef area (9800 m2), this works out to 43 ± 9 g m-2 h-1 during

161

day and 22 ± 7 g m-2 h-1 at night. We estimate that this zooplankton supply to JDN, based on the length of the reef cross-section aligned with current (143 m) and current velocity of 0.33 m s-1, would have been completely replenished by current every 7 minutes.

5.5. Discussion

This study reveals dynamic patterns in the behaviour of schooling zooplanktivores around benthic structure, at both artificial and natural rocky reefs. Through measuring variation in the distribution and spatial characteristics of fish schools, we have gained insight into the behaviour of reef-associated schooling fish. Two schooling species dominated the patterns in our study, Trachurus novaezelandiae and Atypicthys strigatus, which were the only species observed in high abundance on the remote cameras. As A. strigatus was less abundant than

T. novaezelandiae, and because A. strigatus generally maintains close proximity to reef structure (Champion et al., 2015), it is likely that the broad patterns we observed in MBES data across the artificial reef complex were primarily driven by T. novaezelandiae. This is consistent with other studies of community composition on artificial reefs in the same region (Scott et al., 2015; Smith et al., 2017).

The large zooplankton flux measured at this artificial reef, and the rate at which it was replenished by the prevailing current, suggests that these zooplanktivores are unlikely to be limited by overall prey availability. While we acknowledge that localised depletion of zooplankton by fish could have impacted our calculations of zooplankton delivery rates, the values we recorded were still relatively high. It is generally accepted that temperate reefs receive more than adequate delivery of zooplankton to achieve maximum growth rates in planktivorous reef fishes, however, this is not always the case (Anderson & Sabado, 1995).

The 43 g m-2 h-1 we calculated for zooplankton delivery to the artificial reef was adequate to sustain the population, considering that an adult (34 g) A. strigatus consumes only 0.77 g

162

day-1 (Champion et al., 2015). If these fish can only visually forage for 12 hours a day, each square metre of reef could potentially support up to 670 individual (or 23 kg) A. strigatus assuming all plankton passing through the reef are consumed, and that zooplankton density is uniform across the reef. Although these assumptions are unlikely (Kiflawi & Genin, 1997), they provide evidence that fish assemblages on large coastal artificial reefs in this region are unlikely to be limited by zooplankton availability.

Despite the large abundance of zooplankton, there was evidence that intra-school competition for preferential foraging opportunities helped structure the distribution of fish schools. Despite major differences in the morphology of artificial reefs and natural rocky reefs, there were similarities in the response of school distribution to variation in benthic structure and water current. In both habitats, schools demonstrated an affinity for vertical relief and a bias towards the upstream side of reefs. This upstream bias suggests fish distribution may be influenced by intra-school prey depletion (Paxton et al., 2019), as fish compete for preferential access to prey delivered by the current. There may be value in exploring other explanations of upstream bias, such as flow dynamics and its influence on the energetics of swimming, or whether these observed patterns reflect innate behaviours which are unaffected by daily variation in prey availability.

5.5.1. Foraging behaviour drives spatial distribution around benthic structure

Some diver-based studies have detected localised zooplankton depletion at multiple locations along a gradient in predation pressure (Kingsford & MacDiarmid, 1988; Motro et al., 2005), although our coarser vessel-based sampling approach was unable to attribute any variation in zooplankton density to factors we tested, including time of day (Ohlhorst,

1982). This study may have been unable to detect small levels of depletion due to sampling large volumes (~50 m3) in mid-water, rather than adjacent to reef structures (which was

163

impractical with a towed plankton net). However, fish schools exhibited distributional behaviours that were indicative of competition for access to prey. Our models indicated that fish showed a tendency to orient to the upstream side of benthic hard structure, possibly to gain ‘first’ access to zooplankton, before it could be accessed by other reef organisms.

This tendency for schooling zooplanktivores to forage upstream of benthic structure has been observed in several studies (Hobson & Chess, 1978; Bray, 1980; Kingsford &

MacDiarmid, 1988; Forrester, 1991; Paxton et al., 2019). As zooplankton are delivered by current, fish have a limited window or ‘reactive volume’ within which to notice and respond to drifting prey items (Kiflawi & Genin, 1997). Fish positioned back from the leading edge of a school have a reduced reactive volume, as their field of vision is obstructed by fish upstream of their position (O'Brien, 1979). They also experience reduced prey density, as fish upstream feed ahead of them, picking off the larger, more visually apparent and thus likely more nutritious prey items (Forrester, 1991). Fish that position themselves at the leading edge of the school may stand to improve their feeding rate. Thus, while the biomass of zooplankton delivered to large reefs may not limit fish abundance, competition for zooplankton may be a key process structuring the distribution of fish around the reef.

Another reason why fish schools may prefer the upstream aspect is the physics of current flow. Artificial reefs are often constructed with specific orientation and geometry intended to disrupt the flow of currents and generate a counter flow ‘wake region’ downstream (Oh et al., 2011). Such was the case with the designed artificial reefs we studied. This feature could make the downstream side of artificial reefs more attractive to zooplanktivores, however, our observations of upstream orientation refute this. On the upstream side of structure, fish can simply maintain their position while monitoring their reactive volume and waiting for zooplankton to drift toward them (Kiflawi & Genin, 1997). On the downstream side,

164

turbulence may generate unstable flow patterns which make it more energetically costly for zooplanktivores to forage or maintain their position.

5.5.2. Accessing the water column

In temperate regions, where a large proportion of reef fish biomass is supported by pelagic subsidies (Truong et al., 2017; Udy et al., 2019; Zuercher & Galloway, 2019; Holland et al.,

2020b), and where the direction of oceanic currents remains consistent over time, designed reefs arranged as linear arrays with the long axis positioned perpendicular to the prevailing current may be able to support increased local production (Champion et al., 2015). This configuration would increase the horizontal length of the leading edge of reef-associated schools, minimising the effects of localised depletion over the reef and thus enhancing the production of zooplanktivores (Champion et al., 2015).

Incorporating enhanced vertical relief into artificial reef design may achieve a similar outcome, by extending the potential foraging space into the vertical dimension. After sunrise, fish schools on the artificial reef rose in the water column and expanded in vertical thickness so the top of schools was 13 m above the seafloor, while at the natural reef this value was only 8 m. This difference suggests that the additional vertical relief provided by the artificial reef (9 m) was used by schooling zooplanktivores to access a greater vertical extent of the water column.

We observed an analogous behaviour of association with vertical relief at natural reefs, as the probability of occurrence of schooling fish was positively correlated with changes in bathymetry (Davis & Smith, 2017). Our modelling estimated only an 8% probability of school occurrence over completely flat seafloor, versus a 37% probability for the maximum roughness we measured (6 m difference in bathymetry within a 15 ⨯ 15 m area). It is likely that schooling fish take advantage of this natural vertical relief because it allows them to

165

feed higher in the water column, while maintaining proximity to refuge. As schooling zooplanktivores will typically remain close to benthic hard structure for refuge from predators, there may exist a vertical gradient in zooplankton predation pressure, with intense predation pressure just above the seafloor (Motro et al., 2005). By providing additional vertical relief, far above what would typically be found on natural reefs, artificial reefs can overcome this limitation.

Our observations of how schooling fish used the enhanced vertical relief at an artificial reef provides evidence for the effectiveness of simple vertical structures. The reef design incorporated steel tower structures with small cross-braces, but no holes or internal refuges. These towers were sufficient to encourage fish to use a greater proportion of the water column. This finding suggests that a simple modification to reef design, through the inclusion of basic steel towers, could improve reef productivity without significantly inflating material and installation costs. However, this should not diminish the importance of internal spaces incorporated into the main structure of artificial reefs, which are still an important feature for more benthic associated species (Sherman et al., 2002).

5.5.3. The effects of current velocity

Feeding rate, or number of particles of food consumed per unit of time, is influenced by a combination of current velocity and particle density (Kiflawi & Genin, 1997). Kiflawi and

Genin (1997) found that feeding rate in site-attached reef pomacentrids peaked at current velocities between 0.06 to 0.12 m s-1. Rather than reaching an asymptote through saturation, higher current velocity caused feeding rate to decline. As we have already demonstrated that schooling reef zooplanktivores adapt their distribution to improve foraging success, if current velocity can influence feeding rate it seems probable that it can also influence school distribution.

166

At natural reefs the distribution of school thickness was more aggregated during surveys which recorded stronger current velocity. During periods with weak current, schools were more dispersed and widespread and during periods of strong current they were more localised. Since the rate with which a fish’s reactive volume is replenished is dependent on both prey density and current velocity (Kiflawi & Genin, 1997), it is possible for prey items within this volume to be rapidly depleted by competing schoolmates when current is weak.

Consequently, when subject to very weak current, fish may be driven to spread out to increase their individual reactive volume.

Foraging at the leading edge of a school presents an energetic trade-off, as fish subject to an oncoming current must expend energy to maintain position. However, less energy is required to maintain position when behind the leading edge of the school (Chen et al., 2016).

In this case, individual fish may take turns varying their positions within a school to prioritise either feeding or rest, similar to behaviours exhibited by flocking migratory birds

(Maeng et al., 2013; Mirzaeinia & Hassanalian, 2019). However, increased school compaction under increased current velocity may not necessarily be traded off with reduced foraging success, as faster prey flux rates would reduce the likelihood of competitors monopolising prey. It would be useful to study whether prey consumption rates co-vary with school shape and prey flux rates.

While current is not a variable that can be manipulated in the design of artificial reefs, careful consideration should be given to reef placement to ensure reefs are deployed in areas exposed to moderate current. While the optimal current velocity measured by Kiflawi and

Genin (1997) would not necessarily apply, more localised studies may be necessary to determine the optimal current velocity to maximise feeding rates in schooling zooplanktivores to inform the placement of new artificial reefs.

167

5.5.4. Nocturnal behaviour

The bulk of schooling fish recorded at the artificial reef, and to a lesser degree at the natural reef, exhibited demersal distribution at night and were settled near the seabed as a thin layer with wide horizontal extent. These fish aggregated and rose to midwater in the early morning. We did not observe a similar degree of nocturnal behaviour at the natural reef site. As there were significant differences in fish community composition between the two reefs, it is likely that this pattern is at least partially driven by T. novaezelandiae, and to a lesser extent A. strigatus, as they were the dominant species identified at the artificial reef in the early morning.

Some reef zooplanktivores have been observed to feed at night (Gladfelter, 1979) which indicates the nocturnal demersal distribution observed here may be in response to the zooplanktivores feeding at night on epi-benthic zooplankton as they emerged from or returned to the benthic substrate (Galzin, 1987; Myers et al., 2016). This is more likely to occur at the natural reef, as highly abundant nocturnal species, such as Pempheris affinis and

Pempheris multiradiata, emerge to feed on zooplankton at night (Annese & Kingsford, 2005).

Due to their diurnal distribution under natural rock overhangs they would have been absent from our remote video deployments. Alternatively, schooling zooplanktivores may assume this distribution at dusk, while light levels are sufficient to facilitate aggregation, and maintain their relative positions throughout the night in the absence of visual cues. This would provide them a means of conserving energy at night while minimising individual predation risk.

Patterns in school characteristics we observed before and after sunrise represent just two instances in a 24-hour cycle, and as such they may only be representative of the times of observation. For example, Sala et al. (2007) observed the greatest density of fish at an

168

artificial reef late at night and in the early morning, but very low densities in the afternoon.

Thus, in order to understand diel behaviour it may be necessary to observe fish distribution over 24-hour cycles with finer temporal resolution of sampling (Myers et al., 2016).

5.5.5. Conclusion

By investigating fine scale patterns in the distribution of schooling zooplanktivores around temperate reefs we have highlighted important behavioural responses to benthic structure, water currents and diel period. These insights into the behaviour of schooling zooplanktivores can be used to inform improved artificial reef designs for fisheries enhancement. Our main findings indicate reefs with large vertical relief will be more attractive to zooplanktivores, and long reefs oriented perpendicular to the prevailing current will enhance the potential influence of planktonic subsidies. However, more research relying on in-situ sampling of zooplankton at multiple distances to artificial reef structures will be required to validate whether the behaviours we observed correspond with actual gradients in zooplankton density, and even more importantly, in individual consumption rates.

169

6. General discussion

The overarching aim of this thesis was to study patterns in distribution of zooplanktivorous fishes at several levels of spatial scale, highlighting the importance of zooplankton in supporting fish assemblages in temperate coastal ecosystems. The broad expectation that the regional distribution of coastal zooplanktivores would be driven by the sustained influence of pelagic subsidies and their ability to support long-term population viability was largely supported. While there was no clear influence of zooplankton density on zooplanktivore biomass at the coarse scale examined in the latitudinal analysis (Chapter 2), zooplankton density did influence the relative contribution of zooplanktivores to the reef fish community. The productive upwelling region downstream of the East Australian

Current separation zone supported the greatest densities of zooplanktivore biomass along the coast, however temperature (represented in Chapter 2 by latitude and seasonality) was a stronger driver of zooplanktivore distribution. Across the shelf, the influence of zooplankton density was less important than habitat and temperature (Chapter 3), which likely imposed physiological limitations on zooplanktivore distribution in spring.

These results also supported the broad expectation that locally across reef habitats, the distribution of zooplanktivores would be driven by the collective behaviour of fish attempting to maximise their individual fitness. Schooling zooplanktivores shifted position in response to variation in current direction and adapted their schooling behaviour to balance perceived predation risk with individual access to prey (Chapter 5). Overall, these findings highlight the often-unexpected differences in the influence of physical and biological factors on the distribution of organisms across varying levels of scale. The suggestion that physical gradients in temperature are a more important influence on fish distribution than gradients in zooplankton biomass is interesting, suggesting that

170

zooplankton biomass may not limit their fitness, or is just distributed more patchily than temperature in the coastal ocean. It may also be that fish are better able to navigate along more predictable physical gradients, such as temperature, or that they are not generally limited by the availability of prey within the study region.

Zooplanktivores are dependent on resources which are themselves dynamic and unpredictable, and their variable distribution reflects this. By examining the functional role of these fish, rather than their taxonomic status, a better understanding of the relative importance of pelagic subsidies to their environments has been established (Bellwood et al.,

2019). The distributed nature of zooplankton and the passive way in which it is delivered to coastal ecosystems drives the distribution and behaviour of multiple species within a functional group dependent on a shared resource (Blondel, 2003). As a group, however, their reliance on plankton also makes them particularly sensitive to oceanographic variability

(Lluch‐Belda et al., 1992; Lindegren et al., 2013). Climate variability, for example, has been linked to changes in zooplankton size-structure, which can drive fluctuating abundances of sardine and anchovy within major upwelling regions (Ayón et al., 2011; Checkley et al.,

2017). By gaining a better understanding of current continental and shelf-scale patterns in the distribution of zooplanktivores it is now easier to predict how fish communities will respond to environmental change (Blois et al., 2013; Sato et al., 2018; Holland et al., 2020b).

The broad implications of these findings highlight the geographically widespread role of zooplanktivores in capturing and retaining pelagic subsidies within coastal ecosystems and improve our understanding of how they use three-dimensional space to influence survival and fitness.

While previous research indicated zooplanktivores contributed over 40% of fish biomass to rocky reefs in the Sydney region (Truong et al., 2017), this study of latitudinal scale patterns (over hundreds of kms; Chapter 2) has demonstrated that their sizeable

171

contribution to community biomass extends across the southeast coast of mainland

Australia (Figure 6.1), particularly south of the East Australian Current separation zone and as far south as the Bass Strait (Chapter 2) (Holland et al., 2020b). By comparison, the surprising paucity of zooplanktivores and low total fish biomass south of the Bass Strait may reflect the highly seasonal zooplankton availability at higher latitudes (Harris et al.,

1987). The particularly low zooplankton density in winter may impose limitations on the year-round persistence of zooplanktivores at higher latitudes. In order to properly test this hypothesis, better continuous zooplankton biomass data is needed than what is currently available in Australia. Continuous plankton recorder (CPR) data, while internally comparable, poorly approximates absolute zooplankton abundance or biomass (Richardson et al., 2004; McEnnulty et al., 2020).

Figure 6.1 The main distributional patterns documented in each of the four studies of this thesis, including the gradient in reef community trophic composition across 16 degrees of latitude (Chapter 2), the model-predicted coast-to-shelf gradient in zooplanktivore density around Montague Island (Chapter 3), the distribution of school thickness around an artificial reef field near Sydney (Chapter Error! Reference source not found.) and differences in the distribution of zooplanktivore schools b etween night and day at artificial and natural reefs (Chapter 5).

172

While zooplankton density is likely an important factor affecting the long-term suitability of environments for zooplanktivorous fishes over latitudinal scales, at more regional scales

(<10 km) the distribution of zooplankton density may be less important (Chapter 3). In early spring the distribution of pelagic jack mackerel (Trachurus declivis) and blue mackerel

(Scomber australasicus) was found to be more closely linked to the distribution of temperature and bathymetry, with fish demonstrating a consistent preference for the warm waters of the

East Australian Current along the edge of the continental shelf (Figure 6.1). This distribution occurred despite the considerably greater zooplankton density along the coastal fringe. I concluded that this coast-to-shelf gradient in zooplanktivore density may have been influenced by both thermal limitations on fish physiology and a parallel coast-to-shelf gradient in predation pressure. Unfortunately, distinguishing between these possibilities is difficult, particularly considering the patchily distributed nature of zooplankton and the fact that zooplankton was only sampled over one day of a two-day survey. Additional trawls or drifting underwater video would be useful in revealing spatial variation in the size distribution of fish, as it is likely that little penguins (Eudyptula minor) documented in other studies (Carroll et al., 2016; Carroll et al., 2017) have been feeding on juvenile fish inshore of

Montague Island.

Established methods, including underwater visual census and split-beam acoustic surveys, were suitable for measuring fish distribution at the coarse levels of scale required in the first two studies (Chapters 2, 3), however accurately measuring fine-scale (~1 m) school distribution relative to artificial structures and seafloor variability required the development of novel methods (Chapter Error! Reference source not found.). A modern portable m ultibeam echosounder was used to survey the seafloor and adjacent mid-water schools at an artificial reef and the distribution of schools was simulated by developing grid-based methods to quantify distributed school thickness (Figure 6.1).

173

These methods were applied to document variability in the distribution of zooplanktivore schools at artificial and natural reefs during day and night (Chapter 5). At this fine scale, variability in school shape and distribution relative to benthic structure suggested strong intra-school competition for access to prey, and trade-offs between predator-avoidance and feeding opportunities (Figure 6.1). Understanding the behaviour that drives these patterns in spatial distribution can help to inform the design of artificial reefs for habitat restoration and fisheries enhancement. These findings should be considered in a broader global context with the objective of optimising coastal infrastructure to obtain greater benefits from pelagic subsidies.

6.1. Trophic composition along western boundary currents globally

Herbivorous fish are most prevalent in tropical coastal regions, and their diversity, abundance and the impacts they inflict on habitat typically decline towards higher latitudes

(Horn, 1989; Floeter et al., 2004; Longo et al., 2019). The wealth of literature on this topic partially stems from the fact that tropical coral reefs have long been subjects of interest for researchers, and their warm temperatures and high biodiversity have made them particularly attractive habitats to study for decades (Bennett et al., 2016). Governments and institutions globally have established reef monitoring programs to conduct regular biodiversity surveys of coral reefs (Global Coral Reef Monitoring Network, 2014; Mellin et al., 2020), however it has not been until recently that similar fishery-independent datasets with high spatiotemporal resolution have been available for colder temperate regions

(Edgar & Stuart-Smith, 2014).

Far less is known about persistent global trophic patterns along the transition from subtropical to temperate latitudes. In Chapter 2, I established that coastal rocky reefs south of the separation zone of the East Australian Current support very large relative biomass of

174

zooplanktivores (>40% of fish biomass). Other studies conducted along tropicalisation gradients have uncovered similar shifts, favouring planktivory over herbivory with transitions to higher latitude (Holmes et al., 2013; Longo et al., 2019). This study of trophic structure across southeastern Australia highlights the need for similar large-scale and high- resolution monitoring programs for other subtropical to temperate transitions along western boundary currents (e.g. the Kuroshio Current).

Along other western boundary current systems, there is also a need to resolve latitudinal patterns in functional trophic composition. Western boundary currents generate regions of persistent coastal upwelling and enrichment downstream from the location where the main flow diverges from the coast, such as off the KwaZulu-Natal coast in South Africa (Lamont et al., 2016) or the South Atlantic Bight between Florida and North Carolina (Hyun & He,

2010). Although these regions are typically not as productive as the great coastal upwelling zones generated by eastern boundary currents (Chavez & Messié, 2009), my findings support the hypothesis that they can also sustain food webs which are highly dependent on plankton.

The current intensification or ‘spin-up’ of ocean gyres globally is causing more intense eddying and accelerated rates of temperature increase poleward of western boundary current separation zones (Malan et al., in review). This spin-up is also driving separation zones to shift poleward (Yasuda & Kitamura, 2003; Cetina‐Heredia et al., 2014; Oliver &

Holbrook, 2014). The combination of altered circulation patterns and increased warming is already driving species redistributions into higher latitudes (Parmesan, 2006; Poloczanska et al., 2013), affecting the supply of planktonic organisms to coastal reefs (Cetina-Heredia et al., 2015). Yet, it is still unclear what impacts this might have on coastal reef communities heavily dependent on the delivery of plankton, both for energy and recruitment. As many of these regions are experiencing macroalgae loss due to the poleward redistribution of

175

herbivorous fishes (Vergés et al., 2014a) and invertebrates (Ling & Keane, 2018), it is increasingly important to understand the trophic basis that sustains them, and to what degree this loss of benthic productivity will be bolstered by pelagic subsidies.

Regular monitoring will be vital for detecting changes brought on by ocean warming, particularly along these western boundary currents where the distributions of tropical taxa are shifting to higher latitudes (Champion et al., 2018). Given the sizeable resource requirements of such programs, and the influences of the current political climate on scientific funding, this will almost certainly require researchers to embrace and promote citizen science (Silvertown, 2009). If not for the rising interest in citizen science that has driven the success of the Reef Life Survey program, it would have been an insurmountable task to gather the resources necessary to conduct my latitudinal analysis. I believe the continued global expansion of this program (Figure 6.2), or the establishment of similar programs using the same standardised methods for ‘structured’ citizen science (Callaghan et al., 2019) will be vital to establishing a trophic basis for western boundary current ecosystems globally and predicting future impacts of climate change.

176

Figure 6.2 Global distribution of the 3537 Reef Life Survey sites surveyed by divers between 2008 and 2020, from Edgar et al. (2020), Figure 2.

6.2. Coast-to-shelf gradients on continental scales

Like transitions across latitude, transitions from the coast to the edge of the continental shelf can be accompanied by strong gradients in temperature. In Chapter 3 I theorised that blue mackerel (Scomber australasicus) and jack mackerel (Trachurus declivis) were thermally limited within coastal waters in early spring, and likely remained offshore to access the warmer waters of the East Australian Current. This theory was further supported by the seasonality of the commercial Small Pelagics Fishery (being primarily restricted to summer;

Stewart & Ferrell, 2001) and the mean bathymetry of targeted trawls for these species

(~150 m depth). This distribution of fish occurred, despite observations that the greatest densities of zooplankton prey were distributed along the coast.

Coast-to-shelf gradients in zooplankton and phytoplankton density, similar to the one observed in Chapter 3, have been documented in other regions during spring, including the northeast Atlantic (Sourisseau & Carlotti, 2006; Irigoien et al., 2009; Vandromme et al.,

2014) the southwest Atlantic (da Rocha Marcolin et al., 2013; Brandini et al., 2014) and

177

along other parts of southeastern Australia (Figure 6.3) (Schilling et al., In review). This is indicative of a general pattern for regions that experience regular current-driven upwelling onto the shelf (Keister et al., 2009; Briseño-Avena et al., 2020). In these regions, coastal waters are generally more productive than the oligotrophic waters along the shelf edge.

While a mismatch was observed between the distribution of zooplankton along the coast, and the distribution of zooplanktivores at the shelf edge in spring, it is unclear whether this distribution is consistent throughout the year, or whether fish are even limited by prey availability at the shelf edge.

178

In Chapter 2, strong seasonal patterns were detected in the observed density of fish, particularly zooplanktivores, at shallow coastal reefs. The highest densities were observed in autumn, when coastal waters had generally reached their maximum annual temperature, while the lowest densities were observed in spring when temperatures had reached their annual minimum. Synthesising the observations from Chapters 2 and 3 would suggest that schooling zooplanktivores, such as blue mackerel (Scomber australasicus), jack mackerel

(Trachurus declivis) and yellowtail scad (Trachurus novaezelandiae), may undertake a seasonal migration from the coast to the shelf to seek warmer water and reduce thermal limitations on physiological performance as temperatures decline throughout winter.

It is also possible that the outer shelf distribution we measured in spring surveys was a reflection of spawning behaviour (Sexton et al., 2017), as the spawning of these species occurs on the outer shelf and is closely linked to temperature (Neira, 2011). They may subsequently migrate back to the coast to access denser aggregations of prey as spawning concludes and coastal temperatures warm to be closer to their thermal optima. However, while these fish may currently benefit from the accessibility of warmer water offshore, their upper critical thermal limits will likely be tested as water temperatures continue to rise due to climate change (Sandblom et al., 2016).

One significant caveat to these findings is the short duration of the study. Due to the ubiquitously patchy nature of zooplankton distribution, across both space and time, it is possible that our one-day survey only captured a snapshot in time and may not have been representative of average zooplankton distribution. To draw robust conclusions regarding whether there was indeed a mismatch between predators and prey, zooplankton sampling would likely need to be conducted across a much longer time period (e.g. weeks, months) with high temporal resolution.

179

To resolve whether coast-to-shelf migrations were responsible for driving the patterns observed in Chapters 2 and 3, the most logical approach would be to study the spatiotemporal distribution of CPUE from commercial fishery catch data collected throughout the year (Verdoit et al., 2003), however, the Small Pelagics Fishery in New

South Wales is limited to only a few vessels and does not provide the necessary spatial or temporal resolution for such an analysis. To better resolve these patterns, it would be useful to conduct monthly acoustic trawl surveys across multiple transects oriented along the slope of the continental shelf and distributed across latitude. These surveys would require vessels capable of working between shallow coastal waters and the shelf edge, and potentially further offshore. Acoustic data could be used to generate a generalised additive mixed model with a tensor product smoother of bathymetry and month as the main covariate. Fish collection would also facilitate the assessment of reproductive state, which could be used to understand how seasonal cross-shelf migrations are linked to spawning.

This approach could help uncover the ecological basis for seasonal patterns documented in

Chapter 2 and provide greater context for the coast-to-shelf gradients observed in Chapter

3, however, it would require significant resources and funding.

6.3. Decommissioning offshore infrastructure

Coastal regions throughout the world are scattered with oil and gas platforms, many of which are no longer productive and are nearing retirement and eventual decommissioning

(Bull & Love, 2019). These platforms are prevalent throughout the Middle East, Asia-

Pacific, Australia and New Zealand, Europe and the United States, particularly in the Gulf of Mexico where approximately 4000 platforms are currently in operation (Bull & Love,

2019). Some jurisdictions, notably those regulated by the OSPAR convention, the environmental protection legislation regulating the northeast Atlantic, currently require retired platforms to be completely removed and dismantled offsite (Jørgensen, 2012). These

180

structures, however, develop diverse and abundant marine communities during their operational lives, which are lost in the removal process (Macreadie et al., 2011). The process to dismantle this infrastructure is also very costly. Decommissioning 23 platforms off the coast of California, for example, will cost ~US $1.5 billion (TSB Offshore, 2015; Bull &

Love, 2019).

A significantly less impactful and more cost-effective alternative is to repurpose these structures. Thousands of retired reefs in the Gulf of Mexico have already been converted into artificial reefs by removing the topside structure and leaving the submerged support structure, or jacket, in place (Schroeder & Love, 2004). These vertical structures have been found to produce more fish biomass than even the most productive natural habitats by facilitating recruitment and providing substantial adult fish habitat (Claisse et al., 2014). As a result of this high productivity and the considerable cost savings over complete removal, the practice of leave-in-place decommissioning, otherwise known as Rigs-to-Reefs (RtR), has become popular with resource companies, fisheries management authorities, and fishers

(Bull & Love, 2019).

In comparison to the Gulf of Mexico, offshore oil and gas drilling in Australia is a relatively young industry. Oil and gas platforms in Australia are concentrated around sparsely populated regions including the Gascoyne, Pilbara and Kimberley regions of Western

Australia, southwestern Victoria, East Gippsland and the Bass Strait (Figure 6.4a) (Evans et al., 2017). Decisions around how to manage retired platforms in Australia are still a subject of debate (Department of Industry and Science, 2015), and the National Offshore Petroleum

Safety and Environmental Management Authority (NOPSEMA) is currently investigating whether to support RtR in Australia (Bull & Love, 2019). In concurrence with Claisse et al.

(2014), my finding that structures featuring enhanced vertical relief allow resident zooplanktivores to forage across a greater vertical extent of the water column (Chapter 5)

181

provides support for RtR in Australia. As zooplanktivores tend to forage on the upstream side of structure, the greatest productivity benefit would likely be achieved by overturning platforms onto their side, with the longest dimension oriented to intersect the current

(Figure 6.4b) (Champion et al., 2015). In some cases, platforms may need to be towed to locations exposed to more moderate and reliable currents.

Figure 6.4 The distribution of oil and gas wells drilled in Australian waters (a), from Evans et al. (2017), and three different methods for reefing retired platforms, including tow-and-place, topple-in- place and partial removal, from Bull and Love (2019).

6.4. Designing artificial reefs to capture pelagic subsidies

My study of school distribution on artificial reefs (Chapter 5) has highlighted the need to optimise reef designs to provide more favourable foraging opportunities for zooplanktivores, and thus increase the potential production benefit from pelagic subsidies.

Despite a rise in the number of reef deployments globally, designs and locations are seldom evaluated for such objectives (Baine, 2001; Becker et al., 2018). The increasing investment into artificial reefs in Australia, at a cost of ~AU $1 million per reef, also warrants further study into their basal sources of productivity to inform incremental enhancements to their

182

design and ultimately improve their sustainability (Bortone et al., 2011). This study has provided several new insights into fish behaviour, which can be applied to optimise the design of plankton-supported artificial reefs (Table 6.1).

Table 6.1 Key behavioural findings and corresponding design implications for artificial reefs.

Behavioural finding Artificial reef design implications

Greater abundance, area coverage and probability of Multiple structures should be arranged into school occurrence at artificial reefs a ‘reef field’ to promote supporting larger schools of zooplanktivores.

Attraction to seafloor complexity / vertical relief Artificial reefs should incorporate exaggerated vertical relief to promote access to a greater vertical extent of the water column for feeding.

Preference for upstream side of bathymetry / structure Maximising upstream surfaces and minimising potential shadowing of downstream structures should be prioritised.

Schools more aggregated under high current velocity Localised studies should be conducted to determine optimal current velocity to maximise feeding rates in resident zooplanktivores. This should be a consideration in deciding on reef location.

These findings and implications are interesting in isolation but may be of greater use if they can be synthesised into a single conceptual model. Since schools of zooplanktivores were twice as likely to occur on the upstream face of reefs, a design that would maximise the upstream face of the reef, while minimising the potential shadowing effect of the reef face on downstream modules would be most effective (Figure 6.5). A triangular arrangement of modules is suggested, with one side featuring towers and oriented to intersect the prevailing current. Like the artificial reef studied in Chapters Error! Reference source not f

183

ound. and 5 (Reef site: JDN), arranging reef modules into clusters would enable this simple arrangement to be scaled hierarchically to fit any desired footprint. The arrangement of modules in such a configuration would create the ‘reef-field effect’ (Becker et al., 2019) while still minimising downstream shadowing, no matter how large the reef was scaled. More research would likely be required to determine the optimal current exposure, location and spacing of modules within specific coastal environments.

Figure 6.5 Conceptual diagram of an optimised plankton-supported artificial reef. The vector representing current direction extends out of the page.

To supplement this initial research, it would be useful to investigate the effects of environmental conditions, reef design and fishing effort in order to improve the effectiveness of designed artificial reefs and better understand the risks associated with locally unsustainable fishing. This research would provide a scientific basis for the management of artificial reef deployments and their social, economic, and environmental impacts. It could also be useful to relate the biomass and diversity of fish and invertebrate

184

assemblages to the various design features and locations of artificial reefs across a range of environmental and oceanographic attributes. Measuring the residency time of zooplanktivores on artificial reefs, and using stable isotope analysis to quantify the breakdown in planktonic versus benthic sources of carbon supporting harvested fish would also be beneficial (Docmac et al., 2017; Udy et al., 2019). Measuring fishing effort at artificial reefs by employing remote monitoring tools, including hydrophones (Kaplan &

Mooney, 2015; Dinh et al., 2018), long-range cameras (Keller et al., 2016; Becker et al.,

2020) and coastal radar (Chang, 2014), could also help determine how much of the biomass of harvested fish is supported by estimates of on-reef production. This research should be complemented with comparative studies of nearby natural reefs to determine the effects of artificial reefs on growth, reproduction, and survival, and to address concerns that artificial reefs can act like ecological traps (Battin, 2004), attracting and aggregating fish around poorer-quality habitat (Pickering & Whitmarsh, 1997; Hallier & Gaertner, 2008) .

6.5. Concluding remarks

Zooplanktivorous fishes bridge the trophic gap between photosynthesis across the ocean surface and the production of fish biomass, sustaining marine ecosystems and fisheries globally. This thesis has attempted to fill a major knowledge gap regarding the distribution and behaviour of these fish in temperate coastal habitats and highlights the importance of understanding the spatial and environmental context of fisheries ecology. This work has underlined the need to consider a trophic perspective in planning for the impacts of climate change. Finally, it has highlighted the importance of considering the foraging behaviour of fish in the design of artificial reefs and offshore infrastructure in order to increase the contribution of pelagic subsidies to enhancing local production.

185

7. Literature cited

Aarflot, J. M., Aksnes, D. L., Opdal, A. F., Skjoldal, H. R., & Fiksen, Ø. (2019). Caught in broad daylight: Topographic constraints of zooplankton depth distributions. Limnology and Oceanography, 64(3), 849-859. https://doi.org/10.1002/lno.11079

Adler, D., & Murdoch, D. (2020). rgl: 3D Visualization Using OpenGL. R package version 0.100.54. https://CRAN.R-project.org/package=rgl

Aguiar, A., Cirano, M., Pereira, J., & Marta-Almeida, M. (2014). Upwelling processes along a western boundary current in the Abrolhos–Campos region of Brazil. Continental Shelf Research, 85, 42-59. https://doi.org/10.1016/j.csr.2014.04.013

Ajemian, M. J., Wetz, J. J., Shipley-Lozano, B., Shively, J. D., & Stunz, G. W. (2015a). An Analysis of Artificial Reef Fish Community Structure along the Northwestern Gulf of Mexico Shelf: Potential Impacts of "Rigs-to-Reefs" Programs. PloS one, 10(5), e0126354. https://doi.org/10.1371/journal.pone.0126354

Ajemian, M. J., Wetz, J. J., Shipley-Lozano, B., & Stunz, G. W. (2015b). Rapid assessment of fish communities on submerged oil and gas platform reefs using remotely operated vehicles. Fisheries Research, 167, 143-155. https://doi.org/10.1016/j.fishres.2015.02.011

Anderson, T. W., & Sabado, B. D. (1995). Correspondence between food availability and growth of a planktivorous temperate reef fish. Journal of Experimental Marine Biology and Ecology, 189(1-2), 65-76. https://doi.org/10.1016/0022-0981(95)00011-F

Andrewartha, H. G., & Birch, L. C. (1954). The distribution and abundance of animals: University of Chicago press.

Annese, D. M., & Kingsford, M. J. (2005). Distribution, movements and diet of nocturnal fishes on temperate reefs. Environmental Biology of Fishes, 72(2), 161-174. https://doi.org/10.1007/s10641-004-0774-7

Arnould, J. P., & Kirkwood, R. (2007). Habitat selection by female Australian fur seals (Arctocephalus pusillus doriferus). Aquatic Conservation: Marine and Freshwater Ecosystems, 17(S1), S53-S67. https://doi.org/10.1002/aqc.908

Arya, S., Mount, D., Kemp, S., & Jefferis, G. (2019). RANN: Fast nearest neighbour search (wraps ANN library) Using L2 metric. R package version 2.6.1. https://CRAN.R- project.org/package=RANN

Australian Bureau of Statistics. (2017). Australian population grid 2016. http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/3218.02015- 16?OpenDocument

Ayón, P., Swartzman, G., Bertrand, A., Gutiérrez, M., & Bertrand, S. (2008). Zooplankton and forage fish species off Peru: large-scale bottom-up forcing and local-scale depletion. Progress in oceanography, 79(2-4), 208-214. https://doi.org/10.1016/j.pocean.2008.10.023

186

Ayón, P., Swartzman, G., Espinoza, P., & Bertrand, A. (2011). Long-term changes in zooplankton size distribution in the Peruvian Humboldt Current System: conditions favouring sardine or anchovy. Marine Ecology Progress Series, 422, 211-222. https://doi.org/10.3354/meps08918

Aziz, K., Mahmud, M., Hasan, R., Gunathilaka, M., Peter, B., Tajam, J., . . . Kamaruddin, S. (2018). Water column features and seafloor characteristics using simultaneous multibeam water column detection. Journal of Advanced Manufacturing Technology (JAMT), 12(2), 41-52.

Azzurro, E., Franzitta, G., Milazzo, M., Bariche, M., & Fanelli, E. (2017). Abundance patterns at the invasion front: the case of Siganus luridus in Linosa (Strait of Sicily, Central Mediterranean Sea). Marine and Freshwater Research, 68(4), 697-702. https://doi.org/10.1071/MF16024

Bailey, R., & Simmonds, E. (1990). The use of acoustic surveys in the assessment of the North Sea herring stock and a comparison with other methods. Rapports et Procés- Verbaux des Réunions Conseil International pour l’Exploration de la Mer, 189, 9-17.

Baine, M. (2001). Artificial reefs: a review of their design, application, management and performance. Ocean & Coastal Management, 44(3-4), 241-259. https://doi.org/10.1016/S0964-5691(01)00048-5

Bakhtiar, D., Jaya, I., Manik, H. M., Madduppa, H. H., & Febrianto, T. (2020). Analysis of Reef Fish Target Strength through Ex-Situ Measurement Using Acoustic Methods. IOSR Journal of Environmental Science, Toxicology and Food Technology, 14(5). https://doi.org/10.9790/2402-1405025361

Bakun, A. (2006). Wasp-waist populations and marine ecosystem dynamics: navigating the “predator pit” topographies. Progress in oceanography, 68(2-4), 271-288. https://doi.org/10.1016/j.pocean.2006.02.004

Bakun, A., Black, B. A., Bograd, S. J., Garcia-Reyes, M., Miller, A. J., Rykaczewski, R. R., & Sydeman, W. J. (2015). Anticipated effects of climate change on coastal upwelling ecosystems. Current Climate Change Reports, 1(2), 85-93. https://doi.org/10.1007/s40641-015-0008-4

Barnes, C., Maxwell, D., Reuman, D. C., & Jennings, S. (2010). Global patterns in predator- prey size relationships reveal size dependency of trophic transfer efficiency. Ecology, 91(1), 222-232. https://doi.org/10.1890/08-2061.1

Barton, K. (2020). MuMIn: Multi-Model Inference. R package version 1.43.17. https://CRAN.R-project.org/package=MuMIn

Basford, A. J., Feary, D. A., Truong, G., Steinberg, P. D., Marzinelli, E. M., & Vergés, A. (2016). Feeding habits of range-shifting herbivores: tropical surgeonfishes in a temperate environment. Marine and Freshwater Research, 67(1), 75-83. https://doi.org/10.1071/MF14208

Battin, J. (2004). When good animals love bad habitats: ecological traps and the conservation of animal populations. Conservation Biology, 18(6), 1482-1491. https://doi.org/10.1111/j.1523-1739.2004.00417.x

187

Beck, H., Feary, D. A., Nakamura, Y., & Booth, D. J. (2017). Temperate macroalgae impacts tropical fish recruitment at forefronts of range expansion. Coral reefs, 36(2), 639-651. https://doi.org/10.1007/s00338-017-1553-1

Becker, A., Holland, M., Smith, J. A., & Suthers, I. M. (2016). Fish movement through an estuary mouth is related to tidal flow. Estuaries and coasts, 39(4), 1199-1207. https://doi.org/10.1007/s12237-015-0043-3

Becker, A., Smith, J. A., Taylor, M. D., McLeod, J., & Lowry, M. B. (2019). Distribution of pelagic and epi-benthic fish around a multi-module artificial reef-field: Close module spacing supports a connected assemblage. Fisheries Research, 209, 75-85. https://doi.org/10.1016/j.fishres.2018.09.020

Becker, A., Taylor, M., McLeod, J., & Lowry, M. (2020). Application of a long-range camera to monitor fishing effort on an offshore artificial reef. Fisheries Research, 228, 105589. https://doi.org/10.1016/j.fishres.2020.105589

Becker, A., Taylor, M. D., Folpp, H., & Lowry, M. B. (2018). Managing the development of artificial reef systems: The need for quantitative goals. Fish and Fisheries, 19(4), 740- 752. https://doi.org/10.1111/faf.12288

Becker, A., Taylor, M. D., & Lowry, M. B. (2017). Monitoring of reef associated and pelagic fish communities on Australia’s first purpose built offshore artificial reef. ICES Journal of Marine Science, 74(1), 277-285. https://doi.org/10.1093/icesjms/fsw133

Becker, A., Whitfield, A. K., Cowley, P. D., Järnegren, J., & Næsje, T. F. (2013). Potential effects of artificial light associated with anthropogenic infrastructure on the abundance and foraging behaviour of estuary‐associated fishes. Journal of Applied Ecology, 50(1), 43-50. https://doi.org/10.1111/1365-2664.12024

Bellwood, D. R., Goatley, C. H., Brandl, S. J., & Bellwood, O. (2014). Fifty million years of herbivory on coral reefs: fossils, fish and functional innovations. Proc Biol Sci, 281(1781), 20133046. https://doi.org/10.1098/rspb.2013.3046

Bellwood, D. R., Streit, R. P., Brandl, S. J., & Tebbett, S. B. (2019). The meaning of the term ‘function’ in ecology: a coral reef perspective. Functional Ecology, 33(6), 948-961. https://doi.org/10.1111/1365-2435.13265

Bennett, S., Wernberg, T., Connell, S. D., Hobday, A. J., Johnson, C. R., & Poloczanska, E. S. (2016). The ‘Great Southern Reef’: social, ecological and economic value of Australia’s neglected kelp forests. Marine and Freshwater Research, 67(1), 47-56. https://doi.org/10.1071/MF15232

Bennett, S., Wernberg, T., Harvey, E. S., Santana-Garcon, J., & Saunders, B. J. (2015). Tropical herbivores provide resilience to a climate-mediated phase shift on temperate reefs. Ecol Lett, 18(7), 714-723. https://doi.org/10.1111/ele.12450

Benoit-Bird, K. J., & McManus, M. A. (2012). Bottom-up regulation of a pelagic community through spatial aggregations. Biology Letters, 8(5), 813-816. https://doi.org/10.1098/rsbl.2012.0232

188

Benoit‐Bird, K. J., Waluk, C. M., & Ryan, J. P. (2019). Forage Species Swarm in Response to Coastal Upwelling. Geophysical Research Letters, 46(3), 1537-1546. https://doi.org/10.1029/2018GL081603

Bertrand, A., Grados, D., Colas, F., Bertrand, S., Capet, X., Chaigneau, A., . . . Fablet, R. (2014). Broad impacts of fine-scale dynamics on seascape structure from zooplankton to seabirds. Nat Commun, 5(1), 5239. https://doi.org/10.1038/ncomms6239

Bez, N., & Rivoirard, J. (2001). Transitive geostatistics to characterise spatial aggregations with diffuse limits: an application on mackerel ichtyoplankton. Fisheries Research, 50(1-2), 41-58. https://doi.org/10.1016/S0165-7836(00)00241-1

Blamey, L. K., & Bolton, J. J. (2018). The economic value of South African kelp forests and temperate reefs: past, present and future. Journal of Marine Systems, 188, 172-181. https://doi.org/10.1016/j.jmarsys.2017.06.003

Blois, J. L., Williams, J. W., Fitzpatrick, M. C., Jackson, S. T., & Ferrier, S. (2013). Space can substitute for time in predicting climate-change effects on biodiversity. Proceedings of the national academy of sciences, 110(23), 9374-9379. https://doi.org/10.1073/pnas.1220228110

Blondel, J. (2003). Guilds or functional groups: does it matter? Oikos, 100(2), 223-231. https://doi.org/10.1034/j.1600-0706.2003.12152.x

Bohnsack, J. A. (1989). Are high densities of fishes at artificial reefs the result of habitat limitation or behavioral preference? Bulletin of marine Science, 44(2), 631-645.

Bohnsack, J. A., & Sutherland, D. L. (1985). Artificial reef research: a review with recommendations for future priorities. Bulletin of marine Science, 37(1), 11-39.

Bond, T., Partridge, J. C., Taylor, M. D., Cooper, T. F., & McLean, D. L. (2018). The influence of depth and a subsea pipeline on fish assemblages and commercially fished species. PloS one, 13(11), e0207703. https://doi.org/10.1371/journal.pone.0207703

Booth, D. J., & Sear, J. (2018). Coral expansion in Sydney and associated coral-reef fishes. Coral reefs, 37(4), 995-995. https://doi.org/10.1007/s00338-018-1727-5

Bortone, S. A. (2006). A perspective of artificial reef research: the past, present, and future. Bulletin of marine Science, 78(1), 1-8.

Bortone, S. A., Brandini, F. P., Fabi, G., & Otake, S. (2011). Artificial reefs in fisheries management: CRC Press.

Boswell, K. M., Pedersen, G., Taylor, J. C., LaBua, S., & Patterson III, W. F. (2020). Examining the relationship between morphological variation and modeled broadband scattering responses of reef-associated fishes from the Southeast United States. Fisheries Research, 228, 105590. https://doi.org/10.1016/j.fishres.2020.105590

Boswell, K. M., Wells, R., Cowan Jr, J. H., & Wilson, C. A. (2010). Biomass, density, and size distributions of fishes associated with a large-scale artificial reef complex in the Gulf

189

of Mexico. Bulletin of marine Science, 86(4), 879-889. https://doi.org/10.5343/bms.2010.1026

Brainerd, K., & Gregg, M. (1997). Turbulence and stratification on the tropical ocean‐global atmosphere‐coupled ocean‐atmosphere response experiment microstructure pilot cruise. Journal of Geophysical Research: Oceans, 102(C5), 10437-10455. https://doi.org/10.1029/96JC03864

Brandini, F. P., Nogueira Jr, M., Simião, M., Codina, J. C. U., & Noernberg, M. A. (2014). Deep chlorophyll maximum and plankton community response to oceanic bottom intrusions on the continental shelf in the South Brazilian Bight. Continental Shelf Research, 89, 61-75. https://doi.org/10.1016/j.csr.2013.08.002

Brandl, S. J., Tornabene, L., Goatley, C. H., Casey, J. M., Morais, R. A., Côté, I. M., . . . Bellwood, D. R. (2019). Demographic dynamics of the smallest marine vertebrates fuel coral reef ecosystem functioning. Science, 364(6446), 1189-1192. https://doi.org/10.1126/science.aav3384

Brandt, S. (1996). Acoustic assessment of fish abundance and distribution. Fisheries techniques, 2nd edition. American Fisheries Society, Bethesda, Maryland, 385-432.

Bray, D. J., & Gomon, M. F. (2018). Fishes of Australia. http://fishesofaustralia.net.au/

Bray, R. N. (1980). Influence of water currents and zooplankton densities on daily foraging movements of blacksmith, Chromis punctipinnis, a planktivorous reef fish. Fishery Bulletin, 78(4), 829-841.

Breck, J. E. (1993). Foraging theory and piscivorous fish: are forage fish just big zooplankton? Transactions of the American Fisheries Society, 122(5), 902-911. https://doi.org/10.1577/1548-8659(1993)122%3C0902:FTAPFA%3E2.3.CO;2

Brett, J. (1956). Some principles in the thermal requirements of fishes. The Quarterly Review of Biology, 31(2), 75-87.

Brewer, T. D., Cinner, J. E., Green, A., & Pressey, R. L. (2013). Effects of human population density and proximity to markets on coral reef fishes vulnerable to extinction by fishing. Conserv Biol, 27(3), 443-452. https://doi.org/10.1111/j.1523- 1739.2012.01963.x

Brickhill, M., Lee, S., & Connolly, R. (2005). Fishes associated with artificial reefs: attributing changes to attraction or production using novel approaches. Journal of Fish Biology, 67, 53-71. https://doi.org/10.1111/j.0022-1112.2005.00915.x

Briseño-Avena, C., Schmid, M. S., Swieca, K., Sponaugle, S., Brodeur, R. D., & Cowen, R. K. (2020). Three-dimensional cross-shelf zooplankton distributions off the Central Oregon Coast during anomalous oceanographic conditions. Progress in oceanography, 102436. https://doi.org/10.1016/j.pocean.2020.102436

Brodie, S., Hobday, A. J., Smith, J. A., Spillman, C. M., Hartog, J. R., Everett, J. D., . . . Suthers, I. M. (2017). Seasonal forecasting of dolphinfish distribution in eastern

190

Australia to aid recreational fishers and managers. Deep Sea Research Part II: Topical Studies in Oceanography, 140, 222-229. https://doi.org/10.1016/j.dsr2.2017.03.004

Bull, A. S., & Love, M. S. (2019). Worldwide oil and gas platform decommissioning: a review of practices and reefing options. Ocean & Coastal Management, 168, 274-306. https://doi.org/10.1016/j.ocecoaman.2018.10.024

Bulman, C., Althaus, F., He, X., Bax, N., & Williams, A. (2001). Diets and trophic guilds of demersal fishes of the south-eastern Australian shelf. Marine and Freshwater Research, 52(4), 537-548. https://doi.org/10.1071/MF99152

Caldwell, Z. R., Zgliczynski, B. J., Williams, G. J., & Sandin, S. A. (2016). Reef Fish Survey Techniques: Assessing the Potential for Standardizing Methodologies. PloS one, 11(4), e0153066. https://doi.org/10.1371/journal.pone.0153066

Callaghan, C. T., Rowley, J. J. L., Cornwell, W. K., Poore, A. G. B., & Major, R. E. (2019). Improving big citizen science data: Moving beyond haphazard sampling. PLoS Biol, 17(6), e3000357. https://doi.org/10.1371/journal.pbio.3000357

Campbell, M. D., Pollack, A. G., Gledhill, C. T., Switzer, T. S., & DeVries, D. A. (2015). Comparison of relative abundance indices calculated from two methods of generating video count data. Fisheries Research, 170, 125-133. https://doi.org/10.1016/j.fishres.2015.05.011

Canfield, D. E. J., & Watkins, C. E. (1984). Relationships between zooplankton abundance and chlorophyll a concentrations in Florida lakes. Journal of Freshwater Ecology, 2(4), 335-344. https://doi.org/10.1080/02705060.1984.9664611

Carr, M.-E. (2001). Estimation of potential productivity in Eastern Boundary Currents using remote sensing. Deep Sea Research Part II: Topical Studies in Oceanography, 49(1- 3), 59-80. https://doi.org/10.1016/S0967-0645(01)00094-7

Carroll, G., Cox, M., Harcourt, R., Pitcher, B. J., Slip, D., & Jonsen, I. (2017). Hierarchical influences of prey distribution on patterns of prey capture by a marine predator. Functional Ecology, 31(9), 1750-1760. https://doi.org/10.1111/1365-2435.12873

Carroll, G., Everett, J. D., Harcourt, R., Slip, D., & Jonsen, I. (2016). High sea surface temperatures driven by a strengthening current reduce foraging success by penguins. Scientific reports, 6(1), 1-13. https://doi.org/10.1038/srep22236

Castillo, J., Barbieri, M., & Gonzalez, A. (1996). Relationships between sea surface temperature, salinity, and pelagic fish distribution off northern Chile. ICES Journal of Marine Science, 53(2), 139-146. https://doi.org/10.1006/jmsc.1996.0014

Cebrian, J. (1999). Patterns in the Fate of Production in Plant Communities. Am Nat, 154(4), 449-468. https://doi.org/10.1086/303244

Cetina-Heredia, P., Roughan, M., van Sebille, E., Feng, M., & Coleman, M. A. (2015). Strengthened currents override the effect of warming on lobster larval dispersal and survival. Glob Chang Biol, 21(12), 4377-4386. https://doi.org/10.1111/gcb.13063

191

Cetina‐Heredia, P., Roughan, M., Van Sebille, E., & Coleman, M. (2014). Long‐term trends in the East Australian Current separation latitude and eddy driven transport. Journal of Geophysical Research: Oceans, 119(7), 4351-4366. https://doi.org/10.1002/2014JC010071

Champion, C., Hobday, A. J., Tracey, S. R., & Pecl, G. T. (2018). Rapid shifts in distribution and high‐latitude persistence of oceanographic habitat revealed using citizen science data from a climate change hotspot. Global Change Biology, 24(11), 5440-5453. https://doi.org/10.1111/gcb.14398

Champion, C., Hobday, A. J., Zhang, X., Pecl, G. T., & Tracey, S. R. (2019). Changing windows of opportunity: past and future climate-driven shifts in temporal persistence of kingfish (Seriola lalandi) oceanographic habitat within south-eastern Australian bioregions. Marine and Freshwater Research, 70(1), 33-42. https://doi.org/10.1071/MF17387

Champion, C., Suthers, I. M., & Smith, J. A. (2015). Zooplanktivory is a key process for fish production on a coastal artificial reef. Marine Ecology Progress Series, 541, 1-14. https://doi.org/10.3354/meps11529

Chang, S.-K. (2014). Constructing logbook-like statistics for coastal fisheries using coastal surveillance radar and fish market data. Marine Policy, 43, 338-346. https://doi.org/10.1016/j.marpol.2013.07.003

Chavez, F. P., & Messié, M. (2009). A comparison of eastern boundary upwelling ecosystems. Progress in oceanography, 83(1-4), 80-96. https://doi.org/10.1016/j.pocean.2009.07.032

Cheal, A. J., Emslie, M., Aaron, M. M., Miller, I., & Sweatman, H. (2013). Spatial variation in the functional characteristics of herbivorous fish communities and the resilience of coral reefs. Ecol Appl, 23(1), 174-188. https://doi.org/10.1890/11-2253.1

Checkley, D. M. J., Asch, R. G., & Rykaczewski, R. R. (2017). Climate, anchovy, and sardine. Annual Review of Marine Science, 9, 469-493. https://doi.org/10.1146/annurev- marine-122414-033819

Chen, I. C., Hill, J. K., Ohlemuller, R., Roy, D. B., & Thomas, C. D. (2011). Rapid range shifts of species associated with high levels of climate warming. Science, 333(6045), 1024-1026. https://doi.org/10.1126/science.1206432

Chen, S.-Y., Fei, Y.-H. J., Chen, Y.-C., Chi, K.-J., & Yang, J.-T. (2016). The swimming patterns and energy-saving mechanism revealed from three fish in a school. Ocean Engineering, 122, 22-31. https://doi.org/10.1016/j.oceaneng.2016.06.018

Claisse, J. T., Pondella, D. J., Love, M., Zahn, L. A., Williams, C. M., Williams, J. P., & Bull, A. S. (2014). Oil platforms off California are among the most productive marine fish habitats globally. Proceedings of the national academy of sciences, 111(43), 15462-15467. https://doi.org/10.1073/pnas.1411477111

Clarke, J. H. (2006). Applications of multibeam water column imaging for hydrographic survey. Hydrographic Journal, 120, 3.

192

Colbo, K., Ross, T., Brown, C., & Weber, T. (2014). A review of oceanographic applications of water column data from multibeam echosounders. Estuarine, coastal and shelf science, 145, 41-56. https://doi.org/10.1016/j.ecss.2014.04.002

Coleman, M. A., & Kelaher, B. P. (2009). Connectivity among fragmented populations of a habitat-forming alga, Phyllospora comosa (Phaeophyceae, Fucales) on an urbanised coast. Marine Ecology Progress Series, 381, 63-70. https://doi.org/10.3354/meps07977

Coleman, M. A., & Wernberg, T. (2017). Forgotten underwater forests: The key role of fucoids on Australian temperate reefs. Ecol Evol, 7(20), 8406-8418. https://doi.org/10.1002/ece3.3279

Creque, S. M., Raffenberg, M. J., Brofka, W. A., & Dettmers, J. M. (2006). If you build it, will they come? Fish and angler use at a freshwater artificial reef. North American journal of fisheries management, 26(3), 702-713. https://doi.org/10.1577/M05-029.1

Cury, P. M., Boyd, I. L., Bonhommeau, S., Anker-Nilssen, T., Crawford, R. J., Furness, R. W., . . . Paleczny, M. (2011). Global seabird response to forage fish depletion—one- third for the birds. Science, 334(6063), 1703-1706. https://doi.org/10.1126/science.1212928 da Rocha Marcolin, C., Schultes, S., Jackson, G. A., & Lopes, R. M. (2013). Plankton and seston size spectra estimated by the LOPC and ZooScan in the Abrolhos Bank ecosystem (SE Atlantic). Continental Shelf Research, 70, 74-87. https://doi.org/10.1016/j.csr.2013.09.022

Davis, T. R., & Smith, S. D. (2017). Proximity effects of natural and artificial reef walls on fish assemblages. Regional Studies in Marine Science, 9, 17-23. https://doi.org/10.1016/j.rsma.2016.10.007

De Robertis, A., & Higginbottom, I. (2007). A post-processing technique to estimate the signal-to-noise ratio and remove echosounder background noise. ICES Journal of Marine Science, 64(6), 1282-1291. https://doi.org/10.1093/icesjms/fsm112

Deegan, L. A. (1993). Nutrient and energy transport between estuaries and coastal marine ecosystems by fish migration. Canadian Journal of Fisheries and Aquatic Sciences, 50(1), 74-79. https://doi.org/10.1139/f93-009

Delworth, T. L., Rosati, A., Anderson, W., Adcroft, A. J., Balaji, V., Benson, R., . . . Pacanowski, R. C. (2012). Simulated climate and climate change in the GFDL CM2. 5 high-resolution coupled climate model. Journal of Climate, 25(8), 2755-2781. https://doi.org/10.1175/JCLI-D-11-00316.1

Department of Industry and Science. (2015). 2015 Operational Review of the National Offshore Petroleum Safety and Environmental Management Authority. https://www.industry.gov.au/sites/default/files/2019-05/nopsema-operational- review-2015.pdf

Diaz, R. J., Solan, M., & Valente, R. M. (2004). A review of approaches for classifying benthic habitats and evaluating habitat quality. J Environ Manage, 73(3), 165-181. https://doi.org/10.1016/j.jenvman.2004.06.004

193

Dickens, L. C., Goatley, C. H., Tanner, J. K., & Bellwood, D. R. (2011). Quantifying relative diver effects in underwater visual censuses. PloS one, 6(4), e18965.

Dinh, J. P., Suca, J. J., Lillis, A., Apprill, A., Llopiz, J. K., & Mooney, T. A. (2018). Multiscale spatio-temporal patterns of boat noise on US Virgin Island coral reefs. Marine pollution bulletin, 136, 282-290. https://doi.org/10.1016/j.marpolbul.2018.09.009

Dix, M., Abd-Elrahman, A., Dewitt, B., & Nash Jr, L. (2012). Accuracy evaluation of terrestrial LiDAR and multibeam sonar systems mounted on a survey vessel. Journal of Surveying Engineering, 138(4), 203-213. https://doi.org/10.1061/(ASCE)SU.1943- 5428.0000075

Docmac, F., Araya, M., Hinojosa, I. A., Dorador, C., & Harrod, C. (2017). Habitat coupling writ large: pelagic-derived materials fuel benthivorous macroalgal reef fishes in an upwelling zone. Ecology, 98(9), 2267-2272. https://doi.org/10.1002/ecy.1936 dos Santos, L. N., Brotto, D. S., & Zalmon, I. R. (2010). Fish responses to increasing distance from artificial reefs on the Southeastern Brazilian Coast. Journal of Experimental Marine Biology and Ecology, 386(1-2), 54-60. https://doi.org/10.1016/j.jembe.2010.01.018

Dougherty, E. R. (1992). An introduction to morphological image processing. SPIE, 1992.

Dunlop, K. M., Jarvis, T., Benoit-Bird, K. J., Waluk, C. M., Caress, D. W., Thomas, H., & Smith Jr, K. L. (2018). Detection and characterisation of deep-sea benthopelagic animals from an autonomous underwater vehicle with a multibeam echosounder: A proof of concept and description of data-processing methods. Deep Sea Research Part I: Oceanographic Research Papers, 134, 64-79. https://doi.org/10.1016/j.dsr.2018.01.006

Durant, J. M., Hjermann, D. Ø., Ottersen, G., & Stenseth, N. C. (2007). Climate and the match or mismatch between predator requirements and resource availability. Climate research, 33(3), 271-283. https://doi.org/10.3354/cr033271

Edgar, G. J., Barrett, N. S., & Morton, A. J. (2004). Biases associated with the use of underwater visual census techniques to quantify the density and size-structure of fish populations. Journal of Experimental Marine Biology and Ecology, 308(2), 269-290.

Edgar, G. J., Cooper, A., Baker, S. C., Barker, W., Barrett, N. S., Becerro, M. A., . . . Stuart- Smith, R. D. (2020). Reef Life Survey: Establishing the ecological basis for conservation of shallow marine life. Biological Conservation, 252, 108855. https://doi.org/10.1016/j.biocon.2020.108855

Edgar, G. J., & Stuart-Smith, R. D. (2014). Systematic global assessment of reef fish communities by the Reef Life Survey program. Sci Data, 1(1), 140007. https://doi.org/10.1038/sdata.2014.7

Edwards, M., & Richardson, A. J. (2004). Impact of climate change on marine pelagic phenology and trophic mismatch. Nature, 430(7002), 881-884. https://doi.org/10.1038/nature02808

194

El-Haweet, A. (2001). Catch composition and management of daytime purse seine fishery on the Southern Mediterranean Sea coast, Abu Qir Bay, Egypt. Mediterr. Mar. Sci, 2, 119-126. https://doi.org/10.12681/mms.270

Elith, J., H. Graham, C., P. Anderson, R., Dudík, M., Ferrier, S., Guisan, A., . . . Lehmann, A. (2006). Novel methods improve prediction of species’ distributions from occurrence data. Ecography, 29(2), 129-151. https://doi.org/10.1111/j.2006.0906-7590.04596.x

Elmendorf, S. C., Henry, G. H., Hollister, R. D., Fosaa, A. M., Gould, W. A., Hermanutz, L., . . . Lévesque, E. (2015). Experiment, monitoring, and gradient methods used to infer climate change effects on plant communities yield consistent patterns. Proceedings of the national academy of sciences, 112(2), 448-452. https://doi.org/10.1073/pnas.1410088112

Emmett, R. L., & Krutzikowsky, G. K. (2008). Nocturnal feeding of Pacific hake and jack mackerel off the mouth of the Columbia River, 1998-2004: Implications for juvenile salmon predation. Transactions of the American Fisheries Society, 137(3), 657-676. https://doi.org/10.1577/T06-058.1

Erez, J. (1990). On the importance of food sources in coral-reef ecosystems. Ecosystems of the world, 25, 411-418.

ESRI, R. (2011). ArcGIS desktop: release 10. Environmental Systems Research Institute, CA.

Essington, T. E., Moriarty, P. E., Froehlich, H. E., Hodgson, E. E., Koehn, L. E., Oken, K. L., . . . Stawitz, C. C. (2015). Fishing amplifies forage fish population collapses. Proc Natl Acad Sci U S A, 112(21), 6648-6652. https://doi.org/10.1073/pnas.1422020112

Evans, K., Bax, N., & Smith, D. (2017). Australia state of the environment 2016: marine environment, independent report to the Australian Government Minister for the Environment and Energy. https://soe.environment.gov.au/theme/marine- environment

Everett, J., Baird, M., Oke, P., & Suthers, I. (2012). An avenue of eddies: Quantifying the biophysical properties of mesoscale eddies in the Tasman Sea. Geophysical Research Letters, 39(16). https://doi.org/10.1029/2012GL053091

Everett, J. D., Baird, M. E., Roughan, M., Suthers, I. M., & Doblin, M. A. (2014). Relative impact of seasonal and oceanographic drivers on surface chlorophyll a along a Western Boundary Current. Progress in oceanography, 120, 340-351. https://doi.org/10.1016/j.pocean.2013.10.016

Fauchald, P. (2009). Spatial interaction between seabirds and prey: review and synthesis. Marine Ecology Progress Series, 391, 139-151. https://doi.org/10.3354/meps07818

Ferguson, R. (1958). The preferred temperature of fish and their midsummer distribution in temperate lakes and streams. Journal of the Fisheries Board of Canada, 15(4), 607-624. https://doi.org/10.1139/f58-032

Fernandes, P. G. (2009). Classification trees for species identification of fish-school echotraces. ICES Journal of Marine Science, 66(6), 1073-1080. https://doi.org/10.1093/icesjms/fsp060

195

Figueroa-Pico, J., Carpio, A. J., & Tortosa, F. S. (2020). Turbidity: A key factor in the estimation of fish species richness and abundance in the rocky reefs of Ecuador. Ecological Indicators, 111, 106021. https://doi.org/10.1016/j.ecolind.2019.106021

Filbee-Dexter, K., & Wernberg, T. (2018). Rise of turfs: A new battlefront for globally declining kelp forests. BioScience, 68(2), 64-76. https://doi.org/10.1093/biosci/bix147

Floeter, S. R., Behrens, M., Ferreira, C., Paddack, M., & Horn, M. (2005). Geographical gradients of marine herbivorous fishes: patterns and processes. Marine Biology, 147(6), 1435-1447. https://doi.org/10.1007/s00227-005-0027-0

Floeter, S. R., Ferreira, C., Dominici‐Arosemena, A., & Zalmon, I. (2004). Latitudinal gradients in Atlantic reef fish communities: trophic structure and spatial use patterns. Journal of Fish Biology, 64(6), 1680-1699. https://doi.org/10.1111/j.0022- 1112.2004.00428.x

Folpp, H., Lowry, M., Gregson, M., & Suthers, I. M. (2011). Colonization and community development of fish assemblages associated with estuarine artificial reefs. Brazilian Journal of Oceanography, 59(SPE1), 55-67.

Foote, K. G., Chu, D., Hammar, T. R., Baldwin, K. C., Mayer, L. A., Hufnagle, L. C., Jr., & Jech, J. M. (2005). Protocols for calibrating multibeam sonar. J Acoust Soc Am, 117(4 Pt 1), 2013-2027. https://doi.org/10.1121/1.1869073

Foote, K. G., & MacLennan, D. N. (1984). Comparison of copper and tungsten carbide calibration spheres. the Journal of the Acoustical Society of America, 75(2), 612-616. https://doi.org/10.1121/1.390489

Forrester, G. E. (1991). Social rank, individual size and group composition as determinants of food consumption by humbug damselfish, Dascyllus aruanus. Animal Behaviour, 42(5), 701-711. https://doi.org/10.1016/S0003-3472(05)80116-2

Foster, S. D., & Bravington, M. V. (2013). A Poisson–Gamma model for analysis of ecological non-negative continuous data. Environmental and ecological statistics, 20(4), 533-552. https://doi.org/10.1007/s10651-012-0233-0

Freestone, A. L., Osman, R. W., Ruiz, G. M., & Torchin, M. E. (2011). Stronger predation in the tropics shapes species richness patterns in marine communities. Ecology, 92(4), 983-993. https://doi.org/10.1890/09-2379.1

Fréon, P., Gerlotto, F., & Soria, M. (1992). Changes in school structure according to external stimuli: description and influence on acoustic assessment. Fisheries Research, 15(1-2), 45-66. https://doi.org/10.1016/0165-7836(92)90004-D

Froese, R., and, & Pauly, D. (2009). FishBase: a global information system on fishes. http://www.fishbase.org

Gal, G. (1993). The interactions between topography, currents and zooplankton. (Thesis). Hebrew University, Eilat, Israel.

196

Galzin, R. (1987). Structure of fish communities of French Polynesian coral reefs. 11. Temporal Scales. Mar. Ecol. Prog. Ser, 41, 137-145. https://www.jstor.org/stable/24827444

Gee, J. (1989). An ecological and economic review of meiofauna as food for fish. Zoological journal of the Linnean Society, 96(3), 243-261. https://doi.org/10.1111/j.1096- 3642.1989.tb02259.x

Genin, A. (2004). Bio-physical coupling in the formation of zooplankton and fish aggregations over abrupt topographies. Journal of Marine Systems, 50(1-2), 3-20. https://doi.org/10.1016/j.jmarsys.2003.10.008

Genin, A., Lindemann, Y., Roberts, P., & Jaffe, J. (2016). Predation Rates of Zooplankton by Fish Quantified with a Novel Acoustic and Optic System. Paper presented at the American Geophysical Union, Ocean Sciences Meeting 2016, abstract# ME33A-02.

Genin, A., Monismith, S. G., Reidenbach, M. A., Yahel, G., & Koseff, J. R. (2009). Intense benthic grazing of phytoplankton in a coral reef. Limnology and Oceanography, 54(3), 938-951. https://doi.org/10.4319/lo.2009.54.3.0938

Gerlotto, F., Soria, M., & Fréon, P. (1999). From two dimensions to three: the use of multibeam sonar for a new approach in fisheries acoustics. Canadian Journal of Fisheries and Aquatic Sciences, 56(1), 6-12. https://doi.org/10.1139/f98-138

Gili, J.-M., & Coma, R. (1998). Benthic suspension feeders: their paramount role in littoral marine food webs. Trends in ecology & evolution, 13(8), 316-321. https://doi.org/10.1016/S0169-5347(98)01365-2

Gladfelter, W. (1979). Twilight migrations and foraging activities of the copper sweeper Pempheris schomburgki (Teleostei: Pempheridae). Marine Biology, 50(2), 109-119. https://doi.org/10.1007/BF00397815

Glasby, T., & Kingsford, M. (1994). Atypichthys strigatus (Pisces: Scorpididae): An opportunistic planktivore that responds to benthic disturbances and cleans other fishes. Australian journal of ecology, 19(4), 385-394. https://doi.org/10.1111/j.1442- 9993.1994.tb00504.x

Global Coral Reef Monitoring Network. (2014). Status and trends of Caribbean coral reefs: 1970-2012. https://www.iucn.org/content/status-and-trends-caribbean-coral-reefs- 1970-2012

Goetze, J. S., Januchowski‐Hartley, F. A., Claudet, J., Langlois, T. J., Wilson, S. K., & Jupiter, S. D. (2017). Fish wariness is a more sensitive indicator to changes in fishing pressure than abundance, length or biomass. Ecological applications, 27(4), 1178-1189. https://doi.org/10.1002/eap.1511

Graham, N., & Nash, K. (2013). The importance of structural complexity in coral reef ecosystems. Coral reefs, 32(2), 315-326. https://doi.org/10.1007/s00338-012-0984-y

Gray, A. E., Williams, I. D., Stamoulis, K. A., Boland, R. C., Lino, K. C., Hauk, B. B., . . . Kosaki, R. K. (2016). Comparison of Reef Fish Survey Data Gathered by Open and

197

Closed Circuit SCUBA Divers Reveals Differences in Areas With Higher Fishing Pressure. PloS one, 11(12), e0167724. https://doi.org/10.1371/journal.pone.0167724

Gregson, M. A., & Booth, D. J. (2005). Zooplankton patchiness and the associated shoaling response of the temperate reef fish Trachinops taeniatus. Marine Ecology Progress Series, 299, 269-275. https://doi.org/10.3354/meps299269

Grémillet, D., Lewis, S., Drapeau, L., van Der Lingen, C. D., Huggett, J. A., Coetzee, J. C., . . . Ryan, P. G. (2008). Spatial match–mismatch in the Benguela upwelling zone: should we expect chlorophyll and sea‐surface temperature to predict marine predator distributions? Journal of Applied Ecology, 45(2), 610-621. https://doi.org/10.1111/j.1365-2664.2007.01447.x

Grigg, R. W. (1994). Effects of sewage discharge, fishing pressure and habitat complexity on coral ecosystems and reef fishes in Hawaii. Marine ecology progress series. Oldendorf, 103(1), 25-34. https://doi.org/10.3354/meps103025

Guillard, J., Paul, F., Laloë, T., & Brehmer, P. (2011). Three‐dimensional internal spatial structure of young‐of‐the‐year pelagic freshwater fish provides evidence for the identification of fish school species. Limnology and Oceanography: Methods, 9(7), 322- 328. https://doi.org/10.4319/lom.2011.9.322

Hagen, W., Van Vleet, E. S., & Kattner, G. (1996). Seasonal lipid storage as overwintering strategy of Antarctic krill. Marine Ecology Progress Series, 134, 85-89. https://doi.org/10.3354/meps134085

Hallier, J.-P., & Gaertner, D. (2008). Drifting fish aggregation devices could act as an ecological trap for tropical tuna species. Marine Ecology Progress Series, 353, 255-264. https://doi.org/10.3354/meps07180

Hamner, W., Jones, M., Carleton, J., Hauri, I., & Williams, D. M. (1988). Zooplankton, planktivorous fish, and water currents on a windward reef face: Great Barrier Reef, Australia. Bulletin of marine Science, 42(3), 459-479.

Hamner, W. M. (1995). Predation, cover, and convergent evolution in epipelagic oceans. Marine & Freshwater Behaviour & Phy, 26(2-4), 71-89. https://doi.org/10.1080/10236249509378930

Harris, G., Nilsson, C., Clementson, L., & Thomas, D. (1987). The water masses of the east coast of Tasmania: seasonal and interannual variability and the influence on phytoplankton biomass and productivity. Marine and Freshwater Research, 38(5), 569- 590. https://doi.org/10.1071/MF9870569

Harvey, E., Fletcher, D., Shortis, M. R., & Kendrick, G. A. (2004). A comparison of underwater visual distance estimates made by scuba divers and a stereo-video system: implications for underwater visual census of reef fish abundance. Marine and Freshwater Research, 55(6), 573-580. https://doi.org/10.1071/MF03130

Hedgpeth, J. W. (1957). Marine biogeography. Treatise on marine ecology and paleoecology, 1, 359-382.

198

Heip, C. (1995). Eutrophication and zoobenthos dynamics. Ophelia, 41(1), 113-136. https://doi.org/10.1080/00785236.1995.10422040

Hemer, M., Pitman, T., McInnes, K., & Rosebrock, U. (2018). The Australian Wave Energy Atlas Project Overview and Final Report. https://arena.gov.au/knowledge- bank/australian-wave-energy-atlas-project-overview-final-report/

Hijmans, R. J. (2020). raster: Geographic Data Analysis and Modeling. R package version 3.3-13. https://CRAN.R-project.org/package=raster

Hijmans, R. J., Williams, E., Vennes, C., & Hijmans, M. R. J. (2020). geosphere: Spherical Trigonometry. R package version 1.5-10. https://CRAN.R- project.org/package=geosphere

Hilborn, R., Amoroso, R. O., Bogazzi, E., Jensen, O. P., Parma, A. M., Szuwalski, C., & Walters, C. J. (2017). When does fishing forage species affect their predators? Fisheries Research, 191, 211-221. https://doi.org/10.1016/j.fishres.2017.01.008

Hilborn, R., Amoroso, R. O., Bogazzi, E., Jensen, O. P., Parma, A. M., Szuwalski, C., & Walters, C. J. (2018). Response to Pikitch et al. https://doi.org/10.1016/j.fishres.2017.07.025

Hixon, M. A., & Beets, J. P. (1993). Predation, prey refuges, and the structure of coral‐reef fish assemblages. Ecological Monographs, 63(1), 77-101. https://doi.org/10.2307/2937124

Hobday, A. J. (2011). Sliding baselines and shuffling species: implications of climate change for marine conservation. Marine Ecology, 32(3), 392-403. https://doi.org/10.1111/j.1439-0485.2011.00459.x

Hobson, E. S., & Chess, J. R. (1978). Trophic relationships among fishes and plankton in the lagoon at Enewetak Atoll, Marshall Islands. Fish Bull, 76(1), 133-153.

Holbrook, S. J., & Schmitt, R. J. (1988). The combined effects of predation risk and food reward on patch selection. Ecology, 69(1), 125-134. https://doi.org/10.2307/1943167

Holland, M. M., Everett, J. D., Cox, M. J., Doblin, M. A., & Suthers, I. M. (2020a). Pelagic forage fish distribution in a dynamic shelf ecosystem – Thermal demands and zooplankton prey distribution. Estuarine, coastal and shelf science, 107074. https://doi.org/10.1016/j.ecss.2020.107074

Holmes, T. H., Wilson, S. K., Travers, M. J., Langlois, T. J., Evans, R. D., Moore, G. I., . . . Hickey, K. (2013). A comparison of visual‐and stereo‐video based fish community assessment methods in tropical and temperate marine waters of Western Australia. Limnology and Oceanography: Methods, 11(7), 337-350. https://doi.org/10.4319/lom.2013.11.337

199

Holzman, R., Ohavia, M., Vaknin, R., & Genin, A. (2007). Abundance and distribution of nocturnal fishes over a coral reef during the night. Marine Ecology Progress Series, 342, 205-215. https://doi.org/10.3354/meps342205

Holzman, R., Reidenbach, M. A., Monismith, S. G., Koseff, J. R., & Genin, A. (2005). Near- bottom depletion of zooplankton over a coral reef II: relationships with zooplankton swimming ability. Coral reefs, 24(1), 87-94. https://doi.org/10.1007/s00338-004- 0450-6

Horn, M. H. (1989). Biology of marine herbivorous fishes. Oceanogr Mar Biol, 27, 167-272.

Horn, P. L. (1993). Growth, age structure, and productivity of jack mackerels (Trachurus spp.) in New Zealand waters. New Zealand Journal of Marine and Freshwater Research, 27(2), 145-155. https://doi.org/10.1080/00288330.1993.9516553

Hunter, W., & Sayer, M. (2009). The comparative effects of habitat complexity on faunal assemblages of northern temperate artificial and natural reefs. ICES Journal of Marine Science, 66(4), 691-698. https://doi.org/10.1093/icesjms/fsp058

Hwang, D. J., Park, J. S., & Lee, Y. W. (2004). Estimation of fish school abundance by using an echo sounder in an artificial reef area. Korean Journal of Fisheries and Aquatic Sciences, 37(3), 249-254. https://doi.org/10.5657/kfas.2004.37.3.249

Hwang, K., Yoon, E.-A., Lee, K., Lee, H., & Hwang, D.-J. (2015). Multifrequency acoustic scattering characteristics of jack mackerel by KRM model. Journal of the Korean Society of Fisheries and Ocean Technology, 51(3), 424-431. https://doi.org/10.3796/KSFT.2015.51.3.424

Hyun, K. H., & He, R. (2010). Coastal upwelling in the South Atlantic Bight: A revisit of the 2003 cold event using long term observations and model hindcast solutions. Journal of Marine Systems, 83(1-2), 1-13. https://doi.org/10.1016/j.jmarsys.2010.05.014

Innangi, S., Bonanno, A., Tonielli, R., Gerlotto, F., Innangi, M., & Mazzola, S. (2016). High resolution 3-D shapes of fish schools: A new method to use the water column backscatter from hydrographic MultiBeam Echo Sounders. Applied Acoustics, 111, 148-160. https://doi.org/10.1016/j.apacoust.2016.04.017

Integrated Marine Observing System. (2018). IMOS - SRS - MODIS - 01 day - Ocean Colour - SST. https://portal.aodn.org.au/search

Irigoien, X., Fernandes, J. A., Grosjean, P., Denis, K., Albaina, A., & Santos, M. (2009). Spring zooplankton distribution in the Bay of Biscay from 1998 to 2006 in relation with anchovy recruitment. Journal of plankton research, 31(1), 1-17. https://doi.org/10.1093/plankt/fbn096

Jannasch, H. W., & Mottl, M. J. (1985). Geomicrobiology of deep-sea hydrothermal vents. Science, 229(4715), 717-725. https://doi.org/10.1126/science.229.4715.717

Johansen, J., Bellwood, D., & Fulton, C. (2008). Coral reef fishes exploit flow refuges in high-flow habitats. Marine Ecology Progress Series, 360, 219-226. https://doi.org/10.3354/meps07482

200

Johnson, C. R., Banks, S. C., Barrett, N. S., Cazassus, F., Dunstan, P. K., Edgar, G. J., . . . Helidoniotis, F. (2011). Climate change cascades: Shifts in oceanography, species' ranges and subtidal marine community dynamics in eastern Tasmania. Journal of Experimental Marine Biology and Ecology, 400(1-2), 17-32. https://doi.org/10.1016/j.jembe.2011.02.032

Johnson, N. L., Kotz, S., & Balakrishnan, N. (1970). Continuous univariate distributions (Vol. 1): Houghton Mifflin Boston.

Jørgensen, D. (2012). OSPAR’s exclusion of rigs-to-reefs in the North Sea. Ocean & Coastal Management, 58, 57-61. https://doi.org/10.1016/j.ocecoaman.2011.12.012

Jørgensen, T., Løkkeborg, S., & Soldal, A. V. (2002). Residence of fish in the vicinity of a decommissioned oil platform in the North Sea. ICES Journal of Marine Science, 59(suppl), S288-S293. https://doi.org/10.1006/jmsc.2001.1165

Jung, D., Kim, J., & Byun, G. (2018). Numerical modeling and simulation technique in time- domain for multibeam echo sounder. International Journal of Naval Architecture and Ocean Engineering, 10(2), 225-234. https://doi.org/10.1016/j.ijnaoe.2017.08.004

Kang, M.-H. (2011). Semiautomated analysis of data from an imaging sonar for fish counting, sizing, and tracking in a post-processing application. Fisheries and aquatic sciences, 14(3), 218-225. https://doi.org/10.5657/FAS.2011.0218

Kaplan, I. C., Brown, C. J., Fulton, E. A., Gray, I. A., Field, J. C., & Smith, A. D. (2013). Impacts of depleting forage species in the California Current. Environmental Conservation, 40(4), 380-393. https://doi.org/10.1017/S0376892913000052

Kaplan, M. B., & Mooney, T. A. (2015). Ambient noise and temporal patterns of boat activity in the US Virgin Islands National Park. Mar Pollut Bull, 98(1-2), 221-228. https://doi.org/10.1016/j.marpolbul.2015.06.047

Kara, A. B., Rochford, P. A., & Hurlburt, H. E. (2000). Mixed layer depth variability and barrier layer formation over the North Pacific Ocean. Journal of Geophysical Research: Oceans, 105(C7), 16783-16801. https://doi.org/10.1029/2000JC900071

Kaschner, K., Karpouzi, V., Watson, R., & Pauly, D. (2006). Forage fish consumption by marine mammals and seabirds. http://www.academia.edu/download/48429030/Forage_fish_consumption_by_mar ine_mammal20160829-1669-1yvnfaz.pdf

Kaschner, K., Ready, J., Agbayani, E., Rius, J., Kesner-Reyes, K., Eastwood, P., . . . Close, C. (2008). AquaMaps: Predicted range maps for aquatic species. https://www.aquamaps.org/

Keister, J., Peterson, W., & Pierce, S. (2009). Zooplankton distribution and cross-shelf transfer of carbon in an area of complex mesoscale circulation in the northern California Current. Deep Sea Research Part I: Oceanographic Research Papers, 56(2), 212-231. https://doi.org/10.1016/j.dsr.2008.09.004

201

Keller, K., Steffe, A. S., Lowry, M., Murphy, J. J., & Suthers, I. M. (2016). Monitoring boat- based recreational fishing effort at a nearshore artificial reef with a shore-based camera. Fisheries Research, 181, 84-92. https://doi.org/10.1016/j.fishres.2016.03.025

Kennedy, M., & Gray, R. D. (1993). Can ecological theory predict the distribution of foraging animals? A critical analysis of experiments on the ideal free distribution. Oikos, 158-166. https://doi.org/10.2307/3545322

Kieser, R., Mulligan, T., Richards, L. J., & Leaman, B. (1993). Bias correction of rockfish school cross section widths from digitized echo sounder data. Canadian Journal of Fisheries and Aquatic Sciences, 50(8), 1801-1811. https://doi.org/10.1139/f93-202

Kiflawi, M., & Genin, A. (1997). Prey flux manipulation and the feeding rates of reef‐ dwelling planktivorous fish. Ecology, 78(4), 1062-1077. https://doi.org/10.1890/0012-9658(1997)078[1062:PFMATF]2.0.CO;2

Kingsford, M., & MacDiarmid, A. (1988). Interrelations between planktivorous reef fish and zooplankton in temperate waters. Marine ecology progress series. Oldendorf, 48(2), 103- 117.

Klemas, V. (2013). Fisheries applications of remote sensing: an overview. Fisheries Research, 148, 124-136. https://doi.org/10.1016/j.fishres.2012.02.027

Komyakova, V., Chamberlain, D., Jones, G. P., & Swearer, S. E. (2019). Assessing the performance of artificial reefs as substitute habitat for temperate reef fishes: Implications for reef design and placement. Science of the total environment, 668, 139- 152. https://doi.org/10.1016/j.scitotenv.2019.02.357

Korneliussen, R. J., Heggelund, Y., Eliassen, I. K., & Johansen, G. O. (2009). Acoustic species identification of schooling fish. ICES Journal of Marine Science, 66(6), 1111- 1118. https://doi.org/10.1093/icesjms/fsp119

Korneliussen, R. J., & Ona, E. (2002). An operational system for processing and visualizing multi-frequency acoustic data. ICES Journal of Marine Science, 59(2), 293-313. https://doi.org/10.1006/jmsc.2001.1168

Krumhansl, K. A., & Scheibling, R. E. (2012). Production and fate of kelp detritus. Marine Ecology Progress Series, 467, 281-302. https://doi.org/10.3354/meps09940

Kulbicki, M. (1998). How the acquired behaviour of commercial reef fishes may influence the results obtained from visual censuses. Journal of Experimental Marine Biology and Ecology, 222(1-2), 11-30. https://doi.org/10.1016/S0022-0981(97)00133-0

Kumagai, N. H., Garcia Molinos, J., Yamano, H., Takao, S., Fujii, M., & Yamanaka, Y. (2018). Ocean currents and herbivory drive macroalgae-to-coral community shift under climate warming. Proc Natl Acad Sci U S A, 115(36), 8990-8995. https://doi.org/10.1073/pnas.1716826115

Kuriyama, P. T., Zwolinski, J. P., Hill, K. T., & Crone, P. R. (2020). Assessment of the Pacific sardine resource in 2020 for US management in 2020-2021. https://doi.org/10.25923/r2fz-4y79

202

Lamont, T., Van Den Berg, M., & Barlow, R. (2016). Agulhas Current influence on the shelf dynamics of the KwaZulu-Natal Bight. Journal of Physical Oceanography, 46(4), 1323- 1338. https://doi.org/10.1175/JPO-D-15-0152.1

Lampert, W. (1989). The adaptive significance of diel vertical migration of zooplankton. Functional Ecology, 21-27. https://doi.org/10.2307/2389671

Last, P. R., White, W. T., Gledhill, D. C., Hobday, A. J., Brown, R., Edgar, G. J., & Pecl, G. (2011). Long‐term shifts in abundance and distribution of a temperate fish fauna: a response to climate change and fishing practices. Global ecology and biogeography, 20(1), 58-72. https://doi.org/10.1111/j.1466-8238.2010.00575.x

Layton, C., Coleman, M. A., Marzinelli, E. M., Steinberg, P. D., Swearer, S. E., Vergés, A., . . . Johnson, C. R. (2020). Kelp forest restoration in Australia. Frontiers in Marine Science, 7, 74. https://doi.org/10.3389/fmars.2020.00074

Lee, D.-J. (2013). Monitoring of fish aggregations responding to artificial reefs using a split- beam echo sounder, side-scan sonar, and an underwater CCTV camera system at Suyeong Man, Busan, Korea. Korean Journal of Fisheries and Aquatic Sciences, 46(3), 266-272. https://doi.org/10.5657/KFAS.2013.0266

Leitao, F., Santos, M. N., Erzini, K., & Monteiro, C. C. (2008). Fish assemblages and rapid colonization after enlargement of an artificial reef off the Algarve coast (Southern Portugal). Marine Ecology, 29(4), 435-448. https://doi.org/10.1111/j.1439- 0485.2008.00253.x

Li, J., & Heap, A. D. (2011). A review of comparative studies of spatial interpolation methods in environmental sciences: Performance and impact factors. Ecological Informatics, 6(3-4), 228-241. https://doi.org/10.1016/j.ecoinf.2010.12.003

Lindegren, M., Checkley, D. M., Jr., Rouyer, T., MacCall, A. D., & Stenseth, N. C. (2013). Climate, fishing, and fluctuations of sardine and anchovy in the California Current. Proc Natl Acad Sci U S A, 110(33), 13672-13677. https://doi.org/10.1073/pnas.1305733110

Lindenmayer, D. B., Likens, G. E., Krebs, C. J., & Hobbs, R. J. (2010). Improved probability of detection of ecological "surprises". Proc Natl Acad Sci U S A, 107(51), 21957- 21962. https://doi.org/10.1073/pnas.1015696107

Lindfield, S. J., McIlwain, J. L., & Harvey, E. S. (2014). Depth refuge and the impacts of SCUBA spearfishing on coral reef fishes. PloS one, 9(3), e92628. https://doi.org/10.1371/journal.pone.0092628

Ling, S., Johnson, C., Ridgway, K., Hobday, A., & Haddon, M. (2009). Climate‐driven range extension of a sea urchin: inferring future trends by analysis of recent population dynamics. Global Change Biology, 15(3), 719-731. https://doi.org/10.1111/j.1365- 2486.2008.01734.x

Ling, S. D., & Keane, J. P. (2018). Resurvey of the longspined sea urchin (Centrostephanus rodgersii) and associated barren reef in Tasmania. https://eprints.utas.edu.au/28761/

203

Lluch‐Belda, D., Schwartzlose, R., Serra, R., Parrish, R., Kawasaki, T., Hedgecock, D., & Crawford, R. (1992). Sardine and anchovy regime fluctuations of abundance in four regions of the world oceans: a workshop report. Fisheries oceanography, 1(4), 339-347. https://doi.org/10.1111/j.1365-2419.1992.tb00006.x

Logerwell, E. A., & Hargreaves, N. B. (1996). The distribution of sea birds relative to their fish prey off Vancouver Island: opposing results at large and small spatial scales. Fisheries oceanography, 5(3‐4), 163-175. https://doi.org/10.1111/j.1365- 2419.1996.tb00115.x

Longhurst, A. R. (2010). Ecological geography of the sea: Elsevier.

Longo, G. O., Ferreira, C. E., & Floeter, S. R. (2014). Herbivory drives large-scale spatial variation in reef fish trophic interactions. Ecol Evol, 4(23), 4553-4566. https://doi.org/10.1002/ece3.1310

Longo, G. O., Hay, M. E., Ferreira, C. E., & Floeter, S. R. (2019). Trophic interactions across 61 degrees of latitude in the Western Atlantic. Global ecology and biogeography, 28(2), 107-117. https://doi.org/10.1111/geb.12806

Loures, R. C., & Pompeu, P. (2015). Seasonal and diel changes in fish distribution in a tropical hydropower plant tailrace: evidence from hydroacoustic and gillnet sampling. Fisheries Management and Ecology, 22(3), 185-196.

Love, M. S., & York, A. (2005). A comparison of the fish assemblages associated with an oil/gas pipeline and adjacent seafloor in the Santa Barbara Channel, Southern California Bight. Bulletin of marine Science, 77(1), 101-118.

Lowry, M., Becker, A., Folpp, H., McLeod, J., & Taylor, M. D. (2017). Residency and movement patterns of yellowfin bream (Acanthopagrus australis) released at natural and artificial reef sites. Marine and Freshwater Research, 68(8), 1479-1488. https://doi.org/10.1071/MF16351

Lowry, M., Folpp, H., Gregson, M., & Suthers, I. (2012). Comparison of baited remote underwater video (BRUV) and underwater visual census (UVC) for assessment of artificial reefs in estuaries. Journal of Experimental Marine Biology and Ecology, 416, 243-253. https://doi.org/10.1016/j.jembe.2012.01.013

Lowry, M., Glasby, T., Boys, C., Folpp, H., Suthers, I., & Gregson, M. (2014). Response of fish communities to the deployment of estuarine artificial reefs for fisheries enhancement. Fisheries Management and Ecology, 21(1), 42-56. https://doi.org/10.1111/fme.12048

Lucas, A. J., Dupont, C. L., Tai, V., Largier, J. L., Palenik, B., & Franks, P. J. (2011). The green ribbon: Multiscale physical control of phytoplankton productivity and community structure over a narrow continental shelf. Limnology and Oceanography, 56(2), 611-626. https://doi.org/10.4319/lo.2011.56.2.0611

Lukoschek, V., & McCormick, M. (2000). A review of multi-species foraging associations in fishes and their ecological significance. Paper presented at the Proceedings of the 9th international coral reef symposium.

204

Lumpkin, R., Özgökmen, T., & Centurioni, L. (2017). Advances in the application of surface drifters. Annual Review of Marine Science, 9, 59-81.

Mackinson, S., Freeman, S., Flatt, R., & Meadows, B. (2004). Improved acoustic surveys that save time and money: integrating fisheries and ground-discrimination acoustic technologies. Journal of Experimental Marine Biology and Ecology, 305(2), 129-140. https://doi.org/10.1016/j.jembe.2003.10.020

MacLennan, D. N., Fernandes, P. G., & Dalen, J. (2002). A consistent approach to definitions and symbols in fisheries acoustics. ICES Journal of Marine Science, 59(2), 365-369. https://doi.org/10.1006/jmsc.2001.1158

MacLennan, D. N., & Simmonds, E. J. (2013). Fisheries acoustics (Vol. 5): Springer Science & Business Media.

MacNeil, M. A., Tyler, E. H., Fonnesbeck, C. J., Rushton, S. P., Polunin, N. V., & Conroy, M. J. (2008). Accounting for detectability in reef-fish biodiversity estimates. Marine Ecology Progress Series, 367, 249-260. https://doi.org/10.3354/meps07580

Macreadie, P. I., Fowler, A. M., & Booth, D. J. (2011). Rigs‐to‐reefs: will the deep sea benefit from artificial habitat? Frontiers in Ecology and the Environment, 9(8), 455-461. https://doi.org/10.1890/100112

Madin, E. M. P., Madin, J. S., Harmer, A. M. T., Barrett, N. S., Booth, D. J., Caley, M. J., . . . Sweatman, H. P. A. (2020). Latitude and protection affect decadal trends in reef trophic structure over a continental scale. Ecol Evol, 10(14), 6954-6966. https://doi.org/10.1002/ece3.6347

Maeng, J.-S., Park, J.-H., Jang, S.-M., & Han, S.-Y. (2013). A modeling approach to energy savings of flying Canada geese using computational fluid dynamics. Journal of Theoretical Biology, 320, 76-85. https://doi.org/10.1016/j.jtbi.2012.11.032

Magnuson, J. J., Crowder, L. B., & Medvick, P. A. (1979). Temperature as an ecological resource. American Zoologist, 19(1), 331-343. https://doi.org/10.1093/icb/19.1.331

Malan, N. C., Roughan, M., & Kerry, C. G. (in review). Rate of coastal temperature rise adjacent to a warming western boundary current is non-uniform with latitude. https://doi.org/10.1002/essoar.10504392.1

Malcolm, H. A., Gladstone, W., Lindfield, S., Wraith, J., & Lynch, T. P. (2007). Spatial and temporal variation in reef fish assemblages of marine parks in New South Wales, Australia—baited video observations. Marine Ecology Progress Series, 350, 277-290. https://doi.org/10.3354/meps07195

Malcolm, H. A., Jordan, A., Schultz, A. L., Smith, S. D., Ingleton, T., Foulsham, E., . . . Hill, N. (2016). Integrating seafloor habitat mapping and fish assemblage patterns improves spatial management planning in a marine park. Journal of Coastal Research(75), 1292-1296. https://doi.org/10.2112/SI75-259.1

Mann, K. (1973). Seaweeds: their productivity and strategy for growth. Science, 182(4116), 975-981. https://www.jstor.org/stable/1737803

205

Maravelias, C. D. (1999). Habitat selection and clustering of a pelagic fish: effects of topography and bathymetry on species dynamics. Canadian Journal of Fisheries and Aquatic Sciences, 56(3), 437-450. https://doi.org/10.1139/f98-176

Marini, S., Fanelli, E., Sbragaglia, V., Azzurro, E., Del Rio Fernandez, J., & Aguzzi, J. (2018). Tracking Fish Abundance by Underwater Image Recognition. Sci Rep, 8(1), 13748. https://doi.org/10.1038/s41598-018-32089-8

Marras, S., Batty, R. S., & Domenici, P. (2012). Information transfer and antipredator maneuvers in schooling herring. Adaptive Behavior, 20(1), 44-56. https://doi.org/10.1177%2F1059712311426799

Marras, S., Killen, S. S., Lindström, J., McKenzie, D. J., Steffensen, J. F., & Domenici, P. (2015). Fish swimming in schools save energy regardless of their spatial position. Behavioral ecology and sociobiology, 69(2), 219-226. https://doi.org/10.1007/s00265- 014-1834-4

Marton, N., & Steven, A. (2019). Chapter 7: Small Pelagic Fishery. https://www.agriculture.gov.au/abares/research-topics/fisheries/fishery- status/small-pelagic#71-description-of-the-fishery

Matsumura, S., Arlinghaus, R., & Dieckmann, U. (2010). Foraging on spatially distributed resources with sub‐optimal movement, imperfect information, and travelling costs: departures from the ideal free distribution. Oikos, 119(9), 1469-1483. https://doi.org/10.1111/j.1600-0706.2010.18196.x

Maxwell, S. M., & Morgan, L. E. (2013). Foraging of seabirds on pelagic fishes: implications for management of pelagic marine protected areas. Marine Ecology Progress Series, 481, 289-303. https://doi.org/10.3354/meps10255

Mayer, L., Li, Y., & Melvin, G. (2002). 3D visualization for pelagic fisheries research and assessment. ICES Journal of Marine Science, 59(1), 216-225. https://doi.org/10.1006/jmsc.2001.1125

McClanahan, T. R., Graham, N. A., & Darling, E. S. (2014). Coral reefs in a crystal ball: predicting the future from the vulnerability of corals and reef fishes to multiple stressors. Current Opinion in Environmental Sustainability, 7, 59-64. https://doi.org/10.1016/j.cosust.2013.11.028

McEnnulty, F. R., Davies, C. H., Armstrong, A. O., Atkins, N., Coman, F., Clementson, L., . . . Koslow, J. A. (2020). A database of zooplankton biomass in Australian marine waters. Scientific Data, 7(1), 1-9. https://doi.org/10.6084/m9.figshare.12651425

McInnes, A. M., Ryan, P. G., Lacerda, M., Deshayes, J., Goschen, W. S., & Pichegru, L. (2017). Small pelagic fish responses to fine-scale oceanographic conditions: implications for the endangered African penguin. Marine Ecology Progress Series, 569, 187-203. https://doi.org/10.3354/meps12089

McKelvey, D. R., & Wilson, C. D. (2006). Discriminant classification of fish and zooplankton backscattering at 38 and 120 kHz. Transactions of the American Fisheries Society, 135(2), 488-499. https://doi.org/10.1577/T04-140.1

206

McLean, D., Partridge, J., Bond, T., Birt, M., Bornt, K., & Langlois, T. (2017). Using industry ROV videos to assess fish associations with subsea pipelines. Continental Shelf Research, 141, 76-97. https://doi.org/10.1016/j.csr.2017.05.006

McLean, D., Taylor, M., Ospina, A. G., & Partridge, J. (2019). An assessment of fish and marine growth associated with an oil and gas platform jacket using an augmented remotely operated vehicle. Continental Shelf Research, 179, 66-84. https://doi.org/10.1016/j.csr.2019.04.006

Meekan, M., & Choat, J. (1997). Latitudinal variation in abundance of herbivorous fishes: a comparison of temperate and tropical reefs. Marine Biology, 128(3), 373-383. https://doi.org/10.1007/s002270050103

Mellin, C., Peterson, E. E., Puotinen, M., & Schaffelke, B. (2020). Representation and complementarity of the long-term coral monitoring on the Great Barrier Reef. Ecol Appl, 30(6), e02122. https://doi.org/10.1002/eap.2122

Mirzaeinia, A., & Hassanalian, M. (2019). Energy conservation of V-shaped flocking Canada geese through leader and tail switching. Paper presented at the AIAA Propulsion and Energy 2019 Forum.

Misund, O. A., Aglen, A., & Frønæs, E. (1995). Mapping the shape, size, and density of fish schools by echo integration and a high-resolution sonar. ICES Journal of Marine Science: Journal du Conseil, 52(1), 11-20. https://doi.org/10.1016/1054- 3139(95)80011-5

Miyanohana, Y., Ishii, K., & Furusawa, M. (1990). Measurements and analyses of dorsal- aspect target strength of six species of fish at four frequencies. Rapports et Proces- Verbaux des Réunions du Conseil International pour l’Exploration de la Mer, 189, 317- 324.

Monismith, S. G., Davis, K. A., Shellenbarger, G. G., Hench, J. L., Nidzieko, N. J., Santoro, A. E., . . . Martens, C. S. (2010). Flow effects on benthic grazing on phytoplankton by a Caribbean reef. Limnology and Oceanography, 55(5), 1881-1892.

Monk, J., Ierodiaconou, D., Harvey, E., Rattray, A., & Versace, V. L. (2012). Are we predicting the actual or apparent distribution of temperate marine fishes? PloS one, 7(4). https://doi.org/10.1371/journal.pone.0034558

Monk, J., Ierodiaconou, D., Versace, V. L., Bellgrove, A., Harvey, E., Rattray, A., . . . Quinn, G. P. (2010). Habitat suitability for marine fishes using presence-only modelling and multibeam sonar. Marine Ecology Progress Series, 420, 157-174. https://doi.org/10.3354/meps08858

Moore, S. K., & Suthers, I. M. (2006). Evaluation and correction of subresolved particles by the optical plankton counter in three Australian estuaries with pristine to highly modified catchments. Journal of Geophysical Research: Oceans, 111(C5). https://doi.org/10.1029/2005JC002920

Morais, R. A., & Bellwood, D. R. (2019). Pelagic Subsidies Underpin Fish Productivity on a Degraded Coral Reef. Curr Biol, 29(9), 1521-1527 e1526. https://doi.org/10.1016/j.cub.2019.03.044

207

Morais, R. A., Ferreira, C. E. L., & Floeter, S. R. (2017). Spatial patterns of fish standing biomass across Brazilian reefs. J Fish Biol, 91(6), 1642-1667. https://doi.org/10.1111/jfb.13482

Morgan, M. J., & Godin, J. G. J. (1985). Antipredator benefits of schooling behaviour in a cyprinodontid fish, the banded killifish (Fundulus diaphanus). Zeitschrift für Tierpsychologie, 70(3), 236-246. https://doi.org/10.1111/j.1439-0310.1985.tb00515.x

Morton, J. K., Platell, M. E., & Gladstone, W. (2008). Differences in feeding ecology among three co-occurring species of wrasse (Teleostei: Labridae) on rocky reefs of temperate Australia. Marine Biology, 154(3), 577-592. https://doi.org/10.1007/s00227-008-0951-x

Motro, R., Ayalon, I., & Genin, A. (2005). Near-bottom depletion of zooplankton over coral reefs: III: vertical gradient of predation pressure. Coral reefs, 24(1), 95-98. https://doi.org/10.1007/s00338-004-0451-5

Myers, E. M., Harvey, E. S., Saunders, B. J., & Travers, M. J. (2016). Fine‐scale patterns in the day, night and crepuscular composition of a temperate reef fish assemblage. Marine Ecology, 37(3), 668-678. https://doi.org/10.1111/maec.12336

National Marine Science Committee. (2015). National Marine Science Plan 2015-2025: Driving the Development of Australia's Blue Economy. (1925012026). Australian Government. https://www.marinescience.net.au/nationalmarinescienceplan/

Neira, F. J. (2011). Application of daily egg production to estimate biomass of jack mackerel, Trachurus declivis–a key fish species in the pelagic ecosystem of south-eastern Australia. Institute for Marine and Antarctic Studies (IMAS), University of Tasmania.

Neira, F. J., & Keane, J. P. (2008). Ichthyoplankton‐based spawning dynamics of blue mackerel (Scomber australasicus) in south‐eastern Australia: links to the East Australian Current. Fisheries oceanography, 17(4), 281-298. https://doi.org/10.1111/j.1365-2419.2008.00479.x

Neuheimer, A., Thresher, R., Lyle, J., & Semmens, J. (2011). Tolerance limit for fish growth exceeded by warming waters. Nature Climate Change, 1(2), 110. https://doi.org/10.1038/nclimate1084

Neuheimer, A. B., & Taggart, C. T. (2007). The growing degree-day and fish size-at-age: the overlooked metric. Canadian Journal of Fisheries and Aquatic Sciences, 64(2), 375- 385. https://doi.org/10.1139/f07-003

Niku, J., Brooks, W., Herliansyah, R., Hui, F. K., Taskinen, S., & Warton, D. I. (2020). gllvm: Generalized Linear Latent Variable Models. R package version 1.2.2. https://CRAN.R-project.org/package=gllvm

O'Brien, W. J. (1979). The predator-prey interaction of planktivorous fish and zooplankton: recent research with planktivorous fish and their zooplankton prey shows the evolutionary thrust and parry of the predator-prey relationship. American Scientist, 67(5), 572-581. https://www.jstor.org/stable/27849438

208

O'Donnell, C., Mullins, E., Lynch, D., Lyons, K., Connaughton, P., & Power, J. (2019). Celtic Sea Herring Acoustic Survey Cruise Report 2019, 09-29 October, 2019. https://oar.marine.ie/handle/10793/1494

Oh, T. G., Otake, S., & Lee, M. O. (2011). Estimating the effective wake region (current shadow) of artificial reefs. In Artificial Reefs in Fisheries Management (pp. 279-295): CRC Press Boca Raton, FL, USA.

Ohlhorst, S. L. (1982). Diel migration patterns of demersal reef zooplankton. Journal of Experimental Marine Biology and Ecology, 60(1), 1-15. https://doi.org/10.1016/0022- 0981(81)90176-3

Oksanen, J., Blanchet, F. G., Friendly, M., Kindt, R., Legendre, P., McGlinn, D., . . . Wagner, H. (2019). vegan: Community Ecology Package. R package version 2.5-6. https://CRAN.R-project.org/package=vegan

Oliver, E., & Holbrook, N. (2014). Extending our understanding of South Pacific gyre “spin‐ up”: Modeling the East Australian Current in a future climate. Journal of Geophysical Research: Oceans, 119(5), 2788-2805. https://doi.org/10.1002/2013JC009591

Ona, E., & Mitson, R. (1996). Acoustic sampling and signal processing near the seabed: the deadzone revisited. ICES Journal of Marine Science, 53(4), 677-690. https://doi.org/10.1006/jmsc.1996.0087

Pais, M. P., & Cabral, H. N. (2017). Fish behaviour effects on the accuracy and precision of underwater visual census surveys. A virtual ecologist approach using an individual- based model. ecological modelling, 346, 58-69. https://doi.org/10.1016/j.ecolmodel.2016.12.011

Paradis, E., & Schliep, K. (2019). ape: Analyses of Phylogenetics and Evolution. R package version 5.4. https://cran.r-project.org/web/packages/ape

Paramo, J., Bertrand, S., Villalobos, H., & Gerlotto, F. (2007). A three-dimensional approach to school typology using vertical scanning multibeam sonar. Fisheries Research, 84(2), 171-179. https://doi.org/10.1016/j.fishres.2006.10.023

Parmesan, C. (2006). Ecological and evolutionary responses to recent climate change. Annu. Rev. Ecol. Evol. Syst., 37, 637-669. https://doi.org/10.1146/annurev.ecolsys.37.091305.110100

Parrish, C. C., Deibel, D., & Thompson, R. J. (2009). Effect of sinking spring phytoplankton blooms on lipid content and composition in suprabenthic and benthic invertebrates in a cold ocean coastal environment. Marine Ecology Progress Series, 391, 33-51. https://doi.org/10.3354/meps08148

Parsons, D. F., Suthers, I. M., Cruz, D. O., & Smith, J. A. (2016). Effects of habitat on fish abundance and species composition on temperate rocky reefs. Marine Ecology Progress Series, 561, 155-171. https://doi.org/10.3354/meps11927

Pauly, D., Watson, R., & Alder, J. (2005). Global trends in world fisheries: impacts on marine ecosystems and food security. Philosophical Transactions of the Royal Society B: Biological Sciences, 360(1453), 5-12. https://doi.org/10.1098/rstb.2004.1574

209

Paxton, A. B., Taylor, J. C., Peterson, C., Fegley, S. R., & Rosman, J. H. (2019). Consistent spatial patterns in multiple trophic levels occur around artificial habitats. Marine Ecology Progress Series, 611, 189-202. https://doi.org/10.3354/meps12865

Pecl, G. T., Araujo, M. B., Bell, J. D., Blanchard, J., Bonebrake, T. C., Chen, I. C., . . . Williams, S. E. (2017). Biodiversity redistribution under climate change: Impacts on ecosystems and human well-being. Science, 355(6332), eaai9214. https://doi.org/10.1126/science.aai9214

Petrik, C. M., Stock, C. A., Andersen, K. H., van Denderen, P. D., & Watson, J. R. (2019). Bottom-up drivers of global patterns of demersal, forage, and pelagic fishes. Progress in oceanography, 176, 102124. https://doi.org/10.1016/j.pocean.2019.102124

Petrou, K., Kranz, S. A., Trimborn, S., Hassler, C. S., Ameijeiras, S. B., Sackett, O., . . . Davidson, A. T. (2016). Southern Ocean phytoplankton physiology in a changing climate. J Plant Physiol, 203, 135-150. https://doi.org/10.1016/j.jplph.2016.05.004

Philander, S., & Yoon, J. (1982). Eastern boundary currents and coastal upwelling. Journal of Physical Oceanography, 12(8), 862-879. https://doi.org/10.1175/1520- 0485(1982)012%3C0862:EBCACU%3E2.0.CO;2

Phillips, N. D., Reid, N., Thys, T., Harrod, C., Payne, N. L., Morgan, C. A., . . . Houghton, J. D. (2017). Applying species distribution modelling to a data poor, pelagic fish complex: The ocean sunfishes. Journal of biogeography, 44(10), 2176-2187. https://doi.org/10.1111/jbi.13033

Pickering, H., & Whitmarsh, D. (1997). Artificial reefs and fisheries exploitation: a review of the ‘attraction versus production’debate, the influence of design and its significance for policy. Fisheries Research, 31(1-2), 39-59. https://doi.org/10.1016/S0165- 7836(97)00019-2

Pikitch, E., Boersma, P. D., Boyd, I., Conover, D., Cury, P., Essington, T., . . . Pauly, D. (2012). Little fish, big impact: managing a crucial link in ocean food webs. Lenfest Ocean Program, Washington, DC, 108. http://hdl.handle.net/102.100.100/101102?index=1

Pikitch, E. K., Boersma, P. D., Boyd, I. L., Conover, D. O., Cury, P., Essington, T. E., . . . Pauly, D. (2018). The strong connection between forage fish and their predators: A response to Hilborn et al.(2017). https://doi.org/10.1016/j.fishres.2017.07.022

Pikitch, E. K., Rountos, K. J., Essington, T. E., Santora, C., Pauly, D., Watson, R., . . . Conover, D. O. (2014). The global contribution of forage fish to marine fisheries and ecosystems. Fish and Fisheries, 15(1), 43-64. https://doi.org/10.1111/faf.12004

Pinnegar, J. K., & Polunin, N. V. (2006). Planktivorous damselfish support significant nitrogen and phosphorus fluxes to Mediterranean reefs. Marine Biology, 148(5), 1089-1099. https://doi.org/10.1007/s00227-005-0141-z

Pinnegar, J. K., Polunin, N. V., Videler, J. J., & de Wiljes, J. J. (2007). Daily carbon, nitrogen and phosphorus budgets for the Mediterranean planktivorous damselfish Chromis chromis. Journal of Experimental Marine Biology and Ecology, 352(2), 378-391. https://doi.org/10.1016/j.jembe.2007.08.016

210

Pinsky, M. L., Worm, B., Fogarty, M. J., Sarmiento, J. L., & Levin, S. A. (2013). Marine taxa track local climate velocities. Science, 341(6151), 1239-1242. https://doi.org/10.1126/science.1239352

Pitcher, G. C., Probyn, T. A., du Randt, A., Lucas, A. J., Bernard, S., Evers‐King, H., . . . Hutchings, L. (2014). Dynamics of oxygen depletion in the nearshore of a coastal embayment of the southern Benguela upwelling system. Journal of Geophysical Research: Oceans, 119(4), 2183-2200. https://doi.org/10.1002/2013JC009443

Pitcher, T. J. (1986). Functions of shoaling behaviour in teleosts. In The behaviour of teleost fishes (pp. 294-337): Springer.

Pittman, S., Christensen, J., Caldow, C., Menza, C., & Monaco, M. (2007). Predictive mapping of fish species richness across shallow-water seascapes in the Caribbean. ecological modelling, 204(1-2), 9-21. https://doi.org/10.1016/j.ecolmodel.2006.12.017

Poloczanska, E. S., Brown, C. J., Sydeman, W. J., Kiessling, W., Schoeman, D. S., Moore, P. J., . . . Burrows, M. T. (2013). Global imprint of climate change on marine life. Nature Climate Change, 3(10), 919. https://doi.org/10.1038/nclimate1958

Poore, A. G., & Steinberg, P. D. (1999). Preference–performance relationships and effects of host plant choice in an herbivorous marine amphipod. Ecological Monographs, 69(4), 443-464. https://doi.org/10.1890/0012-9615(1999)069[0443:PPRAEO]2.0.CO;2

Pribyl, A. L., Hyde, J. R., Robertson, L., & Vetter, R. (2016). Defining an ideal temperature range for the northern subpopulation of Pacific sardine, Sardinops sagax caeruleus. Environmental Biology of Fishes, 99(2-3), 275-291. https://doi.org/10.1007/s10641- 016-0473-1

Punzo, E., Malaspina, S., Domenichetti, F., Polidori, P., Scarcella, G., & Fabi, G. (2015). Fish detection around offshore artificial structures: preliminary results from hydroacoustics and fishing surveys. Journal of Applied Ichthyology, 31, 48-59. https://doi.org/10.1111/jai.12950

R Core Team. (2020). R: A language and environment for statistical computing (Version 4.0.2): R Foundation for Statistical Computing, Vienna, Austria. https://www.R- project.org/

Reeds, K. A., Smith, J. A., Suthers, I. M., & Johnston, E. L. (2018). An ecological halo surrounding a large offshore artificial reef: Sediments, infauna, and fish foraging. Mar Environ Res, 141, 30-38. https://doi.org/10.1016/j.marenvres.2018.07.011

Reese Robillard, M. M., Payne, L. M., Vega, R. R., & Stunz, G. W. (2015). Best practices for surgically implanting acoustic transmitters in spotted seatrout. Transactions of the American Fisheries Society, 144(1), 81-88. https://doi.org/10.1080/00028487.2014.965343

Reid, D., & Simmonds, E. (1993). Image analysis techniques for the study of fish school structure from acoustic survey data. Canadian Journal of Fisheries and Aquatic Sciences, 50(5), 886-893. https://doi.org/10.1139/f93-102

211

Revelle, W. (2020). psych: Procedures for Personality and Psychological Research. R package version 2.0.8. https://CRAN.R-project.org/package=psych

Richardson, A. J., John, E. H., Irigoien, X., Harris, R. P., & Hays, G. C. (2004). How well does the Continuous Plankton Recorder (CPR) sample zooplankton? A comparison with the Longhurst Hardy Plankton Recorder (LHPR) in the northeast Atlantic. Deep Sea Research Part I: Oceanographic Research Papers, 51(9), 1283-1294. https://doi.org/10.1016/j.dsr.2004.04.002

Ridgway, K. (2007). Long‐term trend and decadal variability of the southward penetration of the East Australian Current. Geophysical Research Letters, 34(13). https://doi.org/10.1029/2007GL030393

Rieucau, G., Boswell, K. M., Kimball, M. E., Diaz, G., & Allen, D. M. (2015). Tidal and diel variations in abundance and schooling behavior of estuarine fish within an intertidal salt marsh pool. Hydrobiologia, 753(1), 149-162. https://doi.org/10.1007/s10750- 015-2202-8

Rilov, G., & Benayahu, Y. (1998). Vertical artificial structures as an alternative habitat for coral reef fishes in disturbed environments. Marine environmental research, 45(4-5), 431-451. https://doi.org/10.1016/S0141-1136(98)00106-8

Rilov, G., & Benayahu, Y. (2000). Fish assemblage on natural versus vertical artificial reefs: the rehabilitation perspective. Marine Biology, 136(5), 931-942. https://doi.org/10.1007/s002279900250

Rilov, G., & Benayahu, Y. (2002). Rehabilitation of coral reef-fish communities: the importance of artificial-reef relief to recruitment rates. Bulletin of marine Science, 70(1), 185-197.

Robertson, D. R. (1982). Fish feces as fish food on a Pacific coral reef. Marine ecology progress series. Oldendorf, 7(3), 253-265. https://doi.org/10.3354/meps007253

Roche, M., Degrendele, K., Vrignaud, C., Loyer, S., Le Bas, T., Augustin, J.-M., & Lurton, X. (2018). Control of the repeatability of high frequency multibeam echosounder backscatter by using natural reference areas. Marine Geophysical Research, 39(1-2), 89- 104. https://doi.org/10.1007/s11001-018-9343-x

Rosenfeld, J. S. (2002). Functional redundancy in ecology and conservation. Oikos, 98(1), 156-162. https://doi.org/10.1034/j.1600-0706.2002.980116.x

Roughan, M., & Middleton, J. H. (2002). A comparison of observed upwelling mechanisms off the east coast of Australia. Continental Shelf Research, 22(17), 2551-2572. https://doi.org/10.1016/S0278-4343(02)00101-2

Roughan, M., & Middleton, J. H. (2004). On the East Australian Current: variability, encroachment, and upwelling. Journal of Geophysical Research: Oceans, 109(C7). https://doi.org/10.1029/2003JC001833

Rudstam, L. G., Hansson, S., Lindem, T., & Einhouse, D. W. (1999). Comparison of target strength distributions and fish densities obtained with split and single beam echo

212

sounders. Fisheries Research, 42(3), 207-214. https://doi.org/10.1016/S0165- 7836(99)00047-8

Rumyantseva, A., Henson, S., Martin, A., Thompson, A. F., Damerell, G. M., Kaiser, J., & Heywood, K. J. (2019). Phytoplankton spring bloom initiation: The impact of atmospheric forcing and light in the temperate North Atlantic Ocean. Progress in oceanography, 178, 102202. https://doi.org/10.1016/j.pocean.2019.102202

Ryan, T. E., Downie, R. A., Kloser, R. J., & Keith, G. (2015). Reducing bias due to noise and attenuation in open-ocean echo integration data. ICES Journal of Marine Science, 72(8), 2482-2493. https://doi.org/10.1093/icesjms/fsv121

Sala, A., Fabi, G., & Manoukian, S. (2007). Vertical diel dynamic of fish assemblage associated with an artificial reef (Northern Adriatic Sea). Scientia Marina, 71(2), 355- 364. https://doi.org/10.3989/scimar.2007.71n2355

Sala, E., Ballesteros, E., Dendrinos, P., Di Franco, A., Ferretti, F., Foley, D., . . . Zabala, M. (2012). The structure of Mediterranean rocky reef ecosystems across environmental and human gradients, and conservation implications. PloS one, 7(2), e32742. https://doi.org/10.1371/journal.pone.0032742

Sandblom, E., Clark, T. D., Grans, A., Ekstrom, A., Brijs, J., Sundstrom, L. F., . . . Jutfelt, F. (2016). Physiological constraints to climate warming in fish follow principles of plastic floors and concrete ceilings. Nat Commun, 7(1), 11447. https://doi.org/10.1038/ncomms11447

Sandin, S. A., & Pacala, S. W. (2005). Demographic theory of coral reef fish populations with stochastic recruitment: comparing sources of population regulation. Am Nat, 165(1), 107-119. https://doi.org/10.1086/426674

Santos, M. N., & Monteiro, C. C. (1997). The Olhao artificial reef system (south Portugal): fish assemblages and fishing yield. Fisheries Research, 30(1-2), 33-41. https://doi.org/10.1016/S0165-7836(96)00567-X

Sassa, C., Yamamoto, K., Tsukamoto, Y., Konishi, Y., & Tokimura, M. (2009). Distribution and migration of age‐0 jack mackerel (Trachurus japonicus) in the East China and Yellow Seas, based on seasonal bottom trawl surveys. Fisheries oceanography, 18(4), 255-267. https://doi.org/10.1111/j.1365-2419.2009.00510.x

Sato, M., Barth, J. A., Benoit-Bird, K. J., Pierce, S. D., Cowles, T. J., Brodeur, R. D., & Peterson, W. T. (2018). Coastal upwelling fronts as a boundary for planktivorous fish distributions. Marine Ecology Progress Series, 595, 171-186. https://doi.org/10.3354/meps12553

Saveliev, A., Mukharamova, S., Chizhikova, N., Budgey, R., & Zuur, A. (2007). Spatially continuous data analysis and modelling. In Analysing ecological data (pp. 341-372): Springer.

Schaafsma, F., David, C., Pakhomov, E., Hunt, B. P., Lange, B., Flores, H., & van Franeker, J. A. (2016). Size and stage composition of age class 0 Antarctic krill (Euphausia

213

superba) in the ice–water interface layer during winter/early spring. Polar Biology, 39(9), 1515-1526. https://doi.org/10.1007/s00300-015-1877-7

Scharf, F. S., Buckel, J. A., McGinn, P. A., & Juanes, F. (2003). Vulnerability of marine forage fishes to piscivory: effects of prey behavior on susceptibility to attack and capture. Journal of Experimental Marine Biology and Ecology, 294(1), 41-59. https://doi.org/10.1016/S0022-0981(03)00260-0

Scheffers, B. R., De Meester, L., Bridge, T. C., Hoffmann, A. A., Pandolfi, J. M., Corlett, R. T., . . . Watson, J. E. (2016). The broad footprint of climate change from genes to biomes to people. Science, 354(6313), aaf7671. https://doi.org/10.1126/science.aaf7671

Schilling, H. T., Everett, J. D., Schaeffer, A., Yates, P., Baird, M. E., & Suthers, I. M. (In review). Vertically resolved zooplankton biomass and size-structure across a continental shelf under the influence of a western boundary current.

Schobernd, Z. H., Bacheler, N. M., & Conn, P. B. (2014). Examining the utility of alternative video monitoring metrics for indexing reef fish abundance. Canadian Journal of Fisheries and Aquatic Sciences, 71(3), 464-471. https://doi.org/10.1139/cjfas-2013- 0086

Schramm, K. D., Harvey, E. S., Goetze, J. S., Travers, M. J., Warnock, B., & Saunders, B. J. (2020). A comparison of stereo-BRUV, diver operated and remote stereo-video transects for assessing reef fish assemblages. Journal of Experimental Marine Biology and Ecology, 524, 151273. https://doi.org/10.1016/j.jembe.2019.151273

Schroeder, D. M., & Love, M. S. (2004). Ecological and political issues surrounding decommissioning of offshore oil facilities in the Southern California Bight. Ocean & Coastal Management, 47(1-2), 21-48. https://doi.org/10.1016/j.ocecoaman.2004.03.002

Scott, M. E., Smith, J. A., Lowry, M. B., Taylor, M. D., & Suthers, I. M. (2015). The influence of an offshore artificial reef on the abundance of fish in the surrounding pelagic environment. Marine and Freshwater Research, 66(5), 429-437. https://doi.org/10.1071/MF14064

Serpetti, N., Baudron, A. R., Burrows, M. T., Payne, B. L., Helaouet, P., Fernandes, P. G., & Heymans, J. J. (2017). Impact of ocean warming on sustainable fisheries management informs the Ecosystem Approach to Fisheries. Sci Rep, 7(1), 13438. https://doi.org/10.1038/s41598-017-13220-7

Sexton, S. C., Ward, T. M., & Huveneers, C. (2017). Characterising the spawning patterns of Jack Mackerel (Trachurus declivis) off eastern Australia to optimise future survey design. Fisheries Research, 186, 223-236. https://doi.org/10.1016/j.fishres.2016.08.029

Shaughnessy, P. D., Briggs, S. V., & Constable, R. (2001). Observations on seals at Montague Island, New South Wales. Australian Mammalogy, 23(1), 1-7. https://doi.org/10.1071/AM01001

214

Shaw, E. (1962). The schooling of fishes. Scientific American, 206(6), 128-141. https://www.jstor.org/stable/24936575

Sheehan, E. V., Bridger, D., Nancollas, S. J., & Pittman, S. J. (2020). PelagiCam: a novel underwater imaging system with computer vision for semi-automated monitoring of mobile marine fauna at offshore structures. Environmental Monitoring and Assessment, 192(1), 11. https://doi.org/10.1007/s10661-019-7980-4

Sherman, R. L., Gilliam, D. S., & Spieler, R. E. (2002). Artificial reef design: void space, complexity, and attractants. ICES Journal of Marine Science, 59(suppl), S196-S200. https://doi.org/10.1006/jmsc.2001.1163

Silvertown, J. (2009). A new dawn for citizen science. Trends Ecol Evol, 24(9), 467-471. https://doi.org/10.1016/j.tree.2009.03.017

Sims, D. W., Witt, M. J., Richardson, A. J., Southall, E. J., & Metcalfe, J. D. (2006). Encounter success of free-ranging marine predator movements across a dynamic prey landscape. Proc Biol Sci, 273(1591), 1195-1201. https://doi.org/10.1098/rspb.2005.3444

Siqueira, A. C., Morais, R. A., Bellwood, D. R., & Cowman, P. F. (2020). Trophic innovations fuel reef fish diversification. Nat Commun, 11(1), 2669. https://doi.org/10.1038/s41467-020-16498-w

Skinner, C., Mill, A., Fox, M., Newman, S., Zhu, Y., Kuhl, A., & Polunin, N. (2021). Offshore pelagic subsidies dominate carbon inputs to coral reef predators. Science advances, 7(8), eabf3792. https://doi.org/10.1126/sciadv.abf3792

Smith, A. D., Brown, C. J., Bulman, C. M., Fulton, E. A., Johnson, P., Kaplan, I. C., . . . Tam, J. (2011). Impacts of fishing low-trophic level species on marine ecosystems. Science, 333(6046), 1147-1150. https://doi.org/10.1126/science.1209395

Smith, G., Zuur, A., & Ieno, E. (2007). Analysing ecological data: Springer.

Smith, J. A., Cornwell, W. K., Lowry, M. B., & Suthers, I. M. (2017). Modelling the distribution of fish around an artificial reef. Marine and Freshwater Research, 68(10), 1955-1964. https://doi.org/10.1071/MF16019

Smith, J. A., Lowry, M. B., Champion, C., & Suthers, I. M. (2016). A designed artificial reef is among the most productive marine fish habitats: new metrics to address ‘production versus attraction’. Marine Biology, 163(9), 188. https://doi.org/10.1007/s00227-016-2967-y

Smith, J. A., Lowry, M. B., & Suthers, I. M. (2015). Fish attraction to artificial reefs not always harmful: a simulation study. Ecology and Evolution, 5(20), 4590-4602. https://doi.org/10.1002/ece3.1730

Soldal, A. V., Svellingen, I., Jørgensen, T., & Løkkeborg, S. (2002). Rigs-to-reefs in the North Sea: hydroacoustic quantification of fish in the vicinity of a “semi-cold” platform. ICES Journal of Marine Science, 59(suppl), S281-S287. https://doi.org/10.1006/jmsc.2002.1279

215

Soler, G. A., Edgar, G. J., Thomson, R. J., Kininmonth, S., Campbell, S. J., Dawson, T. P., . . . Willis, T. J. (2015). Reef fishes at all trophic levels respond positively to effective marine protected areas. PloS one, 10(10), e0140270. https://doi.org/10.1371/journal.pone.0140270

Sourisseau, M., & Carlotti, F. (2006). Spatial distribution of zooplankton size spectra on the French continental shelf of the Bay of Biscay during spring 2000 and 2001. Journal of Geophysical Research: Oceans, 111(C5). https://doi.org/10.1029/2005JC003063

Speers-Roesch, B., Norin, T., & Driedzic, W. R. (2018). The benefit of being still: energy savings during winter dormancy in fish come from inactivity and the cold, not from metabolic rate depression. Proceedings of the Royal Society B: Biological Sciences, 285(1886), 20181593. https://doi.org/10.1098/rspb.2018.1593

Stallings, C. D. (2009). Fishery-independent data reveal negative effect of human population density on Caribbean predatory fish communities. PloS one, 4(5), e5333. https://doi.org/10.1371/journal.pone.0005333

Stanton, T. K., Wiebe, P. H., Chu, D., Benfield, M. C., Scanlon, L., Martin, L., & Eastwood, R. L. (1994). On acoustic estimates of zooplankton biomass. ICES Journal of Marine Science, 51(4), 505-512. https://doi.org/10.1006/jmsc.1994.1051

Stephens, D. (2008). Optimal foraging theory. In Encyclopedia of ecology (pp. 2561-2566).

Stevenson, C., Katz, L. S., Micheli, F., Block, B., Heiman, K. W., Perle, C., . . . Witting, J. (2007). High apex predator biomass on remote Pacific islands. Coral reefs, 26(1), 47- 51. https://doi.org/10.1007/s00338-006-0158-x

Stewart, J., & Ferrell, D. J. (2001). Age, growth, and commercial landings of yellowtail scad (Trachurus novaezelandiae) and blue mackerel (Scomber australasicus) off the coast of New South Wales, Australia. New Zealand Journal of Marine and Freshwater Research, 35(3), 541-551. https://doi.org/10.1080/00288330.2001.9517021

Strelcheck, A. J., Cowan, J. H., & Shah, A. (2005). Influence of reef location on artificial-reef fish assemblages in the northcentral Gulf of Mexico. Bulletin of marine Science, 77(3), 425-440.

Strömberg, K. P., Smyth, T. J., Allen, J. I., Pitois, S., & O'Brien, T. D. (2009). Estimation of global zooplankton biomass from satellite ocean colour. Journal of Marine Systems, 78(1), 18-27. https://doi.org/10.1016/j.jmarsys.2009.02.004

Sunday, J. M., Pecl, G. T., Frusher, S., Hobday, A. J., Hill, N., Holbrook, N. J., . . . Bates, A. E. (2015). Species traits and climate velocity explain geographic range shifts in an ocean-warming hotspot. Ecol Lett, 18(9), 944-953. https://doi.org/10.1111/ele.12474

Suthers, I. M., Taggart, C., Rissik, D., & Baird, M. (2006). Day and night ichthyoplankton assemblages and zooplankton biomass size spectrum in a deep ocean island wake. Marine Ecology Progress Series, 322, 225-238. https://doi.org/10.3354/meps322225

Suthers, I. M., Young, J. W., Baird, M. E., Roughan, M., Everett, J. D., Brassington, G. B., . . . Hassler, C. S. (2011). The strengthening East Australian Current, its eddies and

216

biological effects—an introduction and overview. Deep Sea Research Part II: Topical Studies in Oceanography, 58(5), 538-546. https://doi.org/10.1016/j.dsr2.2010.09.029

Szedlmayer, S. T., & Schroepfer, R. L. (2005). Long-term residence of red snapper on artificial reefs in the northeastern Gulf of Mexico. Transactions of the American Fisheries Society, 134(2), 315-325. https://doi.org/10.1577/T04-070.1

Tassetti, A., Malaspina, S., & Fabi, G. (2015). Using a multibeam echosounder to monitor an artificial reef. International Archives of the Photogrammetry, Remote Sensing & Spatial Information Sciences. https://doi.org/10.5194/isprsarchives-XL-5-W5-207-2015

Taylor, J., & Ebert, E. (2012). Mapping coral reef fish schools and aggregations with high- frequency multibeam and split-beam sonars. Paper presented at the Proceedings of Meetings on Acoustics ECUA2012.

Taylor, M. D., Becker, A., & Lowry, M. B. (2018). Investigating the functional role of an artificial reef within an estuarine seascape: a case study of yellowfin bream (Acanthopagrus australis). Estuaries and coasts, 41(6), 1782-1792. https://doi.org/10.1007/s12237-018-0395-6

Taylor, M. D., Chick, R. C., Lorenzen, K., Agnalt, A.-L., Leber, K. M., Blankenship, H. L., . . . Loneragan, N. R. (2017). Fisheries enhancement and restoration in a changing world. https://doi.org/10.1016/j.fishres.2016.10.004

Theberge, A. E. J., & Cherkis, N. Z. (2013). A note on fifty years of multi-beam. Hydro International, 22. https://www.hydro-international.com/content/article/a-note-on- fifty-years-of-multi-beam

Thierry, J.-M. (1988). Artificial reefs in Japan—a general outline. Aquacultural Engineering, 7(5), 321-348. https://doi.org/10.1016/0144-8609(88)90014-3

Tornabene, L., Valdez, S., Erdmann, M., & Pezold, F. (2015). Support for a ‘Center of Origin’in the Coral Triangle: Cryptic diversity, recent speciation, and local endemism in a diverse lineage of reef fishes (Gobiidae: Eviota). Molecular Phylogenetics and Evolution, 82, 200-210. https://doi.org/10.1016/j.ympev.2014.09.012

Tregenza, T. (1995). Building on the ideal free distribution. Advances in ecological research, 26, 253-307. https://doi.org/10.1016/S0065-2504(08)60067-7

Trenkel, V. M., Mazauric, V., & Berger, L. (2008). The new fisheries multibeam echosounder ME70: description and expected contribution to fisheries research. ICES Journal of Marine Science, 65(4), 645-655. https://doi.org/10.1093/icesjms/fsn051

Truong, L., Suthers, I. M., Cruz, D. O., & Smith, J. A. (2017). Plankton supports the majority of fish biomass on temperate rocky reefs. Marine Biology, 164(4), 73. https://doi.org/10.1007/s00227-017-3101-5

TSB Offshore. (2015). Decommissioning cost update for Pacific OCS region facilities. https://www.bsee.gov/technology-assessment-program-tap/decommissioning-cost- update-for-pacific-ocs-region-facilities

217

Turnbull, J. W., Esmaeili, Y. S., Clark, G. F., Figueira, W. F., Johnston, E. L., & Ferrari, R. (2018). Key drivers of effectiveness in small marine protected areas. Biodiversity and Conservation, 27(9), 2217-2242. https://doi.org/10.1007/s10531-018-1532-z

Udy, J., Wing, S., Connell-Milne, S. O., Durante, L., McMullin, R., Kolodzey, S., & Frew, R. (2019). Regional differences in supply of organic matter from kelp forests drive trophodynamics of temperate reef fish. Marine Ecology Progress Series, 621, 19-32. https://doi.org/10.3354/meps12974

Vabø, R., & Nøttestad, L. (1997). An individual based model of fish school reactions: predicting antipredator behaviour as observed in nature. Fisheries oceanography, 6(3), 155-171. https://doi.org/10.1046/j.1365-2419.1997.00037.x

Vabø, R., Olsen, K., & Huse, I. (2002). The effect of vessel avoidance of wintering Norwegian spring spawning herring. Fisheries Research, 58(1), 59-77. https://doi.org/10.1016/S0165-7836(01)00360-5

Valdivia, A., Cox, C. E., & Bruno, J. F. (2017). Predatory fish depletion and recovery potential on Caribbean reefs. Sci Adv, 3(3), e1601303. https://doi.org/10.1126/sciadv.1601303 van Denderen, P. D., Lindegren, M., MacKenzie, B. R., Watson, R. A., & Andersen, K. H. (2018). Global patterns in marine predatory fish. Nat Ecol Evol, 2(1), 65-70. https://doi.org/10.1038/s41559-017-0388-z

Vandromme, P., Nogueira, E., Huret, M., Lopez-Urrutia, A., González-Nuevo González, G., Sourisseau, M., & Petitgas, P. (2014). Springtime zooplankton size structure over the continental shelf of the Bay of Biscay. Ocean Science, 10(5), 821-835. http://doi.org/10.5194/os-10-821-2014

Verdoit, M., Pelletier, D., & Bellail, R. (2003). Are commercial logbook and scientific CPUE data useful for characterizing the spatial and seasonal distribution of exploited populations? The case of the Celtic Sea whiting. Aquatic Living Resources, 16(6), 467- 485. https://doi.org/10.1016/j.aquliv.2003.07.002

Vergés, A., Doropoulos, C., Malcolm, H. A., Skye, M., Garcia-Piza, M., Marzinelli, E. M., . . . Steinberg, P. D. (2016). Long-term empirical evidence of ocean warming leading to tropicalization of fish communities, increased herbivory, and loss of kelp. Proc Natl Acad Sci U S A, 113(48), 13791-13796. https://doi.org/10.1073/pnas.1610725113

Vergés, A., McCosker, E., Mayer‐Pinto, M., Coleman, M. A., Wernberg, T., Ainsworth, T., & Steinberg, P. D. (2019). Tropicalisation of temperate reefs: implications for ecosystem functions and management actions. Functional Ecology, 33(6), 1000-1013. https://doi.org/10.1111/1365-2435.13310

Vergés, A., Steinberg, P. D., Hay, M. E., Poore, A. G., Campbell, A. H., Ballesteros, E., . . . Wilson, S. K. (2014a). The tropicalization of temperate marine ecosystems: climate- mediated changes in herbivory and community phase shifts. Proc Biol Sci, 281(1789), 20140846. https://doi.org/10.1098/rspb.2014.0846

218

Vergés, A., Tomas, F., Cebrian, E., Ballesteros, E., Kizilkaya, Z., Dendrinos, P., . . . Sala, E. (2014b). Tropical rabbitfish and the deforestation of a warming temperate sea. Journal of Ecology, 102(6), 1518-1527. https://doi.org/10.1111/1365-2745.12324

Villéger, S., Brosse, S., Mouchet, M., Mouillot, D., & Vanni, M. J. (2017). Functional ecology of fish: current approaches and future challenges. Aquatic Sciences, 79(4), 783-801. https://doi.org/10.1007/s00027-017-0546-z

Ward-Paige, C., Mills Flemming, J., & Lotze, H. K. (2010). Overestimating fish counts by non-instantaneous visual censuses: consequences for population and community descriptions. PloS one, 5(7), e11722. https://doi.org/10.1371/journal.pone.0011722

Ward , T. M., & Grammer, G. L. (2018). Commonwealth Small Pelagic Fishery: Fishery Assessment Report 2017 (SARDI Research Report Series No. 982). https://www.afma.gov.au/sites/default/files/uploads/2018/06/2018- Commonwealth-SPF-Report-2017-FINAL-30_05_2018.pdf

Wassmann, P. (1997). Retention versus export food chains: processes controlling sinking loss from marine pelagic systems. Hydrobiologia, 363(1-3), 29-57. https://doi.org/10.1023/A:1003113403096

Watson, D. L., & Harvey, E. S. (2007). Behaviour of temperate and sub-tropical reef fishes towards a stationary SCUBA diver. Marine and Freshwater Behaviour and Physiology, 40(2), 85-103. https://doi.org/10.1080/10236240701393263

Watson, D. L., Harvey, E. S., Anderson, M. J., & Kendrick, G. A. (2005). A comparison of temperate reef fish assemblages recorded by three underwater stereo-video techniques. Marine Biology, 148(2), 415-425. https://doi.org/10.1007/s00227-005- 0090-6

Watson, R., Kitchingman, A., Gelchu, A., & Pauly, D. (2004). Mapping global fisheries: sharpening our focus. Fish and Fisheries, 5(2), 168-177. https://doi.org/10.1111/j.1467-2979.2004.00142.x

Weber, T. C., Peña, H., & Jech, J. M. (2009). Consecutive acoustic observations of an Atlantic herring school in the Northwest Atlantic. ICES Journal of Marine Science, 66(6), 1270-1277. https://doi.org/10.1093/icesjms/fsp090

Wernberg, T., Coleman, M. A., Babcock, R. C., Bell, S. Y., Bolton, J. J., Connell, S. D., . . . Shears, N. T. (2019). Biology and Ecology of the Globally Significant Kelp Ecklonia radiata. In Oceanography and Marine Biology: Taylor & Francis.

Wernberg, T., Smale, D. A., Tuya, F., Thomsen, M. S., Langlois, T. J., De Bettignies, T., . . . Rousseaux, C. S. (2013). An extreme climatic event alters marine ecosystem structure in a global biodiversity hotspot. Nature Climate Change, 3(1), 78-82. https://doi.org/10.1038/nclimate1627

Werner, E. E., Gilliam, J. F., Hall, D. J., & Mittelbach, G. G. (1983). An experimental test of the effects of predation risk on habitat use in fish. Ecology, 64(6), 1540-1548. https://doi.org/10.2307/1937508

219

Whiteway, T. (2009). Australian bathymetry and topography grid. Geoscience Australia, Canberra. https://data.gov.au/data/dataset/australian-bathymetry-and-topography- grid-june-2009

Williams, J. W., & Jackson, S. T. (2007). Novel climates, no‐analog communities, and ecological surprises. Frontiers in Ecology and the Environment, 5(9), 475-482. https://doi.org/10.1890/070037

Wilson, M. F., O’Connell, B., Brown, C., Guinan, J. C., & Grehan, A. J. (2007). Multiscale terrain analysis of multibeam bathymetry data for habitat mapping on the continental slope. Marine Geodesy, 30(1-2), 3-35. https://doi.org/10.1080/01490410701295962

Wogan, G. O., & Wang, I. J. (2018). The value of space‐for‐time substitution for studying fine‐scale microevolutionary processes. Ecography, 41(9), 1456-1468. https://doi.org/10.1111/ecog.03235

Wood, S., & Scheipl, F. (2020). Generalized Additive Mixed Models using 'mgcv' and 'lme4'. R package version 0.2-6. https://CRAN.R-project.org/package=gamm4

Wood, S. N. (2020). mgcv: Mixed GAM Computation Vehicle with Automatic Smoothness Estimation. R package version 1.8-31. https://CRAN.R-project.org/package=mgcv

Woodland, R. J., & Secor, D. H. (2013). Benthic‐pelagic coupling in a temperate inner continental shelf fish assemblage. Limnology and Oceanography, 58(3), 966-976. https://doi.org/10.4319/lo.2013.58.3.0966

Wu, L., Cai, W., Zhang, L., Nakamura, H., Timmermann, A., Joyce, T., . . . Visbeck, M. (2012). Enhanced warming over the global subtropical western boundary currents. Nature Climate Change, 2(3), 161-166. https://doi.org/10.1038/nclimate1353

Wyatt, A. S., Lowe, R. J., Humphries, S., & Waite, A. M. (2013). Particulate nutrient fluxes over a fringing coral reef: Source‐sink dynamics inferred from carbon to nitrogen ratios and stable isotopes. Limnology and Oceanography, 58(1), 409-427. https://doi.org/10.4319/lo.2013.58.1.0409

Yahel, R., Yahel, G., Berman, T., Jaffe, J. S., & Genin, A. (2005a). Diel pattern with abrupt crepuscular changes of zooplankton over a coral reef. Limnology and Oceanography, 50(3), 930-944. https://doi.org/10.4319/lo.2005.50.3.0930

Yahel, R., Yahel, G., & Genin, A. (2005b). Near-bottom depletion of zooplankton over coral reefs: I: diurnal dynamics and size distribution. Coral reefs, 24(1), 75-85. https://doi.org/10.1007/s00338-004-0449-z

Yamano, H., Sugihara, K., & Nomura, K. (2011). Rapid poleward range expansion of tropical reef corals in response to rising sea surface temperatures. Geophysical Research Letters, 38(4). https://doi.org/10.1029/2010GL046474

Yang, H., Lohmann, G., Wei, W., Dima, M., Ionita, M., & Liu, J. (2016). Intensification and poleward shift of subtropical western boundary currents in a warming climate.

220

Journal of Geophysical Research: Oceans, 121(7), 4928-4945. https://doi.org/10.1002/2015JC011513

Yasuda, T., & Kitamura, Y. (2003). Long-term variability of North Pacific subtropical mode water in response to spin-up of the subtropical gyre. Journal of oceanography, 59(3), 279-290. https://doi.org/10.1023/A:1025507725222

Zeileis, A. (2006). sandwich: Robust Covariance Matrix Estimators. R package version 2.5- 1. https://CRAN.R-project.org/web/packages/sandwich

Zuercher, R., & Galloway, A. W. (2019). Coastal marine ecosystem connectivity: pelagic ocean to kelp forest subsidies. Ecosphere, 10(2), e02602. https://doi.org/10.1002/ecs2.2602

Zwolinski, J., Fernandes, P. G., Marques, V., & Stratoudakis, Y. (2009). Estimating fish abundance from acoustic surveys: calculating variance due to acoustic backscatter and length distribution error. Canadian Journal of Fisheries and Aquatic Sciences, 66(12), 2081-2095. https://doi.org/10.1139/F09-138

221

8. Appendices

8.2. Supplementary material for Chapter 2

8.2.1. Supplementary methods

For Figure 2.1, the proportional representation of herbivores and zooplanktivores was calculated from four select studies which examined fish community trophic structure.

Figures were digitised using ImageJ (National Institute of Health, Washington D.C., USA).

Sala et al. (2012), Holmes et al. (2013), and Longo et al. (2019) all examined trophic structure along environmental gradients, even if that was not the specific focus of the paper, while Truong et al. (2017) examined trophic structure in the centre of the latitudinal range of our study. Although Sala et al. (2012) was not associated with a poleward flowing boundary current, it occurred along a tropicalisation gradient originating from the Suez

Canal, which currently allows tropical species from the Red Sea to enter and spread across the Mediterranean Sea.

For Sala et al. (2012), we were provided their original data, which was generated using underwater visual census (UVC). They calculated fish biomass by using taxa-specific length-weight relationships. For consistency, we only examined sites outside of marine protected areas. For each site, we calculated the mean biomass density of zooplanktivores and herbivores. We then calculated the mean of these values (± SE) for sites between 0 to

5°E (Balearic Sea, Zooplanktivores: 6.73 ± 1.00, Herbivores: 1.93 ± 0.73 g m-2, n = 5) and between 25 to 35°E (Aegean Sea and Eastern Mediterranean, Zooplanktivores: 4.97 ± 1.66,

Herbivores: 5.06 ± 1.37 g m-2, n = 5). We then summed the biomass density of herbivores and zooplanktivores and calculated the proportional contribution of each group to the total for display (Balearic Sea, Zooplanktivores: 0.80, Herbivores: 0.20; Aegean Sea and Eastern

Mediterranean, Zooplanktivores: 0.50, Herbivores: 0.50).

222

For Holmes et al. (2013), we digitised their UVC abundance data from their Figure 4. For sites at Ningaloo (22° 26.066′ S 113° 38.827′ E, n = 9) and Rottnest Island (32° 0.399′ S

115° 31.127′ E, n = 6), we summed the abundance of fish classified as ‘Large Croppers’,

‘Scraper/Excavators’ and ‘Small Croppers’ as herbivores and compared this to the abundance of ‘Zooplanktivores’ (Ningaloo, Zooplanktivores: 70.94, Herbivores: 138.40 ind.

250 m-2; Rottnest, Zooplanktivores: 53.68, Herbivores: 18.70 ind. 250 m-2). We then summed the abundances of herbivores and zooplanktivores and calculated the proportional contribution of each group to the total for display (Ningaloo, Zooplanktivores: 0.34,

Herbivores: 0.66; Rottnest, Zooplanktivores: 0.74, Herbivores: 0.26).

For Truong et al. (2017), we digitised their UVC mean relative biomass from their Figure 3.

Similar to Sala et al. (2012), they also calculated biomass using visual length estimates and length-weight relationships. For zooplanktivore relative biomass, we summed the mean relative biomass of ‘Coastal Zooplanktivores’ and ‘Reef Zooplanktivores’ and compared this to the biomass of ‘Reef Herbivores’ (Zooplanktivores: 0.42, Herbivores: 0.16). We then summed the relative biomass of herbivores and zooplanktivores and calculated the proportional contribution of each group to the total for display (Zooplanktivores: 0.72,

Herbivores: 0.28).

For Longo et al. (2019), we digitised their benthic remote camera recordings of feeding pressure from their Figure 2, which they calculated by multiplying the number of bites taken by the weight of the fish (in units of (bites × kg)(2 m2 × 10 min)-1). In this case, they classified zooplanktivores along with omnivores into the same category. We compared the relative feeding pressure from this ‘Omnivores’ group with the summed relative feeding pressure from ‘Scrapers’, ‘Excavators’, ‘Fine Browsers’, ‘Rough Browsers’ and ‘Territorial

Herbivores’. We then calculated mean relative feeding pressure (± SE) across survey locations for these two groups between 15 to 30°S (Zooplanktivores: 0.18 ± 0.10,

223

Herbivores: 0.81 ± 0.10, n = 5), 0 to 5°S (Zooplanktivores: 0.003 ± 0.003, Herbivores: 0.997

± 0.003, n = 3), 10 to 25°N (Zooplanktivores: 0.002 ± 0.002, Herbivores: 0.998 ± 0.002, n =

4) and 25 to 35°N (Zooplanktivores: 0.67 ± 0.33, Herbivores: 0.33 ± 0.33, n = 3).

For the results displayed from our own study, we calculated the mean proportion of total biomass across latitudes ranges from 29 to 31°S (Zooplanktivores: 0.31 ± 0.06, Herbivores:

0.35 ± 0.09, n = 166), 35 to 37°S (Zooplanktivores: 0.68 ± 0.04, Herbivores: 0.10 ± 0.007, n

= 1007) and 43 to 45°S (Zooplanktivores: 0.16 ± 0.03, Herbivores: 0.02 ± 0.004, n = 251).

We then summed these proportions and calculated the proportional contribution of each group to the total for display (29 to 31°S, Zooplanktivores: 0.47, Herbivores: 0.53; 35 to

37°S, Zooplanktivores: 0.87, Herbivores: 0.13; 43 to 45°S, Zooplanktivores: 0.87,

Herbivores: 0.13).

224

8.2.2. Supplementary tables

Supplementary Table 8.2.1 Pairwise Tukey’s Honest Significant Differences for results of two-way ANOVA by latitude bin and by season for site depth, wave exposure, water visibility and for the comparison of multivariate homogeneity of group variance based on benthic habitat composition. Values displayed indicate the difference between means for each combination of seasons or latitude bins. Data range and units are indicated in the title of each set of tables (*: p < 0.05, **: p < 0.01, ***: p < 0.001).

Depth (1.3 - 42 m)

Summer Autumn Winter Spring

*** *** *** Summer 0.00 -0.75 -1.03 -1.00

*** Autumn 0.75 0.00 -0.29 -0.26

*** Winter 1.03 0.29 0.00 0.03

*** Spring 1.00 0.26 -0.03 0.00

29 to 33 33 to 37 37 to 41 41 to 45

*** *** 29 to 33 0.00 3.12 -0.19 3.07

*** *** 33 to 37 -3.12 0.00 -3.31 -0.05

*** *** 37 to 41 0.19 3.31 0.00 3.26

*** *** 41 to 45 -3.07 0.05 -3.26 0.00

225

Exposure (0-1)

Summer Autumn Winter Spring

Summer 0.00 0.00 -0.02 -0.02

Autumn 0.00 0.00 -0.01 -0.02

Winter 0.02 0.01 0.00 0.00

Spring 0.02 0.02 0.00 0.00

29 to 33 33 to 37 37 to 41 41 to 45

* *** 29 to 33 0.00 -0.03 -0.06 0.00

* ** * 33 to 37 0.03 0.00 -0.04 0.03

*** ** *** 37 to 41 0.06 0.04 0.00 0.07

* *** 41 to 45 0.00 -0.03 -0.07 0.00

226

Visibility (2 – 30 m)

Summer Autumn Winter Spring

Summer 0.00 -0.28 -0.34 0.79

** Autumn 0.28 0.00 -0.05 1.07

Winter 0.34 0.05 0.00 1.12

** Spring -0.79 -1.07 -1.12 0.00

29 to 33 33 to 37 37 to 41 41 to 45

*** *** x 29 to 33 0.00 1.76 -1.72 -0.19

*** *** *** 33 to 37 -1.76 0.00 -3.48 -1.95

*** *** ** 37 to 41 1.72 3.48 0.00 1.53

*** ** 41 to 45 0.19 1.95 -1.53 0.00

227

Habitat quadrats (PERMANOVA)

Summer Autumn Winter Spring

Summer 0.00 -0.01 0.04 -0.04

Autumn 0.01 0.00 0.05 -0.03

* Winter -0.04 -0.05 0.00 -0.08

* Spring 0.04 0.03 0.08 0.00

29 to 33 33 to 37 37 to 41 41 to 45

** *** *** 29 to 33 0.00 -0.04 0.13 0.10

** *** *** 33 to 37 0.04 0.00 0.18 0.15

*** *** 37 to 41 -0.13 -0.18 0.00 -0.03

*** *** 41 to 45 -0.10 -0.15 0.03 0.00

228

Supplementary Table 8.2.2 Tabulated seasonal distribution of surveys across all RLS sites included in the analysis. Sites are designated by a unique site identifier provided by RLS (SiteCode) and arranged according to increasing latitude.

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

PIMP1 29.698080 0 4 0 0 4

SI14 29.912740 0 0 3 0 3

SI2 29.923140 0 5 9 1 15

SI23 29.925100 0 5 0 0 5

SI1 29.930440 0 10 9 0 19

SI13 29.933083 0 0 3 0 3

SI22 29.975150 0 2 2 0 4

SI21 30.007510 0 4 0 0 4

SI20 30.017450 0 0 2 0 2

SI6 30.017650 2 6 6 2 16

SI24 30.018760 0 5 6 1 12

SI25 30.151960 0 4 0 0 4

SI19 30.159320 4 4 1 2 11

SI15 30.201020 0 0 1 0 1

SI4 30.201670 0 3 4 0 7

SI17 30.201900 0 4 3 0 7

SI7 30.202280 0 5 2 0 7

SI18 30.202610 0 0 1 0 1

SI3 30.204320 1 4 6 0 11

SI16 30.204510 0 2 3 0 5

229

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

SI8 30.207010 1 2 6 0 9

SI11 30.207740 0 0 1 0 1

SI12 30.241563 0 4 1 0 5

SI9 30.242080 0 4 6 0 10

NSW44 31.572340 0 0 1 0 1

NSW43 31.588080 0 0 1 0 1

NSW42 31.594410 0 0 1 0 1

CG9 31.680700 0 3 0 0 3

CG2 31.681280 0 5 0 0 5

CG10 31.681520 0 4 0 0 4

CG1 31.682070 0 8 0 0 8

CG8 31.682540 0 3 0 0 3

CG3 31.683090 0 4 0 0 4

CG11 31.684060 0 3 0 0 3

CG5 31.693390 0 1 0 0 1

CG4 31.708040 0 2 0 0 2

CG6 31.716990 0 4 0 0 4

CG7 31.725820 0 2 0 0 2

NSW41 31.763820 0 0 1 0 1

NSW40 32.069535 0 0 1 0 1

NSW39 32.074552 0 0 1 0 1

NSW38 32.179170 2 0 0 0 2

230

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

NSW45 32.184900 0 0 0 1 1

NSW33 32.195130 0 0 2 0 2

NSW37 32.203443 0 0 1 0 1

NSW23 32.208839 0 1 0 0 1

NSW35 32.209033 0 1 1 0 2

NSW30 32.274950 0 0 1 2 3

NSW28 32.327160 0 0 1 0 1

NSW27 32.327369 0 0 0 1 1

NSW26 32.429250 0 1 0 0 1

NSW29 32.431837 1 0 0 0 1

NSW32 32.432077 0 0 2 0 2

NSW24 32.432241 0 0 2 0 2

NSW25 32.433744 0 0 1 0 1

NSW31 32.446500 0 0 2 0 2

PS1 32.599170 0 14 0 1 15

PS5 32.599190 0 6 1 0 7

PS51 32.612770 0 2 0 0 2

PS6 32.615270 0 7 1 0 8

PS2 32.618300 0 13 0 1 14

PS50 32.620241 0 2 0 0 2

PS52 32.620940 0 0 1 0 1

PS54 32.628800 0 2 0 0 2

231

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

PS53 32.629000 0 5 0 0 5

PS25 32.682700 1 5 0 0 6

PS34 32.684790 1 2 0 0 3

PS35 32.685510 3 2 1 3 9

PS24 32.687050 1 5 0 0 6

PS3 32.688840 1 18 4 7 30

PS31 32.689160 1 2 4 2 9

PS22 32.704750 0 2 0 4 6

PS41 32.708000 0 5 0 0 5

PS30 32.708828 1 3 2 1 7

PS21 32.709470 2 1 2 0 5

PS20 32.709760 21 29 13 24 87

PS18 32.711610 0 6 0 5 11

PS7 32.713325 10 17 11 4 42

PS17 32.714463 7 38 9 9 63

PS16 32.715150 11 9 1 2 23

PS13 32.718020 8 16 9 5 38

PS40 32.718401 1 2 0 0 3

PS12 32.718640 1 0 0 1 2

PS4 32.740570 6 16 6 9 37

PS11 32.741190 2 0 0 0 2

PS32 32.741460 1 0 1 0 2

232

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

PS42 32.741600 0 1 0 1 2

PS36 32.741700 0 1 0 0 1

PS38 32.744990 0 0 5 2 7

PS10 32.747820 0 1 2 1 4

PS9 32.748810 5 1 2 2 10

PS8 32.749280 0 0 0 2 2

PS44 32.750000 0 1 1 1 3

PS43 32.760000 0 1 0 1 2

PS29 32.787550 1 4 6 2 13

PS28 32.788520 2 2 1 0 5

PS27 32.789201 1 8 2 8 19

PS26 32.790100 1 5 1 0 7

PS15 32.791350 1 0 1 0 2

PS23 32.791630 1 0 3 0 4

NSW21 33.080170 2 0 0 0 2

NSW20 33.082830 2 0 0 0 2

NSW18 33.085237 0 0 0 1 1

NSW22 33.086000 1 2 0 0 3

NSW17 33.086450 0 0 2 0 2

NSW16 33.088501 0 0 1 1 2

NSW15 33.090810 0 0 1 0 1

NSW14 33.092480 1 0 0 0 1

233

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

NSW13 33.446500 0 5 3 4 12

SYD63 33.524360 0 2 1 0 3

SYD62 33.529870 0 2 1 0 3

SYD57 33.578880 2 2 0 0 4

SYD56 33.735239 0 4 1 0 5

SYD54 33.780660 0 0 1 2 3

SYD53 33.782418 0 1 1 0 2

SYD52 33.799269 5 14 0 2 21

SYD51 33.799608 3 10 0 1 14

SYD50 33.800465 4 11 1 3 19

SYD49 33.800970 1 10 0 3 14

SYD48 33.801000 0 9 0 0 9

SYD47 33.806460 0 8 3 0 11

SYD46 33.808495 2 0 0 0 2

SYD45 33.813220 8 3 0 0 11

SYD44 33.814840 0 3 0 0 3

SYD11 33.817290 2 0 0 0 2

SYD42 33.818020 9 17 0 0 26

SYD43 33.818480 7 0 0 0 7

SYD41 33.821400 2 12 0 0 14

SYD40 33.823400 7 11 0 0 18

SYD39 33.824100 2 7 0 0 9

234

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

SYD38 33.828270 2 13 5 0 20

SYD37 33.830360 3 7 0 0 10

SYD36 33.833040 3 14 0 0 17

SYD10 33.837850 2 0 0 0 2

SYD35 33.838800 0 6 5 0 11

SYD34 33.839307 3 6 5 2 16

SYD33 33.839495 0 1 0 0 1

SYD32 33.839753 0 3 0 2 5

SYD31 33.840111 2 4 6 2 14

SYD30 33.841655 2 3 0 2 7

SYD29 33.842820 0 8 5 0 13

SYD7 33.848190 2 0 0 0 2

SYD28 33.849696 3 3 0 0 6

SYD4 33.851330 2 0 0 0 2

SYD1 33.851860 2 0 0 0 2

SYD9 33.852053 2 0 0 0 2

SYD2 33.855290 2 0 0 0 2

SYD6 33.855950 2 0 0 0 2

SYD5 33.857720 3 0 0 0 3

SYD8 33.859660 2 0 2 0 4

SYD3 33.863280 2 0 0 0 2

SYD27 33.894140 0 3 0 0 3

235

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

SYD26 33.897088 0 1 0 0 1

SYD25 33.915622 1 2 0 0 3

SYD24 33.915887 0 4 0 1 5

SYD23 33.916500 3 6 0 0 9

SYD21 33.953801 1 0 0 0 1

SYD22 33.953801 0 1 0 0 1

SYD55 33.968110 0 5 0 0 5

SYD66 33.979529 0 1 0 0 1

SYD20 33.991420 6 5 2 0 13

SYD18 33.992761 4 4 0 0 8

SYD19 33.992931 8 4 0 1 13

SYD17 33.995640 0 0 3 0 3

SYD16 33.997450 0 0 4 0 4

SYD61 33.999220 0 4 0 0 4

SYD58 34.002544 0 5 0 0 5

SYD64 34.041100 0 6 0 0 6

SYD15 34.059032 0 0 3 0 3

SYD14 34.069152 0 3 0 0 3

SYD13 34.069335 0 2 0 0 2

SYD12 34.069990 0 4 2 0 6

NSW12 34.597683 3 3 0 0 6

SYD65 34.598657 0 3 0 0 3

236

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

JBMP-S14 35.016201 0 4 0 0 4

NJB3 35.026666 4 0 0 4 8

JBMP-S13 35.027084 0 4 0 0 4

NJB2 35.029450 2 0 0 6 8

JBMP26 35.045916 6 1 0 4 11

JBMP-S22 35.047798 0 4 0 0 4

JBMP24 35.048750 0 1 0 0 1

JBMP23 35.057600 1 1 0 0 2

NJB4 35.058600 5 0 0 3 8

JBMP-S6 35.059418 0 14 0 0 14

JBMP-S5 35.066101 0 14 0 0 14

JBMP22 35.067560 0 2 0 2 4

JBMP25 35.067560 1 0 0 0 1

JBMP-S8 35.070217 0 8 0 0 8

JBMP21 35.071390 8 4 0 0 12

JBMP20 35.071560 0 3 0 0 3

JBMP-S25 35.072265 0 6 0 0 6

JBMP-S12 35.074690 4 16 0 0 20

JBMP19 35.075940 4 0 0 0 4

JBMP-S4 35.077683 11 16 1 0 28

JBMP17 35.078340 0 0 0 2 2

JBMP-S27 35.078785 0 12 0 0 12

237

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

JBMP16 35.078870 5 0 0 0 5

JBMP15 35.079540 0 1 0 2 3

JBMP14 35.079610 4 0 0 0 4

JBMP-S3 35.079930 5 24 0 0 29

JBMP-S2 35.081420 0 35 0 0 35

JBMP13 35.083880 0 0 1 0 1

JBMP11 35.084920 0 1 1 0 2

JBMP10 35.085130 0 2 3 0 5

JBMP-S26 35.087730 8 28 0 0 36

JBMP47 35.089700 0 0 1 0 1

JBMP-S11 35.090584 3 6 0 0 9

JBMP46 35.099940 0 6 4 0 10

JBMP45 35.111520 0 0 0 1 1

JBMP44 35.112080 1 0 0 0 1

JBMP-S15 35.112717 4 20 0 0 24

JBMP43 35.112950 5 13 0 0 18

JBMP42 35.113790 0 0 1 0 1

JBMP41 35.117700 0 0 1 0 1

JBMP51 35.120370 0 4 0 0 4

JBMP-S1 35.120968 4 12 0 0 16

JBMP40 35.121080 0 0 0 1 1

JBMP-S16 35.121132 8 13 0 0 21

238

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

JBMP-S33 35.122930 0 8 0 0 8

JBMP38 35.123270 0 1 0 0 1

JBMP-S17 35.123932 5 13 0 0 18

JBMP37 35.125700 2 4 1 6 13

JBMP36 35.126200 0 5 0 0 5

JBMP50 35.130330 0 4 0 0 4

JBMP35 35.131100 7 0 0 0 7

JBMP-S18 35.134384 4 19 0 0 23

JBMP34 35.134880 4 8 0 0 12

JBMP49 35.135280 0 1 0 0 1

JBMP6 35.136760 0 4 0 0 4

SJB4 35.139333 4 0 0 4 8

JBMP48 35.158201 0 0 0 1 1

JBMP33 35.173750 0 0 1 4 5

SJB2 35.175550 3 0 0 5 8

SJB1 35.179616 3 0 0 5 8

SJB5 35.197633 8 0 0 0 8

NSW34 35.322110 3 4 0 0 7

NSW19 35.504743 3 6 1 0 10

BMP-S29 35.667960 12 0 0 0 12

BMP-S27 35.685460 12 0 0 0 12

BMP-S28 35.690190 12 0 0 0 12

239

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

BMP-S36 35.712230 12 2 6 0 20

BMP20 35.715740 0 1 0 0 1

BMP-S20 35.720300 6 0 0 4 10

BMP-S19 35.722350 8 0 4 4 16

BMP-S18 35.728110 8 0 0 4 12

BMP27 35.747940 0 4 2 0 6

BMP19 35.748580 0 1 0 0 1

BMP-S10 35.750120 17 1 2 0 20

BMP-S11 35.750730 12 1 2 0 15

BMP-S23 35.771330 12 4 0 4 20

BMP23 35.773920 0 1 0 0 1

BMP18 35.780620 2 0 0 0 2

BMP-S22 35.783890 8 9 0 4 21

BMP-S21 35.800190 8 0 0 6 14

BMP28 35.801830 0 3 2 0 5

BMP-S17 35.804530 8 0 0 6 14

BMP-S16 35.816720 9 1 0 6 16

BMP17 35.819290 0 1 0 0 1

BMP16 35.819640 0 0 0 2 2

BMP1 35.825895 3 7 0 0 10

BMP15 35.829020 0 1 0 0 1

BMP24 35.830020 2 5 0 0 7

240

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

BMP-S15 35.831320 8 1 0 4 13

BMP32 35.832890 2 2 2 0 6

BMP14 35.834550 1 0 0 0 1

BMP-S14 35.834570 10 2 2 4 18

BMP-S13 35.855070 8 6 0 4 18

BMP29 35.862000 0 3 2 0 5

BMP-S37 35.862140 8 0 0 4 12

BMP13 35.907200 0 0 0 2 2

BMP30 35.909160 0 2 2 0 4

BMP12 35.928200 0 0 0 2 2

BMP11 35.983540 0 0 0 2 2

BMP26 35.985560 0 3 2 0 5

BMP10 35.986260 0 0 0 2 2

BMP-S26 35.991570 8 0 1 4 13

BMP9 35.993010 0 0 1 0 1

BMP-S24 36.011520 8 0 0 4 12

BMP-S25 36.012900 8 2 2 4 16

BMP8 36.065810 0 0 1 0 1

BMP7 36.092260 0 6 1 0 7

BMP2 36.092760 3 1 0 1 5

BMP6 36.095940 0 1 0 0 1

BMP25 36.135590 0 4 2 0 6

241

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

BMP5 36.161120 0 1 0 0 1

BMP4 36.162134 1 0 0 0 1

BMP3 36.169320 0 1 0 0 1

BMP22 36.244820 0 6 2 0 8

BMP21 36.246830 0 1 0 0 1

BMP-S30 36.248500 0 4 2 0 6

BMP31 36.261330 0 2 4 0 6

BMP-S33 36.261730 8 1 0 0 9

BMP-S32 36.265150 8 1 0 0 9

BMP46 36.423430 0 2 0 0 2

NSW48 36.423430 0 1 0 0 1

NSW49 36.584720 0 0 0 2 2

NSW11 37.015970 1 0 0 0 1

NSW10 37.016460 6 0 0 0 6

NSW9 37.050220 3 0 0 0 3

NSW8 37.054840 2 0 0 0 2

NSW7 37.068850 2 0 0 0 2

NSW6 37.076080 3 0 0 0 3

NSW5 37.100360 4 0 0 0 4

NSW4 37.104680 2 0 0 0 2

NSW47 37.109680 2 0 0 0 2

NSW46 37.115180 3 2 0 0 5

242

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

NSW3 37.143060 4 0 0 0 4

NSW2 37.218870 4 0 0 0 4

NSW1 37.263340 4 0 0 0 4

CHMP-S2 37.507125 4 1 0 0 5

CHMP-S5 37.507840 4 0 0 0 4

CHMP-S1 37.508530 4 0 0 0 4

CHMP-S6 37.509870 4 0 0 0 4

CHMP-S8 37.519331 4 1 0 0 5

CHMP-S7 37.520670 4 0 0 0 4

VIC15 37.553016 1 0 0 0 1

CHMP-S4 37.553200 4 0 0 0 4

CHMP-S9 37.555640 9 3 0 0 12

VIC16 37.557630 1 0 0 0 1

VIC17 37.558530 1 0 0 0 1

VIC21 37.559166 2 0 0 0 2

CHMP-S3 37.564000 4 2 0 0 6

VIC10 37.573280 6 0 0 0 6

VIC55 37.577583 1 0 0 0 1

VIC19 37.590550 1 0 0 0 1

VIC4 37.618480 0 1 0 0 1

BR9 37.789760 1 1 0 3 5

BR8 37.790384 0 0 0 1 1

243

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

VIC20 37.792700 0 1 0 0 1

BR2 37.795350 3 0 0 1 4

VIC14 37.797183 1 0 0 0 1

BR12 37.798667 0 0 0 1 1

VIC54 37.803050 0 0 0 1 1

BR24 37.803400 0 0 0 1 1

VIC13 37.803613 0 0 0 3 3

VIC12 37.809080 0 0 0 1 1

BR1 37.811883 1 0 0 1 2

BR10 37.818109 6 3 1 4 14

BR11 37.818283 5 4 4 5 18

BR3 37.819050 11 9 7 12 39

BR13 37.819630 1 4 2 5 12

BR4 37.821430 0 2 1 3 6

BR5 37.822300 2 1 0 1 4

BR6 37.823016 1 1 0 0 2

BR7 37.824910 4 0 1 3 8

VIC3 37.870400 2 1 0 0 3

VIC8 37.888380 2 0 0 0 2

VIC9 37.888470 3 3 1 3 10

WP-S4 39.024618 3 0 0 0 3

WP-S27 39.024903 3 0 0 0 3

244

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

WP-S25 39.036100 2 0 0 0 2

WP-S6 39.039600 2 0 0 0 2

WP-S24 39.045659 2 0 0 0 2

WP-S23 39.052144 2 0 0 0 2

WP-S7 39.053778 2 0 0 0 2

WP-S22 39.055796 2 0 0 0 2

WP-S20 39.061851 2 1 0 0 3

WP-S21 39.064201 3 0 0 0 3

WP-S8 39.076409 2 0 0 0 2

WP-S9 39.083874 3 0 0 0 3

WP-S19 39.087712 0 1 0 0 1

WP-S11 39.102352 2 0 0 0 2

WP-S12 39.111800 2 0 0 0 2

WP-S18 39.113091 2 0 0 0 2

WP-S17 39.124107 2 0 0 0 2

WP-S15 39.130012 2 0 0 0 2

WP-S13 39.137843 2 0 0 0 2

BS-S8 39.215149 0 8 0 0 8

KG-S1 39.440670 0 8 0 0 8

KG-S13 39.444778 0 8 0 0 8

KG-S12 39.449000 0 8 0 0 8

KG-S4 39.464569 0 4 0 0 4

245

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

KG-S17 39.465302 0 8 0 0 8

KG-S16 39.473740 0 8 0 0 8

KG-S15 39.492950 0 8 0 0 8

KG-S7 39.494961 0 8 0 0 8

TAS78 40.000710 2 0 0 0 2

TAS77 40.003720 3 0 0 0 3

TAS76 40.011060 2 0 0 0 2

TAS75 40.022690 2 0 0 0 2

TAS74 40.023870 2 0 0 0 2

TAS73 40.024620 2 0 0 0 2

TAS72 40.040060 1 0 0 0 1

TAS71 40.057880 2 0 0 0 2

TAS70 40.082270 2 0 0 0 2

TAS69 40.091460 2 0 0 0 2

TAS67 41.164970 1 0 0 0 1

TAS66 41.869690 0 4 0 0 4

TAS88 41.870940 3 0 0 0 3

TAS65 41.874970 0 3 0 0 3

TAS83 41.875570 2 0 0 0 2

TAS84 41.875630 2 0 0 0 2

TAS85 42.180810 4 0 0 0 4

TAS86 42.234180 3 0 0 0 3

246

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

TAS87 42.255560 3 0 0 0 3

TAS100 42.411930 2 0 0 0 2

TAS64 42.587940 2 0 0 0 2

TAS63 42.592140 2 0 0 0 2

MIR1 42.592780 4 0 0 0 4

MIR-S12 42.602500 6 0 0 0 6

MIR-S15 42.628990 4 0 0 0 4

TAS101 42.646220 2 0 0 0 2

TAS62 42.663580 1 0 0 0 1

TAS61 42.720220 3 0 0 0 3

TAS60 42.730450 2 0 0 0 2

TAS59 42.741850 2 0 0 0 2

DE43 42.832017 2 0 0 0 2

DE44 42.837917 2 0 0 0 2

TAS58 42.842320 1 0 0 0 1

DE45 42.844333 2 0 0 0 2

DE42 42.851810 3 0 0 0 3

DE16 42.853130 1 0 0 0 1

DE46 42.861017 3 0 0 0 3

DE47 42.871910 2 0 0 0 2

DE17 42.875950 2 0 0 0 2

DE31 42.880140 1 0 0 0 1

247

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

DE9 42.881810 0 1 0 2 3

DE48 42.881990 2 1 0 0 3

DE49 42.882490 2 0 0 0 2

DE18 42.888320 2 0 0 0 2

DE40 42.889710 2 0 0 0 2

TAS79 42.898093 1 0 0 2 3

DE10 42.900250 0 0 0 2 2

DE4 42.906110 0 0 0 2 2

DE22 42.913390 1 0 0 0 1

DE11 42.918223 0 0 0 1 1

DE36 42.930420 2 0 0 0 2

DE1 42.940260 0 0 0 2 2

TAS57 42.940380 2 0 0 0 2

TAS55 42.941883 1 0 0 0 1

DE23 42.947520 1 0 0 0 1

DE6 42.949610 0 0 0 2 2

DE19 42.950922 1 0 0 0 1

DE37 42.953130 2 0 0 0 2

DE50 42.953240 0 0 2 0 2

TAS105 42.959647 2 0 0 0 2

DE21 42.963339 0 1 0 0 1

DE3 42.964530 0 0 0 2 2

248

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

DE2 42.966460 0 0 0 2 2

DE39 42.966630 2 0 0 0 2

DE20 42.969910 2 0 0 0 2

DE26 42.971127 4 0 0 0 4

TAS104 42.971519 2 0 0 0 2

TAS54 42.972590 1 0 0 0 1

DE41 42.977440 4 0 0 0 4

DE13 42.977580 3 0 0 0 3

DE30 42.984489 2 0 0 0 2

TAS56 42.985200 5 0 0 0 5

DE15 42.990070 2 0 0 0 2

DE28 42.992140 3 0 0 0 3

DE32 42.994600 3 0 0 0 3

DE27 42.995990 2 0 0 0 2

DE38 42.998370 4 0 0 0 4

DE52 42.999290 0 1 0 0 1

DE14 43.007220 3 0 0 0 3

TAS51 43.010360 2 0 0 0 2

DE7 43.013340 0 0 0 1 1

DE8 43.017200 0 0 0 1 1

DE29 43.020400 3 0 0 0 3

TAS50 43.025820 2 0 0 0 2

249

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

DE34 43.030700 2 0 0 0 2

TAS49 43.032010 2 0 0 0 2

DE25 43.032030 4 0 0 0 4

TAS81 43.042670 0 2 0 0 2

DE5 43.043710 0 0 0 2 2

DE24 43.047870 4 0 0 0 4

TAS53 43.049470 0 0 0 1 1

TAS80 43.052470 0 2 0 0 2

DE53 43.056880 0 0 0 1 1

TAS52 43.057530 0 0 0 1 1

DE12 43.058160 0 1 0 0 1

DE33 43.058432 2 0 0 0 2

DE51 43.060050 0 0 0 1 1

TAS48 43.060850 4 0 0 0 4

TAS102 43.063040 2 0 0 0 2

TAS109 43.065767 1 1 0 0 2

TAS46 43.069950 2 0 0 0 2

TAS45 43.083960 0 0 0 1 1

TAS44 43.084370 1 0 0 0 1

TAS82 43.086090 0 2 0 0 2

TAS43 43.089440 0 0 0 2 2

TAS42 43.092900 0 0 0 1 1

250

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

TAS41 43.099390 1 0 0 0 1

TAS40 43.100220 1 0 0 0 1

TAS39 43.102870 1 0 0 0 1

TAS38 43.104480 1 0 0 0 1

Tas108 43.106618 1 1 0 0 2

TAS47 43.111000 0 1 0 0 1

TAS36 43.111310 0 0 0 1 1

TAS35 43.119280 0 0 0 1 1

TAS34 43.121220 1 0 0 0 1

TAS33 43.121550 0 0 0 1 1

TAS32 43.122910 4 0 0 0 4

TAS31 43.124040 0 0 0 1 1

TAS120 43.129544 2 0 0 0 2

TAS30 43.134940 0 0 0 1 1

TAS29 43.137680 3 0 0 0 3

TAS28 43.138080 3 0 0 0 3

TAS27 43.139130 0 0 0 1 1

TAS26 43.142420 0 0 0 1 1

TAS25 43.145460 1 0 0 0 1

TAS107 43.146980 2 0 0 0 2

TAS37 43.157923 0 1 0 0 1

TAS106 43.188070 1 0 0 0 1

251

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

DE35 43.190985 6 0 0 0 6

TAS24 43.194490 4 0 0 0 4

TAS23 43.207830 1 0 0 0 1

TAS22 43.218000 1 0 0 0 1

TAS21 43.241090 1 0 0 0 1

TAS20 43.249000 1 0 0 0 1

TAS19 43.263480 0 0 0 1 1

TAS18 43.267830 1 0 0 0 1

TAS17 43.270100 0 0 0 1 1

TAS16 43.270450 1 0 0 0 1

NIN-S1 43.272920 0 4 0 0 4

TAS15 43.278160 0 0 0 1 1

TAS14 43.283100 1 0 0 0 1

NI-S1 43.284500 0 4 0 0 4

NIN-S3 43.285340 0 4 0 0 4

PD-S25 43.291130 4 0 0 0 4

NIN-S2 43.293830 0 4 0 0 4

PD-S26 43.297112 4 0 0 0 4

PD-S16 43.297279 4 0 0 0 4

PD-S11 43.299011 4 0 0 0 4

TAS13 43.299740 1 0 0 0 1

PD-S22 43.300629 4 0 0 0 4

252

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

PD-S23 43.302521 4 0 0 0 4

PD-S14 43.303341 8 0 0 0 8

PD-S15 43.304150 4 0 0 0 4

PD-S13 43.309330 4 0 0 0 4

PD-S12 43.309811 4 0 0 0 4

PD-S17 43.312431 8 0 0 0 8

TAS12 43.312840 1 0 0 0 1

PD-S20 43.318138 4 0 0 0 4

TAS103 43.320970 0 1 0 0 1

TAS91 43.321560 0 2 0 0 2

PD-S10 43.322659 4 2 0 0 6

TAS11 43.322860 1 0 0 0 1

PD-S24 43.329430 4 0 0 0 4

BH-S1 43.330558 4 0 0 0 4

TAS94 43.332580 0 2 0 0 2

PD-S18 43.333580 4 0 0 0 4

TAS10 43.333950 2 0 0 0 2

TAS9 43.336570 1 0 0 0 1

BCPD3 43.338690 1 0 0 0 1

TAS8 43.339320 1 0 0 0 1

TAS7 43.339340 1 0 0 0 1

BH-S2 43.339511 2 0 0 0 2

253

SiteCode Latitude (°S) Summer Autumn Winter Spring Total

BCPD2 43.339920 2 0 0 0 2

BCPD1 43.342060 1 0 0 0 1

TAS6 43.346640 1 0 0 0 1

PD-S7 43.361190 4 0 0 0 4

PD-S3 43.363380 4 0 0 0 4

TAS5 43.363440 1 0 0 0 1

PD-S19 43.369438 4 0 0 0 4

TAS4 43.402930 1 0 0 0 1

PD-S4 43.416512 4 0 0 0 4

PD-S21 43.421768 4 0 0 0 4

TAS3 43.438560 1 0 0 0 1

TAS2 43.454064 1 0 0 0 1

TAS1 43.460990 1 0 0 0 1

TAS111 43.521320 0 1 0 0 1

TAS112 43.523480 0 1 0 0 1

BI-S4 43.525250 0 2 0 0 2

TAS110 43.530360 0 1 0 0 1

TAS92 43.588200 0 2 0 0 2

TAS90 43.736310 3 0 0 0 3

Total 1110 1249 321 352 3032

254

Supplementary Table 8.2.3 Tabulated yearly distribution of surveys across all RLS sites included in the analysis. Sites are designated by a unique site identifier provided by RLS (SiteCode) and arranged according to increasing latitude.

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

PIMP1 29.698080 0 0 0 0 0 0 0 0 4 0 4

SI14 29.912740 0 3 0 0 0 0 0 0 0 0 3

SI2 29.923140 5 3 0 2 0 1 0 0 2 2 15

SI23 29.925100 0 1 0 0 0 0 0 0 2 2 5

SI1 29.930440 5 4 0 1 0 1 0 0 4 4 19

SI13 29.933083 0 3 0 0 0 0 0 0 0 0 3

SI22 29.975150 0 0 2 2 0 0 0 0 0 0 4

SI21 30.007510 0 0 3 1 0 0 0 0 0 0 4

SI20 30.017450 2 0 0 0 0 0 0 0 0 0 2

SI6 30.017650 7 0 2 2 0 1 0 0 2 2 16

SI24 30.018760 5 0 0 0 2 1 0 0 2 2 12

SI25 30.151960 0 0 0 0 0 0 0 0 2 2 4

SI19 30.159320 0 4 0 3 0 0 0 0 2 2 11

SI15 30.201020 0 1 0 0 0 0 0 0 0 0 1

SI4 30.201670 2 2 0 0 1 0 0 0 0 2 7

SI17 30.201900 0 3 0 0 0 0 0 0 2 2 7

SI7 30.202280 1 1 0 0 0 1 0 0 2 2 7

SI18 30.202610 0 1 0 0 0 0 0 0 0 0 1

SI3 30.204320 4 2 1 0 0 0 0 0 2 2 11

255

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

SI16 30.204510 0 3 0 0 0 1 0 0 1 0 5

SI8 30.207010 2 4 1 0 0 0 0 0 2 0 9

SI11 30.207740 0 1 0 0 0 0 0 0 0 0 1

SI12 30.241563 0 1 0 0 0 0 0 0 2 2 5

SI9 30.242080 0 4 0 2 0 0 0 0 2 2 10

NSW44 31.572340 0 0 1 0 0 0 0 0 0 0 1

NSW43 31.588080 0 0 1 0 0 0 0 0 0 0 1

NSW42 31.594410 0 0 1 0 0 0 0 0 0 0 1

CG9 31.680700 0 2 0 0 0 0 0 0 1 0 3

CG2 31.681280 0 3 0 0 0 0 0 0 2 0 5

CG10 31.681520 0 3 0 0 0 0 0 0 1 0 4

CG1 31.682070 0 6 0 0 0 0 0 0 2 0 8

CG8 31.682540 0 0 0 0 0 0 0 0 3 0 3

CG3 31.683090 0 2 0 0 0 0 0 0 2 0 4

CG11 31.684060 0 2 0 0 0 0 0 0 1 0 3

CG5 31.693390 0 1 0 0 0 0 0 0 0 0 1

CG4 31.708040 0 2 0 0 0 0 0 0 0 0 2

CG6 31.716990 0 2 0 0 0 0 0 0 2 0 4

CG7 31.725820 0 1 0 0 0 0 0 0 1 0 2

NSW41 31.763820 0 0 1 0 0 0 0 0 0 0 1

NSW40 32.069535 0 1 0 0 0 0 0 0 0 0 1

256

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

NSW39 32.074552 0 1 0 0 0 0 0 0 0 0 1

NSW38 32.179170 0 0 2 0 0 0 0 0 0 0 2

NSW45 32.184900 0 1 0 0 0 0 0 0 0 0 1

NSW33 32.195130 0 2 0 0 0 0 0 0 0 0 2

NSW37 32.203443 0 1 0 0 0 0 0 0 0 0 1

NSW23 32.208839 0 0 1 0 0 0 0 0 0 0 1

NSW35 32.209033 0 1 0 1 0 0 0 0 0 0 2

NSW30 32.274950 0 2 1 0 0 0 0 0 0 0 3

NSW28 32.327160 0 1 0 0 0 0 0 0 0 0 1

NSW27 32.327369 0 1 0 0 0 0 0 0 0 0 1

NSW26 32.429250 0 0 1 0 0 0 0 0 0 0 1

NSW29 32.431837 0 0 1 0 0 0 0 0 0 0 1

NSW32 32.432077 0 2 0 0 0 0 0 0 0 0 2

NSW24 32.432241 0 2 0 0 0 0 0 0 0 0 2

NSW25 32.433744 0 1 0 0 0 0 0 0 0 0 1

NSW31 32.446500 0 1 1 0 0 0 0 0 0 0 2

PS1 32.599170 0 5 0 0 2 1 2 0 2 3 15

PS5 32.599190 0 2 0 0 2 1 0 0 2 0 7

PS51 32.612770 0 0 0 0 0 2 0 0 0 0 2

PS6 32.615270 0 2 0 0 2 1 0 0 3 0 8

PS2 32.618300 0 5 0 0 3 1 2 0 3 0 14

257

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

PS50 32.620241 0 0 0 0 0 2 0 0 0 0 2

PS52 32.620940 0 0 0 0 0 1 0 0 0 0 1

PS54 32.628800 0 0 0 0 0 0 2 0 0 0 2

PS53 32.629000 0 0 0 0 0 0 2 0 0 3 5

PS25 32.682700 0 1 5 0 0 0 0 0 0 0 6

PS34 32.684790 0 0 0 1 0 2 0 0 0 0 3

PS35 32.685510 0 0 0 2 1 4 2 0 0 0 9

PS24 32.687050 0 0 1 2 2 0 0 1 0 0 6

PS3 32.688840 0 6 2 2 2 8 1 2 3 4 30

PS31 32.689160 0 0 1 1 2 2 1 0 2 0 9

PS22 32.704750 0 2 0 3 1 0 0 0 0 0 6

PS41 32.708000 0 0 0 0 2 2 0 1 0 0 5

PS30 32.708828 0 0 1 1 1 1 0 1 0 2 7

PS21 32.709470 0 0 1 3 1 0 0 0 0 0 5

PS20 32.709760 2 10 15 18 11 6 2 6 9 8 87

PS18 32.711610 0 0 2 2 1 5 0 0 1 0 11

PS7 32.713325 0 7 3 4 5 5 1 4 8 5 42

PS17 32.714463 4 13 4 5 5 11 1 3 9 8 63

PS16 32.715150 0 1 4 6 2 4 0 0 4 2 23

PS13 32.718020 0 7 6 3 2 4 2 3 3 8 38

PS40 32.718401 0 0 0 0 1 2 0 0 0 0 3

258

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

PS12 32.718640 0 1 1 0 0 0 0 0 0 0 2

PS4 32.740570 0 4 1 7 4 5 1 4 5 6 37

PS11 32.741190 0 0 2 0 0 0 0 0 0 0 2

PS32 32.741460 0 0 0 2 0 0 0 0 0 0 2

PS42 32.741600 0 0 0 0 0 2 0 0 0 0 2

PS36 32.741700 0 0 0 1 0 0 0 0 0 0 1

PS38 32.744990 0 0 0 2 1 3 0 1 0 0 7

PS10 32.747820 1 0 1 0 2 0 0 0 0 0 4

PS9 32.748810 0 4 1 1 2 0 0 1 1 0 10

PS8 32.749280 0 1 0 1 0 0 0 0 0 0 2

PS44 32.750000 0 0 0 0 1 2 0 0 0 0 3

PS43 32.760000 0 0 0 0 1 1 0 0 0 0 2

PS29 32.787550 2 0 0 1 4 2 0 2 0 2 13

PS28 32.788520 0 0 2 1 1 1 0 0 0 0 5

PS27 32.789201 1 1 2 2 2 3 0 0 4 4 19

PS26 32.790100 0 1 2 1 2 1 0 0 0 0 7

PS15 32.791350 0 0 1 0 1 0 0 0 0 0 2

PS23 32.791630 0 2 1 0 0 1 0 0 0 0 4

NSW21 33.080170 0 2 0 0 0 0 0 0 0 0 2

NSW20 33.082830 0 0 1 1 0 0 0 0 0 0 2

NSW18 33.085237 0 1 0 0 0 0 0 0 0 0 1

259

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

NSW22 33.086000 0 1 2 0 0 0 0 0 0 0 3

NSW17 33.086450 0 2 0 0 0 0 0 0 0 0 2

NSW16 33.088501 0 2 0 0 0 0 0 0 0 0 2

NSW15 33.090810 0 1 0 0 0 0 0 0 0 0 1

NSW14 33.092480 0 1 0 0 0 0 0 0 0 0 1

NSW13 33.446500 1 2 2 0 5 1 0 0 1 0 12

SYD63 33.524360 0 0 0 0 0 0 0 1 0 2 3

SYD62 33.529870 0 0 0 0 0 0 0 1 0 2 3

SYD57 33.578880 0 0 0 0 0 0 2 0 0 2 4

SYD56 33.735239 0 0 0 0 0 0 2 1 0 2 5

SYD54 33.780660 0 3 0 0 0 0 0 0 0 0 3

SYD53 33.782418 0 1 1 0 0 0 0 0 0 0 2

SYD52 33.799269 1 1 3 2 2 2 1 5 2 2 21

SYD51 33.799608 0 0 2 2 2 2 1 0 3 2 14

SYD50 33.800465 2 0 2 4 0 2 2 3 2 2 19

SYD49 33.800970 0 5 0 1 2 0 3 1 1 1 14

SYD48 33.801000 0 3 0 0 0 0 2 1 2 1 9

SYD47 33.806460 0 3 0 3 0 1 1 0 1 2 11

SYD46 33.808495 0 2 0 0 0 0 0 0 0 0 2

SYD45 33.813220 0 5 3 0 0 0 0 3 0 0 11

SYD44 33.814840 0 3 0 0 0 0 0 0 0 0 3

260

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

SYD11 33.817290 0 0 2 0 0 0 0 0 0 0 2

SYD42 33.818020 0 8 4 4 4 1 2 1 0 2 26

SYD43 33.818480 0 7 0 0 0 0 0 0 0 0 7

SYD41 33.821400 0 3 0 0 0 1 3 2 5 0 14

SYD40 33.823400 0 5 0 4 4 1 2 2 0 0 18

SYD39 33.824100 0 3 0 0 0 0 1 2 3 0 9

SYD38 33.828270 0 5 0 4 2 1 2 2 2 2 20

SYD37 33.830360 0 0 3 0 2 1 2 0 2 0 10

SYD36 33.833040 0 0 3 3 4 1 2 2 2 0 17

SYD10 33.837850 0 0 2 0 0 0 0 0 0 0 2

SYD35 33.838800 0 5 0 2 2 0 0 0 1 1 11

SYD34 33.839307 0 5 0 1 1 2 2 2 2 1 16

SYD33 33.839495 0 1 0 0 0 0 0 0 0 0 1

SYD32 33.839753 0 3 0 1 1 0 0 0 0 0 5

SYD31 33.840111 0 5 0 1 2 0 0 2 4 0 14

SYD30 33.841655 0 5 2 0 0 0 0 0 0 0 7

SYD29 33.842820 0 5 0 2 0 1 0 3 0 2 13

SYD7 33.848190 0 0 2 0 0 0 0 0 0 0 2

SYD28 33.849696 0 3 0 1 0 0 0 2 0 0 6

SYD4 33.851330 0 0 2 0 0 0 0 0 0 0 2

SYD1 33.851860 0 0 2 0 0 0 0 0 0 0 2

261

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

SYD9 33.852053 0 0 2 0 0 0 0 0 0 0 2

SYD2 33.855290 0 0 2 0 0 0 0 0 0 0 2

SYD6 33.855950 0 0 2 0 0 0 0 0 0 0 2

SYD5 33.857720 0 0 3 0 0 0 0 0 0 0 3

SYD8 33.859660 0 2 2 0 0 0 0 0 0 0 4

SYD3 33.863280 0 0 2 0 0 0 0 0 0 0 2

SYD27 33.894140 0 0 1 0 2 0 0 0 0 0 3

SYD26 33.897088 0 0 1 0 0 0 0 0 0 0 1

SYD25 33.915622 0 0 2 0 1 0 0 0 0 0 3

SYD24 33.915887 0 0 1 0 1 0 1 0 0 2 5

SYD23 33.916500 0 0 2 0 0 0 1 3 1 2 9

SYD21 33.953801 0 0 1 0 0 0 0 0 0 0 1

SYD22 33.953801 0 0 1 0 0 0 0 0 0 0 1

SYD55 33.968110 0 0 0 2 2 0 0 0 1 0 5

SYD66 33.979529 0 0 0 0 0 0 0 0 1 0 1

SYD20 33.991420 0 5 0 1 0 0 0 2 3 2 13

SYD18 33.992761 0 1 0 0 0 0 0 2 3 2 8

SYD19 33.992931 1 4 0 0 0 1 2 2 3 0 13

SYD17 33.995640 0 3 0 0 0 0 0 0 0 0 3

SYD16 33.997450 0 4 0 0 0 0 0 0 0 0 4

SYD61 33.999220 0 0 0 0 0 0 0 2 0 2 4

262

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

SYD58 34.002544 0 0 0 0 0 0 0 1 2 2 5

SYD64 34.041100 0 0 0 0 0 0 0 4 0 2 6

SYD15 34.059032 0 3 0 0 0 0 0 0 0 0 3

SYD14 34.069152 0 0 1 0 0 0 0 1 0 1 3

SYD13 34.069335 0 0 1 0 0 0 0 1 0 0 2

SYD12 34.069990 0 2 0 0 2 0 0 0 0 2 6

NSW12 34.597683 0 0 1 1 0 0 0 3 0 1 6

SYD65 34.598657 0 0 0 0 0 0 0 3 0 0 3

JBMP- S14 35.016201 0 0 4 0 0 0 0 0 0 0 4

NJB3 35.026666 0 0 0 0 0 0 8 0 0 0 8

JBMP- S13 35.027084 0 0 4 0 0 0 0 0 0 0 4

NJB2 35.029450 0 0 0 0 0 0 8 0 0 0 8

JBMP26 35.045916 1 0 0 0 0 0 10 0 0 0 11

JBMP- S22 35.047798 0 4 0 0 0 0 0 0 0 0 4

JBMP24 35.048750 1 0 0 0 0 0 0 0 0 0 1

JBMP23 35.057600 1 1 0 0 0 0 0 0 0 0 2

NJB4 35.058600 0 0 0 0 0 0 8 0 0 0 8

JBMP- S6 35.059418 0 0 4 4 0 4 0 0 2 0 14

JBMP- S5 35.066101 0 0 4 4 0 4 0 0 2 0 14

263

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

JBMP22 35.067560 2 1 0 0 0 0 0 0 1 0 4

JBMP25 35.067560 1 0 0 0 0 0 0 0 0 0 1

JBMP- S8 35.070217 0 0 4 0 0 4 0 0 0 0 8

JBMP21 35.071390 8 0 0 4 0 0 0 0 0 0 12

JBMP20 35.071560 1 0 2 0 0 0 0 0 0 0 3

JBMP- S25 35.072265 0 0 4 0 0 0 0 0 2 0 6

JBMP- S12 35.074690 0 0 4 4 0 4 0 4 4 0 20

JBMP19 35.075940 4 0 0 0 0 0 0 0 0 0 4

JBMP- S4 35.077683 8 4 4 0 0 0 0 4 8 0 28

JBMP17 35.078340 0 2 0 0 0 0 0 0 0 0 2

JBMP- S27 35.078785 0 0 8 0 0 4 0 0 0 0 12

JBMP16 35.078870 5 0 0 0 0 0 0 0 0 0 5

JBMP15 35.079540 0 2 0 0 0 0 0 0 1 0 3

JBMP14 35.079610 4 0 0 0 0 0 0 0 0 0 4

JBMP- S3 35.079930 0 4 8 0 0 8 0 5 4 0 29

JBMP- S2 35.081420 0 4 9 8 0 8 0 0 6 0 35

JBMP13 35.083880 1 0 0 0 0 0 0 0 0 0 1

JBMP11 35.084920 1 0 1 0 0 0 0 0 0 0 2

JBMP10 35.085130 2 0 3 0 0 0 0 0 0 0 5

264

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

JBMP- S26 35.087730 0 4 9 8 0 0 0 8 7 0 36

JBMP47 35.089700 1 0 0 0 0 0 0 0 0 0 1

JBMP- S11 35.090584 0 0 4 0 0 0 0 3 2 0 9

JBMP46 35.099940 0 1 9 0 0 0 0 0 0 0 10

JBMP45 35.111520 1 0 0 0 0 0 0 0 0 0 1

JBMP44 35.112080 1 0 0 0 0 0 0 0 0 0 1

JBMP- S15 35.112717 0 4 4 4 0 4 0 4 4 0 24

JBMP43 35.112950 6 0 4 4 0 0 0 0 4 0 18

JBMP42 35.113790 0 1 0 0 0 0 0 0 0 0 1

JBMP41 35.117700 1 0 0 0 0 0 0 0 0 0 1

JBMP51 35.120370 0 0 0 4 0 0 0 0 0 0 4

JBMP- S1 35.120968 0 0 5 4 0 0 0 4 3 0 16

JBMP40 35.121080 1 0 0 0 0 0 0 0 0 0 1

JBMP- S16 35.121132 7 0 5 4 0 4 0 1 0 0 21

JBMP- S33 35.122930 0 0 4 4 0 0 0 0 0 0 8

JBMP38 35.123270 1 0 0 0 0 0 0 0 0 0 1

JBMP- S17 35.123932 0 2 4 4 0 4 0 4 0 0 18

JBMP37 35.125700 1 8 2 0 0 1 0 0 1 0 13

JBMP36 35.126200 1 4 0 0 0 0 0 0 0 0 5

265

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

JBMP50 35.130330 0 0 0 4 0 0 0 0 0 0 4

JBMP35 35.131100 7 0 0 0 0 0 0 0 0 0 7

JBMP- S18 35.134384 0 0 5 4 0 4 0 4 6 0 23

JBMP34 35.134880 4 0 4 4 0 0 0 0 0 0 12

JBMP49 35.135280 0 1 0 0 0 0 0 0 0 0 1

JBMP6 35.136760 0 4 0 0 0 0 0 0 0 0 4

SJB4 35.139333 0 0 0 0 0 0 8 0 0 0 8

JBMP48 35.158201 0 0 1 0 0 0 0 0 0 0 1

JBMP33 35.173750 0 5 0 0 0 0 0 0 0 0 5

SJB2 35.175550 0 0 0 0 0 0 8 0 0 0 8

SJB1 35.179616 0 0 0 0 0 0 8 0 0 0 8

SJB5 35.197633 0 0 0 0 0 0 8 0 0 0 8

NSW34 35.322110 0 1 0 1 2 1 2 0 0 0 7

NSW19 35.504743 1 1 0 2 2 2 2 0 0 0 10

BMP- S29 35.667960 0 0 8 0 4 0 0 0 0 0 12

BMP- S27 35.685460 0 0 8 0 4 0 0 0 0 0 12

BMP- S28 35.690190 0 0 8 0 4 0 0 0 0 0 12

BMP- S36 35.712230 0 4 8 2 5 1 0 0 0 0 20

BMP20 35.715740 1 0 0 0 0 0 0 0 0 0 1

266

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

BMP- S20 35.720300 0 2 0 8 0 0 0 0 0 0 10

BMP- S19 35.722350 0 4 4 8 0 0 0 0 0 0 16

BMP- S18 35.728110 0 4 4 4 0 0 0 0 0 0 12

BMP27 35.747940 0 0 0 2 1 2 0 1 0 0 6

BMP19 35.748580 1 0 0 0 0 0 0 0 0 0 1

BMP- S10 35.750120 6 2 4 4 4 0 0 0 0 0 20

BMP- S11 35.750730 1 2 8 0 4 0 0 0 0 0 15

BMP- S23 35.771330 4 8 4 4 0 0 0 0 0 0 20

BMP23 35.773920 1 0 0 0 0 0 0 0 0 0 1

BMP18 35.780620 1 1 0 0 0 0 0 0 0 0 2

BMP- S22 35.783890 0 8 4 6 1 1 0 1 0 0 21

BMP- S21 35.800190 2 4 4 4 0 0 0 0 0 0 14

BMP28 35.801830 0 0 0 2 1 1 0 1 0 0 5

BMP- S17 35.804530 2 0 8 4 0 0 0 0 0 0 14

BMP- S16 35.816720 4 0 8 4 0 0 0 0 0 0 16

BMP17 35.819290 1 0 0 0 0 0 0 0 0 0 1

BMP16 35.819640 2 0 0 0 0 0 0 0 0 0 2

267

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

BMP1 35.825895 2 1 2 2 0 1 0 2 0 0 10

BMP15 35.829020 1 0 0 0 0 0 0 0 0 0 1

BMP24 35.830020 0 0 0 2 1 2 1 1 0 0 7

BMP- S15 35.831320 1 0 8 0 4 0 0 0 0 0 13

BMP32 35.832890 0 0 0 2 1 1 1 1 0 0 6

BMP14 35.834550 1 0 0 0 0 0 0 0 0 0 1

BMP- S14 35.834570 0 0 8 6 1 1 1 1 0 0 18

BMP- S13 35.855070 1 0 8 2 5 1 0 1 0 0 18

BMP29 35.862000 0 0 0 2 1 1 0 1 0 0 5

BMP- S37 35.862140 0 0 8 0 4 0 0 0 0 0 12

BMP13 35.907200 2 0 0 0 0 0 0 0 0 0 2

BMP30 35.909160 0 0 0 2 1 1 0 0 0 0 4

BMP12 35.928200 2 0 0 0 0 0 0 0 0 0 2

BMP11 35.983540 2 0 0 0 0 0 0 0 0 0 2

BMP26 35.985560 0 0 0 2 2 1 0 0 0 0 5

BMP10 35.986260 2 0 0 0 0 0 0 0 0 0 2

BMP- S26 35.991570 1 0 8 4 0 0 0 0 0 0 13

BMP9 35.993010 1 0 0 0 0 0 0 0 0 0 1

BMP- S24 36.011520 0 0 8 4 0 0 0 0 0 0 12

268

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

BMP- S25 36.012900 0 0 8 6 1 1 0 0 0 0 16

BMP8 36.065810 0 1 0 0 0 0 0 0 0 0 1

BMP7 36.092260 1 0 0 2 1 1 0 2 0 0 7

BMP2 36.092760 3 2 0 0 0 0 0 0 0 0 5

BMP6 36.095940 1 0 0 0 0 0 0 0 0 0 1

BMP25 36.135590 0 0 0 2 2 1 0 1 0 0 6

BMP5 36.161120 1 0 0 0 0 0 0 0 0 0 1

BMP4 36.162134 1 0 0 0 0 0 0 0 0 0 1

BMP3 36.169320 1 0 0 0 0 0 0 0 0 0 1

BMP22 36.244820 1 0 0 2 2 2 0 1 0 0 8

BMP21 36.246830 1 0 0 0 0 0 0 0 0 0 1

BMP- S30 36.248500 0 0 0 2 2 1 0 1 0 0 6

BMP31 36.261330 0 0 0 2 2 1 0 1 0 0 6

BMP- S33 36.261730 1 0 8 0 0 0 0 0 0 0 9

BMP- S32 36.265150 1 0 8 0 0 0 0 0 0 0 9

BMP46 36.423430 0 0 0 0 1 0 0 1 0 0 2

NSW48 36.423430 0 0 0 0 0 1 0 0 0 0 1

NSW49 36.584720 0 0 0 0 0 0 2 0 0 0 2

NSW11 37.015970 1 0 0 0 0 0 0 0 0 0 1

NSW10 37.016460 6 0 0 0 0 0 0 0 0 0 6

269

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

NSW9 37.050220 3 0 0 0 0 0 0 0 0 0 3

NSW8 37.054840 2 0 0 0 0 0 0 0 0 0 2

NSW7 37.068850 2 0 0 0 0 0 0 0 0 0 2

NSW6 37.076080 3 0 0 0 0 0 0 0 0 0 3

NSW5 37.100360 4 0 0 0 0 0 0 0 0 0 4

NSW4 37.104680 2 0 0 0 0 0 0 0 0 0 2

NSW47 37.109680 0 0 0 2 0 0 0 0 0 0 2

NSW46 37.115180 0 0 0 4 0 0 0 1 0 0 5

NSW3 37.143060 4 0 0 0 0 0 0 0 0 0 4

NSW2 37.218870 4 0 0 0 0 0 0 0 0 0 4

NSW1 37.263340 4 0 0 0 0 0 0 0 0 0 4

CHMP- S2 37.507125 0 1 0 4 0 0 0 0 0 0 5

CHMP- S5 37.507840 0 0 0 4 0 0 0 0 0 0 4

CHMP- S1 37.508530 0 0 0 4 0 0 0 0 0 0 4

CHMP- S6 37.509870 0 0 0 4 0 0 0 0 0 0 4

CHMP- S8 37.519331 0 1 0 4 0 0 0 0 0 0 5

CHMP- S7 37.520670 0 0 0 4 0 0 0 0 0 0 4

VIC15 37.553016 0 0 0 1 0 0 0 0 0 0 1

CHMP- S4 37.553200 0 0 0 4 0 0 0 0 0 0 4

270

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

CHMP- S9 37.555640 1 0 0 7 1 0 1 2 0 0 12

VIC16 37.557630 0 0 0 1 0 0 0 0 0 0 1

VIC17 37.558530 0 0 0 1 0 0 0 0 0 0 1

VIC21 37.559166 0 0 0 0 0 0 1 0 1 0 2

CHMP- S3 37.564000 0 0 0 4 0 0 0 2 0 0 6

VIC10 37.573280 0 1 0 4 0 1 0 0 0 0 6

VIC55 37.577583 0 0 0 0 0 0 0 0 0 1 1

VIC19 37.590550 0 0 0 0 0 0 1 0 0 0 1

VIC4 37.618480 0 1 0 0 0 0 0 0 0 0 1

BR9 37.789760 0 3 0 0 0 0 0 1 1 0 5

BR8 37.790384 1 0 0 0 0 0 0 0 0 0 1

VIC20 37.792700 0 0 0 0 0 0 1 0 0 0 1

BR2 37.795350 0 1 3 0 0 0 0 0 0 0 4

VIC14 37.797183 0 0 0 1 0 0 0 0 0 0 1

BR12 37.798667 0 0 1 0 0 0 0 0 0 0 1

VIC54 37.803050 0 0 0 0 0 0 1 0 0 0 1

BR24 37.803400 0 0 0 0 0 0 0 1 0 0 1

VIC13 37.803613 0 0 2 0 0 0 1 0 0 0 3

VIC12 37.809080 0 1 0 0 0 0 0 0 0 0 1

BR1 37.811883 0 1 0 1 0 0 0 0 0 0 2

271

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

BR10 37.818109 0 1 1 2 3 2 3 2 0 0 14

BR11 37.818283 4 2 3 1 0 0 0 1 5 2 18

BR3 37.819050 9 8 5 4 4 3 2 1 2 1 39

BR13 37.819630 0 0 0 1 4 2 1 1 2 1 12

BR4 37.821430 1 2 0 0 1 0 0 1 1 0 6

BR5 37.822300 1 0 3 0 0 0 0 0 0 0 4

BR6 37.823016 0 0 1 0 0 0 1 0 0 0 2

BR7 37.824910 4 2 1 1 0 0 0 0 0 0 8

VIC3 37.870400 0 1 0 1 0 0 0 1 0 0 3

VIC8 37.888380 0 1 0 0 0 0 0 1 0 0 2

VIC9 37.888470 1 4 0 0 0 1 1 1 0 2 10

WP-S4 39.024618 0 0 0 0 0 0 0 0 0 3 3

WP-S27 39.024903 0 0 0 0 0 0 1 0 0 2 3

WP-S25 39.036100 0 0 0 0 0 0 0 0 0 2 2

WP-S6 39.039600 0 0 0 0 0 0 0 0 0 2 2

WP-S24 39.045659 0 0 0 0 0 0 0 0 0 2 2

WP-S23 39.052144 0 0 0 0 0 0 0 0 0 2 2

WP-S7 39.053778 0 0 0 0 0 0 0 0 0 2 2

WP-S22 39.055796 0 0 0 0 0 0 0 0 0 2 2

WP-S20 39.061851 0 0 0 0 0 0 0 1 0 2 3

WP-S21 39.064201 0 0 0 0 0 0 1 0 0 2 3

272

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

WP-S8 39.076409 0 0 0 0 0 0 0 0 0 2 2

WP-S9 39.083874 0 0 0 0 0 0 0 0 0 3 3

WP-S19 39.087712 0 0 0 0 0 0 0 1 0 0 1

WP-S11 39.102352 0 0 0 0 0 0 0 0 0 2 2

WP-S12 39.111800 0 0 0 0 0 0 0 0 0 2 2

WP-S18 39.113091 0 0 0 0 0 0 0 0 0 2 2

WP-S17 39.124107 0 0 0 0 0 0 0 0 0 2 2

WP-S15 39.130012 0 0 0 0 0 0 0 0 0 2 2

WP-S13 39.137843 0 0 0 0 0 0 0 0 0 2 2

BS-S8 39.215149 0 0 0 0 0 8 0 0 0 0 8

KG-S1 39.440670 0 0 0 0 0 8 0 0 0 0 8

KG-S13 39.444778 0 0 0 0 0 8 0 0 0 0 8

KG-S12 39.449000 0 0 0 0 0 8 0 0 0 0 8

KG-S4 39.464569 0 0 0 0 0 4 0 0 0 0 4

KG-S17 39.465302 0 0 0 0 0 8 0 0 0 0 8

KG-S16 39.473740 0 0 0 0 0 8 0 0 0 0 8

KG-S15 39.492950 0 0 0 0 0 8 0 0 0 0 8

KG-S7 39.494961 0 0 0 0 0 8 0 0 0 0 8

TAS78 40.000710 2 0 0 0 0 0 0 0 0 0 2

TAS77 40.003720 3 0 0 0 0 0 0 0 0 0 3

TAS76 40.011060 2 0 0 0 0 0 0 0 0 0 2

273

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

TAS75 40.022690 2 0 0 0 0 0 0 0 0 0 2

TAS74 40.023870 2 0 0 0 0 0 0 0 0 0 2

TAS73 40.024620 2 0 0 0 0 0 0 0 0 0 2

TAS72 40.040060 1 0 0 0 0 0 0 0 0 0 1

TAS71 40.057880 2 0 0 0 0 0 0 0 0 0 2

TAS70 40.082270 2 0 0 0 0 0 0 0 0 0 2

TAS69 40.091460 2 0 0 0 0 0 0 0 0 0 2

TAS67 41.164970 0 1 0 0 0 0 0 0 0 0 1

TAS66 41.869690 4 0 0 0 0 0 0 0 0 0 4

TAS88 41.870940 0 0 0 3 0 0 0 0 0 0 3

TAS65 41.874970 3 0 0 0 0 0 0 0 0 0 3

TAS83 41.875570 0 0 0 2 0 0 0 0 0 0 2

TAS84 41.875630 0 0 0 2 0 0 0 0 0 0 2

TAS85 42.180810 0 0 0 4 0 0 0 0 0 0 4

TAS86 42.234180 0 0 0 3 0 0 0 0 0 0 3

TAS87 42.255560 0 0 0 3 0 0 0 0 0 0 3

TAS100 42.411930 0 0 0 0 0 0 2 0 0 0 2

TAS64 42.587940 2 0 0 0 0 0 0 0 0 0 2

TAS63 42.592140 2 0 0 0 0 0 0 0 0 0 2

MIR1 42.592780 0 0 0 0 4 0 0 0 0 0 4

MIR- S12 42.602500 0 0 0 0 4 0 2 0 0 0 6

274

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

MIR- S15 42.628990 0 0 0 0 4 0 0 0 0 0 4

TAS101 42.646220 0 0 0 0 0 0 2 0 0 0 2

TAS62 42.663580 1 0 0 0 0 0 0 0 0 0 1

TAS61 42.720220 3 0 0 0 0 0 0 0 0 0 3

TAS60 42.730450 2 0 0 0 0 0 0 0 0 0 2

TAS59 42.741850 2 0 0 0 0 0 0 0 0 0 2

DE43 42.832017 0 0 2 0 0 0 0 0 0 0 2

DE44 42.837917 0 0 2 0 0 0 0 0 0 0 2

TAS58 42.842320 0 0 1 0 0 0 0 0 0 0 1

DE45 42.844333 0 0 2 0 0 0 0 0 0 0 2

DE42 42.851810 0 1 2 0 0 0 0 0 0 0 3

DE16 42.853130 0 1 0 0 0 0 0 0 0 0 1

DE46 42.861017 0 1 2 0 0 0 0 0 0 0 3

DE47 42.871910 0 0 2 0 0 0 0 0 0 0 2

DE17 42.875950 0 2 0 0 0 0 0 0 0 0 2

DE31 42.880140 0 1 0 0 0 0 0 0 0 0 1

DE9 42.881810 0 3 0 0 0 0 0 0 0 0 3

DE48 42.881990 0 1 2 0 0 0 0 0 0 0 3

DE49 42.882490 0 0 2 0 0 0 0 0 0 0 2

DE18 42.888320 0 2 0 0 0 0 0 0 0 0 2

DE40 42.889710 0 0 2 0 0 0 0 0 0 0 2

275

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

TAS79 42.898093 0 0 1 1 0 0 1 0 0 0 3

DE10 42.900250 0 2 0 0 0 0 0 0 0 0 2

DE4 42.906110 0 2 0 0 0 0 0 0 0 0 2

DE22 42.913390 0 1 0 0 0 0 0 0 0 0 1

DE11 42.918223 0 1 0 0 0 0 0 0 0 0 1

DE36 42.930420 0 0 2 0 0 0 0 0 0 0 2

DE1 42.940260 0 2 0 0 0 0 0 0 0 0 2

TAS57 42.940380 0 2 0 0 0 0 0 0 0 0 2

TAS55 42.941883 1 0 0 0 0 0 0 0 0 0 1

DE23 42.947520 0 1 0 0 0 0 0 0 0 0 1

DE6 42.949610 0 2 0 0 0 0 0 0 0 0 2

DE19 42.950922 1 0 0 0 0 0 0 0 0 0 1

DE37 42.953130 0 0 2 0 0 0 0 0 0 0 2

DE50 42.953240 2 0 0 0 0 0 0 0 0 0 2

TAS105 42.959647 0 0 0 0 0 0 0 1 0 1 2

DE21 42.963339 1 0 0 0 0 0 0 0 0 0 1

DE3 42.964530 0 2 0 0 0 0 0 0 0 0 2

DE2 42.966460 0 2 0 0 0 0 0 0 0 0 2

DE39 42.966630 0 0 2 0 0 0 0 0 0 0 2

DE20 42.969910 0 2 0 0 0 0 0 0 0 0 2

DE26 42.971127 0 0 4 0 0 0 0 0 0 0 4

276

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

TAS104 42.971519 0 0 0 0 0 0 0 1 0 1 2

TAS54 42.972590 1 0 0 0 0 0 0 0 0 0 1

DE41 42.977440 0 0 4 0 0 0 0 0 0 0 4

DE13 42.977580 0 3 0 0 0 0 0 0 0 0 3

DE30 42.984489 0 1 1 0 0 0 0 0 0 0 2

TAS56 42.985200 0 5 0 0 0 0 0 0 0 0 5

DE15 42.990070 0 2 0 0 0 0 0 0 0 0 2

DE28 42.992140 0 0 3 0 0 0 0 0 0 0 3

DE32 42.994600 0 3 0 0 0 0 0 0 0 0 3

DE27 42.995990 0 0 2 0 0 0 0 0 0 0 2

DE38 42.998370 0 0 4 0 0 0 0 0 0 0 4

DE52 42.999290 1 0 0 0 0 0 0 0 0 0 1

DE14 43.007220 0 3 0 0 0 0 0 0 0 0 3

TAS51 43.010360 0 2 0 0 0 0 0 0 0 0 2

DE7 43.013340 0 1 0 0 0 0 0 0 0 0 1

DE8 43.017200 0 1 0 0 0 0 0 0 0 0 1

DE29 43.020400 0 0 3 0 0 0 0 0 0 0 3

TAS50 43.025820 0 2 0 0 0 0 0 0 0 0 2

DE34 43.030700 0 0 2 0 0 0 0 0 0 0 2

TAS49 43.032010 0 2 0 0 0 0 0 0 0 0 2

DE25 43.032030 0 0 4 0 0 0 0 0 0 0 4

277

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

TAS81 43.042670 0 2 0 0 0 0 0 0 0 0 2

DE5 43.043710 0 2 0 0 0 0 0 0 0 0 2

DE24 43.047870 0 0 4 0 0 0 0 0 0 0 4

TAS53 43.049470 1 0 0 0 0 0 0 0 0 0 1

TAS80 43.052470 0 2 0 0 0 0 0 0 0 0 2

DE53 43.056880 1 0 0 0 0 0 0 0 0 0 1

TAS52 43.057530 1 0 0 0 0 0 0 0 0 0 1

DE12 43.058160 1 0 0 0 0 0 0 0 0 0 1

DE33 43.058432 0 0 2 0 0 0 0 0 0 0 2

DE51 43.060050 1 0 0 0 0 0 0 0 0 0 1

TAS48 43.060850 0 4 0 0 0 0 0 0 0 0 4

TAS102 43.063040 0 0 0 0 0 0 2 0 0 0 2

TAS109 43.065767 0 0 0 0 0 0 0 1 0 1 2

TAS46 43.069950 0 0 2 0 0 0 0 0 0 0 2

TAS45 43.083960 1 0 0 0 0 0 0 0 0 0 1

TAS44 43.084370 1 0 0 0 0 0 0 0 0 0 1

TAS82 43.086090 0 2 0 0 0 0 0 0 0 0 2

TAS43 43.089440 2 0 0 0 0 0 0 0 0 0 2

TAS42 43.092900 1 0 0 0 0 0 0 0 0 0 1

TAS41 43.099390 1 0 0 0 0 0 0 0 0 0 1

TAS40 43.100220 1 0 0 0 0 0 0 0 0 0 1

278

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

TAS39 43.102870 1 0 0 0 0 0 0 0 0 0 1

TAS38 43.104480 1 0 0 0 0 0 0 0 0 0 1

Tas108 43.106618 0 0 0 0 0 0 0 1 0 1 2

TAS47 43.111000 1 0 0 0 0 0 0 0 0 0 1

TAS36 43.111310 1 0 0 0 0 0 0 0 0 0 1

TAS35 43.119280 1 0 0 0 0 0 0 0 0 0 1

TAS34 43.121220 1 0 0 0 0 0 0 0 0 0 1

TAS33 43.121550 1 0 0 0 0 0 0 0 0 0 1

TAS32 43.122910 0 4 0 0 0 0 0 0 0 0 4

TAS31 43.124040 1 0 0 0 0 0 0 0 0 0 1

TAS120 43.129544 0 0 0 0 0 0 0 0 0 2 2

TAS30 43.134940 1 0 0 0 0 0 0 0 0 0 1

TAS29 43.137680 0 3 0 0 0 0 0 0 0 0 3

TAS28 43.138080 0 3 0 0 0 0 0 0 0 0 3

TAS27 43.139130 1 0 0 0 0 0 0 0 0 0 1

TAS26 43.142420 1 0 0 0 0 0 0 0 0 0 1

TAS25 43.145460 1 0 0 0 0 0 0 0 0 0 1

TAS107 43.146980 0 0 0 0 0 0 0 2 0 0 2

TAS37 43.157923 1 0 0 0 0 0 0 0 0 0 1

TAS106 43.188070 0 0 0 0 0 0 0 1 0 0 1

DE35 43.190985 0 2 4 0 0 0 0 0 0 0 6

279

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

TAS24 43.194490 0 4 0 0 0 0 0 0 0 0 4

TAS23 43.207830 1 0 0 0 0 0 0 0 0 0 1

TAS22 43.218000 1 0 0 0 0 0 0 0 0 0 1

TAS21 43.241090 1 0 0 0 0 0 0 0 0 0 1

TAS20 43.249000 1 0 0 0 0 0 0 0 0 0 1

TAS19 43.263480 1 0 0 0 0 0 0 0 0 0 1

TAS18 43.267830 1 0 0 0 0 0 0 0 0 0 1

TAS17 43.270100 1 0 0 0 0 0 0 0 0 0 1

TAS16 43.270450 1 0 0 0 0 0 0 0 0 0 1

NIN-S1 43.272920 0 0 0 0 0 4 0 0 0 0 4

TAS15 43.278160 1 0 0 0 0 0 0 0 0 0 1

TAS14 43.283100 1 0 0 0 0 0 0 0 0 0 1

NI-S1 43.284500 0 0 0 0 0 4 0 0 0 0 4

NIN-S3 43.285340 0 0 0 0 0 4 0 0 0 0 4

PD-S25 43.291130 0 0 0 0 0 4 0 0 0 0 4

NIN-S2 43.293830 0 0 0 0 0 4 0 0 0 0 4

PD-S26 43.297112 0 0 0 0 0 4 0 0 0 0 4

PD-S16 43.297279 0 0 0 0 0 4 0 0 0 0 4

PD-S11 43.299011 0 0 0 0 0 4 0 0 0 0 4

TAS13 43.299740 1 0 0 0 0 0 0 0 0 0 1

PD-S22 43.300629 0 0 0 0 0 4 0 0 0 0 4

280

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

PD-S23 43.302521 0 0 0 0 0 4 0 0 0 0 4

PD-S14 43.303341 0 0 0 0 0 8 0 0 0 0 8

PD-S15 43.304150 0 0 0 0 0 4 0 0 0 0 4

PD-S13 43.309330 0 0 0 0 0 4 0 0 0 0 4

PD-S12 43.309811 0 0 0 0 0 4 0 0 0 0 4

PD-S17 43.312431 0 0 0 0 0 8 0 0 0 0 8

TAS12 43.312840 1 0 0 0 0 0 0 0 0 0 1

PD-S20 43.318138 0 0 0 0 0 4 0 0 0 0 4

TAS103 43.320970 0 0 0 0 0 0 1 0 0 0 1

TAS91 43.321560 0 0 0 0 0 2 0 0 0 0 2

PD-S10 43.322659 0 0 0 0 0 4 2 0 0 0 6

TAS11 43.322860 1 0 0 0 0 0 0 0 0 0 1

PD-S24 43.329430 0 0 0 0 0 4 0 0 0 0 4

BH-S1 43.330558 0 0 0 0 0 4 0 0 0 0 4

TAS94 43.332580 0 0 0 0 0 2 0 0 0 0 2

PD-S18 43.333580 0 0 0 0 0 4 0 0 0 0 4

TAS10 43.333950 2 0 0 0 0 0 0 0 0 0 2

TAS9 43.336570 1 0 0 0 0 0 0 0 0 0 1

BCPD3 43.338690 0 0 0 0 1 0 0 0 0 0 1

TAS8 43.339320 1 0 0 0 0 0 0 0 0 0 1

TAS7 43.339340 1 0 0 0 0 0 0 0 0 0 1

281

Site Latitude Code (°S) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total

BH-S2 43.339511 0 0 0 0 0 2 0 0 0 0 2

BCPD2 43.339920 0 0 0 0 2 0 0 0 0 0 2

BCPD1 43.342060 0 0 0 0 1 0 0 0 0 0 1

TAS6 43.346640 1 0 0 0 0 0 0 0 0 0 1

PD-S7 43.361190 0 0 0 0 0 4 0 0 0 0 4

PD-S3 43.363380 0 0 0 0 0 4 0 0 0 0 4

TAS5 43.363440 1 0 0 0 0 0 0 0 0 0 1

PD-S19 43.369438 0 0 0 0 0 4 0 0 0 0 4

TAS4 43.402930 1 0 0 0 0 0 0 0 0 0 1

PD-S4 43.416512 0 0 0 0 0 4 0 0 0 0 4

PD-S21 43.421768 0 0 0 0 0 4 0 0 0 0 4

TAS3 43.438560 1 0 0 0 0 0 0 0 0 0 1

TAS2 43.454064 1 0 0 0 0 0 0 0 0 0 1

TAS1 43.460990 1 0 0 0 0 0 0 0 0 0 1

TAS111 43.521320 0 0 0 0 0 0 0 1 0 0 1

TAS112 43.523480 0 0 0 0 0 0 0 1 0 0 1

BI-S4 43.525250 0 0 0 0 0 2 0 0 0 0 2

TAS110 43.530360 0 0 0 0 0 0 0 1 0 0 1

TAS92 43.588200 0 0 0 0 0 2 0 0 0 0 2

TAS90 43.736310 0 0 0 0 0 3 0 0 0 0 3

Total 327 499 504 379 205 386 156 175 223 178 3032

282

Supplementary Table 8.2.4 Species list of the 163 reef fish taxa which were classified into trophic groups for our trophic structure analysis. The columns a and b represent the Bayesian length-weight parameters for calculating fish biomass (Froese and Pauly 2009).

Family Genus Species Common name Trophic group a b

Sydney Apogonidae Apogon limenus cardinalfish Zooplanktivores 0.01096 3.11000

Apogonidae Siphamia cephalotes Wood's siphonfish Zooplanktivores 0.01549 3.06000

Double-lined Caesionidae Pterocaesio digramma fusilier Zooplanktivores 0.01148 3.16000

Carangidae Trachurus novaezelandiae Yellowtail scad Zooplanktivores 0.01820 2.98000

Kyphosidae Atypichthys strigatus Australian mado Zooplanktivores 0.02291 2.99000

Yellowtail blue Lutjanidae Paracaesio xanthura snapper Zooplanktivores 0.02089 3.06000

Monodactylidae Schuettea scalaripinnis Eastern pomfred Zooplanktivores 0.02512 2.97000

Pempherididae Pempheris affinis Blacktip bullseye Zooplanktivores 0.01622 2.99000

Smallscale Pempherididae Pempheris compressa bullseye Zooplanktivores 0.01622 2.99000

Pempherididae Pempheris multiradiata Bigscale bullseye Zooplanktivores 0.01622 2.99000

283

Family Genus Species Common name Trophic group a b

Plesiopidae Trachinops caudimaculatus Southern hulafish Zooplanktivores 0.00468 3.18000

Plesiopidae Trachinops taeniatus Eastern hulafish Zooplanktivores 0.00468 3.18000

Barrier reef Pomacentridae Amphiprion akindynos anemonefish Zooplanktivores 0.02399 2.97000

Wide-band Pomacentridae Amphiprion latezonatus anemonefish Zooplanktivores 0.02239 2.99000

Pomacentridae Chromis hypsilepis One-spot puller Zooplanktivores 0.01514 2.99000

Reticulate Pomacentridae Dascyllus reticulatus dascyllus Zooplanktivores 0.03090 3.01000

Threespot Pomacentridae Dascyllus trimaculatus dascyllus Zooplanktivores 0.03467 2.99000

Pomacentridae Pomacentrus coelestis Blue damselfish Zooplanktivores 0.01479 2.92000

Pomacentridae Stegastes gascoynei Coral Sea gregory Zooplanktivores 0.01995 2.99000

Serranidae Caesioperca lepidoptera Butterfly perch Zooplanktivores 0.02455 2.98000

Serranidae Caesioperca rasor Barber perch Zooplanktivores 0.00955 2.97000

Serranidae Pseudanthias squamipinnis Orange basslet Zooplanktivores 0.01514 2.92000

284

Family Genus Species Common name Trophic group a b

Acanthuridae Acanthurus dussumieri Pencil surgeonfish Herbivores 0.02951 2.95000

Acanthuridae Acanthurus nigrofuscus Dusky surgeonfish Herbivores 0.02512 2.97000

Orangeblotch Acanthuridae Acanthurus olivaceus surgeonfish Herbivores 0.02570 3.01000

Bluespine Acanthuridae Naso unicornis unicornfish Herbivores 0.02344 2.95000

Acanthuridae Prionurus maculatus Spotted sawtail Herbivores 0.01122 3.04000

Acanthuridae Prionurus microlepidotus Australian sawtail Herbivores 0.01122 3.04000

Aplodactylidae Aplodactylus arctidens Marblefish Herbivores 0.00525 3.15000

Aplodactylidae Aplodactylus lophodon Rock Cale Herbivores 0.00479 3.13000

Kyphosidae Girella elevata Rock blackfish Herbivores 0.01585 3.03000

Kyphosidae Girella tricuspidata Luderick Herbivores 0.01148 3.00000

Kyphosidae Girella zebra Zebra fish Herbivores 0.01585 3.03000

Kyphosidae Kyphosus sydneyanus Silver drummer Herbivores 0.01995 3.02000

285

Family Genus Species Common name Trophic group a b

Odacidae Olisthops cyanomelas Herring Cale Herbivores 0.00692 3.08000

Immaculate Pomacentridae Mecaenichthys immaculatus damsel Herbivores 0.02239 2.99000

Pomacentridae Parma microlepis White Ear Scalyfin Herbivores 0.02239 2.99000

Pomacentridae Parma oligolepis Bigscale scalyfin Herbivores 0.02239 2.99000

Pomacentridae Parma unifasciata Girdled Scalyfin Herbivores 0.02239 2.99000

Pomacentridae Parma victoriae Victorian scalyfin Herbivores 0.02239 2.99000

Siganidae Siganus fuscescens Black rabbitfish Herbivores 0.01380 2.99000

Acanthuridae Paracanthurus hepatus Blue tang Omnivores 0.02291 2.97000

Blenniidae Parablennius tasmanianus Tasmanian blenny Omnivores 0.00741 3.00000

Chaetodontidae Chaetodon flavirostris Dusky butterflyfish Omnivores 0.02455 3.06000

Gunther's Chaetodontidae Chaetodon guentheri butterflyfish Omnivores 0.02291 3.01000

Chaetodontidae Chaetodon kleinii Klein's butterflyfish Omnivores 0.03090 3.05000

286

Family Genus Species Common name Trophic group a b

Chaetodontidae Chelmonops truncatus Eastern Talma Omnivores 0.02188 3.02000

Cheilodactylidae Nemadactylus douglasii Porae Omnivores NA NA

Chironemidae Chironemus marmoratus Eastern kelpfish Omnivores 0.01000 3.05000

Kyphosidae Microcanthus strigatus Stripey Omnivores 0.02291 2.99000

Kyphosidae Scorpis aequipinnis Sea Sweep Omnivores 0.01479 3.00000

Kyphosidae Scorpis lineolata Silver sweep Omnivores 0.01202 3.01000

Bridled Monacanthidae Acanthaluteres spilomelanurus leatherjacket Omnivores 0.02188 2.91000

Brown Monacanthidae Acanthaluteres vittiger leatherjacket Omnivores 0.02188 2.91000

Southern pygmy Monacanthidae Brachaluteres jacksonianus leatherjacket Omnivores 0.02188 2.91000

Black reef Monacanthidae Eubalichthys bucephalus leatherjacket Omnivores 0.02188 2.91000

Gunn's Monacanthidae Eubalichthys gunnii leatherjacket Omnivores 0.01000 3.04000

287

Family Genus Species Common name Trophic group a b

Mosaic Monacanthidae Eubalichthys mosaicus leatherjacket Omnivores 0.02188 2.91000

Southern Monacanthidae Meuschenia australis leatherjacket Omnivores 0.02188 2.91000

Yellow-striped Monacanthidae Meuschenia flavolineata leatherjacket Omnivores 0.02188 2.91000

Six-spined Monacanthidae Meuschenia freycineti leatherjacket Omnivores 0.02188 2.91000

Bluelined Monacanthidae Meuschenia galii leatherjacket Omnivores 0.02188 2.91000

Horse-shoe Monacanthidae Meuschenia hippocrepis leatherjacket Omnivores 0.02188 2.91000

Velvet Monacanthidae Meuschenia scaber leatherjacket Omnivores 0.03090 2.95000

Yellowfin Monacanthidae Meuschenia trachylepis leatherjacket Omnivores 0.02188 2.91000

Fanbelly Monacanthidae Monacanthus chinensis leatherjacket Omnivores 0.02512 2.79000

288

Family Genus Species Common name Trophic group a b

Ocean Monacanthidae Nelusetta ayraudi leatherjacket Omnivores 0.01148 2.81000

Rough Monacanthidae Scobinichthys granulatus leatherjacket Omnivores 0.02188 2.91000

Monodactylidae Monodactylus argenteus Diamondfish Omnivores 0.02951 2.98000

Odacidae Heteroscarus acroptilus Rainbow cale Omnivores 0.00692 3.08000

Odacidae Neoodax balteatus Little weed whiting Omnivores 0.00389 3.12000

Pencil weed Odacidae Siphonognathus beddomei whiting Omnivores 0.00389 3.12000

Eastern smooth Ostraciidae Anoplocapros inermis boxfish Omnivores 0.01995 3.01000

Ostraciidae Ostracion cubicus Yellow boxfish Omnivores 0.05370 2.76000

Plotosidae Cnidoglanis macrocephalus Estuary cobbler Omnivores 0.00457 3.07000

Pomacanthidae Centropyge tibicen Keyhole angelfish Omnivores 0.03388 2.85000

Indo-Pacific Pomacentridae Abudefduf vaigiensis sergeant Omnivores 0.02630 3.01000

289

Family Genus Species Common name Trophic group a b

Pomacentridae Chromis nitida Yellowback puller Omnivores 0.01479 2.98000

Pomacentridae Plectroglyphidodon dickii Dick's damsel Omnivores 0.02455 2.99000

Pomacentridae Pomacentrus australis Australian damsel Omnivores 0.01479 2.98000

Zanclidae Zanclus cornutus Moorish idol Omnivores 0.01738 3.06000

Halfmoon Benthic Balistidae Sufflamen chrysopterum triggerfish invertivores 0.01995 2.95000

Benthic Carangidae Pseudocaranx georgianus Silver Trevally invertivores 0.01445 2.96000

Chevron Benthic Chaetodontidae Chaetodon trifascialis butterflyfish invertivores 0.02089 2.96000

Benthic Cheilodactylidae Cheilodactylus fuscus Red morwong invertivores 0.01202 3.02000

Benthic Cheilodactylidae Cheilodactylus nigripes Magpie perch invertivores 0.01202 3.02000

Benthic Cheilodactylidae Cheilodactylus spectabilis Banded Morwong invertivores 0.01698 3.00000

Benthic Cheilodactylidae Cheilodactylus vestitus Crested morwong invertivores 0.01202 3.02000

290

Family Genus Species Common name Trophic group a b

Benthic Cheilodactylidae Dactylophora nigricans Dusky morwong invertivores 0.00389 3.12000

Benthic Cirrhitidae Cirrhitichthys aprinus Blotched hawkfish invertivores 0.01585 3.01000

Threebar Benthic Diodontidae Dicotylichthys punctulatus porcupinefish invertivores 0.03090 2.89000

Slender-spined Benthic Diodontidae Diodon nicthemerus porcupine fish invertivores 0.03090 2.89000

Benthic Enoplosidae Enoplosus armatus Old wife invertivores 0.00389 3.12000

Common silver Benthic Gerreidae Gerres subfasciatus belly invertivores 0.01479 3.07000

Benthic Gerreidae Parequula melbournensis Silverbelly invertivores 0.01413 3.02000

Benthic Gobiidae Istigobius hoesei Hoese's Sandgoby invertivores 0.00891 3.08000

Benthic Kyphosidae Tilodon sexfasciatus Moonlighter invertivores 0.02291 2.99000

291

Family Genus Species Common name Trophic group a b

Eastern blue Benthic Labridae Achoerodus viridis groper invertivores 0.01995 2.99000

Benthic Labridae Anampses caeruleopunctatus Diamond wrasse invertivores 0.01000 3.06000

Benthic Labridae Anampses neoguinaicus Blackback wrasse invertivores 0.01000 3.06000

Blackspotted Benthic Labridae Austrolabrus maculatus wrasse invertivores 0.01995 2.99000

Benthic Labridae Cirrhilabrus punctatus Finespot wrasse invertivores 0.01660 2.95000

Benthic Labridae Coris dorsomacula Pinklined wrasse invertivores 0.01000 3.06000

Benthic Labridae Coris picta Comb wrasse invertivores 0.00501 3.12000

Eastern king Benthic Labridae Coris sandeyeri wrasse invertivores 0.01000 3.06000

Castelnau's Benthic Labridae Dotalabrus aurantiacus wrasse invertivores 0.01000 3.04000

292

Family Genus Species Common name Trophic group a b

Benthic Labridae Eupetrichthys angustipes Snakeskin wrasse invertivores 0.01000 3.04000

Checkerboard Benthic Labridae Halichoeres hortulanus wrasse invertivores 0.00955 3.08000

Benthic Labridae Halichoeres nebulosus Nebulous wrasse invertivores 0.01000 3.08000

Bluestreak cleaner Benthic Labridae Labroides dimidiatus wrasse invertivores 0.00447 3.13000

Benthic Labridae Notolabrus fucicola Banded wrasse invertivores 0.00977 3.07000

Crimsonband Benthic Labridae Notolabrus gymnogenis wrasse invertivores 0.01995 2.99000

Benthic Labridae Notolabrus tetricus Bluethroat wrasse invertivores 0.01995 2.99000

Southern Maori Benthic Labridae Ophthalmolepis lineolatus wrasse invertivores 0.00447 3.13000

Benthic Labridae Oxycheilinus bimaculatus Little Maori wrasse invertivores 0.01995 2.77000

293

Family Genus Species Common name Trophic group a b

Benthic Labridae Pictilabrus laticlavius Senator wrasse invertivores 0.00977 3.05000

Benthic Labridae Pseudolabrus guentheri Gunther's wrasse invertivores 0.00977 3.05000

Benthic Labridae Pseudolabrus luculentus Luculent wrasse invertivores 0.00977 3.05000

Benthic Labridae Pseudolabrus mortonii Rosy wrasse invertivores 0.00977 3.05000

Benthic Labridae Stethojulis bandanensis Redspot wrasse invertivores 0.01023 2.99000

Benthic Labridae Stethojulis interrupta Brokenline wrasse invertivores 0.01000 3.06000

Painted rainbow Benthic Labridae Suezichthys arquatus wrasse invertivores 0.00447 3.14000

Blunt-headed Benthic Labridae Thalassoma amblycephalum wrasse invertivores 0.01316 3.04000

Benthic Labridae Thalassoma lunare Moon wrasse invertivores 0.01259 2.96000

294

Family Genus Species Common name Trophic group a b

Yellow-brown Benthic Labridae Thalassoma lutescens wrasse invertivores 0.01000 3.04000

Blackbarred Benthic Labridae Thalassoma nigrofasciatum wrasse invertivores 0.00977 3.04000

Benthic Latridae Latridopsis forsteri Bastard trumpeter invertivores 0.01202 3.04000

Benthic Mullidae Parupeneus multifasciatus Banded goatfish invertivores 0.01318 3.11000

Blacksaddle Benthic Mullidae Parupeneus spilurus goatfish invertivores 0.01288 3.04000

Bluestriped Benthic Mullidae Upeneichthys lineatus goatfish invertivores 0.01072 3.10000

Bluespotted Benthic Mullidae Upeneichthys vlamingii goatfish invertivores 0.01023 3.09000

Benthic Ostraciidae Aracana aurita Striped cowfish invertivores 0.01995 3.01000

Benthic Pentacerotidae Pentaceropsis recurvirostris Longsnout boarfish invertivores 0.01995 3.01000

295

Family Genus Species Common name Trophic group a b

Benthic Pinguipedidae Parapercis ramsayi Spotted grubfish invertivores 0.00646 3.10000

Benthic Scorpaenidae Centropogon australis Fortescue invertivores NA NA

Halfbanded Benthic Serranidae Hypoplectrodes maccullochi seaperch invertivores 0.00955 2.97000

Benthic Sparidae Acanthopagrus australis Yellowfin bream invertivores 0.01288 3.03000

Benthic Sparidae Chrysophrys auratus Snapper invertivores 0.02399 2.95000

Benthic Sparidae Rhabdosargus sarba Tarwhine invertivores 0.02138 2.94000

Eastern striped Benthic Terapontidae Pelates sexlineatus grunter invertivores 0.01380 3.02000

White-spotted Benthic Tetraodontidae Arothron hispidus pufferfish invertivores 0.03890 2.83000

Benthic Tetraodontidae Canthigaster callisterna Clown toby invertivores 0.02818 2.94000

296

Family Genus Species Common name Trophic group a b

Benthic Tetraodontidae Tetractenos glaber Smooth toadfish invertivores 0.02692 2.88000

Benthic Trachichthyidae Optivus agastos Violet roughy invertivores 0.01072 3.03000

Benthic Trachichthyidae Trachichthys australis Southern roughy invertivores 0.01660 3.05000

Benthic Tripterygiidae Forsterygion varium Variable threefin invertivores 0.00676 3.08000

Aulopidae Latropiscis purpurissatus Sergeant baker Piscivores 0.00437 3.13000

Bluestriped Blenniidae Plagiotremus rhinorhynchos fangblenny Piscivores 0.00617 3.14000

Blenniidae Plagiotremus tapeinosoma Piano fangblenny Piscivores 0.00457 3.01000

Carangidae Seriola lalandi Yellowtail kingfish Piscivores 0.01820 2.93000

Dinolestidae Dinolestes lewini Long-finned pike Piscivores 0.00389 3.12000

Fistulariidae Fistularia commersonii Smooth flutemouth Piscivores 0.00240 2.94000

Goldspotted Haemulidae Plectorhinchus flavomaculatus sweetlips Piscivores 0.01698 3.02000

297

Family Genus Species Common name Trophic group a b

Moridae Lotella rhacina Largetooth beardie Piscivores 0.00363 3.13000

Muraenidae Gymnothorax prasinus Green moray eel Piscivores 0.00098 3.10000

Platycephalidae Platycephalus fuscus Dusky flathead Piscivores NA NA

Cook's Scorpaenidae Scorpaena cardinalis scorpionfish Piscivores 0.01778 3.03000

Eastern red Scorpaenidae Scorpaena jacksoniensis scorpionfish Piscivores 0.01778 3.03000

Southern red Scorpaenidae Scorpaena papillosa scorpionfish Piscivores 0.01230 3.03000

Serranidae Acanthistius ocellatus Eastern wirrah Piscivores 0.00933 2.97000

Serranidae Epinephelus fasciatus Blacktip rockcod Piscivores 0.01175 3.03000

Serranidae Epinephelus undulatostriatus Maori rockcod Piscivores 0.01175 3.04000

Serranidae Hypoplectrodes nigroruber Banded seaperch Piscivores 0.00933 2.97000

298

Supplementary Table 8.2.5 Summary of trophic group mean biomass and coefficients of variation in biomass for each 1-degree latitude bin. In this case, ‘All groups’ represents total fish biomass and ‘All latitudes’ represents the mean and CV in biomass across all latitudes. The number of surveys used to calculate each row is represented by ‘n’.

Mean observed biomass (g m-2) Coefficient of variation

Latitude bin (°S) zoo herb omni ben.inv pisc All zoo herb omni ben.inv pisc All groups n

29.5 61 30 87 41 0 220 3.17 1.40 5.24 2.75 2.93 2.86 53

30.5 73 117 33 18 0 241 1.99 2.63 1.81 1.30 1.83 1.53 113

31.5 212 9 260 30 6 517 2.04 1.48 2.74 1.72 3.93 1.94 43

32.5 91 15 5 29 4 143 3.43 4.14 1.81 1.24 2.39 2.30 558

33.5 41 19 3 14 2 79 1.78 2.81 1.79 1.24 2.96 1.25 443

34.5 35 10 6 24 3 78 1.95 1.62 1.86 1.57 3.21 1.23 34

35.5 80 16 5 18 2 121 1.93 1.80 1.94 1.11 2.76 1.43 912

36.5 199 35 25 38 7 304 1.38 1.28 2.36 0.95 1.88 1.04 95

37.5 31 12 8 42 11 105 3.40 0.99 1.32 0.96 2.96 1.24 243

38.5 NA NA NA NA NA NA NA NA NA NA NA NA 0

299

Mean observed biomass (g m-2) Coefficient of variation

Latitude bin (°S) zoo herb omni ben.inv pisc All zoo herb omni ben.inv pisc All groups n

39.5 32 7 6 17 1 63 1.62 0.86 1.28 0.73 5.51 0.92 110

40.5 12 7 18 15 0 52 1.91 1.73 2.53 0.50 4.21 1.27 20

41.5 0 1 5 4 3 12 2.63 1.76 0.91 0.87 2.85 0.78 15

42.5 2 1 1 11 2 17 2.19 3.85 2.36 1.48 4.62 1.27 142

43.5 2 0 0 10 1 14 3.29 2.83 2.34 2.13 3.84 1.68 251

All latitudes 64 18 11 21 3 118 2.91 4.11 9.90 1.48 3.80 2.19 3032

300

Supplementary Table 8.2.6 Relative variable importance (RVI) scores for each of the nine tested models. RVI is the sum of the Akaike weights over all models including the explanatory variable. Mean RVI was calculated as the mean of the RVI scores for each variable across the biomass models and across the biomass proportion models.

Total Zooplanktivores Herbivores Omnivores Ben.inv Mean Variable fish biomass biomass biomass biomass RVI biomass t2(SiteLat,Month) 1.00 1.00 1.00 1.00 1.00 1.00

Depth 1.00 0.79 1.00 0.89 0.99 0.93 ln(Population) 1.00 0.21 0.96 1.00 1.00 0.83 ln(MeanAbundZoo) 0.97 1.00 0.28 0.16 0.20 0.52 ln(Chl) 1.00 1.00 0.05 0.14 0.04 0.45

Zooplanktivores Herbivores Omnivores Ben.inv Mean Variable proportion proportion proportion proportion - RVI t2(SiteLat,Month) 1.00 1.00 0.97 1.00 - 0.99 ln(Population) 1.00 0.98 0.88 1.00 - 0.97

Depth 1.00 1.00 1.00 0.13 - 0.78 ln(MeanAbundZoo) 1.00 0.98 0.05 0.78 - 0.70 ln(Chl) 1.00 0.77 0.09 0.18 - 0.51

301

8.2.3. Supplementary figures

Supplementary Figure 8.2.1 Variation in (a) wave exposure, (b) site depth and (c) visibility of surveys for each combination of four seasons and four-degree latitude bins. Lower and upper box boundaries indicate 25th and 75th percentiles, respectively. The line inside the box represents the median. The lower and upper error lines represent 1.5 times the inter-quartile range, with outliers (filled points) occurring outside this range. Letter codes above boxplots represent the results of Tukey’s pairwise comparisons in compact letter display format. Plots that share one or more of the same characters are not statistically different. Note that (b) and (c) have log-transformed y-axes.

302

Supplementary Figure 8.2.2 Nonmetric multidimensional scaling plot generated using Bray-Curtis dissimilarity generated from a matrix of proportionate benthic cover for each combination of seasons and four-degree latitude bins. Centroids of each distribution are plotted in red. Note that the centroid position varies only minimally across seasons within each latitude bin, indicating no bias to survey sites with different benthic cover in different seasons. There is more variation in centroid position across latitude, corresponding to expected latitudinal differences in benthic cover.

303

Supplementary Figure 8.2.3 Categorised mean proportion of benthic cover from photo quadrats for each combination of seasons and four-degree latitude bins. The 36 categories for benthic cover were aggregated into nine functional groups to simplify plot appearance.

304

Supplementary Figure 8.2.4 Seasonal survey coverage tabulated by one-degree latitude bins across the duration of the survey data. Note the log-transformed y-axis. For the most part, latitudes with inconsistent survey coverage across seasons have more regular coverage in adjacent bins. The exception is at the highest latitudes, where winter and spring survey coverage is lacking.

305

Supplementary Figure 8.2.5 Yearly survey coverage tabulated by one-degree latitude bins across the duration of the survey data. Note the log-transformed y-axis. Survey coverage is very regularly spread across years for the mid-latitudes however there are some gaps in coverage in the lowest and highest latitudes.

306

Supplementary Figure 8.2.6 Survey coverage and validation for five very regularly sampled sites (PS4, PS13, PS17, PS7, PS20) in the Port Stephens area (32.5 to 33 °S), with (a) representing the distribution of survey dates across the five sites and (b) mean monthly total fish biomass represented by the height of each bar. Stacked bars represent the proportionate contribution of each trophic group to the monthly mean total fish biomass. Error bars represent standard error of the mean for total fish biomass. Numbers above each error bar indicate the number of unique surveys used to calculate totals and proportions for each corresponding bar. This bar plot represents a localised validation of the large-scale seasonal patterns observed over the latitudinal range of the study. The validity of the observed pattern is supported by the high temporal resolution of surveys at these five sites.

307

Supplementary Figure 8.2.7 Length frequency distributions for all fish included in our analyses, for each combination of season and four-degree latitude bin. Median values are indicated by the dashed red line and number displayed on each panel.

308

Supplementary Figure 8.2.8 Mean generalised additive mixed model (GAMM) relationships for (a) Zooplanktivore biomass; (b) Herbivore biomass; (c) Omnivore biomass (d) Benthic invertivore biomass and (e) Total fish biomass. The rug plot for each panel represents the distribution of data available for each variable. The y-axes represent the natural log-transformed biomass response (in grams m-2). Only covariates included in the best model for each response are shown.

309

Supplementary Figure 8.2.9 Mean generalised additive mixed model (GAMM) relationships for (a) Zooplanktivore proportion; (b) Herbivore proportion; (c) Omnivore proportion and (d) Benthic invertivore proportion. The rug plot for each panel represents the distribution of data available for each variable. The y-axes represent the natural log-transformed biomass response (in grams m-2). Only covariates included in the best model for each response are shown.

310

8.3. Supplementary material for Chapter 3

8.3.1. Supplementary methods

8.3.1.1. Acoustic processing methods The 2016 and 2017 acoustic data were adjusted post collection, based on calibrations which were conducted on 17 August 2016 and 08 June 2018, respectively. Parameters such as transmission power and pulse duration were kept the same between voyages and gain adjustments drifted less than 0.3 dB from the 2016 calibration.

In both years, the EK60 transducers were synchronised and set to a pulse duration of 2.048 ms for the 18, 38 and 70 kHz and 1.024 ms for the 120, 200 and 333 kHz transducers with a ping rate of 1.11 Hz. Transmission power was set to 2000 W for the 18 and 38 kHz, 750 W for the 70 kHz, 250 W for the 120 kHz, 105 W for the 200 kHz and 40 W for the 333 kHz transducers.

In Echoview, seabed returns were manually verified and removed along with the first 5 m of water column below the transducer face to eliminate surface noise and ring down. Data was cleaned for removing background noise, impulse noise and transient noise prior to echo integration following the methods of De Robertis and Higginbottom (2007) and Ryan et al.

(2015) through built-in functions in Echoview. Ping times and geometry were matched across all frequencies so that echoes from each frequency could be compared. Grid cells were sufficiently large to avoid issues of frequency-dependent sampling volume (Korneliussen &

Ona, 2002).

8.3.1.2. Chlorophyll a methods

Chlorophyll a (Chl-a) was modelled from fluorescence by using laboratory-measured Chl-a determinations made with a Turner fluorometer (Turner Designs, San Jose, United States).

A separate regression was performed for each survey. Water samples for Chl-a were

311

collected from the ship’s flowthrough seawater system at intervals during each voyage and matched to the geometric mean of Chl-a fluorescence calculated over ten-minute intervals, centred at the time of underway water sampling and fluorometric Chl-a measurement (2016: coeff. = 0.28, int. = -0.93, n = 48, R2 = 0.85, p < 0.001) or high-performance liquid chromatography estimate (HPLC; 2017: coeff. = 1.09, int. = -2.69, n = 14, R2 = 0.97, p <

0.001).

312

8.3.2. Supplementary tables

Supplementary Table 8.3.1 The number of replicates used to calculate mean values for NASC across each polygon in the 2016 (n 2016) and 2017 (n 2017) surveys.

Polygon n 2016 n 2017

N2.1 177 101

N2.2 181 55

N2.3 219 105

N1.1 170 146

N1.2 418 171

N1.3 221 66

S1.1 115 117

S1.2 243 130

S1.3 43 29

S2.1 200 127

S2.2 272 188

S2.3 236 170

313

Supplementary Table 8.3.2 The four final GLMs generated through forward-stepwise AIC-based selection. Column names are as follows: df = residual degrees of freedom, Est = coefficient estimate, SE = standard error of the estimated coefficient, t = t-value. The original coefficient estimates test statistics are provided and the heteroskedasticity and autocorrelation consistent estimates (suffix: HAC) are also provided to indicate a more conservative test accounting for serial autocorrelation.

Model formula df Predictor Est SE t Pr(>|t|) SE HAC t HAC Pr(>|t|) HAC

NASC.2016 ~ bathymetry + 2838 Intercept -12.18 2.14 -5.69 < 0.001 *** 2.77 -4.40 < 0.001 *** temperature bathymetry 0.05 0.00 15.24 < 0.001 *** 0.01 9.82 < 0.001 ***

temperature 0.56 0.13 4.38 < 0.001 *** 0.15 3.70 < 0.001 ***

NASC.2017 ~ bathymetry + 1569 Intercept -2.28 0.89 -2.56 0.011 * 2.44 -0.93 0.350 . temperature bathymetry 0.03 0.00 18.68 < 0.001 *** 0.00 6.23 < 0.001 ***

temperature 0.34 0.06 5.57 < 0.001 *** 0.17 2.01 0.044 *

314

Model formula df Predictor Est SE t Pr(>|t|) SE HAC t HAC Pr(>|t|) HAC

NASC.2017 ~ bathymetry + 1075 Intercept -6.94 1.17 -5.94 < 0.001 *** 3.00 -2.31 0.021 * temperature + log10(zooplankton biomass) bathymetry 0.02 0.00 11.00 < 0.001 *** 0.00 4.79 < 0.001 ***

temperature 0.72 0.09 7.69 < 0.001 *** 0.23 3.08 0.002 **

log10(zooplankton) -0.24 0.14 -1.76 0.079 . 0.25 -0.96 0.335 .

Zooplankton Biomass ~ 682 Intercept 3.14 1.18 2.66 0.008 ** 1.93 1.62 0.105 . log10(chlorophyll) + bathymetry + temperature log10(chl) 2.20 0.53 4.17 < 0.001 *** 1.19 1.84 0.066 .

bathymetry -0.01 0.00 -5.18 < 0.001 *** 0.00 -2.63 0.009 **

temperature 0.42 0.08 4.95 < 0.001 *** 0.14 2.90 0.004 **

315

8.3.3. Supplementary figures

Supplementary Figure 8.3.1 A sample of echograms from the 2016 (a) and 2017 (b) surveys, along with relative frequency response plots (to the right of each corresponding echogram). Red boxes indicate the section of the echogram that was integrated to produce the relative frequency response plot.

316

Supplementary Figure 8.3.2 Proportion of fisheries logbook catch by weight from midwater trawls in the Montague Island region for August to September 2016 to 2019. Numbers at the base of bars indicate the number of trawls used to calculate mean proportions. Missing bars indicate that fishing operations were not conducted in the region during the corresponding time period.

317

Supplementary Figure 8.3.3 Cells classified as fish for the 2016 (a) and the 2017 (c) surveys and the mean frequency responses (relative to Sv38) for cells which were classified as fish for the 2016 (b) and 2017 (d) surveys.

318

Supplementary Figure 8.3.4 Vertical profile of temperature (a) and zooplankton biomass (c) recorded by the undulating towed-body during the northern leg of the 2017 survey, and interpolated profiles generated using ordinary kriging of the raw temperature (b) and zooplankton biomass (d) data.

319

Supplementary Figure 8.3.5 Smoothed density estimates for the temperature distribution of commercial midwater trawls weighted by CPUE (kg hour-1) for Trachurus declivis (a) and Scomber australasicus (b). The red vertical line indicates the peak in the temperature distribution for each panel and the blue line indicates the mean water temperature for the region (averaged between 100 to 200 m depth) for days when fishing operations occurred.

320

Supplementary Figure 8.3.6 Spatially paired NASC values for each of the 2016 and 2017 surveys (n = 1107, r = 0.40). Points are coloured by bathymetry. Points arranged in a linear pattern along the x-axis represent positive NASC values in 2016 with corresponding zero NASC in 2017 and vice versa for the y-axis.

321

8.4. Supplementary material for Chapter 4

8.4.1. Supplementary tables

Supplementary Table 8.4.1 School detection settings that were used for the validation of the school thickness variable with EK80 data to isolate and extract fish schools.

Parameter Value (m)

Min total school length 3

Min total school height 2

Min candidate length 1

Min candidate height 2

Max vertical linking distance 5

Max horizontal linking distance 5

Distance mode GPS distance

322

8.4.2. Supplementary figures

Supplementary Figure 8.4.1 Sensitivity analysis for the derivation of school thickness with the difference method, conducted using a single survey of the split-beam data and varying the minimum Sv threshold from -65 to -40 dB. Panels represent the distribution of school thickness values across 1 m horizontal intervals for 12 unique fish schools.

323

Supplementary Figure 8.4.2 Sensitivity analysis for the derivation of school thickness with the sum method, conducted using a single survey of the split-beam data and varying the vertical resolution from 0.1 to 2 m. Panels represent the distribution of school thickness values across 1 m horizontal intervals for 12 unique fish schools. Note, all boxplots for each unique school contain the same number of values.

324

Supplementary Figure 8.4.3 GAMM prediction for the MBES school thickness distribution model for six variations of the k-parameter for the smooth term, with (a) k=5, (b) k=10, (c) k=25, (d) k=50, (e) k=75, (f) k=100. Black squares indicate the locations of the concrete reef modules.

325

Supplementary Figure 8.4.4 GAMM prediction for the binomial model representing the decline in probability of target occurrence with distance from the nearest reef module.

326

Supplementary Figure 8.4.5 Spatial profiles of the 68 fish schools considered in the validation with split-beam data. Note that panels have vertical exaggeration of the y-axis relative to the x-axis. Panels are arranged by the correlation of school thickness (using the sum method) and backscatter in descending order and panels are labelled by survey (letter) and by the order in which they were recorded within each survey (number). Correlation coefficient (R) indicated in the top left corner of each panel. Text in blue indicates p < 0.05 for the correlation test, while text in red indicates p > 0.05.

327

Supplementary Figure 8.4.6 Correlation of school thickness with backscatter for the 68 fish schools considered in the validation with split-beam data. Panels are arranged by the correlation of school thickness (using the sum method) and backscatter in descending order and panels are labelled by survey (letter) and by the order in which they were recorded within each survey (number). Correlation coefficient (R) indicated in the top left corner of each panel. Text in blue indicates p < 0.05 for the correlation test, while text in red indicates p > 0.05.

328

Supplementary Figure 8.4.7 Standard error estimate for the MBES school thickness distribution model predictions around the artificial reef complex. Note the greatest error occurs at the periphery of the surveyed area.

329

Supplementary Figure 8.4.8 Semivariogram for the MBES school thickness distribution model.

330

Supplementary Figure 8.4.9 Processed school thickness distributions (1 m resolution) for each of the five MBES surveys and for the mean distribution and standard error of the mean across the five surveys.

331

8.5. Supplementary material for Chapter 5

8.5.1. Supplementary methods

For the study of diel effects on school characteristics, an image analysis process based on

Reid and Simmonds (1993) was used to extract fish schools from gridded school thickness data. Initially a ‘blur’ filter, essentially a weighted mean function with a 3 ⨯ 3 kernel, was applied to each transect to remove background noise (Supplementary Figure 8.5.5) caused by zooplankton and isolated fish, using the following kernel weighting recommended in

Reid and Simmonds (1993):

1 2 1 2 1 2 1 2 1

This was followed by binary thresholding of the blurred data, using a threshold value selected by examining the frequency histogram of school thickness. In this case, 2 m was selected as the minimum school thickness threshold because it excluded the largest spike in frequency to result from noise. Approximately 50% of transects had a mean value above this threshold.

To further isolate schools we then applied an erosion filter to this binary data, which takes the minimum value of an unweighted 3 ⨯ 3 kernel. We subsequently applied a dilation, which takes the maximum value of an identical unweighted kernel. The erosion followed by dilation steps essentially remove a layer of pixels, and then add a layer of pixels back to each object, respectively (Reid & Simmonds, 1993). This removes isolated nonzero pixels and smooths school edges, further cleaning the data and eliminating individual targets and returns not likely to be aggregations. Following this, objects in the binary data were converted to polygons, which were used to extract the original unprocessed data. In this

332

way, aggregations were defined and isolated so that statistics, such as perimeter and area, could be calculated for each aggregation.

333

8.5.2. Supplementary tables

Supplementary Table 8.5.1 Regression estimates for the three GLS models describing the total abundance, species richness and Shannon diversity of schooling fish around the artificial and natural reef sites. Select column names as follows: DF – total degrees of freedom, R DF – residual degrees of freedom, SE Mod – model standard error, ΔAIC date – ΔAIC with date as the only predictor, p – p-value from ANOVA of specified model and model with date as only predictor, Estimate – coefficient estimate, SE term – standard error of the specified term, Pr(>|t|) – p-value of the specified term.

Response DF R DF SE Mod ΔAIC date p Predictor Estimate SE Term t-value Pr(>|t|)

Total abundance 50 44 142.09 4.27 0.012 * intercept 187.88 52.47 3.58 <0.001 ***

reef JDN 103.84 42.84 2.42 0.020 *

date 2019-08-14 -228.90 67.74 -3.38 0.002 **

date 2019-09-26 -191.60 67.74 -2.83 0.007 **

date 2019-10-31 -169.50 67.74 -2.50 0.016 *

date 2019-11-08 -230.40 67.74 -3.40 0.001 **

334

Response DF R DF SE Mod ΔAIC date p Predictor Estimate SE Term t-value Pr(>|t|)

Species richness 50 44 0.94 26.3 <0.001 *** intercept 2.42 0.35 6.98 <0.001 ***

reef JDN -1.64 0.28 -5.79 <0.001 ***

date 2019-08-14 0.00 0.45 0.00 1.000 .

date 2019-09-26 -0.20 0.45 -0.45 0.660 .

date 2019-10-31 0.40 0.45 0.89 0.380 .

date 2019-11-08 -1.10 0.45 -2.46 0.018 *

Shannon diversity 50 44 0.23 34.9 <0.001 *** intercept 0.42 0.08 4.92 <0.001 ***

reef JDN -0.48 0.07 -6.93 <0.001 ***

date 2019-08-14 0.25 0.11 2.31 0.026 *

date 2019-09-26 0.02 0.11 0.15 0.882 .

date 2019-10-31 0.27 0.110 2.510 0.016 *

date 2019-11-08 -0.11 0.11 -0.99 0.329 .

335

Supplementary Table 8.5.2 Regression estimates for the eight GLS models describing the area per school, proportion of area with schools, volume per school, schools hectare-1, P/A ratio, A/V ratio, school height above bottom and school thickness of schooling fish around the artificial and natural reef sites. Select column names as follows: DF – total degrees of freedom, R DF – residual degrees of freedom, SE Mod – model standard error, ΔAIC date – ΔAIC with date as the only predictor, p – p-value from ANOVA of specified model and model with date as only predictor, Estimate – coefficient estimate, SE term – standard error of the specified term, Pr(>|t|) – p-value of the specified term.

Response DF R DF SE Mod ΔAIC date p Predictor Estimate SE Term t-value Pr(>|t|)

Area per 110 102 1438.6 35.4 <0.001 *** intercept -386.75 556.560 -0.69 0.489 . school (m2) diel day 106.41 379.760 0.28 0.780 .

reef site JDN 2715.67 445.890 6.09 <0.001 ***

diel day : reef site JDN -1615.14 596.410 -2.71 0.008 **

Proportion 110 102 0.097 83.6 <0.001 *** intercept -0.05 0.038 -1.28 0.204 . of area with schools diel day 0.02 0.026 0.74 0.459 .

reef site JDN 0.30 0.030 10.09 <0.001 ***

diel day : reef site JDN -0.16 0.040 -3.97 <0.001 ***

336

Response DF R DF SE Mod ΔAIC date p Predictor Estimate SE Term t-value Pr(>|t|)

Volume 110 102 9712.8 30.5 <0.001 *** intercept -4502.68 3756.785 -1.20 0.234 . per school (m3) diel day 1097.82 2562.642 0.43 0.669 .

reef site JDN 14424.62 3006.755 4.80 <0.001 ***

diel day : reef site JDN -2978.94 4024.077 -0.74 0.461 .

Schools 110 102 2.177 10.4 <0.001 *** intercept 3.55 0.845 4.19 <0.001 *** hectare-1 diel day -1.82 0.580 -3.14 0.002 **

reef site JDN -1.12 0.684 -1.64 0.103 .

diel day : reef site JDN 0.42 0.910 0.46 0.647 .

School 110 102 1.578 60.0 <0.001 *** intercept 4.80 0.617 7.77 <0.001 *** height above bottom diel day 1.37 0.428 3.20 0.002 ** (m) reef site JDN 0.27 0.502 0.53 0.598 .

diel day : reef site JDN 2.57 0.667 3.85 <0.001 ***

337

Response DF R DF SE Mod ΔAIC date p Predictor Estimate SE Term t-value Pr(>|t|)

School 110 102 1.544 79.3 <0.001 *** intercept 1.65 0.615 2.68 0.009 ** height (m) diel day 1.00 0.431 2.32 0.022 *

reef site JDN 0.53 0.504 1.05 0.298 .

diel day : reef site JDN 3.52 0.666 5.29 <0.001 ***

P/A ratio 110 102 0.355 14.1 <0.001 *** intercept 1.22 0.138 8.85 <0.001 ***

diel day -0.20 0.096 -2.08 0.041 *

reef site JDN -0.35 0.112 -3.14 0.002 **

diel day : reef site JDN 0.09 0.149 0.62 0.535 .

A/V ratio 110 102 0.093 22.1 <0.001 *** intercept 0.40 0.036 11.05 <0.001 ***

diel day -0.01 0.025 -0.37 0.714 .

reef site JDN -0.02 0.029 -0.82 0.417 .

diel day : reef site JDN -0.10 0.039 -2.64 0.010 **

338

Response DF R DF SE Mod ΔAIC date p Predictor Estimate SE Term t-value Pr(>|t|)

Distance 556 548 30.562 11.1 <0.001 *** intercept 104.91 9.774 10.73 <0.001 *** from reef centre (m) diel day -6.24 5.916 -1.05 0.292 .

reef site JDN -15.21 6.395 -2.38 0.018 *

diel day : reef site JDN -4.88 9.207 -0.53 0.597 .

339

8.5.3. Supplementary figures

Supplementary Figure 8.5.1 Transect patterns for surveying school distribution (a) and diel effects on school characteristics (b), with (a) aligned into the prevailing swell and (b) along lines of constant longitude, starting with red arrow, and ending with the blue arrow. Red lines represent the first set of transects and blue lines represent the second set of transects. Dashed lines indicate sections of the vessel track which were excluded from analysis. The black diamonds indicate the reef centre, or the reef structures in the case of artificial reefs. All data outside grey boxes were excluded from analyses.

340

Supplementary Figure 8.5.2 A schematic diagram of the camera assembly used for drifting and benthic remote video deployments (from Smith et al., 2017).

341

Supplementary Figure 8.5.3 Examples of raster surfaces used to generate explanatory variables on a per-survey basis for the artificial reefs (a) and the natural reef (b) models. The examples provided are from site OAR (a) and N1 (b).

342

Supplementary Figure 8.5.4 Semivariograms generated from spatially referenced residuals for the final artificial reef (a) and natural reef (b) models.

343

Supplementary Figure 8.5.5 An illustrated example of the six-step image analysis process for isolating fish schools from rasterised multibeam echosounder data. Pixel size is 1 m.

344