Trophic Structure Through the History of Exploitation in New Zealand Fish Communities

Trophic structure through the history of exploitation in

New Zealand fish communities

______

A thesis

submitted in partial fulfilment

of the requirements for the Doctor of Philosophy

diploma in Marine Sciences

University of Otago

Leonardo Maia Durante

______

University of Otago

2020 Abstract of a thesis submitted in partial fulfilment of the

requirements for the Doctor of Philosophy diploma in Marine Sciences

Trophic structure through the history of exploitation in New Zealand fish communities

Leonardo Maia Durante

Marine ecosystems have been under a considerable amount of stress due to human activities, especially after industrialization. The abundance and ecological interactions of exploited fishes are influenced by fishing activities, climate change, and natural environmental variability, but there is a lack of knowledge on how these community parameters have changed over time. In the present study, temporal shifts in the trophic structure and composition of an important commercial fish community along the east coast of the South Island of New Zealand was investigated. The Marine Trophic Index (MTI) and species composition of the reported commercial fisheries landings for the whole country were used to identify periods of fisheries expansion and shifts in the trophic architecture of the landings. A stable isotope approach was used to estimate resource use and trophic level of fishes, comparing these parameters spatially and temporally. It was demonstrated that it is possible to access carbon and nitrogen stable isotope values from preserved fishes, for both bulk tissues and specific amino acids, which enabled this work to access ecological information from specimens stored in museum samples. The results demonstrate that fisheries industrialization occurred between 1970 and 2000 in New Zealand, increasing MTI values and shifting the relative landings composition from inshore to offshore and deep-water species over time. Regional comparisons demonstrated that modern fish communities from Otago relied more heavily on macroalgae production when compared to those from Kaikoura, linked to higher fishery yields in the former. When analyzing the same parameters from museum specimens collected before the full expansion of industrialized fisheries, modern species inhabiting the outer shelf and slope also showed a reduced reliance on macroalgae production, together with an increased trophic level. When considering continuous changes in the trophic parameters, environmental factors and prey abundance had a significant relationship with the variation of trophic parameters for most fish assemblages. Trophic parameters of the inner shelf assemblage also showed a relationship with MTI throughout time, indicating the earlier effects of coastal fisheries. Although Kaikoura and modern communities showed a wider niche breadth compared to Otago and historical samples, respectively, tarakihi was the only species to show both regional and temporal trends. Tarakihi stocks in the region have been overexploited and are sensitive to habitat degradation, which could be linked to the shift in trophic structure observed here. Furthermore, temporal changes in community composition towards a relatively higher abundance of lower trophic level species were observed from the analysis of long-term trawl survey data. Species with decreasing abundance that presented shifts in trophic parameters with time, such as red cod, had their trophic level and resource use occupied by species with increasing abundance, like spiny dogfish. The present study demonstrates how the trophic structure and composition of important commercial communities have changed over time and how they are related to natural and anthropogenic effects, with the potential to alter regional fishery yields. This study highlights how uncovering ecological baselines before human impacts is an important step towards the implementation of an ecosystem-based management approach to fisheries resources.

Keywords: Quota Management System, fisheries, exploitation, isotopes, isotopic ecology, compound specific isotopic analysis, amino acids, history, ecosystem-based management, sustainability, kelp forest, food web, resource use, trophic level, museum collection, oceanography, fish community, ecological baselines, fishing down marine food webs

Life is only a snapshot of the ever-changing nature

Acknowledgements

I would like to thank my family for all the support throughout my studies, especially my mom who taught me from a young age the importance of education in a person’s life. She showed me that I had the capabilities to follow my dreams, as long as I was willing to put in the necessary effort to achieve them. I would like to give special thanks to my sister Leticia and my wife Erica, which are my biggest examples of dedication and endurance, both professionally and in their personal lives, I could not have completed this thesis without their support.

I would also like to show my gratitude to all my supervisors, for all the time they have put into the development of this thesis and myself as a researcher. I would like to express an especial thanks to Steve Wing, who took me as a Ph.D. student and provided financial, professional, and moral support for the completion of this thesis.

I thank all the staff from NIWA, Fisheries New Zealand and Sea Around Us, as well as Stina Kolodzey, Tomo, Rex, Tania King, Monique Ladds, Sorrel O'Connell-Milne, Max Williams and Alex Connolly who assisted with data and sample acquisition for this work. I would also like to show my gratitude to Neil Bagley, Mike Beentjes, Rosemary Hurst, Rich Ford, Peter Russel, Rob Smith, Carl Struthers, Clive Roberts and Andrew Stewart for their expertise, contribution, and support of this work.

Technical support was provided by Robert van Hale, Dianne Clark, Doug Mackie, Linda Groenewegen, Reuben Pooley and from the Departments of Marine Science, Chemistry and Zoology at Otago University, which I am grateful for. Monetary support was provided from a National Science Challenge: Sustainable Seas (4.1.1 Ecosystem Connectivity) grant to Stephen Richard Wing, while I was additionally supported by a University of Otago Ph.D. scholarship.

Table of Contents

Acknowledgements ...... 5 List of Tables ...... 11 List of Figures ...... 13 General introduction ...... 18 Chapter 1 Shifting trophic architecture of marine fisheries in New Zealand: implications for guiding effective ecosystem-based management ...... 23 Abstract ...... 23 1.1. Introduction ...... 24 1.1.1. New Zealand and global fisheries ...... 24 1.1.2. A brief overview of the fishery and management history in New Zealand ...... 28 1.1.3. Fish landings and ecological trends of New Zealand fish stocks ...... 31 1.2. Methods ...... 34 1.3. Results ...... 37 1.4. Discussion ...... 43 Chapter 2 Oceanographic transport along frontal zones forms carbon, nitrogen, and oxygen isoscapes on the east coast of New Zealand: implications for food web studies ...... 51 Abstract ...... 51 2.1. Introduction ...... 52 2.2. Methods ...... 57 2.2.1. Samples and data collection ...... 57 2.2.2. Sample preparation and laboratory analysis ...... 58 2.2.3. Data analysis and statistical framework ...... 59 2.3. Results ...... 61 2.3.1. Oceanography of the ECSI ...... 61 2.3.2. ECSI marine isoscapes ...... 67 2.4. Discussion ...... 71 2.4.1. Ocean connectivity along ECSI ...... 71 2.4.2. ECSI marine isoscapes ...... 74 Chapter 3 Regional differences in the trophic structure of fish communities in the east coast of the South Island of New Zealand ...... 79 Abstract ...... 79 3.1. Introduction ...... 80 3.2. Methods ...... 85 6

3.2.1. Interannual variability in isotopic values within a sea perch population ...... 85 3.2.2. Variability in resource use and trophic level of commercial fish species ...... 85 3.2.3. Differences in community niche breadth between Otago and Kaikoura ...... 90 3.2.4. Amount of coastal primary producers supporting commercial catches ...... 90 3.3. Results ...... 92 3.3.1. Interannual variability in isotopic values within a sea perch population ...... 92 3.3.2. Variability in resource use and trophic level of commercial fish species ...... 92 3.3.3. Differences in community niche breadth between Otago and Kaikoura ...... 96 3.3.4. Amount of coastal primary producers supporting commercial catches ...... 98 3.4. Discussion ...... 99 Chapter 4 Effects of fixatives on stable isotopes of fish muscle tissue: implications for trophic studies on preserved specimens ...... 106 Abstract ...... 106 4.1. Introduction ...... 108 4.2. Methods ...... 113 13 15 4.2.1. Meta-analysis of effects of formalin and alcohol on δ Cm and δ Nm ...... 113 4.2.3. Lipid content analysis...... 114 4.2.4. Stable isotope analysis ...... 115 4.2.5. Amino acid stable isotope analyses ...... 115 4.2.6. Data treatment and statistical framework ...... 116 4.2.7. Applying the results ...... 117 4.3. Results ...... 118 4.3.1. Literature meta-analysis ...... 118 4.3.2. Lipid content and isotopes of control samples ...... 119 13 15 4.3.3. Effects of fixation and preservation on δ Cm and δ Nm ...... 121 4.3.4. Model selection for preservation effects ...... 123 4.3.5. Compound-specific Amino acid analysis ...... 124 4.3.6. Applying the results ...... 126 4.4. Discussion ...... 128 Chapter 5 Shifts in the trophic structure of commercial fishes after New Zealand industrialized fisheries expansion ...... 133 Abstract ...... 133 5.1. Introduction ...... 134 5.2. Methods ...... 141 5.2.1. Sample collection and bulk isotope analyses ...... 141 5.2.2. Changes in community niche width ...... 142 7

5.2.3. Changes in species trophic level and resource use ...... 144 5.2.4. The effects of temperature, prey abundance and fisheries ...... 145 5.2.5. Accounting for the effects of latitude and specimens’ size ...... 146 5.3. Results ...... 147 5.3.1. Changes in community niche width ...... 147 5.3.2. Changes in species trophic level and resource use ...... 149 5.3.3. The effects of temperature, prey abundance and fisheries ...... 152 5.3.4. Accounting for the effects of latitude and specimens’ size ...... 155 5.4. Discussion ...... 156 5.4.1. Mid slope assemblage ...... 157 5.4.2. Slope assemblage ...... 157 5.4.3. Outer shelf assemblage...... 159 5.4.4. Inner shelf assemblage ...... 160 5.4.5. CSIA-AA estimates ...... 161 5.4.6. Caveats ...... 162 5.4.7. Implications for the ecology and management of fish communities ...... 163 Chapter 6 Long-term shifts in the structure and length-frequency of fish communities and the relationship with temperature, fisheries and life history ...... 166 Abstract ...... 166 6.1. Introduction ...... 167 6.2. Methods ...... 171 6.2.1. Data collection ...... 171 6.2.2. Data processing ...... 172 6.2.3. Species assemblage composition and size structure ...... 172 6.2.4. Trophic level composition of the fish community ...... 174 6.2.5. The effect of fisheries and environmental variables ...... 174 6.2.6. The effect of species life history traits ...... 176 6.3. Results ...... 178 6.3.1. Species assemblage composition and size structure ...... 178 6.3.2. Trophic level composition of the fish community ...... 186 6.3.3. The effect of fisheries and environmental variables ...... 187 6.3.4. The effect of species life history traits and habitat use ...... 188 6.4. Discussion ...... 191 General discussion ...... 196 Appendices ...... 203 Appendix A.1.1. Publication associated with this thesis ...... 203

Appendix A.1.2. Data on all Quota Management System fish species used in chapter one. Source: Fisheries New Zealand and Fishbase ...... 204 Appendix A.2.1. Details on the global multimodel approach ...... 205 Appendix A.2.2. Isosurfaces of oceanographic variables along the east coast of the South Island during austral summer of 2017/18 at maximum fluorescence depth. Stations are represented by black dots ...... 207 Appendix A.2.3. Isosurfaces of oceanographic variables along the east coast of the South Island during austral summer of 2018/19 at maximum fluorescence depth. Stations are represented by black dots ...... 208 Appendix A.2.4. Linear relationship between NOx and phosphate (orthophosphate) values from surface water samples collected along the east coast of the South Island during the austral Summers of 2017/18 and 2018/19. Linear equation: NOx = -14.55 + 7.38 * Phosphate, n = 72, r2 = 0.94 and p < 0.0001. Concentrations are given in μg.L- ...... 209 Appendix A.2.5. Maximum fluorescence depth (meters) measured during the 2017/18 (a) and 2018/19 (b) sampling seasons along the east coast of the South Island. Data was displayed using the DIVA interpolation. Stations are represented by black dots ...... 210

Appendix A.2.6. Summary fit of the average of models with lowest AICc (delta < 2) predicting surface δ13C, δ15N and δ18O along the east coast of the South Island. Relative importance is the proportion of occurrence of that variable in the top models ...... 211 Appendix A.2.7. Nonparametric comparisons of δ15N of SPOM at surface for each pair of statistical area using Wilcoxon’s test. * represents p-values lower than 0.05 ...... 212 Appendix A.2.8. Nonparametric comparisons of δ15N of SPOM at MF depth for each pair of statistical area using Wilcoxon’s test. * represents p-values lower than 0.05 .... 213 Appendix A.2.9. Nonparametric comparisons of δ18O of seawater from the whole east coast of the South Island for each pair of depth sampled using Wilcoxon’s test. * represents p-values lower than 0.05 ...... 214 Appendix A.4.1. Publication associated with this thesis ...... 215 Appendix A.4.2. CSIA-AA sample preparation extracted and modified from Sabadel et al., (2016) ...... 216 Appendix A.4.3. Meta-data from published literature on the effects of formalin fixation and formalin preservation on bulk carbon and nitrogen isotopes ...... 222 Appendix A.5.1. Results from general linear models between bulk isotope values, year of collection, specimens’ total length and latitude from chapter 5 data ...... 224 Appendix A.5.2. Results from general linear models between proportion of pelagic basal organic matter supporting a fish’s det (%SPOM), trophic level (TL) and year of collection, specimens’ total length and latitude from chapter 5 data ...... 225 Appendix A.5.3. Sample sizes by species, period and assemblage used in the analyses of bulk isotope data (δ13C and δ15N), trophic parameters (%SPOM and trophic level) and compound specific isotopic analysis of amino acids (CSIA-AA) in chapter 5 ...... 226 Appendix A.6.1. Distribution of annually averaged commercial catch (kg/ha) from all species and fishing methods between 2007 and 2017 reported to Fisheries New Zealand. Values were combined in ten catch intensity groups, with values from 1 (low intensity) to 10 (high intensity). A detailed description of the methods for the mapping of this dataset can be found in appendix A.6.2. Geographic limit of the east coast of the South

Island trawl survey is depicted as the white dashed line. Source: Fisheries New Zealand ...... 227 Appendix A.6.2. Details on the creation of the Ministry of Primary Industries (MPI) fishing intensity map between 2007 and 2017, extracted from the metadata of the Geodatabase Raster Dataset ...... 228 Appendix A.6.3. Life history traits of species used in the present study. Predominantly species of RATs: two saddle rattail (Coelorinchus biclinozonalis), Oblique banded rattail (Coelorinchus aspercephalus) and Bollons rattail (Coelorinchus bollonsi). Source: Fishbase.org ...... 230 References ...... 231

List of Tables

Table 1.1. Description of major habitats of fish species under the Quota Management System in New Zealand waters. Descriptions were compiled from Fishbase and the Food and Agriculture Organization of the United Nations online glossaries ...... 35 Table 2.1. Temperature, salinity and depth distributions of the main water masses along the east coast of the South Island retrieved from the literature ...... 60 13 Table 2.2. Summary fit of the average of models with lowest AICc (delta < 2) predicting δ C, δ15N and δ18O along the east coast of the South Island. Relative importance is the proportion of occurrence of that variable in the top models ...... 68 Table 2.3. Average ± SE of δ15N and δ13C of SPOM samples from both sampling periods, and δ18O of water for different compartments along the ECSI ...... 70 Table 3.1. Maximum total length, assemblage, habitat type, depth distribution and common prey items of the seven commercial fish species analyzed in the present study ..... 84 Table 3.2. Average ± standard error of total length of specimens analyzed in the present study, with number of samples and location of collection. WCR stands for Western Chatham Rise ...... 86 Table 3.3. δ15N and δ13C average ± standard error of the macroalgae species collected from the Otago and Kaikoura coast and respective sample sizes ...... 87 Table 3.4. Linear relationship between head length (HL) and total length for each species analyzed in the present study, with respective summary of fit ...... 90 Table 3.5. General linear model results describing the relationship between δ13C and δ15N values, the proportion of macroalgae supporting each species’ diet (%Macroalgae) and it’s estimated trophic level (TL), with latitude and fishes’ total length as the independent variable for each species. Model parameters are given for total model (latitude and total length effect), while p values are also reported for each effect separately ...... 93 Table 4.1. Comparison of means of lipid content, nitrogen and carbon isotope and carbon isotope after lipid extraction along each species dorsal musculature. Values after ± symbol are one standard deviation. Letters represent different groups within each species (t = 2.13, α = 0.05), different letters represent different group means ..... 121 13 15 13 Table 4.2. Summary of two-way repeated measures ANOVA results for δ Cm, δ Nm, δ CAA 15 and δ NAA. P values marked with * were considered significant ...... 122 Table 4.3. Summary of results applied to raw data analyzed in this work. Raw values were taken after FORMALIN+ALCOHOL2, while correction equations were used on the corrected values column. For trophic levels estimated from AA, corrected values were taken from CONTROL. The difference column is the result of corrected minus raw values. Average values are shown ± one standard deviation for mixing model results and propagation of error for the AA data after Dale et al., (2011). All values are given in ‰ ...... 127 Table 5.1. Maximum total length, habitat, depth distribution, assemblage group and common prey items of the fish species analyzed in the present study ...... 139 Table 5.2. Relationship between total length and head length measurements obtained during the collection of fish specimens for the present study ...... 141

Table 5.3. Average ± standard error of contribution from pelagic production to a fish’s diet (%SPOM) and estimated trophic level (TL) of commercial species collected from the east coast of the South Island during historical and modern periods. t values from PERMANOVA comparison between periods are shown with its respective significance ...... 151 Table 5.4. Average ± standard error of δ13C of essential amino acids and trophic level (TL) 15 15 estimated from δ NGlx - δ NPhe of commercial species collected from the east coast of the South Island during historical and modern periods. t values from PERMANOVA comparison between periods are shown with its respective significance ...... 151 Table 5.5. Results of general linear model between estimated trophic level (TL), the contribution from pelagic production to a fish’s diet (%SPOM), estimated δ15N of 15 the basal organic matter (δ NSource); and Southern Oscillation Index (SOI), water temperature recorded from the University of Otago Portobello Marine Laboratory (PML SST), Munida sightings and New Zealand Marine Trophic Index (NZ MTI), for each fish assemblage ...... 154 Table 6.1. Summary of frequency of occurrence (FO, %), average catch rate (CR, kg.km-2), preferred depth (PD, meters) and average length (AL, cm) for each species in each one of the three periods. AL from 1992-1996 also included values taken in 1991 ...... 179 Table 6.2. Results from between-periods pairwise PERMANOVA and PERMDISP analysis of the community composition for the whole region and at different depth categories ...... 182 Table 6.3. Average catch rate (kg.km-2) ± SE for each species at specific depth groups (30– 100, 100–200 and 200–400 m) for each period analyzed (1992–96, 2007–09 and 2012–18). Periods separated by different letters represent significant different average catch rates within the same depth category (Student’s t-test) ...... 185

List of Figures

Figure 1.1. Map of New Zealand Exclusive Economic Zone (EEZ) divided into Fishery Management Areas (FMAs). Detail is shown for the continental shelf off the east coast of the South Island, where the Fisheries New Zealand trawl surveys and most of the FMA 3 inshore fisheries occur. Source: LINZ data service ...... 28 Figure 1.2. Estimated catches of Quota Management System (QMS), QMS group, non- identified and non-QMS fish species from Sea Around Us (SAU) and reported QMS catches from the Food and Agriculture Organization of the United Nations (FAO) and Fisheries New Zealand (FNZ) for the 65 years from 1950 to 2014. The QMS group and non-identified species catches are largely estimated discards ..... 37 Figure 1.3. Trends of Quota Management System (QMS) fish species catches from New Zealand waters from 1931 to 2015, split by marine habitats (a), the number of species landed (b), and Bray-Curtis community similarity and community similarity derivates among years (c). Data source: Fisheries New Zealand and the Food and Agriculture Organization of the United Nations. Habitat type data: Fishbase.org ...... 39 Figure 1.4. Relationships between Marine Trophic Index (MTI) of the commercial catch of Quota Management System fish species among years (a) and with trends in the total commercial catch (b) in New Zealand waters for the 84 years from 1931 to 2015 for all species (black line) and excluding hoki (grey line). Lines on panel (a) are fitted locally estimated scatterplot smoothing regression with a 95% confidence interval (grey). Data source: Fisheries New Zealand (FNZ) and Food and Agriculture Organization of the United Nations (FAO). Trophic level data: Fishbase.org ...... 40 Figure 1.5. Marine Trophic Index (MTI) calculated by year for fish species from commercial landings and Fisheries New Zealand trawl surveys in the east coast of the South Island since 1991. Graphs were plotted using data from 20 species caught from within the Fishery Management Area 3 (FMA 3) and surrounding FMAs (plot a: y = 2.98 + 0.25x, r2 = 0.009) and the 10 species caught inside the FMA 3 in both datasets (plot b: y = 0.33 + 0.91x, r2 = 0.16). Linear fit is represented by the black dotted line, with 95% confidence intervals in gray and 1:1 relationship plotted as a continuous line ...... 40 Figure 1.6. Relationships between the percentage of average catch since 1991 by year from reported landings and Fisheries New Zealand trawl surveys for ling (plot a: y = 12.16 + 0.91x, r2 = 0.2), sea perch (plot b: y = 83.53 + 0.18x, r2 = 0.03), and spiny dogfish (plot c: y = 97.53 + 0.03x, r2 = 0.0003) inside the Fishery Management Area 3. Linear fit is represented by the black dotted line, with 95% confidence intervals in gray and 1:1 relationship plotted as a continuous line ...... 41 Figure 1.7. Average trophic level of the commercial fishing catches per year from the West Coast of the United States (a), East Brazilian Shelf (b) and North Sea (c). Grey and black lines represent periods of increase and decrease of MTI values, respectively. Shaded regions represent the 95% confidence interval of the smoothed line. Data source: searoundus.org ...... 42 Figure 1.8. Weighted maximum length of Quota Management System fish species that comprised catches between 1931 and 2015. Yearly lengths were weighted by landings weight. Data source: Fisheries New Zealand and the Food and

Agriculture Organization of the United Nations. Maximum length data: Fishbase.org ...... 44 Figure 2.1. Map of New Zealand with the most important surface features influencing processes on the east coast of the South Island, based on Chiswell et al., (2015). STW (Subtropical Water), dUC (d’Urville Current), STF (Subtropical Front), mSTW (modified Subtropical Water), NW (Neritic Water) and SASW (Subantarctic Surface Water). The Southland Current (SC) and the region where the transport of mixed water masses is known as the Southland Front (SF) are also shown. Arrows inside orange lines represent the main flow southward from the Tasman Front. Light gray shades represent depths from 10 to 1000 meters ...... 55 Figure 2.2. Oceanographic transects during 2017/18 and 2018/19 cruises with T-S diagram using CTD casts with main water masses represented. NW (Neritic water), mSTW (modified Subtropical Water), SASW (Subantarctic Surface Water), SAMW (Subantarctic Mode Water), AAIW4 (Antarctic Intermediate Water with South formation) and AAIW1&3 (Antarctic Intermediate Water with North formation). Otago8 and Kaikoura4 casts from 2017/18 are not shown ...... 57 Figure 2.3. T-S diagram for CTD vertical transects along the east coast of the South Island during 2017/18 (a) and 2018/19 (b) sampling seasons, with indication of the different water masses. NW (Neritic water), mSTW (modified Subtropical Water), SASW (Subantarctic Surface Water), SAMW (Subantarctic Mode Water), AAIW4 (Antarctic Intermediate Water with South formation) and AAIW1&3 (Antarctic Intermediate Water with North formation)...... 62 Figure 2.4. Temperature profile and its deviation at depth (deltaT) of a subset of the CTD casts from offshore Otago (OT8) and Akaroa (AK4 and AK5) from 2017/18 and 2018/19 sampling seasons. Details to the temperature thermostad indicative of SAMW are shown between the dashed lines ...... 63 Figure 2.5. Isosurfaces of oceanographic variables along the east coast of the South Island during austral summer 2017/18 at the sea surface ...... 65 Figure 2.6. Isosurfaces of oceanographic variables along the east coast of the South Island during austral summer 2018/19 at the sea surface ...... 66 Figure 2.7. Average integration of fluorescence values between surface and 150 meters for each transect. Bar colors correspond to data collected during 2017/18 and 2018/19 sampling season, with its respective standard errors ...... 67 Figure 2.8. TS plot with colored labels representing δ18O values of samples collected from surface to 1000 meters from the whole ECSI. Dots inside the dashed polygon represent samples taken below 150m ...... 76 Figure 3.1. Map of the New Zealand marine environment with fisheries management statistical areas referenced in the present study as Kaikoura (18), Western Chatham Rise (20), Akaroa (22) and Otago (24). White represents New Zealand landmass and grey shades are identifying depths between 0 and 150, 150 to 1000, 1000 to 2000 and deeper than 2000 meters ...... 83 Figure 3.2. Average ± standard error of the contribution of organic matter derived from macroalgae to a fish’s diet (%Macroalgae) and the trophic level (TL) for each species collected from Otago (24), Akaroa (22), Western Chatham Rise (20) and Kaikoura (18). Blank columns correspond to missing data. The variance of the total length from individuals collected at regions did not differ for any species (Wilcoxon’s test, p > 0.05), except for tarakihi and red cod collected from region

22, which were significantly smaller and larger than for the other regions, respectively ...... 95 Figure 3.3. Average trophic level (TL) and relative contribution of organic matter derived from macroalgae (%Macroalgae) to a fish’s diet (0 to 1) for an exploited fish community sampled from Kaikoura (grey symbols) and Otago (black symbols). Error bars represent ± 1 standard error ...... 96 Figure 3.4. Isotopic niche space represented by δ13C and δ15N from fish communities sampled from Otago (black) and Kaikoura (red) circled with their respective Bayesian inferred ellipses. Dotted lines represent the minimum convex hull of each community ...... 97

Figure 3.5. Standard ellipse areas corrected for small sample sizes (SEAc) from isotopic space represented by δ13C and δ15N from fish communities from Otago and Kaikoura. Fish species were split into inner shelf (IS), outer shelf (OS) and slope (SP) assemblages. Black dots are the modes of the Bayesian SEAc values with 50, 75 and 95% credible intervals in grey bars. Red crosses are the maximum likelihood estimates of SEAc ...... 97

Figure 3.6. Standard ellipse areas corrected for small sample sizes (SEAc) from isotopic space represented by δ13C and δ15N from the outer shelf fish assemblage from Otago and Kaikoura represented by barracouta (BAR), red cod (RCO) and tarakihi (TAR). Black dots are the modes of the Bayesian SEA values with 50, 75 and 95% credible intervals in grey bars. Red crosses are the maximum likelihood estimates of SEAc ...... 98 Figure 3.7. Commercial estimated reported landings per unit of effort (kg x fishing events-1) between 1990 and 2017 for Otago and Kaikoura regions for five different species. Source: Fisheries New Zealand ...... 105 Figure 4.1. Shifts in isotope values of fish muscle tissue with time after formalin fixation (a) and ethanol preservation (b) experiments retrieved from literature. Equation and R2 are shown on top of each graph. Values reported in ‰ ...... 119 Figure 4.2. Relationship between lipid concentration in percentage and carbon to nitrogen ratio in fish muscle tissue from samples of O. tshawytscha, P. colias and S. brama, F(1,28) = 0.923, p < 0.001...... 120 13 15 Figure 4.3. δ Cm and δ Nm average per treatment for each one of the fish species (n = 10). Treatments consisted of dried tissues (time 0), after one month under formalin fixation (FORMALIN), under three months (FORMALIN+ALCOHOL) and one year of alcohol preservation (FORMALIN+ALCOHOL2) after formalin treatments. Error bars represent one standard deviation. Values reported in ‰ . 122 13 15 Figure 4.4. Relationship between predicted and actual δ Cm and δ Nm values from preserved 13 fish muscle (n = 60), using the empirical equation developed in this study. δ Cm values are corrected to lipid-free values. Values for FORMALIN+ETHANOL and FORMALIN+ETHANOL2 were included. Dashed line around the relationship represents 95% confidence interval. The horizontal line represents the mean of the data group. Data from O. tshawytscha (OT), P. colias (PC) and S. brama (SB) were included ...... 123 13 Figure 4.5. δ CAA values of 11 amino acids of muscle tissue from S. brama and P. colias at CONTROL (0 months), after FORMALIN+ETHANOL (4 months) and FORMALIN+ETHANOL2 (12 months) treatments. Values reported in ‰ ...... 125

15 Figure 4.6. δ NAA values of 11 amino acids of muscle tissue from S. brama and P. colias at CONTROL (0 months), after FORMALIN+ETHANOL (4 months) and FORMALIN+ETHANOL2 (12 months) treatments. Values reported in ‰ ...... 126 Figure 5.1. Map of New Zealand marine environment with the location of historical and modern fish samples analyzed in the present study. White represents New Zealand landmass and grey shades identify depth bands of 0 to 500, 500 to 1000, 1000 to 2000 and deeper than 2000 meters ...... 138 Figure 5.2. Isotopic niche space represented by corrected δ13C and δ15N values from fish communities of the east coast of the South Island sampled at different periods. Black and red circles represent specimens from historical and modern periods, respectively, with solid lines representing Bayesian 95% ellipses and dotted lines representing the minimum convex hull of each community ...... 147

Figure 5.3. Standard ellipse areas corrected for small samples sizes (SEAc) from isotopic space represented by δ13C and δ15N values from fish communities of the east coast of the South Island sampled at different periods: historical (1; black bar) and modern (2; grey bar). Fish species were split into assemblages, representing inner shelf (IS), outer shelf (OT), slope (SP) and mid slope (MS) habitats. Black dots represent the mode value, with their respective 50, 75 and 95% credible intervals. Red crosses represent the calculated maximum likelihood value ...... 148

Figure 5.4. Standard ellipse areas corrected for small samples sizes (SEAc) from isotopic space represented by δ13C and δ15N values from outer shelf (a) and slope (b) fish assemblages collected at different periods: historical (1; black bar) and modern (2; grey bar). Black dots represent the mode value, with their respective 50, 75 and 95% credible intervals. Red crosses represent the calculated maximum likelihood value ...... 148 Figure 5.5. Bayesian inferred δ13C range for the outer shelf assemblage (a) and mean distance to the centroid (b) and δ13C range (c) for the slope assemblages of historical and modern samples. Black dots represent the mode value, with their respective 50, 75 and 95% credible intervals. Red crosses represent the calculated maximum likelihood value ...... 149 Figure 5.6. Average of estimated trophic level vs contribution from pelagic production to a fish’s diet (%SPOM) for specimens collected from the east coast of the South Island of New Zealand in historical (H) and modern (M) periods, with respective standard errors. Arrows point from historical to modern species values. Inner shelf (a), outer shelf (b), slope and mid slope (c) assemblages were plotted separately ...... 150 Figure 5.7. Linear regressions between the average contribution from pelagic production to a fish’s diet (%SPOM) and δ13C values of essential amino acids (a) and trophic level estimates from bulk isotopes and compound-specific isotopic analysis of amino acids (CSIA-AA) (b). Grey and black symbols represent historical and modern specimens, respectively, with their linear regression lines and standard error bars ...... 152 Figure 5.8. Yearly deviation of the average of Munida sightings, Southern Oscillation Index (SOI), the temperature recorded from University of Otago Portobello Marine lab (PML SST) and New Zealand Marine Trophic Index (NZ MTI) from 1953 to 2018 ...... 153 Figure 6.1. Map with the location of trawling stations (30–400 m) inside the Canterbury Bight used in the present study divided by time periods. White represents land, while

shades of grey represent depth of 0–100, 100–200 200–400 and deeper than 400 meters ...... 170 Figure 6.2. Dendrograms from hierarchical cluster analysis of species catch rates in the whole study region for the different year periods of 1992-1996, 2007-2009 and 2012- 2018. Clusters were based on Bray-Curtis similarities of catch rates (kg.km-2) between species. Species codes can be found in Table 6.1 and Appendix A.6.3 180 Figure 6.3. Non-metric multi-dimensional scaling plots of Bray-Curtis similarity resemblance matrices between species for each period analyzed. Groups identified in Figure 6.2 are depicted inside shapes with their respective letter names. Species codes can be found in Table 6.1 and Appendix A.6.3 ...... 181 Figure 6.4. Range and preferred depth (horizontal line) values between periods analyzed for 15 fish species with the largest differences between periods. Crosses mark the average between minimum (5% of cumulative catch), maximum (95% of cumulative catch) and preferred depth (50% of cumulative catch). Based on catch rates (kg.km-2) ...... 183 Figure 6.5. Average scaled population numbers of elephant fish (a) and dark ghost shark (b) for the three time periods analyzed, 1991-1996, 2007-2009 and 2012-2018 ...... 186 Figure 6.6. Weighted average trophic level of each tow between 1992 and 2018 for different depth categories. Size of the dots represent total catch rate in that trawl station for 31 species of fish combined, except rattails, ranging from 11 to 36885 kg.km-2. Lines are linear fits with shaded 95% confidence interval. The averages ± standard error of the weighted average trophic levels for 1992–96, 2007–09 and 2012–18 were of 4.07 ± 0.01, 3.97 ± 0.01 and 3.93 ± 0.01 for 30–100 m, 4.09 ± 0.01, 3.94 ± 0.02 and 3.88 ± 0.01 for 100–200 m and 4.02 ± 0.03, 4 ± 0.04 and 3.91 ± 0.03 for 200–400 m, respectively ...... 187 Figure 6.7. Boxplot of adjusted r2 values from distance based linear models and redundancy analysis of fishing and environmental variables explaining the long-term variability in catch rates (a) and length-frequency (b) of commercial fish species along the east coast of the South Island. Outliers represented as dots. Note the different y-axes scales ...... 188 Figure 6.8. Explanatory power of temperature, sediment and fishing on the variation of relative abundance of species in different mean preferred temperature groups, estimated through the adjusted r2 from distance based linear models ...... 189 Figure 6.9. Explanatory power of fishing, temperature and their interaction on the variation in the length-frequency of species in different trophic level categories, estimated through the adjusted r2 from redundancy analysis ...... 190

General introduction

The exploitation and use of aquatic resources can be traced through the history of mankind. Prehistorical evidence indicates the consumption of marine fish 40,000 years ago (Hu et al., 2009; Muñoz & Casadevall, 1997; O’Connor et al., 2011) and freshwater organisms as early as 1.95 Ma (Archer et al., 2014). The use of both coastal and pelagic resources was linked with the development of tools and ability to catch fish, as well as the need for increased protein resources with the formation of modern human groups during the late stone age (Klein et al., 2004; O’Connor et al., 2011). With the advent of centralized societies, fish commodities became important to most ancient civilizations, including Greek (1500 BC), Egyptian (1300 BC) and Mongolian (1306). For these groups, the importance of fish products was commonly registered in painted artifacts, with rich details given to larger, most important and abundant fish species (Begossi & Caires, 2015).

More recently, at the end of the 19th and beginning of the 20th century, fluctuations in commercial fish catches in Europe started to emerge as an issue for the fishing industry and scientists. Early studies explained this variation solely as a result of movements of adult fishes (Heincke, 1898). The variability of commercial landings was thought to be due to the distribution of the population and not overall production (Solemdal & Sinclair, 1989), and fishery resources were considered to be virtually unlimited (Huxley, 1881). This paradigm began to change following novel studies demonstrating the importance of population age structure and early life stages to fish production (Dahl, 1907; Hjort, 1914). Throughout the 20th century, it became evident that studies on population dynamics, stock assessment and oceanography were essential to inform sustainable catch yields and ensure productivity of important fisheries (Nature Publishing Group, 1943). At that time most studies focused on Atlantic herring (Clupea harengus, Clupeidae) and Atlantic cod (Gadus morhua, Gadidae), in response to high demand for fishing products, especially cured herring from Western Europe (Newland, 1999).

Late in the 20th century, studies were directed towards a holistic understanding of the marine environment, with the idea that fisheries could be managed using the knowledge of ecosystem processes (Fluharty et al., 1999). The concept of ecosystem-based management (EBM), a holistic approach considering the ecosystem as a whole when applying management to natural resources (Fluharty et al., 1999), was first examined during the establishment of a management policy in the Great Lakes of North America in the 1970s (Caldwell, 1970). Although almost half a century old, some of Caldwell’s thoughts sound contemporary: “…a dollar crisis or a Far Eastern war may offer politically defensible but ecologically invalid

18 arguments for delaying efforts to save the Great Lakes from death by pollution. Today there may be higher political priorities, but, ecologically, tomorrow may be too late.” In the same article, Caldwell gave one of the first definitions of ecosystem-based management, which is described as having a mission to identify, protect and manage the ecosystem, but also preserving human welfare. The importance of suitable and healthy habitat to fisheries sustainability is highlighted in the primary law governing fisheries management in United States, first passed in 1976 (Gutiérrez, 2007), but it would take a couple of decades for it to be broadly implemented in aquatic ecosystems. This transition happened in part because of the later recognition of depletion of fisheries resources and the need for effective management (Fluharty et al., 1999), here called ecosystem-based fishery management. Since then, EBM has been slowly applied globally, with good results towards ecosystem plans (Marshall et al., 2018) and implementation of milestones (NOAA, 2019). Likewise, research grants supporting investigations and implementations of EBM were recently made available in New Zealand (MBIE, 2014).

Marine communities are subject to natural (e.g. El Niño and Pacific Decadal Oscillations) and anthropogenic (climate change, habitat alteration, sedimentation) environmental changes, which can affect habitat suitability and prey abundance as well as influencing life history traits, distributions, and connectivity between populations (Durante et al., 2018; Valdés et al., 2008). Harvest can have similar impacts on fish communities, altering size structure, growth rates, age at maturity and modifying trophic structure (Blanchard et al., 2011; Durant & Hjermann, 2017; Fenberg & Roy, 2008; Frank & Leggett, 1994; Hendry et al., 2008; Palkovacs et al., 2012; Worm, 2005). Moreover, overexploitation can have faster impacts on the structure of fish communities than any other stressor (Jennings et al., 1999; Stenseth & Dunlop, 2009), reported being 50% faster than other human impacts and 300% quicker than those occurring in natural undisturbed systems (Darimont et al., 2009). The result of such changes have the potential to alter how species interact with the environment around them and each other (Audzijonyte et al., 2013; Fenberg & Roy, 2008). For example, biomass declines of a high trophic level species can have the effect of increasing the species' trophic level, since less biomass would be needed to support the community, allowing fishes to feed on higher trophic level prey species (Pinkerton et al., 2015b). The indirect effects of fishing, especially trawling, can also have a deleterious effect on non-target species and bottom communities, contributing to habitat degradation and reduced fish prey availability (Clark et al., 2016; Jennings et al., 2001; Tuck et al., 2017). Moreover, shifts in the food web structure of fish communities can have an impact on ecosystem-wide processes

19 and services, such as nutrient distribution and algal blooms (Brazner et al., 2011; Casini et al., 2008).

Shifts in the average trophic level of global landings of commercial fisheries have been used to identify dramatic changes in the composition of exploited communities over the last century. The concept known as “fishing down marine food webs” describes a decrease of the mean trophic level of fish landings over time (Pauly et al., 1998). The effect is a result of reduced abundance of top predators and increased landings of lower trophic level species due to overfishing and catch selectivity. Abundance of lower trophic level species may be also increased by the release of predatory pressure on prey species (Caddy & Garibaldi, 2000). As indicated by Caddy and Garibaldi (2000), reduction in mean trophic level of commercial catches can be also a result of environmental changes, market preference and available technology, therefore such relationships should be evaluated carefully. Shifts of fishing gear can be quite rapid, substituting techniques completely in a scale of decades and affecting the ecosystems in different ways over time (Greenstreet et al., 1999). This phenomenon is now well recognized in the literature and when applied with the right guidelines it can be used to identify impacted and overexploited fish communities (Caddy, 1998).

Because of the effects of long-term anthropogenic activities on the exploited fish communities and lack of past ecological data, managers are unable to set management goals based on the pre-impacted system in many regions of the world. For example, current management policies for Atlantic Cod on Canada’s Scotian Shelf references the stock biomass at 0.3% of pre-industrialized fisheries (Rosenberg et al., 2005). The shifting baseline syndrome has been identified as the process of accepting new fisheries baselines (abundance, size, species composition and trophic structure) for each new generation of fisheries scientists (Pauly, 1995). This syndrome results in a lack of informed reference points and promotes scientific and management policies adjustment to new ecological states with implications to the exploited community (McAfee et al., 2020). Nevertheless, the environment necessary for a fish to feed, grow and reproduce, including its associated food web structure, should be preserved for sustaining fishery’s yields (Thrush & Dayton, 2010). Therefore, it is important to develop and apply new methodologies that aim to uncover the ecological history of the exploited communities to fully implement EBM approaches.

Trophic level estimates for almost all New Zealand QMS fishes are available on Fishbase.org (Froese & Pauly, 2017), a global dataset of fish species, including information on taxonomy, life history, diet, genetics, and ecology. Trophic level estimates are generated from diet studies or ecological models, but over 60% of the diet studies for New Zealand’s

QMS species were made in Australia or outside New Zealand, indicating a lack of region- specific knowledge of the ecology of many important commercial species of fish in New Zealand waters. Isotopic analysis can be used to resolve temporal and spatial variability of trophic relationships, providing a way to estimate the trophic level and resource use of species of interest, including large and highly mobile species (Dale et al., 2011; Hobson et al., 1994; Popp et al., 2007). Carbon and nitrogen isotopes from a fish’s muscle tissue, expressed as a deviation from a standard δ13C and δ15N, are the most frequently used for tracking organic matter source use and estimate the trophic level of animals (Ramos & González-Solís, 2012), providing long-term information of a fish’s diet, ranging from several weeks to months (Fry, 2006; Mont’Alverne et al., 2016; Skinner et al., 2017). While δ15N in bulk muscle tissue increases substantially when organic matter is transferred between consumers, making it a good indicator of a specimen’s trophic level, δ13C presents small trophic fractionation and thus better indicates reliance on basal sources of organic matter (Post, 2002). These metrics can be used to resolve the effects of anthropogenic impacts on the interactions between organisms and the environment (Bearhop et al., 2004; Hussey et al., 2014). The use of trophic level estimates from different areas is a common practice during large scale ecological analysis. These approaches facilitate the study of large species throughout the vast seascape, but fish diets and trophic interactions are known to vary in space and time (Polis et al., 1997; Wainright et al., 1993). To estimate these parameters, one needs to know the δ13C and δ15N values of the primary producers supporting that specific food web, which also vary spatially and temporally (Kurle & McWhorter, 2017; Trueman et al., 2017). Isotopic maps of marine primary producers, or marine isoscapes, have been used in previous studies to estimate trophic structure, movements and delineating stocks of fish species (MacKenzie et al., 2014; Wing et al., 2012). Furthermore, isotopic analysis of specific amino acids (AA) from a single muscle tissue sample has been used to remove the need for these isotopic baselines. While δ15N of amino acids that fractionates along the food web (trophic AA) represents a fish’s trophic level, other amino acids present no fractionation (source AA) and indicate δ15N values from the basal organic matter supporting its diet (Sabadel et al., 2016; Whiteman et al., 2019). δ13C of amino acids can be used in the same fashion, with some AA presenting fractionation during its synthesis in the fish’s body (non-essential AA) and others retaining δ13C from its basal organic source (essential AA) (Larsen et al., 2013; Whiteman et al., 2019). Therefore, the combined use of isotopic values from trophic, source, essential and non-essential amino acids can provide information regarding a fish’s trophic level and resource use without complete knowledge of isotopic signatures of basal organic matter throughout the seascape.

Although some studies in New Zealand have investigated the trophic dynamics of important fish species and communities (Pinkerton, 2011; Pinkerton et al., 2008), few have addressed questions of shifts in these dynamics over time (MacDiarmid et al., 2016; Pinkerton et al., 2015b) and none of these have applied direct trophic level estimates for the species of interest. The aim of this thesis was to investigate long-term shifts in the trophic structure of the exploited fish community on the east coast of the South Island, in order to reconstruct ecological baselines for the management of these resources. Here long-term commercial fisheries data were analyzed to examine possible effects of fishing pressure in the composition of the fish community in New Zealand and to better understand periods of large impacts due to fisheries expansion. A method to access isotopic information from preserved fish muscle tissue from museum collections was developed, to reconstruct ecological information from past communities under different fishing pressures. The natural distribution of δ13C and δ15N in the marine suspended particulate organic matter and its relationship with the local oceanography was investigated. δ13C and δ15N values from the most abundant macroalgae species were also investigated. The results provided isotopic baselines of pelagic and coastal primary producers for food web analysis along the whole study region. Short term (up to 3 years) and long-term (89 years) variation in the trophic structure of the exploited fish community was investigated using bulk and amino acid stable isotopic techniques. To investigate the drivers of those changes, results were compared with trends of fishing history, prey abundance and climate variables. Using data from the east coast of the South Island trawl surveys (MacGibbon et al., 2019), patterns of changes in the abundance and size distribution of the commercial fish species in the past 28 years were also analyzed. The present thesis is the first study to provide a comprehensive analysis of shifts in the trophic structure and composition of commercially important marine fish communities, considering well resolved historical baselines and linking the results to the history of fish exploitation in New Zealand. The results of the present study provide important insights into how the trophic ecology of fish communities can change in space and time due to natural and anthropogenic stressors.

“To facilitate the peer-reviewed publication of the findings presented in this thesis, each chapter was written as a stand-alone scientific paper, therefore there is some information overlap used throughout the thesis, including the definition of terms and methodology. Chapter one and four have been published in peer-reviewed scientific journals, which can be found in Appendix A.1.1 and A.4.1, respectively.”

Chapter 1 Shifting trophic architecture of marine fisheries in New Zealand: implications for guiding effective ecosystem-based management

“This chapter has been published in the peer-reviewed scientific journal “Fish and Fisheries” on the 22nd of April 2020, with contributions from Michael Peter Beentjes and Stephen Richard Wing (Appendix A.1.1).”

Abstract

New Zealand has led the world in the restoration of marine fisheries since the introduction of the Quota Management System in 1986, but challenges remain in minimizing the ecosystem-level effects of industrialized fishing. Existing long-term fisheries datasets from 1931 to 2015 were analyzed to resolve trends in important ecological properties of major exploited fish communities in New Zealand. Increases in community dissimilarities of catch composition in 1931 and 1972, followed by increasing total landings, highlight major expansions of fishing grounds and exploited species during these periods. Mirroring global patterns, the remarkable rise in fishing power, demand and generation of new markets in New Zealand have all contributed to this expansion. Marine Trophic Index (MTI) values of landings have decreased together with total catch after the year 2000, reflecting smaller catches with a higher composition of lower trophic level species in recent years. Differences in the relative abundance of species estimated between fisheries dependent and fisheries independent data were observed, where high-value species displayed better agreement in relative abundance between datasets. Despite being under a quota management system, temporal development of MTI values relative to the timing of the industrial expansion of fisheries was remarkably similar to those observed in the North Sea and Brazil, with a single expansion and decline. MTI values presented better long-term stability in the United States fisheries analyzed. Analysis of long-term data and the development of well-resolved ecological baselines will be the first step towards applying EBM to New Zealand fisheries, in keeping with global trends in fisheries management.

1.1. Introduction

1.1.1. New Zealand and global fisheries

Variability in the abundance of commercial catches of fish has been studied for centuries (Huxley, 1881), especially in highly productive and highly populated regions (Heincke, 1898; Newland, 1999). Globally, the abundance of fishing stocks has displayed declines over the past several decades (Worm et al., 2009), with changes in commercial fish assemblages identified for important fishing grounds in northwestern Europe (Moyes & Magurran, 2019), the northeastern Pacific (Levin et al., 2006) and on the Newfoundland/Labrador Shelf (Gomes et al., 1995). Long-term impacts of fisheries have included overexploitation of target species, bycatch, and habitat destruction, fundamentally altering marine ecosystems worldwide (Hinz et al., 2017; Jackson, 2001; MRAG & IEEP, 2007; Pauly et al., 1998; Worm et al., 2009). As an example, North Sea demersal fish stocks that had shown signs of recovery during World War II in Europe (1939-1945) were already declining again by 1947 (Beverton & Holt, 1993). The need for more effective management systems, or more sustainable fishing practices in the Northeast Atlantic, stimulated studies on maximum sustainable yield (MSY), which became a key concept in fisheries management (Graham, 1934; Russell, 1931; Schaefer, 1991), enshrined in maritime law.

Independent of current catch shares and regulations, fisheries are largely managed as a series of single-species stocks in many countries, without adequate consideration of trophic interactions and ecological complexity across the seascape. Consequently, single-species management disempowers managers from accounting for vital interactions within the marine community networks that drive community dynamics and food web structure. For example, European fleets fishing in the North Sea use nonselective gear that often results in large amounts of bycatch (Cotter et al., 2002), therefore common metrics used in single-species management, such as MSY are not suitable as biological reference points (Mackinson et al., 2009). While the concept of MSY remains enshrined in many of the laws that govern fisheries exploitation, its usefulness has been reduced to the recognition of “hard limits” to avoid over-exploitation of single species (Mace, 2001), and it has been proven repeatedly to be unsuitable for multi-species management (Eddy et al., 2016; Larkin, 1977; Walters et al., 2005).

Among the wide array of environmental stressors that affect the ecology of marine animals, fisheries mortality has resulted in the fastest observed changes in species composition, trophic structure and population vital rates within marine communities globally

(Fenberg & Roy, 2008). Here, the consequences of “fishing down marine food webs” are broadly recognized in the literature (Fung et al., 2015; Pauly et al., 1998; Shannon et al., 2014) and are often associated with dramatic changes in community assemblages and trophic interactions within and between species (Myers et al., 2007). Based on these observations, the alternative of considering system-level standards and biological reference points that incorporate important trophic relationships has been posed as an important step towards the more holistic approaches espoused in ecosystem-based management (Thrush & Dayton, 2010). Accordingly, understanding the resolution of, and the drivers of long-term changes in the trophic structure of exploited communities are critical for developing ecosystem-level standards to support ecosystem-based management (Pauly et al., 2003; Pikitch et al., 2004), both within New Zealand and globally.

New Zealand has the ninth-largest Exclusive Economic Zone (EEZ) in the world (Migiro, 2018), with an area of 4.3 million square kilometers, or 15.4 times its landmass. Since 1978 when the EEZ was declared, New Zealand has held sovereignty to manage exploitation, exploration and conservation of biodiversity in the EEZ and an extended extensive area for which it has rights and responsibilities including the Ross Sea dependency. Hence management decisions made in New Zealand can have a broad impact on global marine systems. According to the Food and Agriculture Organization of the United Nations (FAO), between 2013 and 2015, the New Zealand fishing industry was responsible for 0.4% of the global capture production of aquatic organisms, a small fraction compared to many countries with larger global fishing fleets (Marchal et al., 2016). For example, Halpern et al., (2008) estimated that 41% of the global ocean was moderately or heavily impacted by anthropogenic stressors, while New Zealand falls within the low-impact category.

New Zealand has demonstrated aspirations for sustainable management of fisheries starting with its first fishery legislation in 1866 and continuing through the implementation of the Quota Management System (QMS) in 1986 (Fisheries New Zealand, 2019). As a corollary, food harvested from the sea in New Zealand has enjoyed a global reputation for being safe and sustainably produced (Marine Stewardship Council, 2013, 2018). Nevertheless increased pressures for intensification of harvest practice, lack of adequate data for stock assessments, unwanted consequences of some management policies on the loss of biodiversity and overharvest have challenged this reputation in recent years as the environmental footprints of fisheries have become more apparent (Simmons et al., 2016; Slooten et al., 2017; Torkington, 2016). To more fully address the detrimental effects of bycatch, discards, habitat degradation and loss of trophic complexity on marine communities,

New Zealand has moved towards incorporating the principles of ecosystem-based management into fisheries policy. The application of ecosystem-based management to fisheries in New Zealand became a clear goal in 2018, when the Minister of Fisheries, Hon. Stuart Nash stated during the Māori Fisheries Conference that “New Zealand has committed to moving towards an ecosystem approach to fisheries management by 2020, as it is one of our targets under the Convention on Biological Diversity”. Tantamount to the effective application of this approach is a well-resolved understanding of how regulatory measures and the trophic structure of marine fish communities have changed during the development of industrialized fishing in New Zealand.

According to the Food and Agriculture Organization of the United Nations (FAO), in 2016, 59.6 million people were engaged in either fishery or aquaculture activities. In the same year, total fish production for both activities peaked at 170.9 million tonnes, increasing more than 800% since 1950. This growth did not only follow the nutritional needs of growing populations, but it also provided products for a higher per capita consumption over the past decades. Today, the world’s per capita fish supply is around 19.8 kg per year, 212% of the value in 1960, roughly half provided from farms and half through wild-caught fisheries, 85% coming from marine environments (FAO, 2020). With consumption increasing and a global stagnation of wild fisheries production in the 1990s, it became clear to the fishing industry and governance that the sustainability and productivity of fisheries needed to be addressed, especially when the majority of the production comes from relatively few species (Monteiro, 2017). Accordingly, one of the international trends is to shift to considering multispecies assemblages in an ecosystem approach to management (Worm et al., 2009).

Here an account of how management policies have evolved during the modern history of fisheries exploitation in New Zealand was provided, along with analyses of time series of total fish landings by habitat, and changes in the trophic level and species composition of the catch over time. As a case study with which to compare to international trends, New Zealand has several interesting characteristics, including isolated stocks with no shared international borders, early development of a quota-based system and a relatively short history of industrialized fishing. Here FAO, Fisheries New Zealand (FNZ, formerly part of Ministry for Primary Industries) and Sea Around Us (SAU, Simmons et al., 2016) databases were used to compare different sources of information on the catch history of New Zealand’s fisheries during the progression of changes in management policy. Key features of these datasets and analyses include the resolution of a time series of the discard rate in fisheries, the changing composition of exploited species landings and the expanding scope of exploited habitats

26 represented in the fisheries catch. Species composition of fisheries landings was used to estimate the average of trophic level per-unit-biomass, given the assumptions of static trophic levels for each species. In order to place the findings of the present study in a more global context, the trophic level of the commercial catch from fishing grounds in the Americas and Europe were compared to the ones found for New Zealand.

Finally, to investigates the limitations of these catch record data, the trophic level per- unit-biomass and percentage of average catch by year between fisheries-dependent and fisheries-independent data from the east coast of the South Island were compared (ECSI, Figure 1.1). The ECSI is an important region for New Zealand’s commercial and recreational fisheries, managed as a single Fishery Management Area (FMA 3, Figure 1.1) for most QMS species. The region has a coastline of 1,490 km, a total area of 205,000 km2 and comprises two of the three biggest fishery ports in New Zealand (Lyttelton and Timaru) (Fisheries New Zealand, 2016). The continental shelf along the east coast is heterogeneous, comprised of rocky reefs in the north and south, and a sandy gently sloping bottom extending approximately 40 km offshore along Canterbury Bight, allowing for the development of an extensive inshore trawl fishery. As a consequence, FMA 3 has been subject to the highest trawling efforts in New Zealand waters since the expansion of industrialized fishing, measured on Catch Effort Landing Returns (CELRs) (Baird et al., 2011). Although important for both commercial and recreational fisheries, the inshore fisheries have been poorly monitored and have had low coverage of observers, discard and bycatch assessment levels, increasing the prevalence of unknown stock status for many species (Mace et al., 2014). Despite these inadequacies in reporting, standardized scientific trawl surveys have taken place since 1991 by Fisheries New Zealand, providing an opportunity to compare fisheries- independent data with reported fishery landings (MacGibbon et al., 2019).

1.1.2. A brief overview of the fishery and management history in New Zealand

From the earliest development of island-based societies in the Pacific Ocean, the economic, cultural and spiritual importance of the marine environment have been blended, resulting in communities that guarded and valued the seascape and its resources. For example, in Polynesian culture, a wealth of knowledge and lore has been intimately linked with the sea, across which people voyaged to settle in New Zealand, at least 700 years ago 28

(Irwin & Walrond, 2016). The myth describing the origin of New Zealand (Aotearoa) is told to be a result of a fishing trip taken by the demigod Māui-pōtiki, where the South Island represents the fishing canoe (Te Waka-a-Māui) and the North Island the large fish caught during the voyage (Te Ika-a-Māui, Johnson and Haworth, 2004).

With the growth of the Māori population in New Zealand and the rising scarceness of terrestrial animals such as moa, smaller birds and kiore (Pacific rats), marine organisms provided the main protein sources (Haami, 2008; Johnson & Haworth, 2004). Māori had access to a wide range of different aquatic environments and were able to catch kaimoana (food from the sea) with an equally wide range of techniques. Flounder/pātiki (Pleuronectiformes), smelt/paraki (Retropinna retropinna, Retropinninae) and whitebait/maeha (Galaxiidae) were commonly caught in nets, while kahawai (Arripis trutta, Arripidae) and barracouta/mangā, (Thyrsites atun, Gempylidae) were captured using bone hooks and flax lines from canoes (Best, 2005; Smith, 2013). Hāpuku (Polyprion oxygeneios, Polyprionidae), blue cod/pākirikiri (Parapercis colias, Pinguipedidae) and smaller reef species were also important in the Māori diet (Lockerbie, 1940). Māori explored both freshwater and marine environments around New Zealand, with species such as long and short finned eels/tuna (Anguilla spp., Anguilliformes) becoming important dietary and cultural components of kaimoana (Paulin, 2007). A fishing economy based on exchange had been developed before the European discovery of New Zealand in 1642 (Wilson, 2005) and the arrival of Capt. Cook in 1769 (Cook, 1769). At that time approximately 85,000 Māori were living in New Zealand (Bargh, 2016) and fishing grounds were managed and occupied by local iwi (hapū) customarily, using a variety of methods including rāhui (local closures).

After the Treaty of Waitangi, an agreement signed by representatives of Māori leadership and the British Crown in 1840, fishing pressure increased with the mass immigration of Europeans who used fishing methods and gear, such as iron hooks and larger vessels, that were new to New Zealand. Although Article 2 of the Treaty granted Māori full exclusive and undisturbed possession of their lands and activities, including fisheries, the growing European population soon disregarded both the letter and intent of the treaty, and local fisheries management through guardianship (kaitiakitanga/tino rangatiratanga) was largely lost (Bargh, 2016; Johnson & Haworth, 2004; Lock & Leslie, 2007). Fishing pressure and the demands on the marine environment continued to grow in the twentieth century, particularly after inshore fisheries became deregulated in 1963 and the establishment of the 12 nautical mile Territorial Sea in 1965. The lack of restrictions and the increased investments by the government for intensification of harvest supported the expansion of industrialized

29 fisheries and fishing effort on coastal grounds, especially on established resources, such as Bluff oysters (Jackson, 2001). Together with the domestic fleet, foreign fleets from Japan, Russia, and Korea were exploiting fisheries resources within the New Zealand Territorial Sea before the declaration of the EEZ in 1978 (Fenaughty & Bagley, 1981).

With some inshore fisheries showing signs of overexploitation during the 1960s and 1970s, especially coastal species such as rock lobster/koura and pawharu (Jasus edwardsii and J. verreauxi, Palinuridae), snapper/tāmure (Chrysophrys auratus, Sparidae ), scallops/tipa (Zygochlamys delicatula, Pectinidae), rig/makō (Mustelus lenticulatus, Triakidae), and school shark/tope (Galeorhinus galeus, Triakidae), there was an increase in political pressure to develop new ways to better manage New Zealand’s fishery resources (Johnson & Haworth, 2004). As a direct result, the world’s first extensive quota management system was implemented in 1986 to sustainably manage all significant commercial species within New Zealand’s EEZ (Clark et al., 1988). The objective was to promote stewardship of fisheries resources among quota owners, by allocating Individual Transferable Quotas (ITQ) in perpetuity, based on catch histories. The required constant nature of ITQs changed in 1990 to a proportion of the total allowable catch (TAC), as previously the government was required to compensate reductions of quota for stressed species, provide quota for recreational fishers and support the development of new fishing grounds (Clark et al., 1988; Sissenwine & Mace, 1992). The regime comprised 27 species at its inception, including commercially important species, such as barracouta, blue cod, flatfishes (Pleuronectidae), hoki (Macruronus novaezelandiae, Merlucciidae), hāpuku, ling/hokarari (Genypterus blacodes, Ophidiidae), oreos (Oreosomatidae), orange roughy (Hoplostethus atlanticus, Trachichthyidae), red cod/hoka (Pseudophycis bachus, Moridae), snapper, tarakihi (Nemadactylus macropterus, Cheilodactylidae) and common warehou/warehou (Seriolella brama, Centrolophidae) (Newell, 2004), reaching a total of 65 species by 2004 (Bess, 2005).

Fish stock boundaries for each species were established based on the ten existing Fishery Management Areas (FMA) from the sub-Antarctic waters in the south to the Kermadec Islands in the north as far out as the 200 nm EEZ (Figure 1.1). Boundaries of fish stocks (QMAs) were considered to be the same as those of the FMA, part of an FMA, or a group of FMAs, based loosely on available knowledge of fish distribution, life history, and population connectivity. For example, barracouta stocks were designated as follows: BAR 1 includes FMAs 1 to 3; BAR 5, FMAs 5 and 6; BAR 4, FMA 4; BAR 7, FMAs 7 to 9; and BAR 10, FMA 10 (Fisheries New Zealand, 2019). Using catch history information, as well as biological reference points based on maximum sustainable yield (MSY) for a few species, a

TAC and Total Allowable Commercial Catch (TACC) was allocated to each stock (Fisheries New Zealand, 2019; Lock & Leslie, 2007). Subsequently, stock assessments were conducted at variable intervals for the most commercially valuable species. The quota was converted into Annual Catch Entitlements (ACE), which provided catching rights to fishers for each species over the fishing year.

The New Zealand system has been criticized because of its lack of flexibility in the quota setting process, with few changes in quota allocations since the inception of the QMS (Torkington, 2016). The system was also deemed unsuitable as a multi-species allocation strategy, where multi-species maximum sustainable yield (MSMSY) or optimum yield (OY) is the basis for management (Dalton et al., 2018; Worm et al., 2009). In 2019, the New Zealand QMS included 100 species allocated into 638 stocks, 70% of which were finfish. Although almost half of these stocks lacked significant commercial potential, they represented a large component of the marine ecosystems within the New Zealand EEZ and comprised a large component of unreported catches.

Although the QMS has accomplished some of its original objectives, promoting the stewardship of fisheries resources by decentralizing ITQs proved to be difficult in the prevailing market conditions that favored monopolies. Quotas have aggregated over time to the most efficient and profitable groups, and in 2019 were owned by a small number of companies that monopolize the industry (e.g. Torkington, 2016). The expansion of exploitation across marine habitats over the years, combined with the removal of species, community assemblage shifts, bycatch, and discards have resulted in distinct changes in both the ecological footprint of fisheries exploitation and in the structure of the marine ecosystem surrounding New Zealand (Pinkerton et al., 2015b).

1.1.3. Fish landings and ecological trends of New Zealand fish stocks

New Zealand enjoys several advantages regarding flexibility to improve the management of fisheries, being isolated and having a relatively short history of industrialized fishing. The expansion of New Zealand fisheries is characterized by years of heavy exploitation by international fleets before the EEZ and the QMS were introduced, expansion of the domestic fleet into deepwater habitats, and subsequent declines in several important stocks (Fenaughty & Bagley, 1981; Walrond, 2006). Despite this early history, in 2016 New Zealand was regarded internationally to have a relatively functional management plan including a clear commitment towards developing ecosystem-based management (EBM) (Cryer et al., 2016). Nevertheless, scientific support for fisheries management lags far behind 31 many other developed nations. For example, when in 2019 a “Web of Science” search using the topic “Fisheries Ecology” by country was made, only 2.88% of these studies originated from New Zealand, significantly lower than those from Brazil (5.49%) and Australia (11.14%) and not following the rising trend in published studies observed worldwide. The relatively small volume of published literature compared to other countries may be a result of relevant research in unpublished reports by Crown Research Institutes, a decrease in fisheries research funding since the 1990s (Mace et al., 2014), or the current institutional position that fisheries impacts upon marine resources are sufficiently managed by current management acts, requiring no further comment (Alder et al., 2010). The resulting scarcity of independent scientific research and assessment advice on QMS species has resulted in relatively fixed TACCs over time for a vast majority of stocks, since the introduction of the QMS in 1986 (Mace et al., 2014). Consequentially the QMS has treated New Zealand fisheries as if they had fixed production over time, with only rare changes in the allocation of catch entitlements.

In 2016, FNZ reported a commercial catch of 414 thousand tonnes of marine organisms, comprising 170 species, most of them managed under the QMS (Fisheries New Zealand, 2019). The management of each of these species was by quota-based controls on single-species stocks, with a few combined species that are difficult to target independently. For example, the distinct species hāpuku and bass (P. americanus) are managed together under the name hāpuku & bass (species code HPB). Reporting of annual catches has rarely surpassed the allocated TACCs, assumed to be a result of the carryover aspect of the ACEs, which allows fishers to fish a proportion of uncaught quotas in the next season (Marchal et al., 2016). However, the observed pattern might, in part, be the result of unreported discards, or more likely in this case, stock productivity well below the allocations.

An attempt to quantify the extent of discards and bycatch in New Zealand fisheries has recently been made (Simmons et al., 2016). In some cases, discarded fishes were estimated to comprise a majority of the fishery catch at sea. For example, throughout the 64-year time series of modeled discard rates, discards comprised over a third of the reported catch during 61 years (Simmons et al., 2016). Unwanted catches may have been discarded by commercial fishers due to a lack of ACE to cover the bycatch and/or high deemed values, which is a fee paid by fishers for landings exceeding the allocated ACE. Status of many less abundant or low-value species caught as bycatch is unknown from fisheries landing data. The result has been that the full effects of fisheries discard and bycatch on the marine ecosystems are largely unresolved, though their toll on marine biodiversity has become increasingly apparent in recent years (Slooten et al., 2017). Understanding the trophic and ecosystem effects of

32 fisheries exploitation on relevant species was part of the responsibility of the New Zealand Fisheries Act of 1996 and FAO’s best management practices (FAO, 2008), as well as the strategy on sustainable exploitation by Fisheries New Zealand (Ministry of Primary Industries, 2017). One of the key issues in this regard has been changes to the marine food web structure as a result of overexploitation of key higher trophic level species (Pauly et al., 1998). The use of Marine Trophic Index (MTI) and other ecological indices to evaluate community changes have become increasingly applied in New Zealand to understand how exploited communities have changed over time (Pinkerton et al., 2017). Here the importance of resolving spatial variability in food web structure and temporal variability in trophic level for key species has been highlighted (Pinkerton et al., 2017).

1.2. Methods

To investigate trends in reported fish landings over time, reported landings and estimated discards data from between 1950 and 2014 were extracted from Sea Around Us (SAU, Simmons et al., 2016), Fisheries New Zealand (Fisheries New Zealand, 2019) and Food and Agriculture Organization (FAO, 2020) databases. Datasets from FNZ also included catch reports from the period between 1931 and 1950. While FNZ and FAO datasets were comprised of reported catches and official landings, SAU has also included estimated discards, which were calculated through interviews, observations and archival data (Simmons et al., 2016). To assess the uncertainties and efficacy of the SAU dataset, and to identify trends in species removal and community change, these data were split into four categories: non-identified, non-QMS, QMS group (potentially QMS species, but lacking identification), and QMS identified species. Most of the discard estimates by SAU are allocated to the first three groups. FAO and FNZ datasets comprised reports on 67 common QMS species or species codes. Species codes were used to group difficult to identify species, usually with similar trophic level and life history. To compare datasets, total catch from each of the three data sources were plotted against time. To investigate how fisheries have expanded and which ecosystems were most exploited in terms of biomass, datasets were combined, and fish species were split into assemblages, based on habitat associations. The following eight habitat groups were used, after Fishbase (Froese & Pauly, 2017): reef-associated, pelagic- oceanic, pelagic, demersal, bathydemersal, benthopelagic, bathypelagic and pelagic-neritic (Table 1.1). Changes in species richness of the catch during the expansion of industrial fisheries in New Zealand were also investigated by plotting the number of species landed by habitat over time, independent of biomass. Total landings were square-root transformed and a Bray-Curtis similarity index was calculated between each year from 1931 to 2016 time series data. This transformation was necessary to investigate changes of landings compositions comprised of species with different abundances, catch rate and therefore total landings. Community similarities and derivates of similarities between successive years were plotted against time. While similarities were calculated in comparison to the year 1931, derivatives were calculated by the difference between the similarity values of the successive years in the series. Large variations in the values of the similarity derivates indicate a change in species composition of the catch, while stable exploitation of communities could be identified by similarity derivates close to zero for consecutive years.

Table 1.1. Description of major habitats of fish species under the Quota Management System in New Zealand waters. Descriptions were compiled from Fishbase and the Food and Agriculture Organization of the United Nations online glossaries Marine habitat Description Reef-associated Individuals that live and feed on reefs Demersal Individuals that live and feed near the ocean floor and benthic fauna Pelagic Individuals that live and feed at surface or middle depths not associated with the bottom. Pelagic-neritic Individuals that live and feed in open ocean above the continental shelf drop off, around 200 meters Pelagic-oceanic Individuals that live and feed in open ocean beyond the continental shelf drop off, with depths of at least 100 meters Benthopelagic Individuals that live and feed on close to the ocean floor but also in mid and surface waters Bathydemersal Individuals that live or feed below the 200 meters zone Bathypelagic Individuals that live or feed between 1000 and 4000 meters

The weighted average trophic level of the catch, or MTI, was used to investigate trophic level shifts of catches in New Zealand waters over time (Equation 1.1), after Pauly (1998), where TL represents trophic level and L the landed weight of species i. Trophic level estimates for the 67 species were extracted from Fishbase (Froese & Pauly, 2017) and used to calculate MTIs for every year with the combined landings dataset, resulting in the MTI for New Zealand waters. When more than one estimate of trophic level was available from Fishbase an average of the values was used. MTI was then plotted against time and total landings to identify trends in landings composition (Pauly & Watson, 2005). Details of species data extracted from Fishbase are available in Appendix A.1.2.

푛 ∑푖=1(푇퐿ᵢ ∗ 퐿ᵢ) ( 푀푇퐼 = 푛 ∑푖=1 퐿ᵢ 1.1)

To verify the use of MTI in fish community analyses using fisheries data, MTI was calculated with fisheries-independent data from the ECSI trawl survey (MacGibbon et al., 2019) and compared with the dataset from fisheries landings. The ECSI inshore trawl survey is a standardized survey with the objective to estimate fish abundance, distribution, and length and age composition of important species along the continental shelf and slope (10 to 400 meters of depth) of Otago, Canterbury Bight, Akaroa, and Pegasus Bay region (Figure 1.1). Surveys were carried out during May and June in 1991–94, 1996, 2007, 2008–2009, 2012, 2014, 2016 and 2018. While the surveys recorded every fish species caught, annual

35 commercial landings data were gathered from the whole FMA 3 for only ten species: tarakihi, giant stargazer (Kathetostoma giganteum, Uranoscopidae), spiny dogfish (Squalus acanthias, Squalidae), dark ghost shark (Hydrolagus novaezealandiae, Chimaeridae), sea perch (Helicolenus percoides, Sebastidae), ling, blue cod, hāpuku, hoki and silver warehou (Seriolella punctata, Centrolophidae). Commercial fisheries data on additional species, including gurnard (Chelidonichthys kumu, Triglidae), elephant fish (Callorhinchus milii, Callorhinchidae), red cod, jack mackerels (Trachurus spp., Carangidae), rig, common warehou, leatherjacket (Meuschenia scaber, Monacanthidae), rough skate (Zearaja nasuta, Rajidae), smooth skate (Dipturus innominatus, Rajidae) and barracoota, were available when surrounding FMAs were included in the analysis (Fisheries New Zealand, 2019). Those species were chosen due to the availability of data from both datasets and represent the majority of the community biomass. Relationships between the percentages of the average catch of fisheries-dependent and independent scientific data between 1991 and 2018 were investigated with correlations between the datasets. Without market or TACC influence and unreported discards, a correlation between the two datasets is expected.

To compare the results found in the present study with exploitation patterns found in other major fishing grounds, MTI values collected between 1950 and 2014 from the North Sea, Eastern Brazil and the West Coast of the United States were retrieved from SAU (Pauly et al., 2020) and plotted with time. While the Brazilian fishing industry is mostly unregulated and continues to present signs of expansion (Fiedler et al., 2017; Perez et al., 2003), commercial fisheries in the US and North Sea are well-developed and studied. Comparing the time course of MTI among regions provides the basis for resolving how differences in management can have contrasting effects on the exploited community structure. While wild fish exploitation in EU countries are managed through the Common Fishery Policy, US fisheries are managed under the Magnusson Stevens Act. The latter incorporates higher uncertainty in policy, have higher influence from advisory councils, undertakes scientific evaluation of 90% of the country’s stocks, presents a lower TAC consumption from catches and protects fish essential habitat (Battista et al., 2018; Chrysafi & Kuparinen, 2016; Gutiérrez, 2007). Comparison among MTI trends over time provides a quantitative basis for comparing each region’s fishing and management history. These data contribute to our understanding of the application of long-term fishery data to management plans for whole communities of fish, helping to place patterns observed in New Zealand in a global perspective.

1.3. Results

Independent time series of estimated fish catch and reported landings data from inside New Zealand’s EEZ, from SAU, FNZ and FAO are presented in Figure 1.2. The temporal patterns in SAU catch data closely followed those from QMS fish species reported by FNZ and FAO. Nevertheless, a large mismatch in values between FNZ and FAO datasets due to interannual differences, especially before the year 2000, was observed. SAU data indicate a significantly higher catch in every single year by including estimated discards (QMS group and non-identified species), unidentified at the species level. Therefore, data from SAU is vital for assessing total catch and the proportion of discards by weight, but it does not increase the knowledge of individual species catch and landings, habitat exploitation or trophic composition of the catch. While FNZ and FAO data are comparable, especially after 1995, FAO included several highly migratory species absent on FNZ datasets, such as bigeye tuna (Thunnus obesus, Scombridae) and porbeagle shark (Lamna nasus, Lamnidae).

Figure 1.2. Estimated catches of Quota Management System (QMS), QMS group, non-identified and non-QMS fish species from Sea Around Us (SAU) and reported QMS catches from the Food and Agriculture Organization of the United Nations (FAO) and Fisheries New Zealand (FNZ) for the 65 years from 1950 to 2014. The QMS group and non-identified species catches are largely estimated discards

Shifts in gear use, market demand, and industry expansion had an important impact on how fishing pressure was distributed over the EEZ. Using FNZ combined with FAO data and general habitat use from Fishbase (Froese & Pauly, 2017) we explored trends in reported landings in New Zealand since 1931, focusing on shifts in the composition of target species and habitats exploited by the fishery fleet (Figure 1.3). The time series demonstrated that the expansion in types of habitats exploited and the increasing numbers of landed species covaried with the community similarity derivate values (Figure 1.3). Here, two major fishery expansions between 1931 to 1970 and 1972 to 2015 are indicated by increases in the number of species and catch landed, resolved by large differences in community similarities among years. Stagnation of these expansions was indicated by decreasing variation and lowered values in community similarity derivates (Figure 1.3c).

Reported landings data in terms of biomass from FNZ and FAO from 1931 to 2015 were combined with trophic level estimates from Fishbase.org (Froese & Pauly, 2017) for each QMS species. The MTI time series closely matched the pattern observed in the QMS species landings, by species and habitat (Figures 1.3a, 1.3b, and 1.4a). MTI values increased during periods of fisheries expansion, related to the inclusion of new habitats and target species, declining for well-established fisheries (Figure 1.4a). The relationship between MTI and total reported landings presented a concave downward slope, reflecting a decrease in both biomass and trophic level of the commercial catch following a peak of those parameters in the late 1990s (Figure 1.4b). Because hoki is a high trophic level species with large commercial landings, it can dominate the average trend of the whole community. Thus, Figure 1.4 shows the pattern based on data including all species and separately, excluding hoki.

A comparison of MTI between 1991 and 2018, calculated from reported landings and trawl survey data, is shown in Figure 1.5. MTI from both fishery-dependent landing data and fisheries-independent trawl survey data presented positive linked signals. When catches from fisheries-dependent and -independent data are compared by individual species (Figure 1.6), it becomes clear how the two datasets diverge. High-value species (e.g. ling) present a much more coherent relationship between average abundance and total landings, compared to a low value (sea perch) and mostly bycatch (spiny dogfish) species which diverge strongly between the two data sets (Figure 1.6).

Figure 1.3. Trends of Quota Management System (QMS) fish species catches from New Zealand waters from 1931 to 2015, split by marine habitats (a), the number of species landed (b), and Bray- Curtis community similarity and community similarity derivates among years (c). Data source: Fisheries New Zealand and the Food and Agriculture Organization of the United Nations. Habitat type data: Fishbase.org

Figure 1.4. Relationships between Marine Trophic Index (MTI) of the commercial catch of Quota Management System fish species among years (a) and with trends in the total commercial catch (b) in New Zealand waters for the 84 years from 1931 to 2015 for all species (black line) and excluding hoki (grey line). Lines on panel (a) are fitted locally estimated scatterplot smoothing regression with a 95% confidence interval (grey). Data source: Fisheries New Zealand (FNZ) and Food and Agriculture Organization of the United Nations (FAO). Trophic level data: Fishbase.org

Figure 1.5. Marine Trophic Index (MTI) calculated by year for fish species from commercial landings and Fisheries New Zealand trawl surveys in the east coast of the South Island since 1991. Graphs were plotted using data from 20 species caught from within the Fishery Management Area 3 (FMA 3) and surrounding FMAs (plot a: y = 2.98 + 0.25x, r2 = 0.009) and the 10 species caught inside the FMA 3 in both datasets (plot b: y = 0.33 + 0.91x, r2 = 0.16). Linear fit is represented by the black dotted line, with 95% confidence intervals in gray and 1:1 relationship plotted as a continuous line

Figure 1.6. Relationships between the percentage of average catch since 1991 by year from reported landings and Fisheries New Zealand trawl surveys for ling (plot a: y = 12.16 + 0.91x, r2 = 0.2), sea perch (plot b: y = 83.53 + 0.18x, r2 = 0.03), and spiny dogfish (plot c: y = 97.53 + 0.03x, r2 = 0.0003) inside the Fishery Management Area 3. Linear fit is represented by the black dotted line, with 95% confidence intervals in gray and 1:1 relationship plotted as a continuous line

MTI values from 1950 to 2014 resolved large differences in the time course of fisheries expansion between the North Sea, the Eastern Brazilian Shelf and the West Coast of the US (Figure 1.7). While MTI values from the North Sea and Brazil showed one peak and decline as fisheries expanded, the peak occurred much earlier in the former, after the late 1960s, than in Brazil where peaks in MTI were not seen until after 2002. Patterns in the MTI of US commercial landings followed two distinct periods of fisheries expansion, first in the late 1960s followed by a decline in the early 1980s and a later peak of MTI in the late 1990s (Figure 1.7). After these periods, the US MTI has presented a decrease and have been 41 currently increasing since 2006. Each of these changes in MTI records a major expansion of commercial fishing into new communities and habitats, but also reflects the country’s ability to avoid fishing down their exploited communities.

Figure 1.7. Average trophic level of the commercial fishing catches per year from the West Coast of the United States (a), East Brazilian Shelf (b) and North Sea (c). Grey and black lines represent periods of increase and decrease of MTI values, respectively. Shaded regions represent the 95% confidence interval of the smoothed line. Data source: searoundus.org 42

1.4. Discussion

New Zealand’s fisheries history can be divided into five distinct periods as indicated by the analysis of catch history and fish community data considered in the present study. From 1931 to 1950 a coastal fishery developed, targeting the fish communities and abundance once exploited by local Māori. Fisheries during this period were composed of reef-associated coastal species and catches were generally small. In early fisheries, the exploitation of the coastal community did not reflect an increase in MTI (Figure 1.4), due to the low average trophic level of the coastal fish community, although the fishery did include several high trophic level predators, such as hāpuku and sharks (Johnson & Haworth, 2004). Species composition of the landed catch varied largely during this period, reflecting highly diverse fisheries and a fleet composed of small vessels with reduced fishing power. Between 1951 and 1961 the expansion of coastal fisheries by the local fleet was complete, resulting in a steady supply of fish landings (Figure 1.3c), together with a small decrease in MTI (Figure 1.4a). The decline in MTI was likely due to a decrease in the abundance of high trophic level species in coastal waters. After 1961 there was an expansion in fishing grounds, assisted by new fishing techniques and large international and national fleets exploiting the marine resources further offshore, increasing catches of benthopelagic, pelagic-oceanic and bathypelagic species. The exploitation of these new habitats allowed the increase of landings of larger sized fish species in 1965, decreasing to past levels after 1979 (Figure 1.8). Expansion by international fleets and joint ventures were partially motivated by the establishment of New Zealand’s 12 nautical mile Territorial Sea in 1965, granting fishing rights only to the national fleet in these coastal waters. Incentives were also created by the government to expand the number of exploited (or preferred) species, including vessel subsidies and the construction of coastal freezing plants (Johnson & Haworth, 2004). During this period, the size and mean trophic level of the catch rose to its peak, probably assisted by the declaration of the New Zealand EEZ in 1978 and the QMS in 1986. Similarly, total landings were maintained during the expansion of North Sea fishing grounds, despite declining fish stocks and dwindling catch rates (Cardinale et al., 2015). In New Zealand, while QMS implementation used fishers’ catch histories to allocate quotas, which could have inflated some of the catches before 1986, exclusion of fishing grounds to international fleets after the introduction of the EEZ created a peak in landings before 1978, probably associated with the future restrictions. From 1991 to 1999, an increase in benthopelagic habitat exploitation due to large catches of southern blue-whiting (Micromesistius australis, Gadidae) and hoki resulted in total QMS fish landings surpassing half a million tonnes on six occasions by 1999 (see Figure 1.2). Finally, after the year 2000, derivates of community similarities of 43 landings stabilized close to zero, indicating the composition of a fully exploited community (Figure 1.3c), followed by a steady decline in both total weight and MTI starting in the early 2000s (Figure 1.4b).

Figure 1.8. Weighted maximum length of Quota Management System fish species that comprised catches between 1931 and 2015. Yearly lengths were weighted by landings weight. Data source: Fisheries New Zealand and the Food and Agriculture Organization of the United Nations. Maximum length data: Fishbase.org

The patterns observed in New Zealand are consistent with trends in the global fishing expansion since the 1950s (Swartz et al., 2010) and provide evidence for “fishing down the food web” in New Zealand fisheries. When fisheries exploitation impacts the structure and function of marine food webs, this has important consequences for the energetics of fisheries production as well as the biodiversity of marine ecosystems (Pauly et al., 1998). Although the community analyses presented in the present study did not include the estimated discards by SAU, these are of great importance to access the full impact of fisheries exploitation on the marine ecosystem. These discards are thought to be a mix of low-value species and species with high deemed values, a consequence of fishers having insufficient quota to cover extra catches, which also promotes ‘high-grading’ of quota species. The SAU data indicate that unreported discards have been a major component of New Zealand catches and this poses serious management problems in the context of ecosystem-based management of fish communities.

The analysis of the east coast of the South Island trawl survey data allowed direct comparisons between MTI and the proportional reported landings by year between fisheries-

44 dependent (reported landings data) and independent data (research trawl surveys) (Figures 1.5 and 1.6). MTI analysis that included 20 species from areas within and outside FMA 3 displayed no clear relationship between datasets (Figure 1.5a). This was probably because the fisheries-independent data were acquired inside FMA 3, while reported commercial landings for several species were sourced from FMA 3 and surrounding areas. However, when only species with data sourced exclusively from FMA 3 were analyzed, MTI from both sources followed the same general trend (Figure 1.5b). While scientific surveys were standardized spatially and temporally, fisheries landings data were affected by exclusions of bycatch and discards, as well as changes in target species and power associated with changes in fishing gear, technological advances, engine power, range and number of fishing events (Bishop, 2006; Chrysafi & Kuparinen, 2016). The volume of discarded catch in New Zealand waters, which were not reported in landings data (Figure 1.2), comprised up to one-third of the fish caught by biomass. Accordingly, divergence in calculations of MTI in the time series of fisheries landings and fisheries-independent trawl surveys is expected. For example, when the proportion of average catches by year are compared for individual species, relationships differ together with the value and abundance of each species. Catches of ling, a species with high-value but low abundance, had much higher coherence between data sources when compared to abundant, low-value species such as sea perch and spiny dogfish (Figure 1.6). Sea perch catch had very low correlations, while spiny dogfish catch exhibited no clear relationship between fisheries-independent and fisheries-dependent datasets.

While ling has the fourth-highest export value of fisheries commodities in New Zealand (only below arrow squid, hoki and rock lobster -stats.govt.nz), there is no current export of spiny dogfish, and sea perch is only sold locally in fish markets. Spiny dogfish are usually caught as bycatch in other fisheries (Fisheries New Zealand, 2019) and discarded at sea, or sold as nonspecific ‘fish and chips’ (Manning et al., 2004). The low values and the hardship of hauling and processing those species, especially spiny dogfish, creates an incentive for discard, reflected in the species-specific discrepancies found between landings and trawl survey datasets. Similar trends have been discussed in the literature for US waters (Fox & Starr, 1996). Fox & Starr (1996) suggested that these discrepancies are likely related to discarding of unwanted species, especially during a multi-species fishery with strict quota and by-catch regulations, such as high deemed values. The disjunction between fisheries- dependent and -independent data have been investigated in New Zealand waters with similar trends observed as those resolved in the present study (Horn & Ballara, 2018), but there are some limitations when comparing these datasets. While fisheries operate throughout the year targeting fish aggregations, research surveys are carried out at the same of each year and 45 follow a random stratified sampling design across the seascape (Francis, 1984). Consequentially fisheries-independent surveys are the best available tool to quantify fish community structure, being free of the inherent and unresolved biases in fishery-dependent data. However, a smaller scale study outside New Zealand that successfully matched both fishing and scientific activities in space and time has demonstrated a much better agreement between these types of datasets in some situations (Fox & Starr, 1996). Therefore, conclusions about community structure drawn from comparisons between these types of datasets should be reached with full resolution of the bias and historical context of the fish communities being considered (Costello et al., 2012). Similarly, methods used by SAU to compile MTI values for the North Sea, Eastern Brazilian Shelf and the West Coast of the US were different than those used in the present study, including the addition of unreported catches (Pauly et al., 2020). Therefore, although trends in the time series among regions can be investigated using these datasets, absolute MTI values should not be compared.

When the QMS was first introduced in 1986, New Zealand’s fish communities had already been impacted by decades of unquantified fishing effort, and ecological baselines had likely shifted from an unexploited state. For example, Fenaughty and Bagley (1981) reported a catch estimate of 52,000 tonnes of fish along the east coast of the South Island by a Japanese fleet in 1977, summing to 140,000 tonnes when including the entire international and domestic fleets. These landings were eight times larger than the current catch levels for the same region (FMA 3; 17,627 tonnes) and well above the multispecies TACC (around 21,000 tonnes) (Fisheries New Zealand, 2019). Early reports highlight that before the introduction of the QMS and stock assessment studies, fish communities, benthic fauna, and the seafloor were likely heavily impacted by fishing pressure, while catches were largely unreported.

In their seminal paper on ecosystem-based management of fisheries, Thrush and Dayton (2010) argued that: “Appropriate management must have a fundamental understanding of the ecological conditions [or baselines] in the absence, or at least reduction, of fishing pressure”. Resolution of ecological baselines in the absence of fishing should include useful ecological indicators that can be incorporated into management goals as biological reference points (Arkema et al., 2014). The unrecorded early patterns in fishing along the east coast of the South Island highlight the importance of recognizing patterns in exploitation that have occurred in the fish community over the past century, previous to regular landings reports. Accordingly, there is a distinct need for the development of fisheries-independent information to understand shifts in community composition, trophic

46 structure and stock distribution leading up to the current industrialized fishing regime. These data become even more important with the recent more holistic appreciation of the impacts of multiple stressors on the marine environment. The synergy between global and local scale anthropogenic impacts can affect natural processes, biological resources and ultimately ecosystem services that we depend upon (Singh et al., 2017).

Fisheries induced shifts in community structure can have dramatic impacts on ecosystem function, affecting species trophic levels (Popp et al., 2007), food web dynamics and reproduction (Jack & Wing, 2010; Persson & Hansson, 1999), organic matter fluxes, energetics and distribution of biomass (Udy et al., 2019c, 2019a), and ultimately catch rates (Hinz et al., 2017). In the present study, average trophic levels of individual species were retrieved from a large online database (Fishbase.org). While these estimates are the best available data on individual species, the values are subject to geographical and temporal variability. For example, fisheries induced changes in size distributions, community structure, distribution of stocks among habitats and food web structure can strongly influence intraspecific changes in trophic level (Shackell et al., 2010). Studies using long-term survey data from New Zealand’s Chatham Rise, a major region for offshore fisheries, have demonstrated systematic decreases in mean trophic level of individual species as well as the proportion of piscivorous and number of large fishes in the food web (Pinkerton, 2010; Tuck et al., 2009). In this context, preservation of intact trophic structure within fish communities is necessary to maintain fluxes of organic matter and nutrients through food webs providing resilience of marine ecosystems, and should be considered a vital component of essential fish habitat (Blanchard et al., 2011; Thrush & Dayton, 2010; Trueman et al., 2014). Understanding food web ecology of commercially important species and its spatial and temporal variation remains an extant challenge (Udy et al., 2019c; Ward et al., 2016). Nevertheless, the principles of ecosystem-based management, with focus on the preservation of intact food web structure have been tested at regional scales both internationally and in New Zealand waters (Lassalle et al., 2011; Sponaugle, 2010; Wing & Jack, 2013), with successful management outcomes (Jack & Wing, 2013; Nicoll, 2018; Wing & Jack, 2014).

Starting in the 1930’s New Zealand’s fisheries have gone through two major expansions, the first development of coastal fisheries culminating in the 1970s and then an offshore expansion coincident with the establishment of the EEZ. While lack of enforcement, scientific knowledge and liquidity of the QMS prevented full reporting of catches in the past, those data have been used as management baselines to support important management decisions throughout the history of New Zealand’s fisheries. The present study has

47 demonstrated how the trophic architecture of New Zealand fisheries has shifted through time and the time frames for the expansion of exploitation across marine habitats in the New Zealand EEZ. Fisheries dependent data were found to be biased towards smaller reported fish mortality due to unreported discards, which agreed with the independent comparison between fisheries and scientific datasets for the past 29 years in FMA 3. While the next steps in applying EBM to fisheries in New Zealand will likely focus on measures to reduce bycatch and discard rates, changes in the trophic structure of the exploited community also need to be understood and considered in the management framework, to maintain trophically intact fish communities. The analysis of changes in trophic levels of reported landings provides an important step towards understanding the full ecological footprint of commercial fishing in New Zealand. Unlike most countries, New Zealand has full sovereignty over the extent of its EEZ, without a need for negotiations with neighboring countries. This allows unilateral control over the delivery of environmental advice, fisheries management plans, stakeholder communication and quota allocation. In contrast, fishery grounds inside the Greater North Sea are shared between nine nations (Belgium, Denmark, France, Germany, Netherlands, Norway, Sweden, England and Scotland), and 6600 fishing vessels (ICES, 2018), which can make fishery management decisions and quota allocation difficult. Moreover, international interests can take precedent over the need to address localized scientific data and environmental issues, as seen during the implementation of the Multiannual Management Plan for the EU’s Western waters in 2018 (Shanahan, 2019).

The comparison of the time course of MTI in the fisheries of New Zealand, the North Sea, the West Coast of the United States and in Eastern Brazil resolved important similarities in the patterns of fisheries expansion and decline. More recently industrialized fisheries in Eastern Brazil and New Zealand displayed a late peak in MTI after the year 2000 followed by a steady decline in MTI (Figure 1.7). In contrast fisheries in the North Sea and the West Coast of the United States showed early, in the case of the North Sea, and multiple periods of expansion and decline in the case of the US fishery. These differences are closely related to the history of industrialized fisheries expansion and management in each region but record a fundamental emergent property of exploited fish communities. Analysis of North Sea fisheries management in 1987 found a lack of long-term policy planning and scientific advisory regarding new policies and the future of fisheries yields, as well as little attention paid to multispecies stock management (Gulland, 1987). Previous analysis of the commercial landings from the Northeast Atlantic fisheries reported a decrease in piscivore to zooplanktivore ratio coupled with a decrease in landings from 1950 to 1997 (Caddy & Garibaldi, 2000). These results are consistent with the decrease of MTI of commercial 48 landings due to the removal of high trophic level species through fisheries in that region (Pauly et al., 1998). These patterns are commonly found in fully developed fisheries and indicate a shift in the trophic structure of the exploited community, coupled with a reduced abundance of key species and the need for the rebuilding of stocks (Worm et al., 2009). Despite management differences, commercial landings in the North Sea increased after 1970, similar to the pattern observed in New Zealand, but both fisheries underwent steep decreases of landings in more recent times following peaks in MTI (ICES, 2018). In contrast, some countries have employed more stringent management policies regarding overfishing. For example, in the United States, long-term plans ensuring the sustainability of fisheries in specific areas were facilitated by the creation of regional management councils, which work directly with their scientific committees and advisory panels (Gulland, 1987; Gutiérrez, 2007). Furthermore, marine fisheries are mandated by law to define ‘overfishing’ and rebuild overfished stocks within 10 years in US waters (Gutiérrez, 2007; Patrick & Cope, 2014), which have resulted in several fisheries-related lawsuits and several fluctuations in landings over time (Link & Marshak, 2019). These policies could be partly responsible for the observed temporal pattern of landings and MTI on the west coast of the US EEZ (Caddy & Garibaldi, 2000), which display two periods of recovery in the 1990s and today, following peaks and declines (Figure 1.7). In regions where fisheries are still expanding to offshore and deeper habitats, such as on the Brazilian shelf, an increase in landings of large piscivorous species can still be observed (Perez et al., 2003; Pincinato & Gasalla, 2019), followed by a general increase in MTI (Figure 1.7). But larger landings of demersal fishes in recent years and scarcity of long-lived species in local markets could represent early signs of already impacted communities (Pincinato & Gasalla, 2019), which can be seen as a decrease in MTI values in the most recent years (Figure 1.7).

Applying effective ecosystem-based management presents a major extant challenge in the world’s now fully exploited fisheries ecosystems. This is particularly difficult when catches and reporting have been influenced by market demands and management has been based on unresolved and quickly shifting ecological baselines. Although the importance of an ecosystem-based approach to fisheries is recognized internationally (Crowder et al., 2008; Garcia et al., 2003), the global overall performance in applied EBM to fisheries remains below “adequate” (Pitcher et al., 2009). The present case study on the history of New Zealand fisheries provides several important results that contribute to a general framework for the goal of informing effective management of whole fish communities. Specifically, the application of effective ecosystem-based management requires accurate information about fish community composition, structure and trophic dynamics, both for fisheries’ present and 49 past states. In this context, fisheries independent data are essential for resolving historic baselines as well as changes to both community structure and trophodynamics in fisheries ecosystems. Here, New Zealand fisheries provide an excellent case study to investigate the effects of exploitation on fish communities managed under a quota system. Inherent biases associated with catch data, not only because of how changing fishing power affects catch-per- unit-effort data but also driven by differences in value applied to different species, or grades within a quota system, can incentivize waste in the landed catch, and under-reporting of fisheries mortality in an ecosystem. Temporal analysis of changes in community composition, structure and the expansion of fisheries across available habitats provides a quantitative method for assessing expansion and the ecological footprint of fisheries. Here, time series derivatives of community similarity, habitat associations and community MTI provide a suite of useful metrics for assessing the stress that communities come under as fisheries expand. As seen for New Zealand and overseas waters, accurate information on the composition and trophic structure of fish communities, history of exploitation and management systems are invaluable for resolving past influences of change in trophic structure and ecological baselines as a framework for assessing ecological function in exploited systems. The utility of these metrics to inform ecosystem-based management will depend on the quality of the commercial fisheries data, including proper identification of bias related to change in market demand and quota through the comparison with fisheries- independent datasets. The results of the present analysis and resolution of the history of New Zealand’s fisheries provide a valuable case study for informing effective ecosystem-based management, and the global aspiration of restoring the diversity and function of exploited fish communities.

Chapter 2 Oceanographic transport along frontal zones forms carbon, nitrogen, and oxygen isoscapes on the east coast of New Zealand: implications for food web studies

Abstract

Characterizing the spatial distribution of 18O in seawater and 15N and 13C of the particulate organic pool can be useful in identifying isotopic discriminators of major water masses and serve as a baseline for food web and other ecological studies. Nevertheless, we do not know how those isoscapes are distributed and interact with physical oceanographic features at relevant scales for large parts of the global ocean. Here the physical structure of the water column as well as seawater δ18O and suspended particulate organic matter δ15N and δ13C collected during research cruises in the austral summer of 2017/18 and 2018/19 along the east coast of the South Island of New Zealand were analyzed. Seasonal differences in inferred circulation, oceanographic features, and the interaction of Subtropical (STW) and Subantarctic water masses (SAMW and SASW) in the study area during the two sampling periods were described. Using a model averaging approach, the relationships between oceanographic variables and the 18O, 15N and 13C isoscapes were investigated. Results demonstrate that different environments along the east coast are connected via northward transport of STW near the surface and SAMW at depth (220-500 m). The northward transport of STW was associated with nutrient inputs and deepening of the maximum fluorescence depth. The spatial distribution of isotopic values was primarily linked to patterns of nutrient input (δ15N), coastal productivity (δ13C) and water mass type (δ18O), with nitrate and silicate concentrations being the best predictors of isotopic values. The present study was the first to resolve the distribution of δ15N and δ13C isotopes at a regional scale along the east coast of the South Island, δ18O in New Zealand Waters and to investigate the extent to which these distributions were reflected by the physical oceanography. δ15N values demonstrated significant differences between regions sampled, while 18O had distinct values for different depths. Average values of 18O, 15N and 13C for different fishery management areas and depths are presented to support future ecological studies along the eastern coast of New Zealand.

2.1. Introduction

Information on the natural distribution of stable isotopes has been used in ecogeochemistry to understand movements, diet and stock differentiation in animals (McMahon et al., 2013). The techniques developed have proven particularly useful when applied to the study of marine environments, especially when the focal species are cryptic or highly migratory, or biological processes occur over large temporal and spatial scales (Ramos & González-Solís, 2012). Isotopic values from a consumer can be retrieved from a single tissue sample, providing a cost-effective method for understanding trophic ecology (Popp et al., 2007). Common studies in ecogeochemistry include using both nitrogen (δ15N) and carbon (δ13C) isotopic values from animal tissues to identify foraging behavior or migration patterns (McMullin et al., 2017; Trueman et al., 2012). δ15N values display large fractionation between trophic levels (between 1.4 and 5.6‰), resulting in large differences in δ15N between predator and prey and making δ15N a reliable tool for estimates of trophic level, which is an estimate of how high an individual is feeding within a food web (Jack & Wing, 2011; Post, 2002). In contrast, δ13C fractionation remains around 0.3 and 0.5‰, preserving δ13C values from primary producers throughout the food web, and providing an important tool to estimate contributions of alternate organic matter sources to the food web (Fry, 2006; Jack et al., 2009; McCutchan et al., 2003). Because δ15N and δ13C of a consumer’s tissues vary relative to those of their prey (i.e. the higher the baseline the higher the consumer’s signature), it is critical to understand the processes that influence δ15N and δ13C of primary producers at the base of food webs (e.g. Cornelissen et al. 2018). δ15N and δ13C of suspended particulate organic matter (SPOM) have been observed to vary across space, time and relative to environmental processes in marine systems (Kurle & McWhorter, 2017). Consequentially, in order to interpret changes in δ15N and δ13C of a consumer’s tissues one must first resolve the isotopic landscape, or isoscapes, and the mechanisms driving isotopic variation along environmental gradients (Cherel et al., 2008; West et al., 2010; Wing et al., 2007, 2008, 2012).

Although some researchers have successfully mapped δ13C and δ15N at ecologically relevant scales (Barnes et al., 2009; Jennings & Warr, 2003; Wing et al., 2012), results for New Zealand waters rely heavily on oceanographic modeling for information about spatial variability in δ13C and δ15N of SPOM (Magozzi et al., 2017). The spatial scales resolved by such models are generally too coarse (~100 km) to provide useful information for regional and coastal studies (Graham & Bury, 2019). Most of the challenges of building regional isoscapes are linked to variability in physical processes in the coastal zone, where inputs from

52 terrestrial and macroalgal productivity may be high, both in the grazing and particulate food webs (Udy et al., 2019c, 2019b; Wing & Jack, 2012). While environmental, and therefore isotopic variation, in the open ocean is related to large spatial and temporal scale processes, coastal dynamics can have finer scale effects on isotopic values of basal organic matter sources, incorporating detrital and macroalgal inputs, leading to distinct spatial heterogeneity (McLeod et al., 2010a; Miller et al., 2006). Processes such as freshwater runoff, sediment resuspension and local eutrophication can hamper one’s ability to understand and predict the natural distribution of isotopes in coastal waters. Therefore, to understand isotopic variability in coastal waters one needs to relate isotopic values to variables influenced by coastal processes. For example, dissolved silica (SiO2) is transported to coastal zones through riverine runoff and represents one of the major nutrients stimulating diatom productivity in marine habitats (Treguer et al., 1995). SiO2 concentrations have the potential to present steep gradients in the continental shelf of the east coast of the South Island of New Zealand (ECSI), since the region displays up to two times more SiO2 yield in coastal catchments than the world’s average (Dürr et al., 2011) and have been used to study coastal processes along the ECSI (Vincent et al., 1991).

Stable oxygen isotope values (δ18O) are also a valuable tracer for the origins of oceanic waters and mixing with freshwater sources. δ18O is only known to fractionate during evaporation/precipitation in a Raleigh distillation process and during freeze and thaw events in sea ice formation (Frew et al., 1995). Moreover, many marine organisms have skeletal material composed of calcium carbonate, which occurs in the alternate forms of calcite and aragonite (CaCo3). For a majority of organisms, the oxygen bonded in the calcium carbonate originates from ambient water (Tohse & Mugiya, 2008). Oxygen from water masses with distinct δ18O values can be incorporated into the calcium carbonate structure and therefore information on 18O can be used to infer temporal patterns in environmental conditions experienced by organisms (Killingley, 1980; Nelson et al., 1989; Tallis et al., 2004). Coastal waters with riverine inputs from different catchments can receive freshwater with different δ18O values (Baisden et al., 2016; Dansgaard, 1964) and have the potential to be applied in investigations of oceanographic processes and spatial ecology of marine organisms. However, differences among δ18O sources must be considered in the context of the natural distribution of δ18O in the study region and its relationship with oceanographic processes.

The transport of particles, nutrients and other subsidies among marine habitats can strongly affect local productivity and enhance biodiversity and abundance of consumers (Polis et al., 1997). Understanding how different habitats are connected by oceanographic

53 processes at the mesoscale is therefore an important step to understand the mechanisms supporting the nearshore marine ecosystem. Furthermore, by identifying the oceanographic variables driving variability in local isoscapes, one may be able to use more conventional oceanographic data (e.g. temperature, salinity and nutrient data) and remote sensing to predict δ13C and δ15N values of particles and δ18O of seawater in specific regions. These models can become cost-effective tools to investigate the ecology, movement and distribution of important marine species. For example, the spatial distribution of δ13C of pelagic primary producers has been successfully used to study the mixing of global stocks of three species of tuna along a 16-year period (Logan et al., 2020).

The ECSI encompasses a large range of marine environments and oceanographic features. Unique features include the high nutrient low chlorophyll Subantarctic Surface Water (SASW) 35-55 km offshore of the southeastern coastline, the most nearshore extent for the Subtropical Frontal Zone (STFZ) worldwide (Jones et al., 2013), and numerous submarine canyons along the shelf break (100-200 m of depth) offshore of the Otago and Kaikoura peninsulas. Offshore of the ECSI coast, normally warm and salty Subtropical Water is modified through mixing with a less saline coastal water mass, referred to as Neritic Water (NW), producing modified STW (mSTW). NW receives freshwater inputs from terrestrial and river runoff, most notably from Fiordland along with the Waiau, Mataura and Clutha rivers (Hawke, 1992). Both NW and mSTW travel northward, forming a distinct low-salinity coastal (1-16 km offshore) and high-salinity offshore band (16-30 km) respectively (Figure 2.1) (Murdoch et al., 1990). Around 30 km from the coast, mSTW interacts and mixes with offshore SASW, forming a band of enhanced temperature and salinity gradients along the ECSI known as the Subtropical Frontal Zone (STFZ), referred to locally as the Southland Front (SF) and associated with the northward flow of the Southland current (Chiswell et al., 2015; Hopkins et al., 2010; Sutton, 2003). The interactions between SASW and mSTW take place along the entire ECSI and can influence water properties down to 800 m depth (Heath, 1981; Vincent et al., 1991). New Zealand’s unique bathymetry, with steep slopes, as well as submarine canyons close to shore, act to bathymetrically steer features such as the STFZ as they interact with the coast (Morris et al., 2001; Shaw, 2001). These topographic features lead to the STFZ becoming relatively stable along the ECSI (Hopkins et al., 2010; Smith et al., 2013), with much of the variability in the position of the STFZ observed after the front detaches from the shelf break near 44˚ S, from where it extends seaward from New Zealand.

Figure 2.1. Map of New Zealand with the most important surface features influencing processes on the east coast of the South Island, based on Chiswell et al., (2015). STW (Subtropical Water), dUC (d’Urville Current), STF (Subtropical Front), mSTW (modified Subtropical Water), NW (Neritic Water) and SASW (Subantarctic Surface Water). The Southland Current (SC) and the region where the transport of mixed water masses is known as the Southland Front (SF) are also shown. Arrows inside orange lines represent the main flow southward from the Tasman Front. Light gray shades represent depths from 10 to 1000 meters

Marine environments in this region receive nutrient inputs from several sources. The macronutrients, nitrate and phosphate, are thought to be mainly supplied to this region landward from the SASW, through advection and/or mixing across the STFZ (Hawke & Hunter, 1992; Jones et al., 2013; Van Hale et al., 2010). In contrast, inputs of micronutrients like Fe, Cu and Si are primarily resourced from runoff and sediment resuspension that occurs within NW and STW (Butler et al., 1992; Croot & Hunter, 1998; Jones et al., 2013). The stirring and mixing of these water masses near the STFZ are likely responsible for enhancing local pelagic productivity, characterized by a phytoplankton community with large diatoms in spring and high productivity rates throughout the year (Boyd et al., 1999; Jones et al., 2013). Remnants of this productivity have been recorded in the form of phytodetritus at depths of

1,200 m in the south of Chatham Rise (~ 44˚S and 178˚E), and it was linked to the dynamics of the STFZ (Nodder et al., 2007).

Much like the Chatham Rise, the Kaikoura Canyon (~ 42˚S) is known to be a highly productive marine system and to support large biomass of demersal species, recognized as one of the most productive deep-sea environments on Earth (De Leo et al., 2010; Leduc et al., 2014). At Kaikoura, primary productivity is potentially enhanced due to several oceanographic processes, including nearshore coastal upwelling (Chiswell & Schiel, 2001), inflows of mSTW and/or SASW via the Mernoo Saddle (Chiswell & Schiel, 2001) as well as gravitational transport of sediments from the coast (Canals et al., 2006). Moreover, previous studies have suggested that the productivity around Kaikoura could be influenced by the dynamics of the STFZ via frontal water flowing north through the Mernoo Saddle (Greig & Gilmour, 1992; Shaw & Vennell, 2000; Vincent et al., 1991). A recent study measuring stable isotopes in sediment samples and invertebrates has also demonstrated that the structure of the invertebrate community from the Kaikoura Canyon is shaped by the higher supply of marine-derived organic matter, compared to Canyons on the west coast of the South Island (Leduc et al., 2020). Collectively, these studies demonstrate that oceanographic processes play a vital role in shaping local communities and determining δ15N and δ13C values of primary producers and consumers along the ECSI.

The aim of the present study was to better understand the characteristics of the ECSI isoscapes, their drivers and the extent to which they were influenced by oceanographic processes. The results provide an important framework for using isoscapes as a tool in ecological studies in the region. Here δ13C and δ15N values of SPOM, as well as δ18O of sea water were analyzed and mapped along with temperature, salinity, fluorescence and macronutrient concentrations, in order to understand the spatial variability of those variables and the relationship between oceanographic variables and isotopes. The present study was the first to use isotope values to elucidate ECSI oceanographic processes at the mesoscale (10 - 100km), as well as to investigate variability in the natural distribution of δ13C, δ15N and δ18O in the region.

2.2. Methods

2.2.1. Samples and data collection

All samples were collected during four cruises in the austral summer of 2017/18 (January, February and March 2018) and 2018/19 (November 2018) along the ECSI (Figure 2.2) on board of the RV Polaris II. Vertical profiles of temperature, salinity and fluorescence were collected using CTD casts along four transects, OT (Otago), AK (Akaroa), KA (Kaikoura) and CS (Cook-Strait) using a Seabird SBE-25 (Seabird Scientific®). When time and weather permitted, CTD casts were also carried out at stations between transects, totaling 57 casts. CTD data were filtered and aligned for depth discrepancies following the Seabird Scientific guidelines (Sae-Bird Scientific, 2017) before been averaged in 1-meter bins.

Figure 2.2. Oceanographic transects during 2017/18 and 2018/19 cruises with T-S diagram using CTD casts with main water masses represented. NW (Neritic water), mSTW (modified Subtropical Water), SASW (Subantarctic Surface Water), SAMW (Subantarctic Mode Water), AAIW4 (Antarctic Intermediate Water with South formation) and AAIW1&3 (Antarctic Intermediate Water with North formation). Otago8 and Kaikoura4 casts from 2017/18 are not shown

Water at the surface (~1.5 m), maximum fluorescence depth (MF), 150 and 500 meters were sampled using a rosette equipped with 12 ten-liter Niskin bottles (Ocean Test Equipment®). Triplicate water samples were taken for nutrient analysis, filtered with a 33mm 0.45μm nylon filter (Microlab Scientific®) and immediately frozen. Particles were filtered for δ13C and δ15N analysis of SPOM with a pre-combusted (at 400˚C for 4 hours) 47mm 0.7μm GF/F filter (Whatman®). Water was pre-filtered with a 300 μm mesh to remove large zooplankton, then filtered until filters clogged to ensure enough material for analysis, they were then immediately frozen. A third set of water samples were collected from surface, 150, 500 and 1000 meters for δ18O analysis, consisting of 30 ml of filtered water (33mm 0.45μm nylon filter) kept in hermetic glass vials.

2.2.2. Sample preparation and laboratory analysis

Nutrient samples were analyzed for nitrate/nitrite (NOx), ammonia (NH3), orthophosphate (PO4) and silica (SiO2) with a flow injection analyzer (Lachat QuikChem® 8500 Series 2 FIA) using QuikChem® methods for brackish and saltwater at Portobello Marine Lab – University of Otago. Due to low concentrations of nitrite and its small importance to the dissolved and particulate N pools in open systems (Sigman & Casciotti, - 2001), we considered that NOx was a measure of nitrate (NO3 ) in the study region. Small zooplankton was manually removed from filters with SPOM samples under a dissection microscope, filters were then acidified in a fume with 5% saturated solution of SO2 in water

(H2SO3) for 8 hours to remove inorganic sources of carbon, oven-dried at 60°C overnight, cut in half and packed in an 8x5 mm tin (Sn) capsule for analysis. Filters were analyzed by the Isotrace Laboratory – University of Otago by combustion in an elemental analyzer (Carlo

Erba NC2500) to CO2 and N2. The isotopic compositions of the sample gases were measured by a Europa Scientific 20-20 ANCA Mass Spectrometer operating with continuous flow. Delta values were normalized and reported against the international standards Vienna Pee Dee Belemnite (VPDB) and atmospheric nitrogen (AIR) for δ13C and δ15N respectively. Normalization was made by 3-point calibration with 2 glutamic acid international reference materials and a laboratory EDTA standard (Elemental Microanalysis) for carbon (USGS-40 = −26.2 ‰, USGS-41 = 37.8 ‰, EDTA = -38.93 ‰) and nitrogen (USGS-40 = −4.52 ‰, USGS-41 = 47.57 ‰, EDTA = -0.73 ‰). Analytical precision was checked by comparing results from analyzed quality control standards EDTA-OAS and IAEA MA-A-1 (Copepod) against accepted values. All measured values for the quality control standards were in the range of accepted values. 58

18 δ O from water samples was determined by the method of equilibration with CO2

(Epstein & Mayeda, 1953) by the Isotrace lab – University of Otago. Half of a millimeter of a sample was equilibrated with 12 ml of 0.3% CO2 in helium on a Thermo (Bremen) GasBench preparation unit for 18 hours at 25.0 ± 0.1 °C. The equilibrated gas was measured in 10 replicates with a Thermo Advantage isotope ratio mass spectrometer (IRMS) in continuous- flow mode and values more than 1 standard deviation from the average were excluded. The filtered average was corrected to the international VSMOW-SLAP isotope scale using a three- point calibration provided by three laboratory standards analyzed before and after every batch of 84 samples. In addition, a control sample was measured at every 12th position to correct for instrumental drift. Consensus values for the laboratory standards have been obtained from 6-year internal laboratory calibration records against primary reference materials, VSMOW, GISP and SLAP, external 6-member interlaboratory comparison exercise and by back- calculation from the ~170 member IAEA interlaboratory comparison exercise, WICO2012. The laboratory standards and their consensus values are as follows: ICE (δ18OVSMOW= - 32.097 ± 0.075‰), TAP (δ18OVSMOW = -11.432 ± 0.038‰), SEA (δ18OVSMOW = -0.029 ± 0.036‰).

Isotope values were reported in the delta notation (δ), i.e. the differences from a standard in parts per thousand (‰), according to the following equation:

푅푆푎푚푝푙푒 3 훿푖 = ( − 1) × 10 2.1) 푅푆푡푎푛푑푎푟푑

Where i is the heavier isotope been analyzed and R is the isotopic ratio (abundance of i related to the most abundant isotope) in both sample and standard (McKinney et al., 1950).

2.2.3. Data analysis and statistical framework

To investigate the distribution of distinct water masses, Temperature-Salinity (T-S) diagrams were generated for different transects, and water masses were identified using definitions from the literature (Table 2.1). Temperature, salinity, isotopes and nutrients were mapped along isosurfaces at surface and MF depth, maps were interpolated using DIVA (Data-Interpolating Variational Analysis) algorithm, which is specialized in gridding in situ data, taking into account coastlines and bathymetry (Troupin et al., 2012). Temperature, salinity, density, fluorescence, isotopes and nutrient concentrations at different depths were 59

compared between transects along the coast in order to infer water movement and nutrient transport in the ECSI. To investigate the relationships between isotopic values (18O, 15N and 13C) and the oceanographic variables (longitude, latitude, depth, sampling season,

fluorescence, salinity, temperature, NOx, NH3 and SiO2), a global model was fitted following an information theory approach. This procedure generated all possible linear models with the oceanographic variables dataset, averaging models with top AICc (Akaike Information Criteria corrected for small sample sizes) scores into a single global model (Grueber et al., 2011; Kurle & McWhorter, 2017). Models were fitted for all depths and surface separately; details can be found in Appendix A.2.1. Surface δ15N and δ13C values for the Otago coast were compared to a long term, 16 year old dataset (Van Hale, 2003) to check for seasonal and interannual effects on the values found in the present study, and to test their application in future ecological studies. To identify differences between relevant regions and depths of δ15N and δ13C of SPOM and δ18O of water, a non-parametric Wilcoxon’s test was used. Averages (± SE) of δ15N and δ13C of SPOM, and δ18O of water were calculated and reported for ecologically relevant regions to inform ecogeochemistry studies within the ECSI. Assumptions related to linear models were checked with residual plots (by predicted values, by row and normal quantile). Analyses were carried out using the software Ocean Data View 5.1.7 (Schlitzer, 2018), JMP 14.0.0 (SAS Institute, 2018) and R 3.5.2 (R Core Team, 2020).

Table 2.1. Temperature, salinity and depth distributions of the main water masses along the east coast of the South Island retrieved from the literature Temperature Water mass name Acronym Salinity (PSU) Depth Reference (°C) Neritic Water NW > 12 < 34.6 Surface (Jillett, 1969) Subtropical Water STW > 15 35.1 Surface (Hamilton, 2006) Modified Subtropical mSTW > 12 > 34.6 0-200 m (Jillett, 1969) Water Subantarctic Surface SASW < 12 < 34.5 0-500 m (Jillett, 1969) Water Subantarctic Mode Water SAMW 7.5 - 8 34.3 - 34.55 100-400 m (Hasson et al., 2012) Southern Antarctic (Tsuchiya & Talley, AAIW 4 3.5 - 10 < 34.4 500-1300 m Intermediate Water 1996) Northern Antarctic (Tsuchiya & Talley, AAIW 1&3 3.5 - 10 > 34.45 500-1300 m Intermediate Water 1996)

2.3. Results

2.3.1. Oceanography of the ECSI

During both sampling seasons, the water mass distribution was broadly consistent with descriptions given in the literature (Jillett, 1969; Jones et al., 2013). Overall, coastal surface waters along the ECSI (< 5 m depth) displayed low salinity (< 34.5 PSU) and high temperature values (> 12˚C) (Figure 2.2). These values are consistent with the signature of NW (Jillett, 1969) and were detected between 0-12.4 km of the coast in Otago, 0-10.9 km in Akaroa and 0-2 km north of Kaikoura. Offshore of the coastal NW band, at distances of 13 to 62 km in Otago, 11 to 60 km in Akaroa and 2 to 18.6 km in Kaikoura, temperature and salinity values were similar to values previously reported for mSTW in this region (temperature < 12˚C and salinity > 34.5 PSU (Jillett, 1969)). Signatures of SASW (< 12˚C and < 34.5 PSU) were detected in surface waters offshore of the mSTW band along the Otago coastline and Banks Peninsula but were not found offshore of Kaikoura.

T-S diagrams (Figure 2.3) showed marked variability in water column properties and structure between the 2017/18 and 2018/19 cruises. Coastal and offshore waters were generally cooler and fresher during the 2018/19 cruise (Figure 2.3), with a large decrease in surface temperature between 2017/18 and 2018/19 in Otago (5˚C in OT8) and along the other transects (2˚C in AK4, AK5, KA5 and CS3) (Figure 2.3). A subsurface salinity maximum was recorded in the Subantarctic Zone offshore of Otago (OT8) between 140 and 260 m depth and with a density of 26.8-26.85 kg m-3 during both cruises. Underneath these features, a layer with reduced temperature variability was measured between 230 and 400 m in 2017/18 and 220 to 350 m during the 2018/19 cruise, represented by a thermostad of 7.5-7.7˚C (Figure 2.4). Measurements collected in the Mernoo Saddle, offshore of Akaroa showed a similar feature between 450 and 500 m in the 2017/18 (AK4 and AK5) and 2018/19 (AK5 only) cruises. The density of these features lay between 26.85 and 26.90 kg m-3 and seems likely to indicate the presence of Subantarctic Mode Water (SAMW) traveling northward in the Southland Current (Hasson et al., 2012; Morris et al., 2001).

Figure 2.3. T-S diagram for CTD vertical transects along the east coast of the South Island during 2017/18 (a) and 2018/19 (b) sampling seasons, with indication of the different water masses. NW (Neritic water), mSTW (modified Subtropical Water), SASW (Subantarctic Surface Water), SAMW (Subantarctic Mode Water), AAIW4 (Antarctic Intermediate Water with South formation) and AAIW1&3 (Antarctic Intermediate Water with North formation).

To understand how oceanographic variables vary spatially along the ECSI, isosurfaces 13 15 18 of T, S, fluorescence, NOx, NH3, PO4, SiO2, δ C, δ N, C:N and δ O were examined at the surface and the depth of the maximum fluorescence (Figures 2.5 and 2.6; Appendices A.2.2 and A.2.3). NOx and PO4 concentrations were positively correlated at surface throughout the region (r2 = 0.94 and p < 0.0001, Appendix A.2.4), with high concentrations in offshore waters south of Akaroa (OT7, OT8, AK4 and AK5). In contrast, SiO2 and NH3 showed markedly different patterns (Figures 2.5, 2.6; Appendices A.2.2 and A.2.3). In 2018/19 there was a clear increase in SiO2 and NH3 concentrations inshore along the Otago region, with a simultaneous decrease in both δ13C, δ15N and C:N ratio of SPOM and δ18O of seawater (Figure 2.6). Distributions of oceanographic parameters showed similar trends at the surface and MF depth for most variables (Figures 2.5 and 2.6; Appendices A.2.2 and A.2.3). One exception was the advection of cooler surface water through the Mernoo Saddle evidenced by a northward tongue of cold water, which was more evident near the surface. T, S and NOx values at MF depth presented a similar distribution, from the Otago shelf (station OT3) to the south of Kaikoura in the 2018/19 cruise (Appendix A.2.3). This trend was also observed for δ15N values of SPOM. MF depths varied between 8 and 80 m for the whole region (average ±

SE of 26.8 ± 1.84) and were usually located at the depth of the thermocline. MF depths along the region showed variations between the 2017/18 and 2018/19 cruises. While samples from 2017/18 showed a deeper MF depth close to Otago coast (~ 40 m), this layer was much deeper north of Banks Peninsula (35 and 80 m) in 2018/19 than in the previous cruises (Appendix A.2.5). When the integration of fluorescence values was calculated between the surface and 150 m, higher average values were observed for all the transects in the 2018/19 cruise (Figure 2.7), although the differences in fluorescence might be attributed to the high variability observed between stations.

Figure 2.4. Temperature profile and its deviation at depth (deltaT) of a subset of the CTD casts from offshore Otago (OT8) and Akaroa (AK4 and AK5) from 2017/18 and 2018/19 sampling seasons. Details to the temperature thermostad indicative of SAMW are shown between the dashed lines 63

δ18O values from surface waters were very similar for all regions and were usually above 0.2‰, except south of Akaroa where samples encompassed values from -0.32 to 0.49‰. When δ18O values at 150 m were compared, two groups showed similar values along the ECSI, one formed by stations closer to shore, OT3 (0.37‰), Ak-Ka4 (0.33‰), KA5 (0.39‰) and CS3 (0.33‰) and other by stations further offshore, OT4 (0.22‰), AK4 (0.26‰), AK5 (0.22‰), KA4 (0.24‰) and CS2 (0.26‰), although CS2 is placed more inshore than CS3. At 500 m δ18O signatures were similar between samples north of Akaroa during the 2018/19 cruise, Ak-Ka5 (0.15‰), KA4 (0.14‰) and CS3 (0.14‰). Similar values of δ18O were also found at 500 m between offshore Otago (0.05 ± 0.04 in OT4, OT7 and OT8) and Kaikoura (0.11 ± 0.03 in KA3, KA4 and KA5), but values from Akaroa transect were much higher (0.21 ± 0.07 in AK4 and AK5). The same pattern seen for δ18O at 150 m was observed for SiO2 concentrations, where OT3 (6.25 µM), OT4 (6.62µM), AK4 (7.28µM), Ak-Ka4 (6.68µM), Ak-Ka5 (6.82µM), KA4 (6.41µM), KA5 (7.22µM) and CS3 (6.2µM) displayed similar values, but this pattern was not observed in any other depth sampled.

Figure 2.5. Isosurfaces of oceanographic variables along the east coast of the South Island during austral summer 2017/18 at the sea surface

Figure 2.6. Isosurfaces of oceanographic variables along the east coast of the South Island during austral summer 2018/19 at the sea surface

)

- Average sum of fluorescence (mg.m fluorescence of sum Average

Figure 2.7. Average integration of fluorescence values between surface and 150 meters for each transect. Bar colors correspond to data collected during 2017/18 and 2018/19 sampling season, with its respective standard errors

2.3.2. ECSI marine isoscapes

Variables included in the global multimodel approach were temperature, salinity, sampling season, latitude, longitude, fluorescence, ammonia, NOx and SiO2. Correlations between variables at different depths were tested using a Pearson correlation coefficient (r), where minimum (-1) and maximum (1) values indicate a perfect negative and positive correlation between pairs of variables. Concentrations of NOx and PO4 were correlated for water samples collected at the surface, MF depth and 150m throughout the whole region (r < -

0.67), therefore PO4 was excluded from the multimodel fitting. Isotopic signatures of SPOM ranged from -2.88 to 9.11‰ for δ15N and -27.77 to -15.72‰ for δ13C throughout the whole region, with a larger variation of δ15N south of Banks Peninsula and δ13C northwards (Figures 2.5 and 2.6; Appendices A.2.2 and A.2.3), while δ18O varied between -0.32 and 0.55‰.

Five out of 512 models were within the lowest AICc models (delta AICc < 2) and were averaged for δ13C and δ15N prediction, the number of models for δ18O within the two lowest

AICc was of eight. When analyzing only samples at surface, the number of averaged models 13 18 15 were higher for δ C (38) and δ O (10) and lower for δ N (4). Latitude, longitude, NOx, 13 SiO2 and temperature were good predictors for δ C including all depths (Table 2.2), while 13 15 δ C values at the surface were only predicted by SiO2 at a p < 0.1 level. Predictors of δ N values included in the top five models were longitude, NOx, SiO2, temperature, depth and 67

sampling season. When only surface samples were analyzed, NOx was the only predictor for 15 18 δ N (p < 0.0001). δ O values among all depths were best predicted by NOx, SiO2, 15 18 temperature and salinity, similarly to δ N, δ O at the surface were only predicted by NOx (p < 0.0001). A table with the results from models at surface can be seen in Appendix A.2.6.

13 Table 2.2. Summary fit of the average of models with lowest AICc (delta < 2) predicting δ C, δ15N and δ18O along the east coast of the South Island. Relative importance is the proportion of occurrence of that variable in the top models

δ13C Estimate Adjusted SE Relative importance p-value (Intercept) -23.96 0.16 - < 0.0001

Latitude -1.69 0.61 1.00 < 0.01 Longitude 2.56 0.60 1.00 < 0.0001

NOx 2.33 0.67 1.00 < 0.0005 SiO2 -1.78 0.46 1.00 < 0.0005 Temperature 1.87 0.61 1.00 < 0.005 Depth 0.13 0.45 0.20 0.77 Fluorescence 0.05 0.16 0.20 0.78 Salinity -0.03 0.12 0.20 0.81 Ammonia 0.03 0.14 0.20 0.82 δ15N (Intercept) 6.19 0.20 - < 0.0001 Latitude 0.09 0.34 0.20 0.79 Longitude 1.59 0.45 1.00 < 0.0005 NOx -6.20 0.92 1.00 < 0.0001 SiO2 2.01 0.58 0.60 < 0.001 Temperature -2.27 1.10 0.20 < 0.05 Depth 5.20 1.00 1.00 < 0.0001 Fluorescence -0.37 0.39 0.60 0.34 Salinity 0.11 0.22 0.40 0.61 Sampling season -1.66 0.51 1.00 < 0.005 18 δ O (Intercept) 0.30 0.02 - < 0.0001 Latitude 0.03 0.03 0.50 0.35 Longitude 0.01 0.02 0.25 0.72 NO -0.35 0.08 1.00 < 0.0001 x SiO2 0.09 0.04 1.00 0.05 Temperature -0.22 0.09 0.75 < 0.01 Fluorescence -0.04 0.05 0.50 0.48 Salinity 0.04 0.02 1.00 < 0.05 Sampling season -0.06 0.04 0.75 0.19

Although the averaged model indicates that latitude and longitude were good predictors for δ13C, linear regressions only from values collected at 150 m showed significant weak linear relationships between the variables, slope = 0.36 ± 0.13, r2 = 0.24, p =0.01, n = 26 for latitude and slope = 0.47 ± 0.17, r2 = 0.25, p = 0.01, for longitude. Conversely, δ13C 2 displayed a weak negative relationship with SiO2 at MF depths (r = 0.32, p < 0.0001, n = 53),

13 2 while NOx was significantly correlated with δ C at 500 m (slope = -0.97 ± 0.28, r = 0.67, p =0.01, n = 8). No relationship between temperature and δ13C was identified.

Depth was a good global predictor for δ15N values but showed no linear relationship in the whole region (r2 = 0.06, p = 0.002, n = 170). Longitude also displayed a high predictive power for δ15N among all depths, with an increase of 1.12 ± 0.24‰ (n = 73, r2 = 0.24 and p < 0.0001), 1.26 ± 0.25‰ (n = 57, r2 = 0.31 and p < 0.0001), 0.97 ± 0.2‰ (n = 26, r2 = 0.49 and p < 0.0001) and 0.49 ± 0.22‰ (n = 11, r2 = 0.36 and p = 0.05) per degree of longitude at surface, MF, 150 and 500 meters, respectively. The same trend was seen for temperature, which varied together with depth and longitude values. δ15N had a negative linear 2 2 relationship with NOx at the surface (r = 0.85, p < 0.001, n = 73) and MF depths (r = 0.65, p 15 < 0.001, n = 53). SiO2 also covaried with δ N at the MF depth but presented a weak linear relationship (slope = -0.63 ± 0.2, r2 = 0.16, p = 0.003, n = 53) and a skewed residual normal quantile plot, which might hinder the linear comparison of those variables.

Temperature and salinity were good predictors of δ18O (Table 2.2), but linear regressions at different depths demonstrate that these relationships are only significant at 150 m for temperature (slope = -0.03 ± 0.006, r2 = 0.18, p < 0.001, n = 102), and at the surface (slope = 0.29 ± 0.05, r2 = 0.34, p < 0.001, n = 63) and at 150 m (slope = 0.54 ± 0.07, r2 = 0.69, p < 0.001, n= 26) for salinity. δ18O values also had a weak negative linear relationship with 2 NOx between all depths (r = 0.18, p < 0.001, n = 102). The same trend was seen for SiO2, but they were only significant at 500 m (r2 = 0.48, p = 0.04, n = 9).

Significant differences in isotopic values of SPOM between south and north of Banks Peninsula were observed for δ15N at surface and at MF depth (Wilcoxon test, p < 0.05, Appendices A.2.7 and A.2.8). Average ± SE of δ15N values were significantly smaller in Otago and South Akaroa (statistical areas 24 and 22) than in North Akaroa and Kaikoura (statistical areas 20 and 18) (Table 2.3). δ13C of SPOM and δ18O of seawater at equal depths did not display significant differences between regions, but differences in δ18O values were observed between surface (0.28 ± 0.02‰), subsurface (0.27 ± 0.03‰ at 150 meters) and deep waters (0.13 ± 0.04‰ at 500 and 0.1 ± 0.07‰ at 1000 meters) (Appendix A.2.9). To verify the long term stability of isotopic values of SPOM collected in the present study and its utility for ecological analysis, δ15N and δ13C values from surface SPOM sampled around Otago were compared with a 3-year long, multi-seasonal study along the same transect (Van Hale, 2003). 15 Although δ N values were significantly higher in the present study, F1,220 = 16.61, p < 0.001, both δ15N (3.32 ± 0.79‰) and δ13C (-23.42 ± 0.54‰) average values from the present study

69 felt within the ranges of samples collected by Van Hale et al (2003) (3.20 ± 0.25‰ and -23.77 ± 0.2‰, respectively).

Table 2.3. Average ± SE of δ15N and δ13C of SPOM samples from both sampling periods, and δ18O of water for different compartments along the ECSI

Region (statistical area) Depth (m) δ15N δ13C δ18O Marlborough Sounds (17) 0 5.93 ± 0.11 -23.4 ± 1.47 0.36 ± 0.04 30 5.86 ± 0.44 -25.48 ± 0.74 - Kaikoura (18) 0 7.02 ± 0.15 -23.33 ± 0.48 0.31 ± 0.02 30 6.9 ± 0.21 -22.97 ± 0.46 - 150 7.74 ± 0.31 -22.89 ± 0.38 0.28 ± 0.02 500 7.74 ± 0.25 -24.52 ± 0.29 0.11 ± 0.02 North Akaroa (20) 0 7.19 ± 0.29 -23.34 ± 0.82 0.35 ± 0.05 30 7.59 ± 0.41 -22.92 ± 1.38 - 150 7.38 -20.72 0.39 500 7.39 -21.47 0.30 South Akaroa (22) 0 3.8 ± 0.72 -23.77 ± 0.38 0.24 ± 0.04 30 3.92 ± 0.83 -23.99 ± 0.37 - Offshore Akaroa (23) 0 6.95 ± 1.09 -21.5 ± 0.67 0.35 ± 0.07 30 7.24 ± 1.01 -22.04 ± 0.46 - 150 7.07 ± 0.36 -23.74 ± 0.44 0.26 ± 0.02 500 7.64 ± 0.17 -23.93 ± 0.23 0.21 ± 0.07 Otago (24) 0 2.99 ± 0.72 -23.49 ± 0.48 0.21 ± 0.04 30 2.62 ± 0.6 -24.08 ± 0.56 - 150 4.68 ± 0.63 -24.65 ± 0.3 0.2 ± 0.04 300 6.38 ± 0.11 -25.23 ± 0.37 - 500 6.07 -25.03 0.05

2.4. Discussion

Understanding the variability and drivers of δ15N and δ13C from the pelagic organic pool is the first step to enable studies of isotopic ecology across any seascape. Contributions from coastal environments from freshwater runoff play important roles in the marine organic pool and must be considered. Accordingly, isotopic studies should consider all important baselines when modeling complex food webs. Similarly, resolving the spatial variation of T, S, δ18O and macronutrients in seawater can help one to elucidate connections among environments through water mass movements and determine important subsidies of organic matter among habitats.

2.4.1. Ocean connectivity along ECSI

Surface water temperature values were highest during the 2017/18 austral summer, probably due to the effect of the most intense heatwave ever recorded in the region (Salinger et al., 2019). The heatwave resulted in an increase of an average of 2˚C in SST in coastal New Zealand waters, which was recorded in the present study, peaking in January 2018 at 3.2˚C. These effects were also coupled with a negative Oceanic Niño Index (ONI), that could be related to the higher evaporation rates and decrease in rainfall and presumably freshwater inflows to the ECSI in January 2018 (Macara, 2015), potentially explaining the relatively high surface salinity values observed during the first sampling period (Figures 2.3, 2.5 and 2.6). High surface temperatures measured during the 2017/18 austral summer, mainly in coastal regions, in some cases confounded the identification of water masses solely by temperature. Nevertheless, information on variability in salinity across the study region indicated the reduction of NW influence over the inner continental shelf during the dry summer of 2017/18 compared to 2018/19 measurements (Otago Regional Council, 2020). While freshwater from rainfall has a more negative δ18O value than seawater in New Zealand (Baisden et al., 2016), oceanic waters along the ECSI receive important SiO2 inputs from freshwater runoff, primarily the Clutha River, in the Otago region (Jones et al., 2013; Vincent et al., 1991). High 18 concentrations of SiO2 and negative values of δ O close to the Otago coast in 2018/19 indicated the influence of freshwater runoff into the adjacent ocean. These features were coupled with a peak in ammonia concentrations, which could have been caused by export from local estuaries and can have implications for primary production on the shelf (O’Connell-Milne et al., in review). Elevated rainfall during the 2018/19 sampling season could also have favored the input of freshwater and the export of nutrients from estuarine and

71 riverine systems (Otago Regional Council, 2020), contributing to the elevated nutrient concentrations observed.

Water masses are transported northward along the ECSI, both near the surface (Shaw & Vennell, 2000) and at depth (Greig & Gilmour, 1992). In the present study, δ15N, fluorescence, NOx and PO4 values near the surface and at MF depth demonstrated a similar distribution and extending from the Canterbury Bight, around Banks Peninsula inshore of the Mernoo Saddle, with this feature reaching as far north as Kaikoura. These features were characterized by temperature and salinity values as a band of mSTW. Similarly, water masses 18 from Otago at 150 m displayed similar densities, δ O and SiO2 concentrations when compared to waters at Akaroa, Kaikoura, and Cook-Strait. The observations indicated a continuous flow of mSTW along the ECSI inshore of the Mernoo Saddle, at least from surface waters to 150 m. It is also notable that there was a northward decrease of surface and subsurface nutrient concentrations that coincided with increased δ15N of SPOM, suggesting nutrient intake by the phytoplankton community (Sigman & Casciotti, 2001). Therefore, northward transport of NOx and PO4 by the band of mSTW appeared to be particularly important for primary producers in the regions north of Akaroa.

There was evidence for year-to-year variations in primary production along the ESCI, which may have been driven by seasonal and/or oceanographic-induced variations in water column stratification. During summer 2017/18, there were lower integrated values of fluorescence along the length of ECSI, from Otago to Cook Strait at depths between surface and 150 m compared to summer 2018/19 (Figure 2.7). It was also notable that waters in these regions were more strongly stratified during summer 2017/18, inferred by the MF depth (Appendix A.2.5). These year-to-year differences in patterns of productivity and stratification may be related to the seasonal nature of the research cruises conducted during the present study. While the 2017/18 sampling season occurred in January, February and March of 2018, the 2018/19 cruise took place in November 2018, representing the end and start of the stratified summer period, respectively. It also seems possible that differences in patterns of oceanic productivity observed between years could have been driven by the varying influence of mSTW on stratification in the waters north of Akaroa. North of Akaroa, eddies from the Wairarapa current (part of the southward flow from the Tasman Front) have been demonstrated to drive a southward influx of warm, stratified, low-nutrient STW close to Kaikoura coast (Greig & Gilmour, 1992), whilst reducing northward transport of nutrient- richer mSTW and SASW via Mernoo Saddle. Oceanic productivity can be reduced by such a shift toward a stratified, warm and nutrient-poor environment, through inhibition of vertical

72 nutrient fluxes (Da Silva et al., 2017; Goldman et al., 1996). Thus, a reduction of northward transport of higher nutrient mSTW, driven by both seasonal and oceanographic processes, might have reduced nutrient concentrations and therefore productivity north of Akaroa during summer 2017/18 (Vincent et al., 1991). Similarly, the 2017/18 marine heatwave (Salinger et al., 2019) could have caused the observed increase in stratification, leading to weak vertical mixing and low nutrient flux, resulting in lower productivity levels. Chiswell and Sutton (2020) suggested that high SST north of Akaroa is related to low sea surface chlorophyll and that these patterns are correlated with shallower winter mixing layers, driven by ocean warming. Because of these processes, it was also indicated that primary productivity will likely decrease north of Akaroa due to ocean warming under future climate scenarios, with either neutral or positive changes in productivity in the southern regions and along the STF (Chiswell & Sutton, 2020). These processes could have a large effect on the local food webs along the coast. For example, intrusion of SAW through Mernoo Saddle and its interaction with STW have been pointed as an important factor driving the abundance of both low (Mills et al., 2008) and high (Guerra, 2018) trophic level species along the coast of Kaikoura. The results from the present study suggest that productivity in the region was enhanced by northward transport of mSTW through the Mernoo Saddle, and that this could potentially be impaired by global (atmosphere-ocean interactions), regional (ocean circulation) and local (freshwater runoff) dynamics, which were likely to have impacted the magnitude and distribution of primary production.

Below the upper layer of the ocean, there was further evidence for connectivity along the ECSI. CTD casts evidenced a thermostad, a layer centred near 350 m depth with constant temperature, at stations offshore Otago (OT8) and Akaroa (AK4 and AK5). While the bottom of this feature was measured between 350 and 400 m in Otago, its range comprised deeper layers (450-500 m) offshore of Akaroa. The density value of these layers lay between 26.85 and 26.9 kg.m-3, although deep waters from Kaikoura showed similar values of density, a thermostad was not identified for those transects. This feature observed offshore of Otago and Kaikoura seems unlikely to be explained by an intrusion of mSTW given its low temperature (7.3-7.7˚C) and relatively low salinity (34.3-34.5 PSU). Instead, the structure and water column properties are consistent with the presence of SAMW being advected northward along the ECIS via the Southland Current (Hasson et al., 2012). During the 2018/19 cruises, the thermostads related to the temperature and salinity values of SAMW were not observed within the Mernoo Saddle (AK5, Figure 2.4). During the same sampling season, the thermostad identified from Otago was much shallower than during the 2017/18 cruise. These results suggest that SAMW transport northwards connects the deeper habitats 73 between Otago and Akaroa, but the amount of connectivity between the regions at the depth of SAMW may vary year to year. Although the transport of SAMW northward along the south of the South Island and the ECSI has previously been studied by Morris et al. (2001) and modeled by Hasson et al. (2012), the new results appeared to be the first time that the SAMW water mass had been observed continuously as far north as Akaroa. Although I was unable to compare those results with δ18O values at these depths, higher δ18O values observed in Akaroa than those found in Otago and Kaikoura indicated that water masses at or deeper than 500 m did not flow northward through the Mernoo Saddle, which is consistent with the local bathymetry (~500 m of depth). δ18O values from offshore Otago at 500m were more negative than those from Akaroa and Kaikoura, indicating that deep waters were not traveling north through the Saddle. CTD measurements and δ18O taken at and below 500m from these stations (OT8, OT7, KA4 and KA5) indicated that water masses differed between OT and KA transects. Although it is recognized that AAIW was the water mass sampled in both regions (Table 2.1), AAIW4 presented in Otago has a southern and fresher formation between 50 and 55˚S than AAIW1&3 found in Kaikoura at 30 to 35˚S (Chiswell et al., 2015), explaining the differences observed in the present study.

2.4.2. ECSI marine isoscapes

When δ13C values of the particulate pool from all depths were amalgamated, δ13C could be predicted by few linear models using latitude, longitude, SiO2, NOx and temperature, 13 but SiO2 was the only variable with significant prediction power for δ C values of the 13 particulate pool at the surface. δ C displayed a linear relationship with SiO2 at MF depths, 13 latitude and longitude at 150m and NOx at 500m. These results suggested that δ C of SPOM at the surface and subsurface are controlled by coastal processes, such as freshwater input from runoff, inferred by its relationship with SiO2 (Treguer et al., 1995), also affected by SiO2 utilization by primary producers (Boyd et al., 1999). At 150 m these values varied primarily with location along the coast, while for samples at 500 m nutrient availability become an important covariate for δ13C. δ13C values of SPOM are usually well correlated with temperature in large datasets (Barnes et al., 2009), but this relationship changes when samples from different depths are pulled together for analysis, which requires sampling of environmental data along the water column (MacKenzie et al., 2014). In systems where productivity is driven by upwelling and therefore large temperature gradients, δ13C of SPOM is expected to vary with those processes (Kurle & McWhorter, 2017). Models describing δ13C variability in the Southern Ocean, including New Zealand waters, have also found that 74

13 δ C values of SPOM could be predicted by the rate of CO2 absorption by phytoplankton cells lacking active transport mechanisms (Francois et al., 1993). The observed process was largely parameterized by temperature and highly dependent on the composition and growth stage of the phytoplankton community (Espinasse et al., 2019). The present study did not find relationships between δ13C of SPOM and temperature along the study region. Although the lack of phytoplankton community composition data, the findings demonstrate that δ13C variation in the region was mainly driven by freshwater runoff, mixing between different water masses and particle decay at depth.

When all depths were entered in the multimodel approach, δ15N values of SPOM covaried with NOx, longitude, SiO2, temperature, depth and sampling season, while only NOx was considered a good predictor at the surface. δ15N values increased with depth and longitude but decreased with temperature, while δ15N showed a significant decrease with the increase of SiO2 (MF depth) and NOx values (surface and MF depth). Month of sampling was also important in predicting δ15N, although signatures did not differ between 2017/18 and 2018/19 cruises at a p < 0.05 level for any depth sampled. These results indicated that like the results of δ13C analysis, the relationship between δ15N of SPOM and environmental variables 15 differed with depth. At surface and subsurface, δ N was largely predicted by NOx alone, on the other hand, δ15N from deeper waters had better relationships with longitude and temperature values. In the Southern Ocean, δ15N of SPOM have been predicted to increase - with the intake of NO3 by phytoplankton at surface (Somes et al., 2010), causing - fractionation of N and decreasing NO3 concentrations (Altabet, 2001; DiFiore et al., 2009). 15 The strong negative relationship between δ N of SPOM and NOx in the present study corroborates those findings, but at a smaller scale. In the Otago region, NOx is thought to be advected onshore from offshore SASW (Van Hale et al., 2010), causing an increase in productivity within the STFZ. The data suggested that because of the northward flow, phytoplankton communities are transported as far north as Kaikoura, where NOx concentrations are lower and δ15N higher and more static. Therefore, regions with low nutrient concentration but that receive water from regions with high productivity, such as the surface waters north of Banks Peninsula, present a much higher δ15N values than the coastal Otago region. Therefore, transport of water through the Mernoo Saddle can have an impact on oceanographic conditions and ecological processes around Kaikoura region, as suggested in previous studies (Bradford & Roberts, 1978; Heath, 1981; Shaw & Vennell, 2000). At greater depths, δ15N values are influenced by other processes, such as the sinking of particles, followed by bacterial decay or intrusion of SPOM from other water masses (Pantoja et al., 2002). 75

Surface water values for δ18O were more positive and presented a smaller range of values north of Banks Peninsula (Figures 2.5 and 2.6). The multimodel approach resulted in 18 NOx, SiO2, temperature and salinity as the best predictors for δ O at all depths, but only NOx 18 in surface waters. Regressions showed a positive linear relationship between δ O and NOx values at all depths, while salinity (surface and at 150m), temperature (at 150m) and SiO2 (at 18 500m) were correlated with δ O at specific depths. NOx advected from SASW could be used as a proxy for water mass differentiation in the study region, especially south of Akaroa (Van Hale et al., 2010). This differentiation was followed by δ18O values, which were more negative in the SASW. While the observed variation was also related to salinity values at the surface, both salinity and temperature displayed a relationship with the δ18O values below 150m (Figure 2.8). In surface and subsurface waters the observed relationship only coincided with salinity values, underlining the utility of δ18O to identify freshwater inputs in coastal systems (Bigg & Rohling, 2000).

Figure 2.8. TS plot with colored labels representing δ18O values of samples collected from surface to 1000 meters from the whole ECSI. Dots inside the dashed polygon represent samples taken below 150m

The similarity between δ15N and δ13C average values of SPOM observed during the present study and by Van Hale, (2003) demonstrated that the isotopic values presented here are adequate for long-term food web studies in the east coast of the South Island, especially when samples from tissues with slow isotopic turnover rates are compared. Moreover, previous studies have demonstrated that seasonality might have a low effect on δ13C of phytoplankton in the Southern Ocean, responsible for a variation between 2 and 4‰ north of 60˚ S (Magozzi et al., 2017). Because of the inshore-offshore variability in isotope values reported here, and since latitude and longitude had a large effect in predicting the overall values of isotopes, spatial differences of isotopic values should be carefully considered when identifying baselines for ecological studies. To provide isotopic baselines of δ15N, δ13C and δ18O at useful scales, the ECSI was divided into different regions according to statistical areas used in regional scale fishery data collection by the New Zealand Ministry of Primary Industries (Fisheries New Zealand, 2019). These isotopic values represent the average of all particulates sampled in the water, including bacterial, small zooplankton and detritus, which can differ from phytoplankton signatures and can hamper food web studies. The observed effect would be increased in regions influenced by coastal processes and continental inputs (Harmelin-Vivien et al., 2008), therefore isotopic values of SPOM should be considered carefully before applied into ecological studies, especially at the regional scale. The ratio between carbon and nitrogen composition (C:N) of SPOM can be used to identify the primary source of organic matter in a sample, with values of zooplankton and bacteria samples ranging between 3 and 6 and terrestrial materials higher than 12 (Savoye et al., 2003). C:N values of samples taken in the present study ranged between 5.52 and 9.92 with an average ± SE of 6.84 ± 0.52, indicating a majority of phytoplankton organic matter.

The highest primary production along the Otago coast occurs in the mSTW (Jones et al., 2013). Periods of high productivity were measured during late spring and early summer and were dominated by microphytoplankton (> 20 μm), limited mainly by nitrate and silicate concentrations. In fact, small nitrate:phosphate ratios corresponded to high chlorophyll-a values measured in that study for NW and mSTW, while values of nitrate:silicate had a positive relationship with chlorophyll-a in the SASW (Jones et al., 2013). In the present study, small values of nitrate:silicate were measured in the SASW (< 5) at surface and MF depth, indicating small primary production in that region, while nitrate:phosphate varied between 0 and 25 (7.17 ± 0.89) at NW and mSTW. Nitrate:phosphate values smaller than 13 were used to indicate samples that were representative of high primary production south of Akaroa within the NW and mSTW water masses at surface and MF depths, resulting in an average ± SE of 5.45 ± 0.33‰ for δ15N and -24 ± 0.25‰ for δ13C. These isotopic values 77 better indicate the pelagic production throughout the region when it is not limited by nutrients, representing a better isotopic baseline for food web studies when compared to raw values of δ15N and δ13C from SPOM. Samples collected north of Akaroa had much less variation in nitrogen:phosphate (0 to 16.5) and isotopic signatures decreased by only 2% when only nitrate:phosphate ratio values below 13 were analyzed.

The present study was the first to investigate the distribution of δ15N and δ13C of SPOM at regional scale along the South Island and the first to study the spatial variability in δ18O values of seawater in New Zealand. Through continuity of temperature, salinity and δ18O values, as well as the decrease in nutrient concentrations followed by the northward increase in δ15N of SPOM, the connectivity of surface and subsurface marine habitats along the ECSI and through the STFZ was demonstrated, from the Otago shelf break to Kaikoura canyons. Deeper environments (200 to 500 m) were also connected along the ECSI, evidenced by water masses with similar density, δ18O values along the length of the ECSI and periodic intrusion of SAMW through the Mernoo Saddle. Linear models demonstrated that while δ15N values of SPOM were linked to nutrient availability in the whole ECSI, δ13C and δ18O were mainly affected by coastal processes, such as freshwater runoff, and had distinct values for the different water masses sampled. Average δ15N and δ13C values of SPOM, as well as δ18O of seawater for ecologically relevant areas along the ECSI were provided for future isotopic ecology studies in the region. These results provide an important step towards applying isotopes to ecological and oceanographic studies in the region.

Chapter 3 Regional differences in the trophic structure of fish communities in the east coast of the South Island of New Zealand

Abstract

Stock management of exploited fish species usually does not consider the resources necessary for each population to sustain its biomass, neither its spatial nor temporal variability. These variations can influence species trophic relationships, with the potential to alter fisheries yields. Therefore, understanding the natural variation of the trophic structure of exploited populations and communities is an important step to inform fisheries management based on an ecosystem approach. Isotope analysis of basal organic matter sources and muscle tissue of consumers was used to investigate regional variation of trophic structure and niche breadth of exploited fish communities along the east coast of the South Island, New Zealand. Isotope values, as well as trophic level and the proportion of macroalgae supporting a fish’s diet were highly variable along the study sites, displaying linear relationships with latitude and total length for most fish species. Community trophic structure and fisheries yields were compared among two key statistical management areas in the southernmost (Otago) and northernmost (Kaikoura) regions of the study area. Otago had higher commercial fishery yields and catch-per-unit-effort than Kaikoura, supported by a community with smaller niche breadth (evidenced by the δ15N and δ13C biplot), higher trophic redundancy and species occupying lower trophic levels and relying more heavily on macroalgal production. When species were analyzed separately, tarakihi (Nemadactylus macropterus, Cheilodactylidae) was the only species to show a significant larger niche breadth in Kaikoura compared to that observed in Otago, indicating higher trophic diversity that could be a result of limited resources and habitat degradation. This is the first study to investigate the spatial and temporal variability in the trophic structure and niche space of commercial fishes in the east coast of the South Island. The results of the present study highlight the importance of considering differences in resource use and community structure in the management of fisheries resources.

3.1. Introduction

A species’ niche can be described as the use of resources in an ecosystem (Elton, 1927; Grinnell, 1904; Pocheville, 2015), including the effect of the environment on the species and, in turn, the species' effect on the environment (Polechová & Storch, 2019), as well as all ecological interactions among organisms. In a food web perspective, while the fundamental niche (potential resource use) is determined by intrinsic factors such as morphology and behavior, the realized niche (actual resource use) largely depends on extrinsic factors acting on resource availability, environmental change and species interactions (Bolnick et al., 2003). Trophic interactions among individuals and populations will be affected by both intrinsic and extrinsic factors and are likely to vary in space and time (Fry et al., 1999). Changes in resource availability and trophic interactions can alter fish’s abundance, distribution and by extension, fisheries yields (Thrush & Dayton, 2010). To classify and study the trophic structure of fish communities, researchers have used estimates of species' trophic levels (TL) and the contribution of basal organic matter sources supporting their diets (Davis & Wing, 2012; Karlsson et al., 2019; Phillips et al., 2014; Pinkerton et al., 2008; Udy et al., 2019a). These metrics can be used to study the effects of anthropogenic impacts at population and community levels (Bearhop et al., 2004; Hussey et al., 2014; Jack et al., 2009). As an example, the variability of a consumer’s TL and resource use among individuals can be linked to resource abundance and availability, with implications for population management (Jack & Wing, 2011; Wing & Jack, 2014). In addition, fish communities with low trophic redundancy have been identified from overfished regions (Matich et al., 2017), and can present lower resilience (Peterson et al., 1998) and increased vulnerability to anthropogenic impacts (Petchey et al., 2008).

The natural distribution of stable isotopes has been used to estimate TL of fish species for decades (Fry, 2006). Isotopic analysis of muscle tissue samples, which have low turnover rates, provide averages from the consumer’s diet from several weeks to months (Hesslein et al., 1993; Mont’Alverne et al., 2016; Skinner et al., 2017; Suring & Wing, 2009). Therefore, large variation in isotope values from muscle tissue among individuals indicates a consistent diet or/and resource use variability in a specific population (Fry et al., 1978). Carbon and nitrogen isotopic ratios, expressed as a deviation from a standard (δ13C and δ15N), are the most frequently used for tracking organic matter sources and estimating trophic levels of animals (Ramos & González-Solís, 2012). While trophic level estimates rely on isotopic fractionation of nitrogen between trophic levels (δ15N in bulk muscle tissue increases substantially when organic matter is transferred between consumers; 15N = 1.4 to 5.6‰),

δ13C has relatively small fractionation with trophic exchanges (13C ~ 0.5‰), so is more precise for estimating mixtures of basal organic matter supporting a food web (McCutchan et al., 2003; Mill et al., 2007; Post, 2002). In order to estimate these parameters, one needs to know the δ13C and δ15N values of the primary producers supporting that specific food web, which can vary spatially and with time (Reed et al., 2016; West et al., 2010; Wing et al., 2008). Furthermore, diet, and by extension TL of a fish can be limited by its body size or gape size, which are often correlated with TL and should be taken into account in ecological studies (Dalponti et al., 2018; Ladds et al., 2020).

While TL and the proportion of basal organic matter supporting a fish’s diet are trophic parameters of an individual, normally averaged at a population level, community-wide Layman’s metrics can be used to compare niche space from distinct communities across whole ecosystems (Layman et al., 2007; Udy et al., 2019c, 2019b). For example, ranges of δ15N and δ13C values provide information on community TL and diversity in resource use, with the trophic diversity in a food web being estimated from the total area of the minimum convex hull of the δ15N and δ13C biplot. Similarly, the mean Euclidian distance from each species’ δ15N and δ13C data cloud to the centroid of the whole food web (CD) is a metric of average trophic diversity that is also affected by trophic packing (Layman et al., 2007). The mean nearest neighbor distance (MNND) and the standard deviation (SDNND) of the same plot can be used as proxies for trophic packing and its evenness, respectively. Layman’s metrics have been used to study the effects of seasonality (Abrantes et al., 2014), species introduction (Jackson et al., 2012) and exploitation (Saporiti et al., 2014) on diet diversification and niche breadth of aquatic communities. When used together with approaches that take into account differences in isotope values at the base of the food web (e.g. mixing models), Layman’s metrics are a powerful tool that can be used to evaluate changes in niche space and trophic relationships within communities and populations (Layman et al., 2012). This is especially true when studying trophic relationships of large communities with incomplete biological data (Bowes et al., 2017; Dammhahn & Kappeler, 2014).

Fish communities are affected by both chronic events such as overfishing, nutrient input, ocean warming and acidification, and acute events such as hurricanes, earthquakes, and floods (Udy et al., 2019a). These events can cause environmental shifts and drive fish distribution and behavior (Bailey & Secor, 2016; Doney et al., 2012; Harding & Jellyman, 2015; Matthews et al., 2014), altering the way individuals and populations interact within a community and the environment. Among known chronic stressors, exploitation is one of the

81 most studied, with examples of impacts on the trophic structure at individual, population and community level (Hinz et al., 2017; Pauly et al., 1998; Thurstan et al., 2010). These effects can be localized and change population dynamics and species interactions and are not likely to be detected from routine research surveys at stock scales (Bradford-Grieve et al., 2007; McClatchie et al., 2005). Furthermore, spatial and temporal variation of trophic relationships are rarely taken into account in management, but can influence a species year-class strength and ultimately fisheries yields (Casini et al., 2010). Considering the stability and variability of food web structure is therefore necessary for the assessment of species conservation status, vulnerability and to identify areas of conservation priority for exploited species.

The east coast of the South Island hosts two of the largest fisheries ports in New Zealand (Lyttelton and Timaru) and is an important region for commercial and recreational fisheries, managed as a single management area for most exploited species (Fisheries New Zealand, 2019). The continental shelf is composed of rocky reefs in the Otago (~ 46˚ S) and Kaikoura (~ 42˚ S) regions, and a sandy bottom extending approximately 40 km offshore along Canterbury Bight, facilitating inshore trawl fishery activities (Figure 3.1). With most of the exploited species managed in this area as a single stock and the recent recognition for the need of a local marine reserve network (MPI, 2020), this region is a good case study to investigate spatial and temporal differences in the trophic structure of the exploited fish communities. Data from common commercial fish species with contrasting ecological traits from the east coast of the South Island were analyzed (Table 3.1). These species made up of over 25% of the export value of wild fisheries from New Zealand in 2018, excluding processed fish products (Stats NZ, 2019), and are the basis of important recreational fishing grounds around Otago and Kaikoura (Fisheries New Zealand, 2019). While blue cod (Parapercis colias, Pinguipedidae), ling (Genypterus blacodes, Ophidiidae), sea perch (Helicolenus percoides, Sebastidae) and tarakihi (Nemadactylus macropterus, Cheilodactylidae) are managed inside the east coast of the South Island as a single stock, the Fishery Management Area 3 (FMA 3), barracouta (Thyrsites atun, Gempylidae) stocks are overseen along the east coast of the South Island and the Chatham Rise (Fisheries New Zealand, 2019). For other species such as red cod (Pseudophycis bachus, Moridae) and slender jack mackerel (Trachurus murphyi, Carangidae), the management area of a single stock comprises most of the exclusive economic zone south of Kaikoura. Slender jack mackerel is also managed together with two other species, T. declivis and T. novaezelandiae (Carangidae). While most species are managed as a single stock inside the east coast of the South Island, there is a lack of knowledge on the variation of niche space and trophic structure of the exploited community. 82

Figure 3.1. Map of the New Zealand marine environment with fisheries management statistical areas referenced in the present study as Kaikoura (18), Western Chatham Rise (20), Akaroa (22) and Otago (24). White represents New Zealand landmass and grey shades are identifying depths between 0 and 150, 150 to 1000, 1000 to 2000 and deeper than 2000 meters

The aim of the present study was to understand the variations in trophic structure of an exploited community, focusing on comparisons between fish communities in the Otago and Kaikoura regions. The current study is the first to investigate the spatial and temporal variations of the trophic structure of the exploited fish community in the southeastern region, representing a benchmark for the management of these resources based on an ecosystem approach. Variation in δ15N and δ13C values in fish muscle tissue, estimates of TL and percentage of organic matter derived from macroalgae supporting food webs (%Macroalgae) based on isotopic mixing model analysis were used to answer the following questions: (1) do

83 isotope values of a sedentary fish species collected from Otago vary among years? (2) Do isotope values, resource use and trophic level have a relationship with latitude and specimens’ length, and how those trophic parameters vary among different species, regions and between years? (3) Are niche breadth and trophic diversity different between communities collected from Otago and Kaikoura? (4) How do differences in the trophic structure between Otago and Kaikoura relate to fisheries yields in each region?

Table 3.1. Maximum total length, assemblage, habitat type, depth distribution and common prey items of the seven commercial fish species analyzed in the present study Maximum Species Habitat Common prey items Reference total length (mm) Blue cod 600 Inner shelf* Fish (47%) (Graham, 1939) Parapercis Demersal Crustaceans (23%) (Russell, 1983) colias 0 - 150 m Tunicates (17%) (Carbines & Mckenzie, 2001) Cephalopods (Fisheries New Zealand, 2019) Gastropods Personal observation Bivalves Worms Slender Jack 700 Inner shelf* Crustaceans (55%) (Fisheries New Zealand, 2019) Mackerel Pelagic-oceanic Salps (36%) (Konchina, 1992) Trachurus 0 - 500 m Fish (11%) murphyi Cuttlefish Barracouta 2000 Outer shelf* Crustaceans (77%) (Fisheries New Zealand, 2019) Thyrsites Benthopelagic Fish (18%) (Graham, 1939) atun 0 - 400 m Squid (9%) (O’Driscoll, 1998) (Stevens et al., 2011) Tarakihi 700 Outer shelf* Chitons (Godfriaux, 1974) Nemadactylus Demersal Gastropods (Graham, 1939) macropterus 0 - 486 m Bivalves (Russell, 1983) Crustaceans (McKenzie et al., 2017) Echinoderms (Annala, 1988) Polychaetas Cephalochordates Red cod 900 Outer shelf* Crustaceans (79%) (Stevenson, 2004) Pseudophycis Demersal Fish (25%) (Cohen et al., 1990) bachus 0 - 700 m Molluscs (Fisheries New Zealand, 2019) Worms (Graham, 1939) Echinoderms (Edgar & Shaw, 1995) Sea perch 470 Slope* Crustaceans (62%) (Paulin, 1989) Helicolenus Demersal Fish (18%) (Smith, 1998) percoides 150 - 500 m Worms (Graham, 1939) Ling 2000 Slope* Fish (65%) (Nielsen et al., 1999) Genypterus Bathydemersal Crustaceans (37%) (Stevenson, 2004) blacodes 0 - 1000 m Cephalopods (3%) (Dunn et al., 2010) Ophiuroids (Graham, 1939) (Kailola et al., 1993) * Species assemblage according to Beentjes et al. (2002) and Francis et al. (2002) Percentage of occurrence of food items retrieved from Stevens et al. (2011)

3.2. Methods

3.2.1. Interannual variability in isotopic values within a sea perch population

Sea perch is a sedentary fish species widely distributed in New Zealand waters (Bentley et al., 2014; Lawton et al., 2010). To access natural interannual variation of isotopes in a local population, sea perch were collected South of the Otago Peninsula (-45.865 S, 170.753) during three research cruises during the austral Summer in 2015/16 (n=17), 2017/18 (n=18) and 2018/19 (n=17) aboard of the RV Polaris II. Specimens were frozen after collection and total length, as well as wet weight, were recorded at the University of Otago's Portobello Marine Laboratory (PML). A muscle tissue sample from the dorsal musculature was taken from every fish for isotopic analysis. Dissection and laboratory analysis of muscle tissues followed procedures described below (section 3.2.2). Isotope values, length and weight were compared between years with ANOVA, and linear regressions were used to investigate the relationship between length and weight, and isotope values for each sampling year.

3.2.2. Variability in resource use and trophic level of commercial fish species

Sample collection - To investigate spatial and temporal differences in resource use and trophic levels of commercial fishes, seven species of fish were collected along the east coast of the South Island, inside the Fishery Management Area 3 between 2017 and 2018 (Table 3.2). While some specimens were collected during scientific cruises aboard the RV Polaris II, others were provided by recreational and commercial fishers, and from the Canterbury Bight/Pegasus Bay, and Chatham Rise trawl surveys aboard of the RV Kaharoa and RV Tangaroa, respectively (MacGibbon et al., 2019; Stevens et al., 2018). These multiple sources of samples allowed for comparisons of trophic parameters of important commercial species over several habitats and across a large latitudinal range along the ECSI (Table 3.2 and Figure 3.1). Specimens collected aboard research vessels were measured for total length and head length, then frozen and taken to the laboratory for muscle tissue sampling. A muscle tissue sample of approximately 1cm3 from the dorsal musculature was taken from every fish for isotopic analysis.

Table 3.2. Average ± standard error of total length of specimens analyzed in the present study, with number of samples and location of collection. WCR stands for Western Chatham Rise

Samples Common name Scientific name Average total length ± SE (mm) Region (collection year) size Ling Genypterus blacodes 849.8 ± 37.8 19 Otago (2017, 2018) WCR (2018) Sea perch Helicolenus percoides 330.4 ± 22.9 38 Otago (2017, 2018) Kaikoura (2017, 2018) Tarakihi Nemadactylus macropterus 313.1 ± 12.4 59 Otago (2018) Akaroa (2018) Kaikoura (2018) Blue cod Parapercis colias 386.1 ± 6.9 56 Otago (2017, 2018) Kaikoura (2017, 2018) Red cod Pseudophycis bachus 546.5 ± 10.2 41 Otago (2017, 2018) Akaroa (2018) Kaikoura (2017, 2018) WCR (2018) Barracouta Thyrsites atun 682.8 ± 12.7 30 Otago (2017) Akaroa (2018) Kaikoura (2018) Slender jack mackerel Trachurus murphyi 501.3 ± 8.6 22 Otago (2018) Kaikoura (2018)

To calculate the isotope values of coastal primary producers to be used in the estimation of fishes’ trophic parameters, the most common macroalgae species were collected from sites in Otago and Kaikoura-Cape Campbell region throughout Summer 2017/18 and 2018/19 (Table 3.3). They were collected between the surface and 10 meters of depth and comprised the predominant species found in each site. Macroalgae were washed with deionized water to remove salt and epibionts and were then frozen for later analysis. To generate single δ13C and δ15N values for macroalgae, an average of the isotopic value of each species was first calculated for each site sampled, then an overall average among sites within a region of each species was computed. δ13C and δ15N macroalgae values for each region were then calculated as the average between all species’ overall average, assuming an equal contribution of the most common macroalgae species at each region to the final isotopic values. The average of isotopic values from suspended particulate organic matter were retrieved from δ13C and δ15N surface isoscapes south of Akaroa, between Akaroa and Cape Campbell and for deep regions offshore Akaroa and Kaikoura (Chapter 2). Due to the high heterogeneity in δ13C and δ15N values south of Akaroa, samples of new pelagic production were inferred from the nitrate:phosphate ratio of nutrients in seawater, after and Jones et al. (2013) (see Chapter 2).

Table 3.3. δ15N and δ13C average ± standard error of the macroalgae species collected from the Otago and Kaikoura coast and respective sample sizes Region Site Species δ¹⁵N δ¹³C Sample size Otago Karitane Carpophylum flexiosum 7.9 ± 0.09 -14.36 ± 0.13 3 Cystophora retroflexa 7.69 ± 0.13 -15.23 ± 0.38 2 Cystophora scalaris 7.93 ± 0.14 -14.08 ± 0.21 3 Macrocystis pyrifera 9.15 ± 0.36 -16.78 ± 1.05 8 Sargassum sinclairii 8.96 ± 0.13 -15.93 ± 0.9 3 Ulva sp. 9.25 ± 0.12 -15.18 ± 0.82 8 Undaria pinnatifida 9.35 ± 0.23 -18.17 ± 0.86 8 Aramoana spit Cystophora retroflexa 8 ± 0.31 -19.3 ± 0.72 2 Cystophora scalaris 7.02 ± 0.13 -18.71 ± 0.43 4 Durvillaea spp. 7.04 ± 0.42 -9.21 ± 1.96 6 Macrocystis pyrifera 7.51 ± 0.16 -18.78 ± 0.38 9 Marginariella boryana 8.32 ± 0.95 -16.16 ± 0.41 6 Ulva sp. 7.52 ± 0.32 -18.88 ± 0.3 10 Undaria pinnatifida 7.93 ± 0.39 -20.29 ± 0.71 3 Among site average 8.11 ± 0.21 -16.5 ± 0.77 Among species average* 8.1 ± 0.21 -15.93 ± 0.94 Kaikoura- Cape Campbell- Carpophyllum maschalocarpum 8.39 ± 0.23 -11.58 ± 0.86 5 Cape Lighthouse Durvillaea poha 6.55 ± 0.37 -11.61 ± 0.71 5 Campbell Durvillaea willana 8.27 ± 0.11 -11.45 ± 0.61 5 Hormosira banksii 8.78 ± 0.32 -8.66 ± 0.59 5 Ulva sp. 7.42 ± 0.21 -17.23 ± 0.67 5 Kaikoura North bay Carpophyllum maschalocarpum 8.28 ± 0.12 -10.19 ± 0.54 5 Cystophora retroflexa 9.68 ± 0.02 -10.94 ± 0.18 3 Eckonia radiata 8.97 ± 0.27 -15.58 ± 0.64 4 Lessonia variegata 8.65 ± 0.12 -15.73 ± 0.2 3 Ulva sp. 8.16 ± 0.16 -17.92 ± 0.46 4 Kaikoura South bay Carpophyllum maschalocarpum 6.46 ± 0.37 -9.21 ± 0.5 3 Cystophora retroflexa 8.55 ± 0.43 -12.4 ± 0.17 4 Eckonia radiata 7.52 ± 0.28 -17.85 ± 0.92 4 Ulva sp. 7.65 ± 0.14 -16.76 ± 0.44 5 Kaikoura-First Bay Carpophyllum maschalocarpum 8.84 ± 0.18 -13.24 ± 1.43 5 Durvillaea poha 7.62 ± 0.25 -10.52 ± 0.69 5 Durvillaea willana 8.25 ± 0.14 -10.59 ± 0.79 5 Macrocystis pyrifera 9.05 ± 0.3 -16.83 ± 0.39 5 Ulva sp. 8.66 ± 0.1 -16.68 ± 0.46 5 Oaru-North Carpophyllum maschalocarpum 7.14 ± 0.12 -10.65 ± 0.63 5 Durvillaea poha 6.12 ± 0.35 -10.09 ± 0.55 5 Durvillaea willana 6.51 ± 0.38 -12.31 ± 1.45 5 Ulva sp. 7.67 ± 0.17 -17.62 ± 0.73 5 Okiwi-South Durvillaea willana 6.5 ± 0.34 -11.43 ± 0.51 5 Ulva sp. 7.41 ± 0.08 -17.44 ± 0.32 5 Waipapa South Carpophyllum maschalocarpum 9.1 ± 0.19 -14.51 ± 0.54 5 Ulva sp. 9.08 ± 0.2 -20.9 ± 1.05 5 Among site average 7.93 ± 0.19 -14.15 ± 0.65 Among species average* 8.22 ± 0.25 -13.46 ± 1.1 *Used in the mixing models

Laboratory analysis - All muscle tissues and macroalgae samples were oven-dried at 60 °C for 72 h and crushed with mortar and pestle until they yielded a fine and homogenous powder. All equipment was cleaned with low-linting Kimwipes® and ethanol and air-dried

87 between samples to avoid cross-contamination. Between 0.8 and 1.2 mg of each muscle tissue sample and 3 to 4 mg of macroalgae were packed in 3x5 mm tin capsules and analyzed by combustion in an elemental analyzer (Carlo Erba NC2500) to CO2 and N2, as described in chapter 2 (section 2.2.2). All measured values for the quality control standards were in the range of accepted values. Replicates from one sample of fish muscle tissue were analyzed in every run so results could be corrected for between run variability.

Because lipids are depleted in δ13C compared to protein and carbohydrates, changes in lipid content in the samples can have a large effect in δ13C values that are not linked to ecological changes. In the present study the following normalization equation was applied to account for differences in lipid concentrations between all samples, after Post et al. (2007):

δ13CNormalized = δ13CUntreated – 3.32 + 0.99 * C:N 3.1)

13 13 Here, δ CUntreated represents raw δ C values and C:N the carbon to nitrogen ratio of the muscle tissue analyzed; this normalization is suitable for aquatic organisms and was generated with the same range of C:N values from samples in the present study (Post et al., 2007).

Trophic parameter estimations - To estimate TL and the relative contributions of organic matter from different source pools, a two-step iterative procedure based on bulk isotope values was used, following Phillips and Gregg (2001). First an approximation of the contribution of each organic matter source to a fish’s δ13C value was calculated from plotted δ15N vs. δ13C. The results were then used to estimate the corresponding δ15N of the mixture 15 of organic matter sources supporting each specimen (δ NSource) (Jack & Wing, 2011), assuming the same contribution from the δ15N pool. TL was then calculated for each fish 15 15 using its own δ N signature (δ NConsumer):

TL = [(δ15NConsumer − δ15NSource)/TDF] 3.2)

The result of the equation was then iterated back into the mass balance model until a stable solution was obtained for both the mixture of organic matter sources and TL for each specimen. A trophic discrimination factor for aquatic environments of 0.5‰ (SE 0.17) was used for δ13C (McCutchan et al., 2003). Because all species analyzed in the present study

88 consume a large array of prey items, including both vertebrates and invertebrates (Table 3.1), the trophic discrimination factor of δ15N (TDF) was estimated to be 2.3‰ (SE 0.28) for all species based on an average for aquatic environments (McCutchan et al., 2003). Trophic discrimination factors, both 13C and 15N, are known to vary between taxa and environments (Newsome et al., 2010), however, as the same species were being compared between regions, errors introduced by the use of an average discrimination factor should not have affected the detection of relative differences in TL and resource use of a species among regions.

Statistical analyses - Fish specimens sourced from recreational and local fishers were often represented by heads only, in these cases total length was estimated by species-specific linear regressions generated by all fish specimens, both collected and released during sampling (Table 3.4). Because latitude, as well as species and total length can have an effect in isotope values of fish muscle, a general linear model was fitted to investigate the effect of those variables on the isotopic values and results from the isotope mass balance models. General linear models were fitted using a least square approach, which is sensitive to outliers and performs well with noisy, collinear and incomplete variables, being suitable for analysis of complex models in chemistry and biology (Wold et al., 2001). Assumptions related to the linear models were checked with residual plots (by predicted values, by row and normal quantile).

Fish specimens were then grouped by region, according to the NZ Fisheries General Statistical Areas (https://koordinates.com/layer/4182-nz-fisheries-general-statistical- areas/metadata/): Otago (024), Akaroa (022), Kaikoura (018) and WCR (020). Interannual and regional differences (among statistical areas) in isotopic value, %Macroalgae and TL were tested using a one-way ANOVA and Tukey-Kramer’s post hoc test. When samples were available, the interaction between region (Otago and Kaikoura, 2 levels, fixed) and year (2017 and 2018, 2 levels, fixed) was also tested using a two-way ANOVA. Shapiro-Wilk test was used to check for data normality, while variance among groups was tested using Levene’s test. When the hypothesis that data came from a non-normal distribution or that groups presented different variances could not be discarded, each pair Wilcoxon’s tests were also applied. Because total length influenced the estimated TL and %Macroalgae for the species analyzed, the largest and smallest fishes were excluded as necessary before group comparisons, until the variance of the total length did not differ between groups (Wilcoxon test, p > 0.05).

Table 3.4. Linear relationship between head length (HL) and total length for each species analyzed in the present study, with respective summary of fit

Common name Scientific name Total length regression equation Sample size r2 p Ling Genypterus blacodes 150.81 + 3.8*HL 9 0.86 0.0003 Sea perch Helicolenus percoides 62.32 + 2.56*HL 56 0.88 <0.0001 Tarakihi Nemadactylus macropterus -13.95 + 4.21*HL 38 0.92 <0.0001 Blue cod Parapercis colias 26.36 + 3.5*HL 84 0.98 <0.0001 Red cod Pseudophycis bachus 127.16 + 3.47*HL 14 0.52 0.0036 Barracouta Thyrsites atun 163.32 + 3.3*HL 32 0.83 <0.0001 Slender Jack Mackerel Trachurus murphyi 54.79 + 3.59*HL 10 0.75 0.001

3.2.3. Differences in community niche breadth between Otago and Kaikoura

Otago and Kaikoura represent the southernmost and northernmost areas in the study region, where samples from all species of interest were available. Therefore, investigations described in sections 3.2.3 and 3.2.4 focuses on these two areas only. To investigate differences in niche breadth and trophic diversity between the fish communities from Otago and Kaikoura, Layman’s metrics based on the δ15N and δ13C values of fish muscle tissue from specimens collected at each region were compared (Layman et al., 2007). To account for uncertainty related to the nature of the data, ellipse-based versions of Layman’s metrics estimated from Bayesian inference were used, with the SIBER package in R (Jackson et al., 2011). The use of ellipse-based metrics is suitable for comparisons between communities with different sample sizes, with improved estimates and reduced uncertainty compared to the original metrics. Minimum convex hull values of the δ13C and δ15N biplot, or isotopic space, can be easily biased by large isotope values from few individuals. The calculation of the Standard Ellipse Areas corrected for small sample sizes instead provides Bayesian distributions that reflect the uncertainty of estimates, resulting in more robust and less biased comparisons of niche breadth. Communities were analyzed as a whole and between specific fish assemblages, according to their habitat use after Francis et al. (2002) and Beentjes et al. (2002) (Table 3.1).

3.2.4. Amount of coastal primary producers supporting commercial catches

Regional differences in coastal resources needed to support local commercial fisheries were calculated by transforming green (ungutted) weight of the reported landed catch per statistical area into dried muscle weight (Venugopal & Shahidi, 1996; Waterman, 2001) and estimating its calorific content per species (Vlieg, 1988). Assuming a trophic efficiency of 10%, equation 3.3 was used to calculate the amount of energy from primary producers needed 90

to support commercial catches, where EPP is the energy from primary producers and ETLn is the energy from the nth trophic level (Udy et al., 2019c). The results were transformed into macroalgae biomass using an average kcal to gram conversion factor for New Zealand species (Lamare & Wing, 2001) and the results of the isotope mixing model from the present study. To remove interannual variability from the analysis, yearly macroalgae mass supporting each species catch were averaged between 2013 and 2017 and compared between regions. Fisheries data were provided by Fisheries New Zealand. All statistical analyses were undertaken with JMP 14.0.0 (SAS Institute, 2018) and R 3.2.6 (R Core Team, 2020).

3.3. Results

3.3.1. Interannual variability in isotopic values within a sea perch population

δ13C and δ15N values from a sea perch population from Otago varied among years 13 15 (δ C; F2,51 = 20.95, p < 0.0001; δ N, F = 37.22, p < 0.0001). A Tukey’s post-hoc test showed that δ13C and δ15N values did not differ significantly between 2017/18 and 2018/19, but isotopic values were significantly higher in 2015/16 at p < 0.05 level (13.4 ± 0.14 versus 12.14 ± 0.13 and 11.88 ± 0.14 for δ15N; and -18.04 ± 0.13 versus -19.18 ± 0.13 and -18.94 ± 0.13 for δ13C). Although fish length and weight were not significantly different between periods, they displayed a positive linear relationship with δ13C and δ15N for all years (p < 0.05), except for δ13C values from samples collected in 2015/16, which had no relationship with length and weight.

3.3.2. Variability in resource use and trophic level of commercial fish species

Effects of latitude, species, and total length - Latitude (p < 0.0001), species (p < 0.0001) and total length (p = 0.0009) significantly contributed to the general linear model 13 2 predicting δ C values of fish muscle tissue (F7,185 = 42.59, r = 0.63 and p < 0.0001). Results and p values were similar for δ15N, but total length was not significant predictors for this variable (F = 27.36, r2 = 0.52 and p < 0.0001). Total length, latitude and species had significant effects on the %Macroalgae to the underlying food web (F = 12.15, r2 = 0.32 and p <0.0001). Total length (p < 0.0001) and latitude (p < 0.0001) had a positive and negative relationship with %Macroalgae, while different species (p < 0.001) had contrasting relationships with %Macroalgae. Similar results were found for the estimated δ15N value of 15 the organic matter baseline (δ NSource) value supporting the underlying food web (F = 72.39, 2 15 r = 0.74 and p < 0.001), but latitude had a positive effect on δ NSource (average slope ± SE = 0.25 ± 0.01, p < 0.0001). Species was the only variable with a significant effect on fish TL (F = 15.28, r2 = 0.38, p < 0.0001).

When general linear models were fitted for single species, latitude had a significant positive effect on the variation in δ13C for all species except barracouta, with slopes ranging from 0.21 ± 0.04 for sea perch to 0.97 ± 0.13 for ling (Table 3.5). Similarly, δ15N values were positively correlated with latitude for all species except blue cod and tarakihi. Barracouta and sea perch were the only species where δ13C varied significantly with total length (p < 0.005), while δ15N values also varied significantly with total length for red cod (slope of 0.004 ± 0.002), sea perch (0.01 ± 0.004) and tarakihi (-0.003 ± 0.002). The contribution of organic

92 matter derived from macroalgae to the food webs supporting ling, barracouta, red cod and sea perch diets displayed a linear negative relationship with latitude and a positive relationship with total length (except for ling), while blue cod and tarakihi had no significant relationships among these variables (Table 3.5). Barracouta and ling displayed a significant increase in estimated TL with latitude (slope of 0.14 ± 0.07, p = 0.047 and 0.23 ± 0.07, p = 0.007 respectively), but red cod, sea perch and tarakihi had small variations in TL with total length (slope of 0.002 ± 0.001, p = 0.006; 0.004 ± 0.002, p = 0.012; and -0.002 ± 0.001, p = 0.017 respectively). Blue cod had no linear relationship between TL and the other variables. While %Macroalgae had no significant relationship with total length for slender jack mackerel, δ13C and TL values displayed a positive (0.01 ± 0.007, p = 0.04) and negative slope (-0.004 ± 0.002, p = 0.05) respectively, from the total length univariate linear model, but tests including latitude for slender jack mackerel were not possible due to lack of precise latitude values provided by fishers.

Table 3.5. General linear model results describing the relationship between δ13C and δ15N values, the proportion of macroalgae supporting each species’ diet (%Macroalgae) and it’s estimated trophic level (TL), with latitude and fishes’ total length as the independent variable for each species. Model parameters are given for total model (latitude and total length effect), while p values are also reported for each effect separately Slender Variable Parameters Barracouta Blue cod Ling Red cod Sea perch Tarakihi jack mackerel1 δ13C r2 0.32 0.52 0.85 0.31 0.73 0.57 0.31 DF (Model, total) 2, 29 2, 34 2, 13 2, 30 2, 35 2, 37 1, 21 F ratio 6.39 17.71 30.23 6.784 45.649 23.426 9.139 Total p value 0.005 < 0.0001 < 0.0001 0.004 < 0.0001 < 0.0001 0.007 Latitude p value 0.471 0.0003 < 0.0001 0.002 < 0.0001 < 0.0001 - Length p value 0.002 0.49 0.299 0.773 < 0.0001 0.058 - δ15N r2 0.3 0.27 0.73 0.24 0.54 0.11 0.08 DF (Model, total) 2, 29 2, 34 2, 13 2, 32 2, 35 2, 37 1, 21 F ratio 5.84 5.85 14.979 4.655 19.458 2.233 1.704 Total p value 0.008 0.007 0.0007 0.017 < 0.0001 0.122 0.207 Latitude p value 0.002 0.246 0.0002 0.01 0.001 0.566 - Length p value 0.424 0.102 0.35 0.023 0.003 0.044 - %Macroalgae r2 0.64 0.004 0.62 0.59 0.64 0.03 0.14 DF (Model, total) 2, 29 2, 34 2, 13 2, 32 2, 35 2, 37 1, 21 F ratio 23.86 0.064 8.812 21.52 29.657 0.629 3.226 Total p value < 0.0001 0.938 0.005 < 0.0001 < 0.0001 0.539 0.088 Latitude p value < 0.001 0.893 0.002 0.001 < 0.0001 0.64 - Length p value 0.002 0.87 0.382 0.002 < 0.0001 0.47 - TP r2 0.17 0.09 0.51 0.28 0.3 0.3 0.18 DF (Model, total) 2, 29 2, 34 2, 13 2, 32 2, 35 2, 37 1, 21 F ratio 2.71 1.564 5.665 5.891 6.926 7.226 4.35 Total p value 0.084 0.225 0.02 0.007 0.003 0.002 0.05 Latitude p value 0.047 0.188 0.007 0.709 0.197 0.13 - Length p value 0.22 0.09 0.373 0.007 0.012 0.017 - 1Total length was the only variable to be included in the model describing the variables of jack mackerel

Regional and temporal differences in trophic parameters - Blue cod and sea perch were collected from both Otago and Kaikoura regions during 2017 and 2018 sampling seasons. The interaction between region (2 levels, fixed) and year (2 levels, fixed) had a significant effect on the %Macroalgae for blue cod diet (F3,40 = 8.59, p = 0.006), while for sea perch, only region had a significant effect on %Macroalgae (F3,28 = 41.15, p < 0.0001). Tukey’s post hoc tests supported the idea that differences in the contribution of organic matter derived from macroalgae were significantly different between the two regions in 2017 and 2018 for sea perch and in-between years in Kaikoura for blue cod. The interaction between region and year had a significant effect on TL for blue cod (F = 4.65, p = 0.037) while region (F = 4.86, p = 0.03) and year (F = 13.18, p = 0.001) had separate effects on sea perch TL. The estimated %Macroalgae for blue cod varied between years for Kaikoura (43 ± 2% for 2017 and 36 ± 1% for 2018, p = 0.007), TL of blue cod was significantly different between 2017 and 2018 in Otago (3.06 ± 0.11 and 2.61 ± 0.07, p = 0.005) and in Kaikoura for sea perch (2.31 ± 0.09 and 3.02 ± 0.18, p =0.004).

Each pairwise comparison with Wilcoxon tests agreed with a majority of the two-way ANOVA results for pairs of variables with unequal variances and non-normal distributions, but showed significant differences in the contribution of organic matter derived from macroalgae to the diet of blue cod between regions in 2017. Significant differences were also found in sea perch %Macroalgae between years in Kaikoura using non-parametric tests, but differences in %Macroalgae between regions were not found in 2018. Significant regional and temporal differences in blue cod TL were identified with Wilcoxon’s test for all year and region combinations at a p < 0.045 level, except for interannual differences in Kaikoura.

When data from 2017 and 2018 were pooled together, all species except blue cod had significant regional differences in the contribution of organic matter derived from macroalgae to the underlying food web (Figure 3.2). The diet of barracouta from Otago had a significantly higher contribution of organic matter derived from macroalgae compared to that recorded from Akaroa and Kaikoura (51 ± 2% versus 42 ± 1.4% and 34 ± 2.7% respectively,

F2,27 =18.36, p < 0.0001). A similar trend was observed for sea perch (53 ± 0.9% for Otago and 42 ± 1.0 for Kaikoura, F1,30 = 60.13, p = 0.0001), ling (62 ± 1.6% for Otago and 46 ±

1.3% for WCR, F1,17 = 56.6, p < 0.0001), red cod (51 ± 1.6% for Otago, 48 ± 1.9% for

Akaroa, 41 ± 1.9% for WCR and 42 ± 1.4% for Kaikoura F3,37 = 7.60, p = 0.0004) and slender jack mackerel (44 ± 2.1% for Otago and 34 ± 2.5% for Kaikoura F1,19 = 9.09, p = 0.007). Tarakihi also displayed significant differences in %Macroalgae (42 ± 2.8% for Otago, 38 ±

3.5% for Akaroa, 44 ± 3.6% for WCR and 32 ± 2.8% for Kaikoura F3,51 = 3.1, p = 0.035), but

94 pair-wise Tukey-Kramer’s post hoc tests only resolved differences between WCR and Kaikoura. Blue cod did not show regional differences in the contribution of organic matter

derived from macroalgae supporting its diet.

) Trophic (TL level

Figure 3.2. Average ± standard error of the contribution of organic matter derived from macroalgae to a fish’s diet (%Macroalgae) and the trophic level (TL) for each species collected from Otago (24), Akaroa (22), Western Chatham Rise (20) and Kaikoura (18). Blank columns correspond to missing data. The variance of the total length from individuals collected at regions did not differ for any species (Wilcoxon’s test, p > 0.05), except for tarakihi and red cod collected from region 22, which were significantly smaller and larger than for the other regions, respectively

TL differed significantly between regions for barracouta, ling, slender jack mackerel and tarakihi. Barracouta specimens collected from the Akaroa region had higher TL (3.27 ± 0.05), when compared to those collected from Otago (2.28 ± 0.06) and Kaikoura (2.75 ± 0.1), F = 73.73, p < 0.0001. Slender jack mackerel had higher TL values in Otago (3 ± 0.06) than in Kaikoura (2.47 ± 0.08), F = 29.46, p < 0.0001. TL of tarakihi was also higher in the south (3.13 ± 0.09 in Otago and 3.33 ± 0.1 in Akaroa) than in the north (2.93 ± 0.08 in Kaikoura and 2.62 ± 0.1 in WCR), F = 8.39 and p = 0.0001. In contrast, ling collected from WCR had 95 higher values of TL (3.12 ± 0.08) compared to that of specimens collected from Otago (2.71 ± 0.09), F = 11 and p = 0.004. In each of these cases, results from a non-parametric Wilcoxon test gave the same results as ANOVA, except for red cod and tarakihi. While red cod showed significant differences in %Macroalgae between Akaroa and Kaikoura (z = 2.56, p = 0.011), tarakihi had significant differences in %Macroalgae values between Otago and Kaikoura (z = 2.38, p = 0.018) and Akaroa and WCR (z = -2.43, p = 0.015) but not Akaroa and Kaikoura. TL of tarakihi was also significantly higher in Kaikoura than in WCR when analyzed with a non-parametric test (z = -2.84, p = 0.005). When results from Otago and Kaikoura (the regions with the largest amount of species sampled) were compared, there was a clear difference in TL of individual species, with a larger percentage of fish diets supported by

organic matter derived from macroalgae in Otago compared to Kaikoura (Figure 3.3).

ophic (TL) level Tr

Relative contribution from macroalgae (%Macroalgae)

Figure 3.3. Average trophic level (TL) and relative contribution of organic matter derived from macroalgae (%Macroalgae) to a fish’s diet (0 to 1) for an exploited fish community sampled from Kaikoura (grey symbols) and Otago (black symbols). Error bars represent ± 1 standard error

3.3.3. Differences in community niche breadth between Otago and Kaikoura

Results from Layman’s metrics analysis suggested differences in niche space for the communities sampled in Otago and Kaikoura. Ranges of δ15N (2.29 and 2.11‰) and δ13C (1.18 and 1.25‰), CD (0.88 and 0.72‰) and SDNND (0.34 and 0.33‰) showed small differences between Otago and Kaikoura communities, with overlapping 95% credible intervals. In contrast, MNND was higher in Kaikoura (0.71‰) than in Otago (0.45‰), although they were also not significantly different when comparing credible intervals. The

Standard Ellipse Area corrected for small sample sizes (SEAc) showed an increase in niche space inferred by isotopic values, from 2.09‰2 in Otago to 2.8‰2 in Kaikoura (Figure 3.4).

Species that inhabit the inner shelf and slope had only small differences in SEAc calculated from Otago and Kaikoura,1.44 to 1.24‰2 and 0.74 to 0.52‰2, respectively. In contrast, 2 assemblages from the outer shelf had much smaller SEAc values in Otago (1.94‰ ) than Kaikoura (2.78‰2), at a 75% credible interval (Figure 3.5). These dissimilarities in metrics of community niche space between regions were largely driven by changes in the niche space of tarakihi (Figure 3.6). While all the species inhabiting the outer shelf showed differences in

SEAc, tarakihi was the only one to be above the 95% credible interval, with an increase of 2 almost 3-fold in SEAc between Otago and Kaikoura (1.24 and 3.31‰ ).

Figure 3.4. Isotopic niche space represented by δ13C and δ15N from fish communities sampled from Otago (black) and Kaikoura (red) circled with their respective Bayesian inferred ellipses. Dotted lines represent the minimum convex hull of each community

3.3.4. Amount of coastal primary producers supporting commercial catches

The Otago region had, on average, higher total commercial fisheries landings of key exploited species (barracouta, blue cod, red cod, sea perch, slender jack mackerel and tarakihi) when compared to landings from Kaikoura between 2013 and 2017, 1162.5 versus 880.1 tonnes. Nevertheless, the biomass of organic matter derived from macroalgae estimated to support those catches was very similar, 26,678.4 versus 27,648.9 tonnes respectively. The biomass of organic matter derived from macroalgae that supported catches of tarakihi, slender jack mackerel and barracouta was different between regions. Accordingly, the ratios of tarakihi’s and slender jack mackerel’s catch to macroalgae biomass supporting its landings were higher by two and four-fold respectively between Kaikoura (0.033 and 0.083) and Otago (0.017 and 0.019). In contrast, barracouta had the reverse trend, with a higher ratio in Otago (0.097) than in Kaikoura (0.049). Sea perch had the highest ratio for all the species but did not show large differences between regions (0.11 for Kaikoura and 0.13 for Otago), similar to blue cod and red cod.

3.4. Discussion

Differences in δ15N and δ13C values of fish muscle tissue, as well as estimated %Macroalgae and TL displayed linear relationships with latitude for all species analyzed, except for blue cod. Values of isotopes and trophic parameters varied between regions and in some cases, among years. Similarly, values of δ15N and δ13C, as well as estimated %Macroalgae and TL were found to have linear relationships with total length for all species 15 analyzed, except blue cod. Although %Macroalgae and δ NSource estimated in the present study had a significant relationship with latitude, fish length and varied between species, TL was not significantly correlated with latitude or fish length. Furthermore, general linear models demonstrated that TL and %Macroalgae of all species, except blue cod, varied with specimens’ size. These results corroborate a large amount of evidence that isotope values and trophic parameters of consumers can present large variations in space and time (Cherel & Hobson, 2007; Trueman et al., 2012). The results also demonstrate that, although specimens of different sizes do rely on different resources and feeding pathways, and variation of TL with length within each species does exist, species commonly occupied a specific range of trophic levels within each community. The observed pattern corroborates earlier studies indicating that although fish TL is related to body size, these patterns are usually driven by ontogenetic shifts and habitat use within populations (Dalponti et al., 2018; Ladds et al., 2020).

After the variation in total length among species was accounted for by the removal of outliers, a decreasing trend in the contribution of organic matter derived from macroalgae to a fish’s diet from south to north was demonstrated. The trend was consistent among all species, except for blue cod where no linear trend was resolved. Differences in TL between regions were more variable, but there was a general trend of higher TL in the northern regions, i.e. WCR and Kaikoura, for barracouta, slender jack mackerel and tarakihi. Ling collected from deeper environments inside the WCR had a significantly higher trophic level than specimens collected from Otago, supporting previous studies that indicated ling feeds at a higher trophic level in deepwater habitats (Dunn et al., 2010).

Regional differences of TL were observed to differ between the sampling years in Otago for blue cod and in Kaikoura for sea perch. Blue cod from Kaikoura displayed a decrease in %Macroalgae between years. Similarly, an increase in TL of sea perch from Kaikoura was observed, while blue cod sampled from Otago decreased in TL between 2017 and 2018. These observed variations could be a result of a regional change in the isotopic values of the basal organic matter sources between years, although previous studies indicate

99 that δ13C and δ15N values of macroalgae and SPOM have long-term stability in these regions (Van Hale, 2003; Vanderklift & Bearham, 2014). Changes in resource use with time was also evidenced by changes in δ15N and δ13C values of muscle tissue samples from a sea perch population collected from Otago in 2015/16 and the later sampling years in 2017/18 and 2018/19. Variation in length and weight of sea perch specimens was not observed among years, while δ13C did not display the positive linear relationship with length and weight observed during the 2017/18 and 2018/19 sampling seasons. Likewise, there was no clear relationship between δ13C and C:N for any given year, indicating that the differences observed here were not related to contrasting lipid contents from different individuals. These results demonstrate that isotope values of fish muscle tissue can present temporal variability that is not linked to regional or individual body condition changes. The higher values of δ13C and δ15N observed in samples collected in 2015/16 indicate a higher contribution of organic matter derived from macroalgae compared to organic matter from SPOM supporting the underlying food web of fish. This pattern indicates a shift in benthic-pelagic coupling processes between years, likely driven by prey resource use and availability in the region during that period (Funes et al., 2018; Hobson et al., 1995). For example, Funes et al. (2018) have shown that different morphological types of squat lobster (Munida gregaria) exploit basal organic matter sources with contrasting δ13C values. M. gragaria can contribute disproportionally to the diet of commercial fishes in Otago (Graham, 1939), including sea perch, and have high variation in abundance in the region (Zeldis & Jillett, 1982). Therefore, temporal changes in its abundance and resource use would likely alter energy transfer and cause the variation in isotope values observed in the present study.

Contributions of organic matter derived from macroalgae were significantly lower in Kaikoura than in Otago for all species, except blue cod. Blue cod have small home ranges as adults and is the only species analyzed in the present study that is found only in coastal waters shallower than 150 meters (Carbines & Mckenzie, 2001; Rodgers & Wing, 2008; Russell, 1983). These results are consistent with the idea that differences in the contribution of organic matter derived from macroalgae affect the trophic ecology of species outside coastal environments (Udy et al., 2019b). These differences challenge the similarity between the coastal marine habitats in Otago and Kaikoura, with a relative lack of coastal-offshore connectivity in the latter. One explanation for the observed pattern would be the reduced export of organic matter from the coastal zone in Kaikoura when compared to Otago. Schiel et al., (2019) reported coastal habitat modification and mass mortality of kelp and grazing invertebrates after the Kaikoura 7.8 magnitude earthquake in 2016, which could have resulted in the patterns seen in the present study. This natural event has resulted in coastal uplifts of 100 between 1 and 6 meters (Hamling et al., 2017) and removal of the benthic community in the region (Mountjoy et al., 2018). Even 19 months later, biophysical conditions remained altered, with low cover of habitat-forming brown kelp and high sedimentation rates (Alestra et al., 2019). Because of the lack of long-term isotopic data on the commercial species, comparing these results with pre-earthquake values is not possible, but the findings corroborate information collected on shifts in habitat use by sperm whales (Physeter macrocephalus, Physeteridae) in Kaikoura after the 2016 earthquake (Guerra et al., 2020). Sperm whales are top predators that feed on large fishes as well as an array of squid species (Guerra, 2018) and typically alter foraging behavior with changes in prey distribution. Therefore, shifts in feeding behavior reported by Guerra et al., (2020) are likely related to changes in fish distribution, linked to the regional differences in resource use observed for the exploited fish community. Habitat fragmentation and reduction of organic matter export by coastal communities would likely drive shifts in %Macroalgae for species outside the reefs when compared to reef-associated species such as blue cod, with impacts on their life histories. Surveys have also demonstrated that relative abundance, size and age structure of blue cod populations in Kaikoura have not significantly changed after the 2016 earthquake (Beentjes & Page, 2018), indicating none or small effects to the local blue cod population.

There were large differences in Layman’s metrics calculated from community samples between Otago and Kaikoura. SEAc indicated that the isotopic niche space of the exploited community sampled from Kaikoura was larger than in the Otago region. The observed pattern could be primarily explained by larger isotopic variability among species from the outer shelf habitats. When SEAc was calculated for each species inhabiting the outer shelf, tarakihi was the only species with significant differences in isotopic niche space between Otago and Kaikoura. These results demonstrate that the variability in the niche space of tarakihi populations between regions was largely responsible for the differences in community SEAc observed. Previous studies have shown an increase in individual niche variation of Arctic charr (Salvelinus alpinus, Salmonidae) due to intense intraspecific competition for a reduced amount of resources (Amundsen, 1995). Tarakihi is a demersal fish that feeds on a broad range of infaunal and epifaunal invertebrates by ingestion of sediment (Godfriaux, 1974), highly susceptible to habitat degradation from anthropogenic (trawling) and natural (earthquake) impacts (Clark et al., 2016; Jennings et al., 2001), which could have caused the increased niche breadth measured in Kaikoura. Distance from past earthquake epicenters was among the most important parameters describing variations of fish prey invertebrates in Canterbury Bight, while their abundance was negatively correlated with fishing effort, reducing prey biomass and productivity by half (Tuck et al., 2017). Trawling is 101 known to have a direct (fish removal) and indirect effect (bottom habitat degradation) on local fish communities, with the potential to adversely affect fisheries yields (Clark et al., 2016; Jennings et al., 2001). Historical data documents large differences in trawling effort between Otago and Kaikoura between 1989 and 2005 (Baird et al., 2011), reflecting the higher historical commercial landings of local demersal species in Kaikoura (Annala, 1988). Furthermore, the most recent stock assessment reported that tarakihi stocks were overfished inside the FMA 3, with biomass significantly below the soft limit (MPI, 2018). Individual and species niche variation have been previously linked to contrasting morphological traits in fish communities (Ingram & Shurin, 2009; Schluter & McPhail, 1992), this is unlikely to be true for tarakihi, due to its non-selective feeding strategy (Godfriaux, 1974; Graham, 1939).

Mean nearest neighbor distance in the isotopic values of individual fish species were 40% lower in Otago than in Kaikoura. Higher values of MNND indicate a more dispersed food webs with less trophic overlap and a potential for an increased number of feeding pathways. These results were mainly due to similar TL displayed by barracouta and sea perch, as well as by tarakihi and slender jack mackerel in Otago but not in Kaikoura. Large omnivores and carnivores that occupy similar lower positions on the food web, but still have different traits, as seen in the present study for Otago, can help to increase food web stability and resilience to anthropogenic stressors (Petchey et al., 2008; Peterson et al., 1998). The decreased trophic overlap estimated for the same community in Kaikoura likely indicates a marine system experiencing resource limitation, with associated decreases in productivity (Bolnick et al., 2003; Jack & Wing, 2011). These trophic patterns are associated with reduced resilience to impacts and more susceptibility to declines due to local extinction and harvesting (Kadoya & McCann, 2015). Although the results of the present study showed remarkable regional differences in the trophic structure of the exploited fish communities, long-term studies are needed to investigate the drivers of these differences.

While the estimated biomass of organic matter derived from macroalgae required to support fisheries was very similar for Otago and Kaikoura, commercial catches were almost 25% higher in Otago. Fishes from Otago also had a higher reliance on macroalgal productivity, which was coupled with lower TL for the most abundant species in the fisheries. Commercial catch per unit of effort (CPUE) of the exploited fish species in Otago and Kaikoura were calculated by dividing the estimated commercial catch by the number of fishing events, between 1990 and 2017, provided by Fisheries New Zealand. Slender jack mackerel was excluded from those comparisons due to extremely large catches before the year 2000, which were probably comprised of a mixture of mackerel species. The total

102 number of fishing events in each region had the same amount of variation among years, but fishing effort in Otago was lower (3289 ± 128 events per year) throughout the whole period when compared to fishing effort in Kaikoura (3947 ± 178 events per year). Averages of the total catch per year were comparable between regions in the time series up until 2005 (1476 ± 196 tons in Otago and 1642 ± 70 tons in Kaikoura), after that total catches decreased to 831 ± 41 tonnes in Kaikoura and to 1093 ± 101 tonnes in Otago. CPUE in Otago were relatively stable over time except for a small decrease after 2004 for tarakihi and sea perch, and apparent recovery in 2011. On the other hand, CPUE in Kaikoura presented a steep decline for all species after the year 2004 (Figure 3.7). Organic matter derived from macroalgae is known to support a high diversity of marine organisms and is especially important supporting the biomass of high trophic level species and top predators (Koenigs et al., 2015; Lusseau & Wing, 2006; McLeod & Wing, 2007). In the present study, species contributing to the highest catches in the southern regions of the study area relied more heavily on macroalgae production and occupied a lower trophic level than those in the north. Predator abundance and distribution are known to vary with prey biomass and availability. Prey biomass can be enhanced by external subsidies (e.g. macroalgae production), increasing local predator abundance beyond what the local resources (e.g. phytoplankton) can support (Polis et al., 1997; Udy et al., 2019c). The higher reliance on coastal resources by the local food web was concomitant to the decrease in trophic level for barracouta and sea perch, likely reflecting increases in the trophic overlap by these species (e.g. Lawton et al. 2010; McLeod et al. 2010). Although the identification of the drivers for the observed differences in trophic structure was not possible, higher reliance on macroalgal productivity was associated with a more stable food web that supported higher fish biomass and fisheries yields (Udy et al., 2019c, 2019b).

While most of the specimens were collected during the spring and summer months, fishes from the Canterbury Bight and Pegasus Bay Trawl survey were collected throughout April and May 2018. While differences in the timing of sampling may have resulted in differences in isotopic values through shifts in baseline values of organic matter source pools, muscle tissue used in the present study has a slow turnover rate (Skinner et al., 2017; Suring & Wing, 2009). Accordingly, the effects of change in isotopic values of baselines are expected to be minimal and within the range of variability in isotopic values of source pools used in the present study (Van Hale 2003, Chapter 2). To estimate the contribution of organic matter derived from macroalgae to the commercial catch in each region, fisheries data from Fisheries New Zealand were used. In this dataset, green weight catches from all fishing methods were estimated by the fishermen for the top 5 or 8 species caught during each fishing 103 trip (depending on fishing method) and data was only compiled if three or more boats were fishing at the time data was collected. Accordingly, the dataset has an inherently low resolution of catch histories. Nevertheless, Fisheries New Zealand data are the only available data with which to compare commercial catches between regions inside the ECSI. These datasets were used to evaluate long-term, common catch trends between regions, while absolute values and single year averages were not analyzed, despite lack of accurate and complete data collection. Although only seven commercial species were analyzed in the present study, they were representatives of the inshore, outer shelf and slope habitats, including demersal (blue cod, red cod, tarakihi, sea perch), benthopelagic (barracouta), pelagic (slender Jack Mackerel) and bathydemersal (ling) species with high commercial importance, both in regards of value and total landings (Fisheries New Zealand, 2019; Stats NZ, 2019). Because there was a relationship between length and isotopic values used to calculate the trophic level of fish, some of the results presented here were dependent on specimen’s size, which should be considered in future comparisons.

The present study is the first to investigate temporal and regional differences in the trophic structure of exploited fish communities along the east coast of the South Island of New Zealand. Here trophic parameters estimated from isotopic analysis of fish muscle tissue varied along the coast and between regions sampled. The observed differences indicated a lower reliance on macroalgae productivity, fisheries yield and trophic overlap in Kaikoura region when compared to a similar fish community in Otago. Community-wide analysis indicated that tarakihi was the only species with larger niche breadth in Kaikoura compared to Otago. This pattern could be explained by the recent overexploitation of tarakihi as well as the chronicle disturbance of benthic habitats from trawling and the Kaikoura earthquake, but long-term studies on changes in the trophic ecology of species are needed to pinpoint these impacts. Here evidence of regional differences in the trophic structure and use of coastal organic matter by an important exploited community managed as a single stock was provided. These results provide a first step towards gathering vital ecological information for the ecosystem-based management of these local fisheries resources.

104

)

events

fishing fishing

x x

Catch per unit of Catch (kg per effort

Figure 3.7. Commercial estimated reported landings per unit of effort (kg x fishing events-1) between 1990 and 2017 for Otago and Kaikoura regions for five different species. Source: Fisheries New Zealand

105

Chapter 4 Effects of fixatives on stable isotopes of fish muscle tissue: implications for trophic studies on preserved specimens

“This chapter has been published in the peer-reviewed scientific journal “Ecological Applications” on the 20th of February 2020, with contributions from Amandine Sabadel, Russel Frew, Travis Ingram and Stephen Richard Wing (Appendix A.4.1).”

Abstract

Isotopic ecology has been widely used to understand spatial connectivity and trophic interactions in marine systems. However, its potential for monitoring an ecosystems’ health and function has been hampered by the lack of consistent sample storage and long-term studies. Preserved specimens from museum collections are a valuable source of tissue for analyses from ancient and pre-modern times, but isotopic signatures are known to be affected by commonly used fixatives. The aim of the present study was to understand the effects of 13 15 13 fixatives on isotopic signatures of bulk tissue (δ Cm and δ Nm) and amino acids (δ CAA and 15 δ NAA) of fish muscle and to provide correction equations for the isotopic shifts. Two specimens of each: blue cod (Parapercis colias), blue warehou (Seriolella brama) and king salmon (Oncorhynchus tschawytscha) were sampled at five locations along their dorsal musculature, at four time periods: (1) fresh, (2) after one month preserved in formalin, (3) three and (4) twelve months fixed in either ethanol or isopropanol. Lipid content was positively correlated with C:N ratio (r² = 0.83) and had a significant effect on δ13C after 15 13 15 treatments, but not on δ N. C:N ratio (for δ Cm) and %N (for δ Nm) from preserved specimens contributed to the most parsimonious linear models, which explained 79% of the 13 15 13 variation due to fixation and preservation for δ C and 81% for δ N. δ CAA were generally 15 not affected by fixatives and preservatives, while most δ NAA showed different signatures 15 between treatments. δ NAA variations did not affect the magnitude of differences between amino acids, allowing scientists to retrieve ecological information (e.g. trophic level) independently of time under preservation. Corrections were applied to the raw data of the 13 15 experiment, highlighting the importance of δ Cm and δ Nm correction when fish muscle tissues from wet collections are compared to fresh samples. The results of the present study

106

13 15 13 15 make it possible to retrieve δ Cm, δ Nm, δ CAA and δ NAA from museum specimens and can be applied to some of the fundamental questions in ecology, such as trophic baseline shifts and changes in community’s food web structure through time.

107

4.1. Introduction

The impacts of anthropogenic activities have become increasingly evident in both coastal and oceanic marine environments, with consequences for the structure and function of marine ecosystems (Harley et al., 2006; Shackell et al., 2010). For example, oceanic warming, eutrophication and overfishing have been linked to dramatic changes in marine ecosystems at decadal to 100-year time scales (Jackson, 2001; Wing & Wing, 2001). Overfishing and climate variations, resulting in stock collapses, has emerged as a primary stressor in these systems. Information gleaned from fish scales and vertebra preserved in archival collections has proved to be a useful resource for resolving changes in food web structure over time and providing a reference point for studies of modern systems (Rosenberg et al., 2005; Valdés et al., 2008). Similarly, stable isotope analyses (SIA) have been used to resolve trophic level decreases of important commercial species at decadal time scales, related to fishing pressure and stock status (Wainright et al., 1993), presenting a powerful tool to access changes in trophic structure in marine systems over longer time scales.

Ecological studies that focus on unraveling the effects of stressors within specific habitats often use nearby “control” or “reference” sites for comparison with impacted sites. The approach relies on the assumption that the effect of a stressor on a given response variable will be greater than the inherent spatial heterogeneity of that variable among sites. The assumption is often violated in field studies where heterogeneity of habitats is high. An alternative approach is to use time series to resolve the impact of a stressor or shift in ecological baseline conditions. In this case, long-term data can be used to resolve ecological changes across time scales that encompass more than a human lifetime or scientific career (Chen et al., 2011). Long-term ecological studies highlight how environmental baselines are important when measuring the extent of stressors, or the potential for recovery of ecosystems (Lotze et al., 2011; Vander Zanden et al., 2003). Finally, incorporating long-term data into the framework of ecological science and management provides an important reference for studies of modern ecological systems (Jackson, 2001; Tallis et al., 2004; Wing & Wing, 2001).

Although there has been an increasing number of studies on food web dynamics and ecological baseline shifts over decadal and 100-year time scales in aquatic environments, most of the work has been carried out in freshwater systems (Kishe-Machumu et al., 2017; Schmidt et al., 2009; Vander Zanden et al., 2003). Trophic data acquired from whole, preserved specimens from past time periods were directly compared with those from specimens collected in modern times. In contrast, investigations in marine systems usually

108 rely on data acquired from otoliths, scales and bones, which can limit the analytical tools available to estimate trophic relationships. Most relevant historical specimens have been fixed in formalin to preserve soft tissues and/or stored in alcohol, but these preservation methods can influence the data, including stable isotope values, that can be obtained from these specimens (Sweeting et al., 2004).

Muscle tissue has relatively slow turnover rates compared to other tissues. Therefore, it is the most common tissue used to identify dietary variations and the routing of organic matter in ecological studies, especially in aquatic environments (Dalerum & Angerbjörn, 2005; Pinnegar & Polunin, 1999). Stable isotopic ratios of carbon and nitrogen acquired from 13 15 samples of whole muscle tissue (δ Cm and δ Nm) are the most commonly used in trophic studies (Ramos & González-Solís, 2012). In the present study trophic level is considered the number of trophic exchanges between basal organic matter sources and a consumer, while trophic position is defined as a combination of its trophic level and organic matter resource 13 use (e.g. % of kelp-derived organic matter). The application of isotopic ecology using δ Cm 15 13 15 and δ Nm rely on an understanding of the isotopic fractionation (δ C, δ N) that occurs 13 between trophic positions. δ Cm has been observed to have only small changes, in the range 13 of 0.4-1.1‰ for δ C (McCutchan et al., 2003) and present different signatures between 13 important primary producers (O’Leary, 1981). Therefore, variability in δ Cm provides a reliable tracer for sources of organic matter supporting food webs (Jack et al., 2009; McLeod 15 15 et al., 2010a). In contrast, δ Nm increases in the range of 1.4-5.6 ‰ for δ N when organic 15 matter is transferred between consumers. Thus, changes in δ Nm relative to the N isotopic 15 baseline, from the mixture of basal organic matter sources in the food web (δ Nbase), is a useful tool to estimate trophic level (McCutchan et al., 2003; Post, 2002). Stable isotopes of fish muscle tissue represent an integration of δ13C and δ15N from the diet over weeks or 13 15 months (Suring & Wing, 2009), hence δ Cm and δ Nm are an effective tool for reconstruction of food web relationships not subject to small scale variability in the diet (Jack et al., 2009; Wing et al., 2012).

Another powerful tool to resolve the flux of organic matter within food webs is the use of Compound-specific Stable Isotopic Analysis of Amino Acids (CSIA-AA). The technique can be used to measure natural isotope abundances of single amino acids, which constitute proteins, making up most muscle tissues (Alberts et al., 2015). Because different amino acids are subject to different metabolic processes, some undergo isotopic fractionation with each trophic transfer while others maintain isotopic ratios close to their original values throughout 15 the food web. The effect is well-described for δ NAA (Chikaraishi et al., 2007, 2009; Popp et

109

15 al., 2007), where δ NAA of ‘source’ amino acids such as phenylalanine, those subject to 15 metabolic processes with little δ NAA change, represent the signature at the base of the food 15 15 web. Analysis of δ NAA can lead to a more accurate estimate of how the δ Nbase of a food web may have changed over time, providing an important complement to studies using 15 isotopic analysis of whole muscle tissues. δ NAA of ‘trophic’ amino acids, such as glutamic acid, which undergoes fractionation due to metabolic transamination, can be compared 15 relative to δ NAA at the base (e.g. phenylalanine) providing an accurate estimate of the trophic level of a consumer. Trophic levels can then be compared among groups independent of information on δ15N of basal organic matter sources (Bradley et al., 2015; Chikaraishi et al., 2009; McMahon & McCarthy, 2016; Popp et al., 2007). Additionally, CSIA-AA of 13 δ CAA has been used to investigate changes in basal organic matter source pools in food web studies (Fantle et al., 1999; Lorrain et al., 2009).

After the discovery of formaldehyde in 1859 and the first paper describing its effective use against bacterial growth (Blum, 1893), it started to be used by museum curators worldwide to preserve whole specimens. Since then, fish specimens have been exhaustively collected around the world, primarily because of the need to register and describe new species. Consequently, museum collections have become a rich source of historical samples, and potentially an important resource of samples for trophic studies.

Most of the fish specimens available from museum wet collections have been fixed in formalin and subsequently preserved in ethanol or isopropanol. It is known that the natural abundances of C and N isotopes change in muscle tissue due to exposure to fixatives (Bosley & Wainright, 1999; Kishe-Machumu et al., 2017). Thus, to compare the trophic ecology of preserved specimens over time using stable isotope analysis, one needs first to understand how fixatives induce isotope fractionation. Correction equations have been generated from empirical studies for other organisms (Bugoni et al., 2008; González-Bergonzoni et al., 2015), 13 15 but relatively few studies have resolved the effects of fixatives on δ Cm and δ Nm in marine fish. Studies have focused on fixative effects on δ13C and δ15N of preserved zooplankton since the 1980s (Mullin et al., 1984), with increased interest in using stable isotope analysis on preserved fish tissues beginning in the late 1990s (Bosley & Wainright, 1999). Although 16 studies have been dedicated to the subject to date, only eight have investigated fixative effects in marine fishes, with a total of 14 species studied.

Penetration rates of fixatives increase with surface area to volume ratio and muscle exposure (Thavarajah et al., 2012). Museums typically maintain their specimens in as intact condition as possible while in contact with the fixatives, but only three studies to date have

110

13 15 used whole fish during the investigations of fixative effects on δ Cm and δ Nm (Arrington & Winemiller, 2002; Bosley & Wainright, 1999; Edwards et al., 2002). Furthermore, only one study has investigated the effect of removing lipids on the isotopic signatures of fixed fish tissues (Kelly et al., 2006). Kelly et al., (2006) demonstrated that lipid-free treatments of muscle tissue from preserved Arctic charr (Salvelinus alpinus, Salmonidae) had different 13 15 δ Cm and δ Nm than those of untreated muscle. Further, studies have highlighted the importance of considering lipid content on isotopic analysis of whole preserved organisms (Ruiz-Cooley et al., 2011; Sweeting et al., 2004), but have not provided statistical effects for 13 15 either the lipid or fixative treatments on the observed isotopic shifts of δ Cm and δ Nm.

15 The long-term effects of fixation in formalin on δ NAA are negligible for fish muscle tissues (Ogawa et al., 2013) and for some zooplankton tissues (Hannides et al., 2009). Nevertheless, only one study to date has tested the effects of formalin fixation followed by 15 alcohol preservation on δ NAA (Hetherington et al., 2019). Here significant fractionation in 15 δ NAA between 0.5 and 4.5 ‰ was reported, but the effects of formalin fixation and alcohol 13 preservation on δ CAA of fish muscle tissues have not been resolved.

The aim of the current study was to understand the effects of formalin fixation, 13 followed by ethanol and isopropanol preservation, on isotopic ratios of bulk tissue (δ Cm and 15 13 15 δ Nm) and amino acids (δ CAA and δ NAA) of fish muscle to generate a statistical relationship that could be used to correct for these effects. The results provide a method to resolve intraspecific changes in trophic position in marine fishes from museum wet collections. The approach was divided into four specific objectives:

13 1) A meta-analysis of studies to date was carried out to investigate how δ Cm and 15 δ Nm changed over time under formalin and ethanol treatments. To use the largest dataset possible and to allow for a comparison between ecosystem types, both marine and freshwater fish species were included. Data on freshwater species comprised a majority of the meta- analysis, which could result in greater variation in tissue composition (regarding isotopes and concentrations of C and N) when compared to the marine species dataset. To account for this possible effect, the effects of preservation and fixation in marine and freshwater species were directly compared before the meta-analysis was carried out.

13 2) C:N ratio has been highlighted as an important determinant of variability in δ Cm and lipid content in muscle tissue (Post et al., 2007), but this has been criticized (Fagan et al., 13 2011). To address this disagreement, the relationships between lipid content, C:N and δ Cm were tested for the samples used in the present study and compared with correction equations in the literature.

111

3) An experiment was carried out to test the effects of formalin fixation and alcohol 13 15 preservation on δ Cm and δ Nm of marine fishes with a range of lipid content. Museum samples were usually preserved either in ethanol or isopropanol after fixation in formalin. To account for different alcohol preservation effects on isotopic ratios, samples were separated 13 into ethanol and isopropanol alcohol treatments and compared their effects on δ Cm and 15 13 15 δ Nm. Treatment, sample location and species effects on isotopic shifts in δ Cm and δ Nm 13 15 were investigated. A statistical correction for the effects of fixatives for δ Cm and δ Nm was 13 then generated. Since lipid content can bias studies applying δ Cm to resolve trophic 13 relationships, lipid-free values were used as the control for the δ Cm in the statistical model.

13 15 4) Lastly, how δ CAA and δ NAA values of eleven amino acids change under different preservatives and over time was tested.

It is hypothesized that shifts in bulk isotope values would not be dependent on time under preservation, that lipid content of the muscle tissue would be an important factor 13 15 covarying with shifts in δ Cm but not in δ Nm, and that C:N ratios from the preserved specimens could be used to account for these deviations. For the CSIA-AA analyses, it is 13 hypothesized that neither the preservation nor the fixation would influence δ CAA, but that 15 they would influence δ NAA. The results provide the first detailed analysis of the effects of 13 15 13 15 formalin fixation and alcohol preservation on δ Cm, δ Nm, δ CAA and δ NAA of marine fish, providing an important step towards accessing trophic information from specimens in the vast scientific resource contained in museum wet collections.

112

4.2. Methods

13 15 4.2.1. Meta-analysis of effects of formalin and alcohol on δ Cm and δ Nm

13 15 Data on formalin and ethanol effects on δ Cm and δ Nm were retrieved from the literature. Only a few studies investigated fixation followed by preservation effects; thus, data were restricted to studies with individual treatments of formalin and ethanol, rather than the treatments in combination. When data was not available from the publication or supplementary documents, values were extracted from graphs using the software Digitize it 4.2.0 (Carrascal, 2016).

4.2.2. Sample collection and experimental design

To standardize the fixation procedure, guidelines provided by the Museum of New Zealand Te Papa Tongarewa were used (Roberts et al., 2015). The guidelines call for fixing specimens for a month in 10% formalin (4% solution of 40% formaldehyde), followed by three weeks of preservation progressively increasing concentrations of ethanol (at 50%, 60% and 70%) or isopropanol (30%, 40% and 50%) with the duration of one week for each concentration increment. The alcohol type is usually chosen by fish size and chemical availability. For the controlled laboratory experiment, three species of fish with muscle lipid content of 0.7 (blue cod, Parapercis colias, Pinguipedidae), 4.4 (blue warehou, Seriolella brama, Centrolophidae) and 7.2 % (king salmon, Oncorhynchus tshawytscha, Salmonidae) were used, based on a previous study that quantified lipid concentrations of whole fish fillets in New Zealand (Vlieg & Body, 1988). These species represent more than 90% of the lipid content spectrum of 55 representative commercial fish species of New Zealand, which ranges from 0.5 to 14.2 with an average of 2.6%. P. colias (PC; n = 2) was collected in the Marlborough Sounds, S. brama (SB; n = 2) was provided by commercial anglers fishing on the Otago coast, and O. tshawytscha (OT; n = 2) was provided by the Dunedin community salmon hatchery. The individuals were sub-sampled at five locations along their dorsal musculature before any fixative treatment, providing a control for initial conditions. Samples were numbered from 1 to 5, with 1 being the closest to the head, 2, 3 and 4 were samples taken in the middle part of the body, and 5 was extracted from the caudal peduncle. Specimens were then placed inside 5-liter jars with 10% formalin for one month before the second set of sub-samples were taken comprising the FORMALIN treatment. One fish of each species was then preserved in ethanol and the other in isopropanol for three months before a third sampling took place (FORMALIN+ALCOHOL treatment). After 11 months in

113 alcohol, the specimens were sampled again (FORMALIN+ALCOHOL2 treatment). During the one-year experimental duration a total of 120 samples were collected. While all samples 13 15 were analyzed for δ Cm, δ Nm, and the proportion of N and C, only the marine 13 15 representatives, P. colias and S. brama, were analyzed for δ CAA and δ NAA of 11 amino acids.

4.2.3. Lipid content analysis

Control samples were analyzed for lipid content using the micro-colorimetric protocol developed by Handel and modified by Inouye and Lotufo (Handel, 1985; Inouye & Lotufo, 2006). The method improves lipid extraction compared to more conventional methods and allows rapid lipid determination of samples as small as 10 mg by using spectrophotometry. 20 mg of each sample was weighed, and lipids were extracted in a reaction tube using a 1:1 chloroform:methanol solution. Tubes were centrifuged to allow separation of liquid (lipids + alcohols) and solid (lipid-free tissue) phases, and the supernatant volume was recorded. Triplicates containing 0.25 ml of the supernatant were extracted and allowed to evaporate over a heat block at 100 °C, then 0.1 ml of sulfuric acid was added to the tube to hydrolyze the organic molecules. The solution was then dried on a heat block, and 2.4 ml of vanillin reagent was added to each tube to promote change in color related to the amount of lipid in the tube. 0.2 ml of each solution was placed in a slot of a 96-well plate before analysis at 490 nm by a SpectraMax M2 Multi-mode Plate Reader Spectrometer (Molecular Devices Corporation, California, US). A 5-point standard curve was produced using samples of commercial oil, with concentrations of 0, 0.2, 0.5, 0.8 and 1 mg/L, which was measured together with the fish samples. Each 96-well plate contained an independent standard curve with the five concentrations of standard oil. An instrument precision (1SD) of 0.009 mg/L and an accuracy of 0.03 mg/L were calculated using sample triplicates and standard values. Lipid concentration was then obtained through regression curves on the SoftMax Pro software (Molecular Devices Corporation, California, US). From the concentrations, the lipid fraction of the muscle tissue samples was calculated as:

푚푔 푆푢푝푒푟푛푎푡푎푛푡 푣표푙푢푚푒 (퐿) ∗ 푙𝑖푝𝑖푑 푐표푛푐푒푛푡푟푎푡𝑖표푛 ( ) ( % 표푓 푙𝑖푝𝑖푑 𝑖푛 푡ℎ푒 푠푎푚푝푙푒 = 퐿 ∗ 100 푆푎푚푝푙푒 푤푒𝑖푔ℎ푡 (푚푔) 4.1)

114

4.2.4. Stable isotope analysis

Lipid-free tissues were washed with deionized water, dried overnight and prepared for isotopic analyses. Samples from all treatments, including a lipid-free CONTROL, were oven- dried at 60 °C for 72 h and crushed with mortar and pestle until they yielded a fine and homogenous powder. All equipment was cleaned with low-linting Kimwipes® and ethanol and air-dried between samples to avoid cross-contamination. Between 0.8 and 1.2 mg of each sample was packed in 3x5 mm tin capsules and analyzed by combustion in an elemental analyzer (Carlo Erba NC2500) to CO2 and N2, as described in chapter 2 (section 2.2.2). All measured values for the quality control standards were in the range of accepted values.

13 To investigate the relationships between C:N and δ Cm, and to generate a correction 13 equation for δ Cm that accounted for lipid concentrations, a linear relationship between these 13 13 variables was fitted (Post et al., 2007; Schmidt et al., 2009). First Δδ Cm= δ Cm lipid-free- 13 δ Cm was calculated for each sample, the results were regressed against C:N values, 13 providing a linear relationship between changes in δ Cm due to lipid concentrations and C:N 13 ratio. This relationship could then be used to correct δ Cm given different lipid concentrations.

4.2.5. Amino acid stable isotope analyses

Amino acids were extracted by hydrolyzing 2.5 mg of the sample with 2 ml 6 M HCl o at 110 C for 24 hours in an N2 atmosphere. An internal standard, norleucine (50 μl of 1 mg/ml solution), was added to monitor the wet chemistry and AA stable isotope values. o Solutes were then dried under a gentle flow of N2 at 60 C and subsequently converted into N- Acetylisopropyl (NAIP) ester derivatives following the protocol described in (Sabadel et al., 2016), modified from (Styring et al., 2012). Details of the derivatization procedure can be 13 15 found in Appendix A.4.2. δ CAA and δ NAA were measured by gas chromatography/combustion/isotope ratio mass spectrometer (GC-C-IRMS), using a Thermo Trace gas chromatograph, the GC combustion III interface, and a Deltaplus XP isotope ratio mass spectrometer (Thermo Fisher Scientific). 200 nl aliquots of derivatized AA were injected, inlet at 270 °C in splitless mode, carried by helium at 1.4 ml min−1 and separated on a VF-35ms column (30 m long, 0.32 mm ID and a 1.0 μm film thickness). The GC program 13 can be found in Appendix A.4.2. For δ CAA measurements, the oxidation reactor was set at o 15 950 °C and the reduction reactor at 200 C; while for δ NAA measurements, the oxidation reactor was set at 980 °C and the reduction reactor at 650 °C and a liquid nitrogen cold trap

115 employed after the reduction reactor. Samples were analyzed in duplicates or triplicates along with amino acid standards of known isotopic composition (measured by EA-IRMS) and bracketing measurement of every two samples. Each run contained no more than 10 samples. 13 13 Raw δ CAA was individually corrected relative to the amino acid δ CAA of the standards to account for the added C and kinetic fractionation introduced during the derivatization procedure (see Appendix A.4.2). As for bulk isotopes, δ values of AA isotopes were reported following the conventional method of expressing δ at natural abundance, in per mil (‰), 13 relative to an international standard: Vienna PeeDee Belemnite (VPDB) for δ CAA and 15 atmospheric N2 for δ NAA (Fry, 2006).

Eleven amino acids from each sample were measured with no peak co-elutions, listed below. In order of elution: alanine (Ala), glycine (Gly), valine (Val), leucine (Leu), isoleucine (Ile), Threonine (Thr), serine (Ser), proline (Pro), aspargine + aspartic acid (Asx), glutamine + glutamic acid (Glu) and phenylalanine (Phe) (see Appendix A.4.2, for a typical chromatogram). Note that during the hydrolysis step, aspargine is converted to aspartic acid (hence the notation Asx) and glutamine is converted to glutamic acid. Methods and analysis were provided by Amandine Sabadel after sample hydrolysis. Precision (1SD) of corrected 13 15 δ CAA ranged from 0 to 1.1 ‰ with a mean of 0.2 ‰, while the precision for δ NAA ranged from 0 to 2.7 ‰ with a mean of 0.5 ‰.

4.2.6. Data treatment and statistical framework

13 15 Shifts of δ Cm and δ Nm through time under preservation were investigated by fitting a linear model to published data collected in the literature review. A Kruskal-Wallis test was 13 15 used to compare the response of δ Cm and δ Nm to treatments in formalin and ethanol between freshwater and marine fishes.

A linear regression was used to test the correlation between lipid content and C:N. The results were used to validate the use of C:N as a proxy for lipid content in the present 13 15 13 15 study. Effects of ethanol and isopropanol on δ Cm, δ Nm, δ CAA and δ NAA were evaluated using a two-way repeated-measures ANOVA, where sample locations were nested within individual specimens and compared with their replicate group.

Paired student’s t-tests were used to investigate the distribution of isotopes and lipid content along fishes’ bodies for each species on CONTROL. Two-way repeated-measures ANOVAs were used to investigate interactions between TREATMENT and SPECIES on bulk and AA isotopic data as described above. To find the most parsimonious model

116

13 15 describing shifts in δ Cm and δ Nm due to the separate treatments, a stepwise fit evaluation was applied to linear models with different combinations of variables. Data were randomly split into ten K-folds for model training and testing, and statistics were averaged between the models. Models explaining isotopic shifts were evaluated using both average r2 of the ten K- folds, Akaike (AICc) and Bayesian Information Criteria (BIC), two common penalized- likelihood methods (Akaike, 1973; Schwarz, 1978). By using both information criteria methods, selecting under or overfitted models was avoided (Dziak et al., 2012). Assumptions related to linear models were checked with residual plots (by predicted values, by row and normal quantile). Statistical tests were carried out using the program JMP pro 14.2.0 (SAS Institute, 2018).

4.2.7. Applying the results

In order to better understand the importance of considering isotopic corrections after fixation, the trophic level and the proportion of kelp-derived organic matter supporting each of the three species used in the experiment were estimated using corrected and non-corrected 13 15 δ Cm and δ Nm values. For TP calculations based on bulk measurements, a mass balance model approach was applied, following (Wing et al., 2012) on suspended particulate organic matter (SPOM) and kelp samples collected from the Otago region, New Zealand, during Summer 2018 and 2019 (unpublished data, but see chapters 2 and 3 for methods). The organic matter source pool dataset provided robust δ13C and δ15N signatures at the base of the local food web, allowing for direct comparisons among species. Fractionation values of 0.5‰ for δ13C and 2.3‰ for δ15N were used between trophic levels and were kept constant for all 15 species (McCutchan et al., 2003). Trophic levels calculated from δ NAA were estimated 15 15 using the relationship between δ Nglu and δ Nphe following the equation from Chikaraishi et al., (2009) and McMahon et al., (2015); details can be found in Appendix A.4.2.

117

4.3. Results

4.3.1. Literature meta-analysis

13 15 δ Cm and δ Nm data from 16 papers comprising 51 fish species were compiled and analyzed. No differences in the magnitude of shifts were observed after formalin fixation 13 15 between freshwater (average of -1.04‰ for δ Cm and +0.24‰ for δ Nm) and marine species 13 15 13 (average of -0.93‰ for δ Cm and +0.28‰ for δ Nm) (Kruskal-Wallis test, δ Cm, Chi square 15 = 1.28, p = 0.26, df = 59; δ Nm, Chi square = 0.0004 , p = 0.98, df = 71). The same was 13 observed after ethanol preservation of freshwater (average of -0.10‰ for δ Cm and +0.38‰ 15 13 15 for δ Nm) and marine species (average of -0.34‰ for δ Cm and +0.45‰ for δ Nm) (Kruskal- 13 15 Wallis test, δ Cm, Chi square = 2.05, p = 0.15, df = 35; δ Nm, Chi square = 0.71, p = 0.4, df 13 15 = 39). Data on the effects of formalin and ethanol on δ Cm and δ Nm from the available literature are presented in Figure 4.1 and in Appendix 4.3.

13 Considering all fish species, formalin fixation decreased δ Cm values (-1.04 ± 0.72‰) 15 while increasing δ Nm (0.24 ± 0.43‰). Ethanol preservation had similar effects, shifting 13 15 δ Cm by -0.04 ± 0.83‰ and δ Nm by 0.4 ± 0.54‰. There was also an overall increase in 13 15 variability of δ Cm and δ Nm (within ± 0.5‰) in the first weeks under formalin and ethanol exposure, dampening their relationship, or lack of it, with the time of preservation. Therefore, 13 15 there were no consistent trends in changes in δ Cm and δ Nm with time following formalin 13 15 (r² < 0.01, F(1,56) = 0.23, p = 0.64 for δ Cm and r² = 0.06, F(1,68) = 4.3, p = 0.04 for δ Nm) 13 and ethanol treatment (r² = 0.03, F(1,32) = 1.13, p = 0.3 for δ Cm and r² <0.01, F(1,36) = 15 0.05, p = 0.1 for δ Nm) in the literature (Figure 4.1).

118

Figure 4.1. Shifts in isotope values of fish muscle tissue with time after formalin fixation (a) and ethanol preservation (b) experiments retrieved from literature. Equation and R2 are shown on top of each graph. Values reported in ‰

4.3.2. Lipid content and isotopes of control samples

Lipid content was positively correlated with C:N ratio and presented low variation around the linear fit, independently of species or sample location (r² = 0.83, F1,28 = 0.923, p < 0.001, Figure 4.2). Lipid concentration within individual fish muscle tissue samples varied from 2.6% for O. tshawytscha to 11.2% for S. brama, while P. colias showed the highest average lipid content (6.6 ± 2.5%) compared to the other species (4.1 ± 1.0% for O. 13 tshawytscha and 4.8 ± 2.5% for S. brama). Δδ Cm displayed a linear relationship with the C:N values (r² = 0.91, F(1,28) = 276.12, p < 0.001):

Δ훿13Cm = −5.37 + 1.58 ∗ 퐶: 푁 4.2)

119

13 This equation can then be used to standardize δ Cm for tissues with different lipid concentrations, after Post et al., (2007) and Schmidt et al., (2009):

13 13 13 훿 Cmcorrected = 훿 Cm + Δ훿 Cm 4.3)

15 δ Nm differed between species but not along the length of fishes’ bodies. In contrast, 13 δ Cm varied between sample locations within the same specimen. After lipid extraction, the 13 variation of δ Cm was reduced, consistent with the strong relationship between lipid content 13 and δ Cm (Table 4.1).

y = -36.124 + 10.808x R 2= 0.83177

6 Lipid concentration (%)

2 3.5 3.6 3.7 3.8 3.9 4 4.1 4.2 4.3

C:N ratio Figure 4.2. Relationship between lipid concentration in percentage and carbon to nitrogen ratio in fish muscle tissue from samples of O. tshawytscha, P. colias and S. brama, F(1,28) = 0.923, p < 0.001

120

Table 4.1. Comparison of means of lipid content, nitrogen and carbon isotope and carbon isotope after lipid extraction along each species dorsal musculature. Values after ± symbol are one standard deviation. Letters represent different groups within each species (t = 2.13, α = 0.05), different letters represent different group means Sample Species Lipid content (%) δ¹⁵N ‰ δ¹³C ‰ δ¹³C ‰ lipid free location 1 5.16 ± 1.55 A 13.86 ± 0.16 A -18.42 ± 0.28 B -17.56 ± 0.21 A 2 4.18 ± 0.93 ABC 13.56 ± 0.03 AB -18.21 ± 0.01 AB -17.62 ± 0.05 A Oncorhynchus 3 3.18 ± 0.80 C 13.45 ± 0.20 B -18.09 ± 0.02 A -17.62 ± 0.05 A tshawytscha 4 3.46 ± 0.69 BC 13.28 ± 0.14 B -18.128 ± 0.05 AB -17.52 ± 0.05 A 5 4.55 ± 0.30 AB 13.36 ± 0.02 B -18.13 ± 0.02 AB -17.52 ± 0.12 A 1 4.43 ± 0.41 D 13.27 ± 0.26 A -18.92 ± 0.02 A -18.34 ± 0.01 A 2 10.39 ± 0.64 A 13.1 ± 0.37 A -19.76 ± 0.18 C -18.45 ± 0.11 A Parapercis 3 8.24 ± 0.43 B 12.97 ± 0.42 A -19.47 ± 0.02 B -18.3 ± 0.22 A colias 4 5.77 ± 0.47 C 12.82 ± 0.10 A -19.14 ± 0.15 A -18.41 ± 0.15 A 5 4.2 ± 0.55 D 12.87 ± 0.25 A -18.9 ± 0.08 A -18.25 ± 0.28 A 1 7.46 ± 5.29 A 14.83 ± 0.62 A -18.88 ± 0.27 A -18.11 ± 0.38 A 2 4.79 ± 2.13 AB 14.78 ± 0.45 A -18.61 ± 0.16 A -18.06 ± 0.32 A Seriolella 3 4.18 ± 1.13 AB 14.74 ± 0.47 A -18.42 ± 0.37 A -18.06 ± 0.31 A brama 4 3.8 ± 1.05 B 14.71 ± 0.45 A -18.27 ± 0.29 A -18.02 ± 0.35 A 5 3.67 ± 0.86 B 14.97 ± 0.41 A -18.25 ± 0.47 A -17.93 ± 0.34 A

13 15 4.3.3. Effects of fixation and preservation on δ Cm and δ Nm

There were no significant differences between ethanol and isopropanol treatments on 13 δ Cm (F1,24 = 0.37, p = 0.55 for 4 months treatment and F1,24 = 0.12, p = 0.73 for 1 year 15 treatment) or δ Nm (F1,24 = 2.82, p = 0.11 for 4 months treatment and F1,24 = 0.81, p = 0.38 for 1 year treatment). Therefore, data from both the isopropanol and ethanol treatments were consolidated under the same ALCOHOL treatment in subsequent analysis. Under formalin and alcohol preservation, fish muscle tissue became enriched in 15N resulting in higher values 15 of δ Nm after formalin fixation and alcohol preservation (>1 ‰, Figure 4.3). After FORMALIN+ALCOHOL2 treatment there was a small shift back toward initial conditions of 15 15 δ Nm, but the net effect was towards higher values of δ Nm. All three species presented the 15 same trends in δ Nm due to formalin and alcohol treatments. All preservation results are within 1‰ of each other, while they reach up to 2‰ compared to the control (Figure 4.3). 13 13 13 For δ Cm there was a trend in depletion of C resulting in more negative values of δ C due 13 to formalin fixation for all species (approximately 1‰). Alcohol treatments shifted δ Cm closer to control values and were indistinguishable from controls for S. brama and P. colias 13 (Figure 4.3). Shifts in δ Cm presented a positive trend with lipid concentration due to the exposure to formalin (r² = 0.21) and FORMALIN+ALCOHOL treatment (r² = 0.34). Shifts in 15 δ Nm under fixation and alcohol preservation had no significant interaction with lipid 15 content. δ Nm shifts were significantly different between SPECIES and TREATMENT but 13 showed no interaction between those variables (Table 4.2). In contrast, δ Cm was

121

significantly affected by the interaction between SPECIES and TREATMENT (F4,54 = 4.14, p = 0.01).

13 15 Figure 4.3. δ Cm and δ Nm average per treatment for each one of the fish species (n = 10). Treatments consisted of dried tissues (time 0), after one month under formalin fixation (FORMALIN), under three months (FORMALIN+ALCOHOL) and one year of alcohol preservation (FORMALIN+ALCOHOL2) after formalin treatments. Error bars represent one standard deviation. Values reported in ‰

13 15 13 Table 4.2. Summary of two-way repeated measures ANOVA results for δ Cm, δ Nm, δ CAA and 15 δ NAA. P values marked with * were considered significant Isotope SPECIES TREATMENT SPECIES~TREATMENT Sample type analyzed DF F p DF F p DF F p δ13C Bulk muscle (2,27) 40.14 <.001* (2,54) 16.34 <.0001* (4,54) 4.14 .01* Alanine (1,18) .03 .86 (2,36) 1.33 .28 (2,36) 13.24 <.0001* Glycine 80.93 <.0001* 4.66 .02* 21 <.0001* Valine 72.91 <.0001* 5.78 .01* 7.85 .001* Leucine 2.11 .02 1.63 .21 1.55 .22 Isoleucine 8.16 .01* .82 .45 7.69 .001* Threonine 41.23 <.0001* .53 .60 .80 .45 Serine 3.47 .08 .91 .41 6.23 .005* Proline 27 <.0001* 2.29 .12 16.81 <.0001* Aspartic acid 20.09 .0003* 1.16 .33 3.64 .04* Glutamic acid .0001 .99 3.52 .04* 2.71 .08 Phenylalanine 14.95 .001* 0.76 .48 3.36 .05* δ15N Bulk muscle (2,27) 127.16 <.0001* (2,54) 504.86 <.0001* (4,54) 1.88 .13 Alanine (1,18) 160.24 <.0001* (2,36) 13.92 <.0001* (2,36) 9.86 .0004* Glycine 42.7 <.0001* 27.24 <.0001* 16 <.0001* Valine 82.71 <.0001* 25.37 <.0001* 25.01 <.0001* Leucine 435.63 <.0001* 17.62 <.0001* 14.4 <.0001* Isoleucine 115.46 <.0001* 25.04 <.0001* 14.41 <.0001* Threonine 104.32 <.0001* 2.23 .12 3.69 0.3* Serine 13.46 .002* 12.10 <.0001* 19.81 <.0001* Proline 44.67 <.0001* 50.17 <.0001* 24.76 <.0001* Aspartic acid 275.46 <.0001* 13.63 <.0001* 9.35 .0005* Glutamic acid 149.69 <.0001* 13.81 <.0001* 10.58 .0002* Phenylalanine 1.98 .18 15.62 <.0001* 17.67 <.0001*

122

4.3.4. Model selection for preservation effects

13 The general linear model that best explained δ Cm of lipid-free control samples 15 13 contained the predictor variables: preserved δ Nm, preserved δ Cm and C:N ratio from the preserved tissue. These models displayed the highest average r2 across the K-folds (0.79 for 15 13 δ Nm and 0.77 for δ Cm), indicating the best fit when random data was used to test the model. Combined AICc and BIC were 2.35 and 6.65% lower than the second-best candidate model, respectively. A general linear model was built with the variables cited above. This model could statistically explain 79% of the total variation due to fixatives, independent of time, where all the terms were considered significant for the final model (p < 0.01) (Figure 4.4 and Equation 4.4):

13 15 Figure 4.4. Relationship between predicted and actual δ Cm and δ Nm values from preserved fish 13 muscle (n = 60), using the empirical equation developed in this study. δ Cm values are corrected to lipid-free values. Values for FORMALIN+ETHANOL and FORMALIN+ETHANOL2 were included. Dashed line around the relationship represents 95% confidence interval. The horizontal line represents the mean of the data group. Data from O. tshawytscha (OT), P. colias (PC) and S. brama (SB) were included

123

13 15 13 Lipid free δ Cm = - 8.42 + 0.07 * preserved δ Nm + 0.76 * preserved δ Cm + 0.97 * preserved C:N 4.4)

15 The general linear model that best explained δ Nm of control samples contained the 15 13 predictor variables: preserved δ Nm, preserved δ Cm and the proportion of N from the preserved tissue. AICc and BIC were 3 and 0.9 % lower than the second-best model candidate, respectively. The model statistically explained 81 % of the total variation due to fixative effects (Figure 4.4 and Equation 4.5). The results were also independent of time, where all the terms were considered significant for the final model (p < 0.02).

15 15 13 δ Nm= 11.25 + 0.71 * preserved δ Nm+ 0.27 * preserved δ Cm + -0.21 * proportion of preserved N 4.5)

4.3.5. Compound-specific Amino acid analysis

13 13 δ Cgly and δ Cser values were lower after isopropanol preservation compared to those 15 15 observed under ethanol preservation (0.66 and 0.45‰ respectively). δ Nval (1.09‰), δ Nasp 15 15 (1.65‰) and δ Nglu (1.80‰) also showed the same trend, while δ Nser displayed higher values (1.23‰). Although those differences were significant at the 95% level, the data was also pulled together for the general analysis, since those differences comprised less than 20% of the variation observed between species (Figures 4.5 and 4.6).

13 The values of δ CAA were not found to be different between treatments (p < 0.05, 13 Figure 4.5). SPECIES had a significant interaction with TREATMENT on δ CAA for all AAs, except Leu, Thr and Glu. Although most AAs showed differences among fish species, 13 Gly, Val and Glu were the only AAs that displayed significant shifts of δ CAA between treatments, with the largest changes displayed by Val (-1.94‰ between CONTROL and FORMALIN+ALCOHOL2 treatment and 1.7‰ between CONTROL and FORMALIN+ALCOHOL1 treatment) (Table 4.2).

For most AAs, enrichment of δ15N remained within 1‰, but on the FORMALIN+ALCOHOL2 treatment, Glu and Asp shifted by an average of 1.6‰ and 1.8‰ respectively. These dissimilarities were not observed in the CONTROL samples. 15 TREATMENT had a significant interaction with SPECIES for all δ NAA (p < 0.05).

To verify if the fixative effects would limit ecological studies applying CSIA-AA on 15 15 preserved specimens, a paired t-test to the subtraction of δ NGlu and δ NPhe between 15 15 treatments was applied. The subtraction of δ NGlu and δ NPhe is a common part of TL

124 estimation of marine organisms using AA values (Chikaraishi et al., 2009). No significant 15 15 difference for the relative values of δ NGlu and δ NPhe between CONTROL and FORMALIN+ALCOHOL (MD = -0.08, SD = 0.49, t(19) = -0.16, p = 0.87) was found, as well as between CONTROL and FORMALIN+ALCOHOL2 (MD = 0.75, SD = 0.45, t(19) = -1.67, p = 0.11), where MD states for mean difference.

Common warehou (Seriolella brama)

-8

-12

-16

-20 Alanine Glycine -24 Valine

d13C of amino-acids Leucine Isoleucine -28 Threonine Serine Proline -32 Aspartic acid Glutamic acid Phenylalanine -36 0 2 Blue cod4 (Parapercis6 8colias)10 12 -5 Months

-10

-15

-20 Alanine Glycine -25 Valine Leucine

d13Cof amino-acids Isoleucine -30 Threonine Serine Proline -35 Aspartic acid Glutamic acid Phenylalanine -40 0 2 4 6 8 10 12

Months 13 Figure 4.5. δ CAA values of 11 amino acids of muscle tissue from S. brama and P. colias at CONTROL (0 months), after FORMALIN+ETHANOL (4 months) and FORMALIN+ETHANOL2 (12 months) treatments. Values reported in ‰

125 3 m 1 C ä

Common warehou (Seriolella brama) 40

10 Alanine Glycine Valine Leucine d15N of amino-acids 0 Isoleucine Threonine Serine -10 Proline Aspartic acid Glutamic acid Phenylalanine -20 0 2 Blue cod4 (Parapercis6 8colias)10 12

Months 30

10 Alanine Glycine Valine 0 Leucine

Isoleucine d15N of amino-acids Threonine Serine Proline -10 Aspartic acid Glutamic acid Phenylalanine

0 2 4 6 8 10 12

Months 15 Figure 4.6. δ NAA values of 11 amino acids of muscle tissue from S. brama and P. colias at CONTROL (0 months), after FORMALIN+ETHANOL (4 months) and FORMALIN+ETHANOL2 (12 months) treatments. Values reported in ‰

4.3.6. Applying the results

15 Differences between estimated trophic levels from corrected and uncorrected δ Nm showed that TL can be biased by as much as one trophic level when corrections shown here are not considered (Table 4.3). Trophic level estimates from AA values did not display any differences between FORMALIN+ALCOHOL2 and CONTROL values and matched with estimates from corrected mixing model results. The same results are seen when kelp-based resource use was estimated, where values were reduced by 20% when raw values were used instead of corrected ones (Table 4.3).

126

Table 4.3. Summary of results applied to raw data analyzed in this work. Raw values were taken after FORMALIN+ALCOHOL2, while correction equations were used on the corrected values column. For trophic levels estimated from AA, corrected values were taken from CONTROL. The difference column is the result of corrected minus raw values. Average values are shown ± one standard deviation for mixing model results and propagation of error for the AA data after Dale et al., (2011). All values are given in ‰ Parameter estimates Species Raw values Corrected values Difference Trophic level (mixing model) S. brama 4.7 ± 0.14 3.69 ± 0.1 -1.01 P. colias 4.02 ± 0.29 3.02 ±0 .11 -1 O. tshawytscha 3.87 ± 0.29 3.11 ± 0.19 -0.76 Trophic level (Glu-Phe) S. brama 4.03 ± 0.44 3.93 ± 0.44 -0.1 P. colias 3.09 ± 0.47 2.99 ± 0.49 -0.1 Proportion of kelp-based carbon S. brama 0.36 ± 0.03 0.52 ± 0.03 0.14 source (mixing model) P. colias 0.29 ± 0.08 0.51 ± 0.01 0.22 O. tshawytscha 0.47 ± 0.06 0.59 ± 0.02 0.12

127

4.4. Discussion

The results of the present study demonstrate that by using a combined approach of SIA of muscle tissue and AAs, one can accurately estimate the trophic position of museum preserved specimens of fish. These data provide the basis for estimating intraspecific changes in tropic position through time for key species of marine fish, an important source of data for reconstructions of food web dynamics and understanding historic baseline conditions. Meta- analysis results demonstrated that both formalin fixation and preservation in ethanol affect the 15 13 values of δ Nm and δ Cm. These effects cannot be ignored if one wants to accurately resolve 15 13 15 δ Nm and δ Cm of preserved fish specimens and model trophic positions. Shifts of δ Nm 13 and δ Cm were not different between marine or freshwater species and did not have any relationship with the amount of time under treatment since most of the effects were shown in the first weeks of contact with ethanol and formalin. These results corroborate those obtained from previous studies (Kelly et al., 2006; Sarakinos et al., 2002) and confirm the rapid effects 15 13 of formalin and ethanol on δ Nm and δ Cm of tissues (Fox et al., 1985).

Time under exposure to preservatives was not a significant factor in the observed 15 13 changes in δ Nm and δ Cm of tissues. In contrast, lipid content was the most important 13 explanatory factor for modeling shifts in δ Cm. C:N was a useful proxy for lipid content in 13 the marine fishes considered here, and an equation for δ Cm correction for samples with different lipid content was provided (Equation 4.3). Previous studies have provided similar 13 corrections for δ Cm due to lipid content in fish muscle tissue, but presented different slopes and intercepts (Post et al., 2007; Schmidt et al., 2009), probably due to differences in the species and range of C:N values analyzed, not suitable for the application in the present work and other studies (Kelly et al., 2006). Furthermore, the relationship between lipid content and C:N found in the present study differs from previously published results (Post et al., 2007).

The relationship between lipid concentration and C:N was found to be population-specific, systematically underestimating lipid content when equations from the literature were applied to other experiments (Fagan et al., 2011). Therefore it is recommended that researchers either develop a correction equation for each specific study, or use a correction equation generated 13 from samples with similar C:N, lipid content and δ Cm to the samples to be analyzed.

13 Lipid content was shown to be an important variable when one wants to correct δ Cm 15 and δ Nm due to preservation, even eliminating some effects of preservatives when lipids are removed (Kelly et al., 2006; Ruiz-Cooley et al., 2011). In the present study, C:N was used as a predictor variable in the models for the effects of both formalin fixation and preservation in 13 alcohol, both ethanol and isopropanol, on δ Cm of fishes. Lipid content results differed from

128 an extensive study of New Zealand marine fishes (Vlieg & Body, 1988). The differences observed in the present study may be due to the use of techniques with lower extraction power in the early study and because lipid concentration varied along each of the specimen’s bodies (Table 4.1). The observed differences in lipid concentrations within a fish specimen were not recorded in previous studies and might be a source of variability in the estimation of its lipid 13 content, as well as its δ Cm.

Because museums use both ethanol and isopropanol during specimen preservation, 15 13 13 15 how those different alcohols altered δ Nm, δ Cm, δ CAA and δ NAA was tested. No 15 significant differences were found between ethanol and isopropanol for changes in δ Nm and 13 13 15 δ Cm following preservation. Differences in δ CAA and δ NAA after preservation in ethanol and isopropanol observed here were small and only found in five of the eleven AAs tested. 13 15 The information from δ CAA and δ NAA can, therefore, be used if one wishes to compare species feeding on distinct trophic levels.

15 Overall, during the preservation experiment, δ Nm increased by almost 1‰, while 13 little shift in δ Cm was found. Using a linear model approach and taking into account C:N 15 13 and the proportion of N values, most of the variation in δ Nm and δ Cm linked to fixation in formalin and preservation in alcohol, both ethanol and isopropanol, could be explained 13 15 (Equation 4.4 and 4.5). Similarly, values of δ CAA and δ NAA from eleven AAs from the 13 different treatments demonstrated that, for most AA, although δ CAA values did not change, 15 δ NAA increased significantly (< 2‰), in agreement with a previous study (Hetherington et 15 al., 2019). Interestingly, δ NAA values of several AAs shifted in similar ways between treatments, maintaining their relationships. These results support the use of relationships 15 between δ NAA to estimate a fish’s trophic level (Chikaraishi et al., 2009). Similarly, the 13 results indicate that δ CAA values can be used to estimate the contribution of alternate basal organic matter sources to the food webs supporting different fishes from samples preserved in museum collections (McMahon et al., 2016). An example of how important it is to consider isotopic corrections for fixation in formalin and preservation in alcohol when retrieving data from preserved specimens was provided, by doing so one can get an accurate measurement 15 13 13 15 for δ Nm, δ Cm, δ CAA and δ NAA comparable to those from fresh specimens (Table 4.3).

Change of the delta values of amino acids between original and chemical-treated molecules is caused by either (1) incorporation of additional carbon or nitrogen, (2) isotopic fractionation associated with degradation during fixation and preservation and/or with non- quantitative reaction during analytical procedures. While there is a range of δ13C values of the various solvents used for the fixation and subsequent preservation of fish samples (δ13C

129 formalin = -41.42‰; δ13C ethanol = -27.84‰; and δ13C isopropanol = -24.50‰), it did not 13 affect the δ CAA values of any of the amino acids measured as the process for fixation/preservation does not affect the C in the amino acid molecules. Thus, as amino acids are effectively extracted from the fish tissues by hydrolysis during sample preparation, it ended up with an unchanged C isotopic signature through time.

15 In contrast, the results indicated enrichment with time of δ NAA. A recent study by 15 Hetherington et al., (2019) showed shifts in δ NAA after preservation for a few AA, but in that case the shifts were attributed to impurities in the samples and the difficulty to separate AAs in the chromatograms. No issues in the chromatograms were found in the present study, and differently of the previous study, almost all AA shifted in the same magnitude and direction. Therefore, the conclusion of Hetherington et al., (2019) does not explain the trends seen here.

The percentage of nitrogen (%N) contained in the ethanol and isopropanol due to impurities, before and after the experiment, were also analyzed. Those values indicate a reduction in %N throughout the experiment, suggesting the movement of molecules containing N in the fish tissues. Although this is true, values of %N were extremely low (0.02 and 0.024% for pure ethanol and isopropanol and 0.015 and 0.01% after the experiment). The lack of a clear mechanism explaining the fractionation of AA and the small amount of transfer of N between the preservatives and the samples also did not explain the trends seen in the present study.

Accounting for what is known, here are a few possible mechanisms of why fractionation of AA could happen:

Formic acid: If not buffered with methanol, a 10% formalin solution can form, with time, formic acid which might have started to hydrolyze preserved fish tissues, possibly causing fractionation. However, as the solution was buffered with 10% methanol, the chances of forming formic acid are quite low, and thus this process was unlikely.

Formalin cross-links: Formalin preserves proteins and cellular organelles in a stepwise process. It penetrates tissues then binds to lysine, tyrosine, asparagine, tryptophan, histidine, arginine, cysteine, glutamine and serine, in all the proteins present in a preserved specimen. It is the reaction between formalin and uncharged reactive amino groups that leads to the formation of cross-links (Milligan & Holt, 1977). If there was an effect of the formalin on all the measured AAs, it would only be noticed for aspartic acid, glutamic acid and serine. However, an enrichment after 1 year for all of AAs measured was observed. So, the change

130

15 observed in δ NAA of preserved fish tissues cannot be due to the direct effect of formalin on proteins.

Under-fixation and tissue degradation (e.g. bacterial reworkings): Under-fixation of tissues may result in cross-links forming only on the exterior of a sample and the center of the sample remaining unfixed. Fixative penetration, binding to amino acids, and cross-linking occur faster at higher temperatures, but the thickness of the sample (whole fish in the experiment) might prevent rapid penetration of the fixative/preservative solvents. If the samples were under-fixated, then bacterial activities could still be happening within the 15 tissues, and enrichment of the tissues in δ NAA would then be possible as AAs with lighter isotopes are preferentially reworked by bacteria. The bacterial activity could therefore help explain the differences between the results presented here and those reported by Hetherington et al., (2019), since samples constituted of isolated pieces of muscle tissue in the earlier study.

Enzyme hydrolysis: Soon after death when the cells start to break down, enzymes are released into the surrounding tissue where they start the process of breaking down the walls of other dead cells (protein autolysis), thus releasing yet more enzymes into the tissues in a process akin to a slow chain reaction (Simmons, 2014). It is possible that some enzymes could be active and breaking down AA in newly collected specimens if there was a significant delay until formalin had penetrated the fish and stopped any enzymatic activities.

Although the present study analyzed several samples from three different species of marine fish, there was relatively low replication at the level of species (2 individuals per species). Low replication of individuals was compensated for by multiple samples collected along each fishes’ body, where every sample differed for both the proportion of carbon and 15 13 nitrogen, C:N ratio and, for the most part, δ Nm and δ Cm. While various samples came from the same specimen, the analysis demonstrated that they encompass a wide range of values for all the explanatory variables analyzed. Here sample values represented the full range that could be found within the New Zealand fish fauna from random sampling. Nevertheless, lack of replication at the species level may have biased the results regarding alcohol type effects, since some variations between isopropanol and ethanol were found for 13 15 the δ CAA and δ NAA of specific amino acids.

13 While the results clearly show a trend in δ Cm shifts due to fixation in formalin, the magnitude of the change could have been influenced by small differences in δ13C of formaldehyde obtained from different sources (Edwards et al., 2002). Although those differences were assumed to be small (under 0.3‰), the present study used formalin and alcohol sourced from the same company and batch throughout the experiment. When

131 measuring isotopes from museum collections, the possible effects of different chemical sources should be considered as a random source of variation, since museums typically have no information on the isotopic signatures of formaldehyde used during fixation. Normality and equality of variances between groups were assumed for all the parametric tests. Although the assumptions were met for most groups (Shapiro-Wilk’s and Levene’s test p > 0.05), sample sizes were small, between 5 and 10 samples, which could have hindered some of the between group comparisons.

The present study was the first to provide an experimental analysis of variation in 15 13 13 15 δ Nm, δ Cm, δ CAA and δ NAA due to fixation in formalin followed by preservation in alcohol for whole marine fishes. With this method, researchers interested in intraspecific 15 13 13 15 changes in trophic position can have access to δ Nm, δ Cm, δ CAA and δ NAA values from extensive museum wet collections of fish. Furthermore, the herein results make it possible for scientists to investigate important ecological and management questions, such as the shifting trophic structure of fish communities (Pauly et al., 1998), and to support ecosystem-based approaches to fisheries management (Thrush & Dayton, 2010), by confidently using isotopic ecology approaches.

132

Chapter 5 Shifts in the trophic structure of commercial fishes after New Zealand industrialized fisheries expansion

Abstract

Studying long-term ecological changes in marine environments is challenging, but understanding how these systems have changed over time is an important steppingstone for the management of natural resources using an ecosystem approach. Here natural stable isotopes were used to investigate temporal changes in the trophic structure of an exploited fish community along the east coast of the South Island of New Zealand. Samples collected before New Zealand’s full expansion of industrialized fisheries were retrieved from museum collections and compared to modern specimens. While some species did not present changes in the trophic parameters analyzed, significant changes were usually towards relying more heavily on pelagic production and feeding at higher trophic levels in the modern period. Wider niche breadths inferred from δ13C and δ15N values were also identified in modern communities inhabiting the outer shelf and slope, mainly due to the higher variability in the resource use by tarakihi, red cod, ling and spiny dogfish. While temporal variability in Munida gregaria abundance and environmental variables had relationships with the trophic parameters from all assemblages (except mid slope), marine trophic index of New Zealand’s commercial fisheries only showed relationships for fishes from the inner shelf. The present study demonstrated that the fish community analyzed underwent changes in the trophic structure over the past century, and that those changes affected each species and habitats differently. Ecological changes presented here likely altered the energy flux along the seascape, demonstrating the need for uncovering ecological baselines before human impacts to guide ecosystem-based management of these resources.

133

5.1. Introduction

Food web stability is recognized as one of the important factors that can help to maintain the health and resilience of aquatic ecosystems (McDonald-Madden et al., 2016; Peterson et al., 1998). While species with unique trophic niches are more vulnerable to local extinctions (Petchey et al., 2008), changes in their trophic level and resource use can have consequences to fish ecology and abundance, ultimately affecting fisheries yields of exploited species (Thrush & Dayton, 2010; Udy et al., 2019b). Therefore, understanding the variability in the trophic structure of commercial species is an important step for the management of these resources. Variability of food web interactions of fishes have been widely studied through the application of dietary, isotopic and modeling approaches (Davis & Wing, 2012; Karlsson et al., 2019; Phillips et al., 2014; Pinkerton et al., 2008; Udy et al., 2019a), but studies investigating long-term changes in those parameters are still rare (Edwards et al., 2010). Therefore, there is a lack of baseline information on the trophic structure for the management of fish communities based on an ecosystem approach, especially in places where aquatic communities have been heavily altered by anthropogenic activities (Auber et al., 2015; Rosenberg et al., 2005; Thurstan et al., 2010).

Investigating the variability of trophic structure in marine systems can be challenging (Mancinelli & Vizzini, 2015), due to patchy historical sampling, species with large home ranges and migrations, as well as rapid changes in prey availability. Stable isotope ecology has proven to be an important tool to assist to answer these pertinent questions. Isotopic analysis offers a repeatable and tested way to estimate trophic level and resource use of a specimen, allowing for comparisons among studies and resolving important temporal and spatial variability of trophic relationships (Logan et al., 2020). As an example, isotopic analysis identified the importance of high trophic level omnivores in stabilizing food webs in marine protected areas in Fiordland, New Zealand (Wing & Jack, 2014). These trophic parameters can be estimated by examining stable isotopes in fish muscle tissues, which provide long-term indicators of diet due to their low turnover rates, averaging the isotope values of a fish’s diet for several weeks to months (Hesslein et al., 1993; Mont’Alverne et al., 2016; Skinner et al., 2017; Suring & Wing, 2009). Stable carbon and nitrogen isotope ratios, expressed as a deviation from a standard (δ13C and δ15N), are the most used for tracking organic matter source, estimating trophic level and studying trophic niche of animals (Ramos & González-Solís, 2012). While trophic level estimates rely on isotopic fractionation of nitrogen (15N between 1.4 and 5.6‰) when organic matter is transferred from a resource to a consumer, δ13C presents small fractionation with trophic exchanges (13C ~ 0.5‰), making

134 it a more precise value for estimating proportions of basal organic matter supporting a food web (McCutchan et al., 2003; Mill et al., 2007; Post, 2002). Because different sources of basal organic matter can have distinct δ13C and δ15N values, a combination of these isotopes has been used in aquatic ecosystems to estimate both trophic level and relative contribution of alternate organic matter sources (Jack & Wing, 2011; Phillips et al., 2005). These metrics classify an organism's interactions with its environment and can be used to study the effects of anthropogenic impacts on trophic relationships (Bearhop et al., 2004; Hussey et al., 2014). To estimate these parameters, one needs to know the δ13C and δ15N values of the primary producers supporting a food web, which can vary spatially and with time (West et al., 2010). Compound-specific stable isotope analysis of amino acids (CSIA-AA) has the potential to eliminate the need for isotope values from primary producers (Chikaraishi et al., 2009; McMahon et al., 2016). Different amino acids fractionate in different ways and can be classified as source (no-to-little fractionation of δ15N), trophic (large fractionation of δ15N), 13 13 13 essential (δ CEAA, no-to-little fractionation of δ C) or non-essential (δ CNEAA, large fractionation of δ13C), allowing researchers to access information regarding a specimen's trophic ecology from a single muscle tissue sample (Whiteman et al., 2019). The technique is especially useful when baseline values for the estimation of trophic level are not known or cannot be sampled, allowing one to investigate changes in trophic parameters at large spatial and temporal scales (Hilton et al., 2006; Lorrain et al., 2009; Quillfeldt & Masello, 2020). 15 15 15 For example, the relationship between δ N of Glutamic acid (δ NGlx, trophic AA) and δ N 15 of Phenylalanine (δ NPhe, source AA) have been used to accurately estimate trophic level of aquatic animals (Chikaraishi et al., 2009).

While trophic level is a property of species, Layman’s metrics measured from a δ15N and δ13C biplot have been used to investigate seasonal effects, as well as exploitation and species introduction on the niche space of aquatic communities of (Abrantes et al., 2014; Jackson et al., 2012; Saporiti et al., 2014). These metrics can indicate variability in trophic level and resource use within communities and populations, serving as a proxy of the diversity and stability of food web structure (Layman et al., 2007; Quevedo et al., 2009), providing a tool to compare community trophic structure between systems (Mancinelli & Vizzini, 2015). This is especially advantageous when studying long-term changes in large communities with incomplete biological data (Bowes et al., 2017; Dammhahn & Kappeler, 2014; Layman et al., 2007).

The lack of long-term biological data collected from most aquatic systems hampers the ability to investigate ecological changes at relevant temporal scales, matching long-term

135 human and natural effects. Museum collections are a large source of fish samples for long- term ecological studies that can occasionally encompass several centuries. Most of the specimens kept by museums have been previously fixed in formalin (usually a 37% aqueous solution of formaldehyde) followed by ethanol or isopropanol preservation (Roberts et al., 2015). Preservatives are known to affect isotope values of fish muscle tissue, and these effects must be corrected before comparison with modern samples (Durante et al., 2020b; Sweeting et al., 2004). Furthermore, specimens’ size can present a positive relationship with trophic level (Dalponti et al., 2018; Ladds et al., 2020), but for practicality, museums tend to keep only small to medium size specimens in their wet collections, skewing historical sample representation towards small individuals.

Although there are difficulties related to retrieving good quality isotopic data of fishes from museum collections, these collections often represent the only available samples to study long-term changes in community trophic structure and its relationship with environmental changes and human activities. As in most countries, New Zealand's marine environment has been altered through anthropogenic impacts such as nutrient inputs, fisheries and climate change, with consequences to the underlying food web (Brown et al., 2010; Crowder et al., 2008; Jennings & Polunin, 1997; Jeppesen et al., 1998; Thrush & Dayton, 2010). For example, changes in water temperature can cause shifts in the occurrence and abundance of important prey species, altering the energy transfer through the environment (Adams et al., 1982; Furlan et al., 2013; Rijnsdorp et al., 2009; Wagner et al., 2013). In New Zealand, pelagic crustaceans such as the squat lobster (Munida gregaria, Munididae) are important for the diet of commercial fishes and pelagic-benthic coupling processes (Funes et al., 2018; Graham, 1939; Zeldis & Jillett, 1982). Munida has been demonstrated to make a disproportionate contribution to the diet of a large array of commercial fishes in New Zealand, from reef-associated to pelagic and deepwater species (Graham, 1939; Zeldis & Jillett, 1982). It presents a pelagic (larvae and post-larvae) and a benthic morphological type, the latter settling after the pelagic phase (Zeldis, 1985). The duration of the pelagic type before settlement depends on environmental conditions and availability of benthic habitat, ranging from absent to a couple of years (Williams, 1973, 1980). Because of the existence of these morphological types and its ability to feed on both pelagic and benthic systems, changes in the occurrence of Munida have the potential to affect the trophic dynamic and transport of energy throughout the seascape. Although Munida shoaling occurrence data have been recorded from the University of Otago Portobello Marine Laboratory (PML) since 1904 (Young, 1925), long-term variation in its occurrence has not been reported for New Zealand waters. 136

While global drivers such as increases in sea surface temperatures can affect the functioning of whole ecosystems, fishing tends to affect communities differently over time due to the selective nature of fishing mortality (Verba et al., 2020). Mortality due to fisheries has been reported as the factor resulting in the fastest observed changes in species composition, trophic structure and population vital rates within marine communities (Fenberg & Roy, 2008). For example, reduced prey abundance due to species removal has been shown to drive the increase in the niche width of fishes (Wang et al., 2019), although similar patterns have been recorded due to reduced habitat productivity (Wing & Jack, 2012). The average trophic level of commercial landings, or Marine Trophic Index (MTI), has been used worldwide as a proxy of the status of fisheries and the trophic composition of the exploited communities (Pauly et al., 1998; Pauly & Watson, 2005; Pinkerton et al., 2017). An increase in MTI values is usually related to fisheries expansion, due to large catches of high trophic level species from newly exploited grounds (Caddy & Garibaldi, 2000; Kleisner et al., 2014). With time, the abundance of high trophic level species decreases and there is a relative increase in catches of lower trophic level species, lowering the MTI, a process termed “fishing down marine food webs” (Pauly et al., 1998). In New Zealand, a large expansion of fishing grounds after 1970 was followed by a decline in both landings and MTI after the year 2000 (Durante et al., 2020a)(Durante et al., 2020a). New Zealand landings and MTI stabilized after this period, indicating the end of the fishing expansion and no overall change in the size of landings. During this expansion, different habitats were exploited at different rates, due to the shift from small scale and coastal to large scale, offshore and deepwater fisheries, mainly composed of large trawlers (Durante et al., 2020a; Johnson & Haworth, 2004). This likely resulted in different impacts over distinct habitats, with inshore species showing earlier signs of overexploitation than offshore communities. Although mass balance models have been used to investigate the human impact on the trophic interactions in the marine ecosystems of New Zealand, comparisons based on long-term biological samples are still lacking (MacDiarmid et al., 2016; Pinkerton et al., 2015).

The east coast of the South Island is an important region for New Zealand’s commercial and recreational fisheries and has been managed as a single management area for most species. Two of the largest New Zealand fishery ports are located in this area (Lyttelton and Timaru) (Fisheries New Zealand, 2016). The north (~42˚ latitude) and south (~46˚ latitude) regions of its continental shelf are rocky reefs, while sandy gently sloping bottom extends 40 km offshore in between (Figure 5.1). This geomorphology facilitated the development of an extensive inshore trawl fishery, resulting in one of the highest trawl fishing effort recorded in the country (Baird et al., 2011). 137

Figure 5.1. Map of New Zealand marine environment with the location of historical and modern fish samples analyzed in the present study. White represents New Zealand landmass and grey shades identify depth bands of 0 to 500, 500 to 1000, 1000 to 2000 and deeper than 2000 meters

Long-term changes on the trophic structure of the exploited fish communities, as well as the effects of climate change, prey abundance and fisheries activities have not been previously studied in New Zealand, presenting a lack of knowledge on ecological baselines for the management of these resources. Here bulk isotope and CSIA-AA data of preserved fish specimens from museum collections were used to investigate shifts in trophic level and proportions of alternate basal organic matter sources of individuals collected between 1919 and 2018. The samples used in the present study comprised commercial fish species ranging from inshore to mid slope habitats (Table 5.1).

138

Table 5.1. Maximum total length, habitat, depth distribution, assemblage group and common prey items of the fish species analyzed in the present study Total length Species (mm) Habitat Common prey items Reference Leatherjacket 310 Inner shelf* Crustaceans (Anderson et al., 1998) Meuschenia Demersal Molluscs (Russell, 1983) scaber 0 - 100 m Sponges Tunicates Echinoderms Algae Blue cod 600 Inner shelf* Fish (47%) (Graham, 1939) Parapercis Demersal Crustaceans (23%) (Russell, 1983) colias 0 - 150 m Tunicates (17%) Personal observation Cephalopods (Carbines & Mckenzie, 2001) Gastropods (Fisheries New Zealand, 2019) Bivalves Worms Gurnard 550 Inner shelf* Crustaceans (50%) (Godfriaux, 1970) Chelidonichthys Demersal Fish (43%) (Fisheries New Zealand, 2019) kumu 0 - 200 m Worms (Stevens et al., 2011)

Elephant fish 970 Inner shelf* Bivalves (Fisheries New Zealand, 2019) Callorhinchus Demersal Jellyfish/hydroids (Anderson et al., 1998) milii 0 - 200 m (Coleman & Mobley, 1984) (Graham, 1939) Common 750 Outer shelf* Salps (97%) (Bagley et al., 1998) warehou Benthopelagic Crustaceans (Fisheries New Zealand, 2019) Seriolella 0 - 400 m Jellyfish/hydroids (Stevens et al., 2011) brama Cephalopods Barracouta 2000 Outer shelf* Crustaceans (77%) (Fisheries New Zealand, 2019) Thyrsites atun Benthopelagic Fish (18%) (Graham, 1939) 0 - 400 m Squid (9%) (O’Driscoll, 1998) (Stevens et al., 2011) Tarakihi 700 Outer shelf* Chitons (Godfriaux, 1974) Nemadactylus Demersal Gastropods (Graham, 1939) macropterus 0 - 486 m Bivalves (Russell, 1983) Crustaceans (McKenzie et al., 2017) Echinoderms (Annala, 1988) Polychaetas Cephalochordates Spiny dogfish 1360 Outer shelf* Fish (Fisheries New Zealand, 2019) Squalus Benthopelagic Squid/octopus (Cortés, 1999) acanthias 0 - 500 m Planktonic invertebrates (Ellis et al., 1996) Crustaceans (Ebert et al., 1992) Jellyfish/hydroids (Fuita et al., 1995) Worms (Demirhan & Seyhan, 2007) Bivalves Giant stargazer 780 Outer shelf* Fish (58%) (Stevenson, 2004) Kathetostoma Demersal Cephalopods (38%) (Sutton, 2004) giganteum 0 - 600 Crustaceans (12%) (Stevens et al., 2011) Continues in the next page

139

Species Total length (mm) Habitat Common prey items Reference Red cod 900 Outer shelf* Crustaceans (79%) (Stevenson, 2004) Pseudophycis Demersal Fish (25%) (Cohen et al., 1990) bachus 0 - 700 m Molluscs (Fisheries New Zealand, 2019) Worms (Graham, 1939) Echinoderms (Edgar & Shaw, 1995) Sea perch 470 Slope* Crustaceans (62%) (Paulin, 1989) Helicolenus Demersal Fish (18%) (Smith, 1998) percoides 0 - 500 m Worms (Graham, 1939) Lookdown 700 Slope* Crustaceans (82%) (Anderson et al., 1998) dory Bathydemersal Fish (20%) (Tracey et al., 2007) Cyttus traversi 200 - 600 m Cnidarians (Forman & Dunn, 2010) Worms (Blaber & Bulman, 1987) Hoki 1020 Slope* Fish (60%) (Fisheries New Zealand, 2019) Macruronus Benthopelagic Crustaceans (43%) (Horn & Sullivan, 1996) novaezelandiae 200 - 600 m Cephalopods (5%) (Blaber & Bulman, 1987) Sponges (Bulman & Blaber, 1986) Tunicates (Clark, 1985) Planktonic invertebrates Hapuka 1500 Slope* Fish (68%) (Paxton et al., 1989) Polyprion Demersal Cephalopods (25%) (Saldanha, 1997) oxygeneious 0 - 850 Crustaceans (18%) (Rojas et al., 1985) Ling 2000 Slope* Fish (65%) (Nielsen et al., 1999) Genypterus Bathydemersal Crustaceans (37%) (Stevenson, 2004) blacodes 0 - 1000 m Cephalopods (3%) (Dunn et al., 2010) Ophiuroids (Graham, 1939) (Kailola et al., 1993) Orange roughy 500 Mid shelf* Crustaceans (58%) (Gordon & Duncan, 1987) Hoplostethus Bathypelagic Fish (41%) (Fisheries New Zealand, 2019) atlanticus 700 - 1500 m Cephalopods (Bulman & Koslow, 1992) Planktonic invertebrates Worms *Species assemblage according to Beentjes et al., (2002) and Francis et al., (2002) Percentage of occurrence of food items were retrieved from Stevens et al., (2011)

These species were responsible for over 60% of the total national commercial landings and export value of finfish in 2018, excluding processed products (Stats NZ, 2019). Management areas and stock delimitation varied among species, from the whole EEZ (e.g. hoki, Macruronus novaezelandiae, Merlucciinae) to the east coast of the South Island (e.g. blue cod, Parapercis colias, Pinguipedidae). Fishes are also managed as single species stocks, based on maximum sustainable yield targets, without considering overall community productivity and trophic interactions, which can lead to deterioration in the ecosystem structure by disproportionally removing larger individuals (Walters et al., 2005). Using stable isotope analysis and Layman’s metrics the present study aimed to answer the following questions: (1) have the niche space of the fish community, as well as (2) the trophic level and resource use of species, changed after the full expansion of New Zealand fisheries? (3) Did changes in trophic level and resource use present a relationship with climate, sea surface temperature, prey abundance and fisheries activities?

140

5.2. Methods

5.2.1. Sample collection and bulk isotope analyses

Sixteen species of fish were collected along the east coast of the South Island between 2017 and 2018 (Table 5.1), comprising the “modern” sample group. Coastal species were collected during scientific cruises onboard the RV Polaris II or provided by recreational anglers, while fishes that inhabit offshore waters were mainly provided by commercial fishers or from the Canterbury Bight and Pegasus Bay trawl survey (MacGibbon et al., 2019). Deeper water species, such as lookdown dory (Cyttus traversi, Cyttidae), orange roughy (Hoplostethus atlanticus, Trachichthyidae) and hoki were also acquired from the Chatham Rise trawl surveys (Stevens et al., 2018). Specimens collected onboard research vessels were measured for total length and head length and were frozen and taken to the laboratory for muscle tissue sampling. Since samples provided by fishermen often consisted of fish heads only, a linear regression between total length and head length was calculated for each species to estimate the total length of those samples (Table 5.2).

Table 5.2. Relationship between total length and head length measurements obtained during the collection of fish specimens for the present study Common name Scientific name Total length regression equation Sample size r2 p Barracouta Thyrsites atun 163.32 + 3.3*HL 32 0.83 <0.0001 Blue cod Parapercis colias 26.36 + 3.5*HL 84 0.98 <0.0001 Common warehou Seriolella brama -109.85 + 5.45*HL 12 0.94 <0.0001 Elephant fish Callorhinchus milii 141.35 + 2.4*HL 9 0.59 0.015 Giant stargazer Kathetostoma giganteum 7.66 + 4.02*HL 6 0.8 0.015 Gurnard Chelidonichthys kumu 36.78 + 3.55*HL 27 0.96 <0.0001 Hapuka Polyprion oxygeneios -40.26 + 2.8*HL 4 0.99 0.0018 Leatherjacket Meuschenia scaber 48.84 + 3.45*HL 29 0.83 <0.0001 Ling Genypterus blacodes 150.81 + 3.8*HL 9 0.86 0.0003 Red cod Pseudophycis bachus 127.16 + 3.47*HL 14 0.52 0.0036 Sea perch Helicolenus percoides 62.32 + 2.56*HL 56 0.88 <0.0001 Spiny dogfish Squalus acanthias 7.19 + 4.79*HL 20 0.93 <0.0001 Tarakihi Nemadactylus macropterus -13.95 + 4.21*HL 38 0.92 <0.0001

To access the stable isotope values and estimate trophic parameters of fish species before the present date, specimens were retrieved from the Otago Museum and the Museum of New Zealand Te Papa Tongarewa collections, spanning 1919 to 1996 and comprising the “historical” sample group. Organizing species into the historical group was necessary due to low sample sizes for specific years from the museum collections. These samples represent the community state before the full expansion of industrialized fisheries in New Zealand in the year 2000, allowing a comparison with fully exploited environments sampled in 2017 and 2018. In order to increase the sample size of the historical group, lookdown dory (n = 1),

141 hapuka (Polyprion oxygeneios, Polyprionidae) (n = 2) and hoki (n = 1) specimens from museums collected before 2012 were added to the historical group. Muscle tissue (1 cm3) was sampled from the dorsal musculature of all specimens for isotope analysis. Tissue samples from museum collections were then treated with deionized water for one week to remove excess preservatives before analysis (Durante et al., 2020b).

Muscle tissues were oven-dried at 60 °C for 72 h and crushed with mortar and pestle until they yielded a fine and homogenous powder. All equipment was cleaned with low- linting Kimwipes® and ethanol and air-dried between samples to avoid cross-contamination. Between 0.8 and 1.2 mg of each sample was packed in 3 x 5 mm tin capsules and analyzed by combustion in an elemental analyzer (Carlo Erba NC2500) to CO2 and N2, as described in chapter 2 (section 2.2.2). All measured values for the quality control standards were in the range of accepted values. Additionally, one sample of fish muscle tissue was analyzed as an internal standard in every run so results could be corrected for in-between run variability.

5.2.2. Changes in community niche width

Because lipids are depleted in δ13C compared to protein and carbohydrates, lipid concentration in the samples can affect δ13C values. This change in δ13C is not linked to the trophic parameters of interest and need to be corrected. In the present study the following normalization equation was applied to all samples collected in 2017 and 2018 to account for variation in lipid concentrations, after Post et al. (2007):

13 13 δ CNormalized = δ CUntreated – 3.32 + 0.99 * C:N 5.1)

13 13 Here, δ CUntreated represents raw δ C measurements and C:N the carbon to nitrogen ratio of the muscle tissue. This normalization is suitable for aquatic organisms with similar range of C:N and δ13C values of samples in the present study (Post et al., 2007). To account for the effects of fixatives and preservatives on the isotope values of muscle tissue, correction equations were applied to all samples from museum collections, for both δ15N and δ13C after (Durante et al., 2020b) (see chapter 4). Equation 4.4 (chapter 4) also normalizes δ13C values due to different lipid content in muscle tissue.

142

δ13C values of primary producers have also been globally influenced by anthropogenic activities since the industrialization period in the 1950s (Keeling, 1979; Verburg, 2007), and to compare with present values those changes need to be taken into account. These processes are known to influence fishes from different trophic levels and have been reported for the 13 Otago region (Sabadel et al., in press). Suess effects, a decrease in δ C of CO2 in the atmosphere linked to fossil fuel emissions, were accounted for using values after Eide et al. 13 (2017), which predicts a decrease in δ CBulk of on average 0.011‰ per year (−0.014 ± 0.001‰ to −0.006 ± 0.001‰) in the ventilated South Pacific Ocean.

After isotope corrections, changes in niche width between periods were then investigated using Layman’s metrics (Layman et al., 2007). These metrics allowed the present study to use isotope values to investigate the size, variation and relationship of niches at the species, assemblage and community level. While ranges of δ15N and δ13C values of an array of samples are proxies of community trophic level and resource use diversity, the total area of the minimum convex hull of the δ15N and δ13C biplot can be calculated to estimate the total trophic diversity, or niche breadth, in a food web (Layman et al., 2007). Similarly, the mean Euclidian distance from the δ15N and δ13C centroid of each species to the food web centroid (CD) indicate the level of trophic diversity, although this metric is also a function of trophic packing between species (Layman et al., 2007). Moreover, trophic packing and its evenness can be estimated by the mean nearest neighbor distance (MNND) and the standard deviation (SDNND) of the same plot, respectively. To account for uncertainty related to the nature of the data and small sample sizes, ellipse-based versions of Layman’s metrics estimated from Bayesian inference were used, with the R package SIBER (Jackson et al., 2011). This approach is more suitable for comparisons between groups with different and small sample sizes, improving estimates and reducing uncertainty compared to the original metrics. For examples, minimum convex hull values of the δ13C and δ15N biplot, or niche breadth, can become biased by large isotope values from few individuals. The calculation of the Standard Ellipse Areas corrected for small sample sizes (SEAc) with the SIBER package instead provides Bayesian distributions that account for the uncertainty of estimates and result in more robust and less biased comparisons. To investigate changes in niche breadth between periods, Layman’s metrics were calculated for historical and modern samples and compared for the whole community and for specific species and fish assemblages (inner shelf, outer shelf, slope and mid slope), after Francis et al. (2002) and Beentjes et al. (2002) (Table 5.1).

143

5.2.3. Changes in species trophic level and resource use

Coastal (macroalgae), and pelagic (phytoplankton) primary producers were used as isotopic baselines in the present study (Udy et al., 2019a; Yorke et al., 2019). Stable isotope values from macroalgae were obtained from Otago, Kaikoura (Chapter 3) and the Marlborough Sounds (Udy et al., 2019b), while phytoplankton values inferred from samples of suspended particulate organic matter (SPOM) from inside the Canterbury Bight, along the Kaikoura coast and the Marlborough Sounds were retrieved from Chapter 2 and from Udy et al., (2019b). Both macroalgae and SPOM samples were collected concomitantly with the collection of specimens from the modern group.

Because variability in isotopic baselines can affect the bulk isotope values of fish muscle tissue, region-specific baselines from the two main primary producers, macroalgae and SPOM, were used to estimate trophic level (TL) and proportion of SPOM supporting a species diet (%SPOM) for each fish. TL was calculated from the specimens’ (consumer) and primary producers’ (source) δ15N values, using a two-step iterative procedure following Phillips and Gregg (2001) and explained in detail in chapter 3:

15 15 TL = 1 + (δ NConsumer − δ NSource)/TDF 5.4)

Because of the differences in diet among species, the trophic discrimination factor (TDF) for δ15N was assumed to be 15N = +3.4‰ for fishes mainly feeding on invertebrates and 2.3‰ for species with reported diet being composed by more than 50% fish (Hussey et al., 2014; McCutchan et al., 2003) (Table 5.1). A trophic discrimination factor for δ13C from aquatic environments of 13C = +0.5‰ (SE 0.17) was used after McCutchan et al., (2003).

Differences in TL and %SPOM between periods were analyzed for each separate species using a PERMANOVA, based on the dissimilarities between periods. Unrestricted permutations (10,000) of the data were used. PERMANOVA is a type of analysis of variance that uses permutation to compute statistical tests instead of statistical tables, and does not assume linearity or normality (Anderson et al., 2008).

To test the accuracy of TL estimates from bulk isotopes, values were compared with TL calculated from CSIA-AA. TL based on amino acid isotope results was calculated based 15 15 on differences between δ NGlx and δ NPhe, following Chikaraishi et al., (2009):

144

15 15 TLGlx-phe = 1 + (δ NGlx - δ NPhe - 3.4)/TDFGlx-Phe 5.5)

15 15 Where 3.4‰ is the difference between δ NGlx and δ NPhe in aquatic cyanobacteria and algae (Chikaraishi et al., 2009). Similar to TLBulk estimate, the trophic discrimination 15 15 factor representing the difference in fractionation per TL of δ NGlx and δ NPhe (TDFGlx-Phe), was chosen according to the fish species’ diet (McMahon et al., 2015). In their feeding experiment study, McMahon and co-authors found that fishes fed on high protein diets, where the AA composition was similar to the fishes’ muscle tissue, presented a smaller TDFGlx-Phe than fishes feeding on an omnivorous diet. Therefore, in the present study a TDFGlx-Phe value of 7.6‰ was used for species with fish contribution to its diet equal or lower than 50% (Chikaraishi et al., 2009), while a value of 5.9‰ was chosen for piscivore species (McMahon et al., 2015). To investigate the potential of δ13C of essential amino acids to track resource 13 13 use, the relationship between δ CEAA and %SPOM was also tested. δ CEAA, i.e. valine (Val), 15 leucine (Leu), isoleucine (Ile), Threonine (Thr), and phenylalanine (Phe), as well as δ NGlx 15 and δ NPhe values from samples were analyzed as described in chapter 4, with details in 13 Appendix A.4.2. Mean precision (1 SD) of δ CEAA was of 0.6‰, while the precision for 15 δ NAA had a mean of 0.4‰.

5.2.4. The effects of temperature, prey abundance and fisheries

General linear models were used to examine the long-term relationship of fishes’ 15 trophic level, %SPOM and the δ NSource estimated from the mixing model with natural variations of climate, sea surface temperature, shifts in prey abundance and fisheries activities. Coefficients were calculated with a least square approach, suitable for analysis of complex models in chemistry and biology, with noisy, collinear and incomplete datasets (Wold et al., 2001).

Deviation of the yearly averages from 1953 to 2018 of the Southern Oscillation Index (SOI) and sea surface temperature measured at the University of Otago Portobello Marine Lab (PML SST) were used as climate variation and sea surface temperature indicators, respectively (Goring & Bell, 1999). SOI was calculated from the difference in atmospheric pressure between Tahiti and Darwin; while low SOI values indicated El Niño events, La Niña conditions were identified by high SOI values. The east coast of the South Island of New Zealand can experience droughts during both anomalies, while La Niña is usually associated with higher temperatures (Mullan, 1996). Although the surface temperature recorded at PML

145 represents a localized proxy of ocean warming, it has been reported to follow large scale processes, such as the overall increase in temperatures in boundary currents associated with changes in wind regimes (Shears & Bowen, 2017). Prey abundance was inferred from the sum of the sightings of Munida from PML in each year (Zeldis, 1985; Zeldis & Jillett, 1982), and MTI calculated for New Zealand commercial fishery landings was used as a proxy for the state of fisheries expansion in a given year, after Durante et al., (2020a). Results from Durante et al., (2020a) comprise the most accurate MTI values available for the New Zealand region, with data amalgamated from Fisheries New Zealand and the Food and Agriculture Organization of the United Nations and annual catches identified to species level (see chapter 1). Because MTI was only calculated until 2015, the value from 2015 was used in the years 2017 and 2018. As shown by Durante et al., (2020a), MTI values for New Zealand have not varied greatly in the most recent years, which justify this approximation when comparing to long-term shifts in MTI. Yearly values of SOI, PML SST, Munida sighting and MTI did not have strong correlations (Pearson’s coefficient correlation from -0.32 to 0.42) and were therefore fit for use in the general linear models. Assumptions related to the linear models were checked with residual plots (by predicted values, by row and normal quantile). All statistical analyses were undertaken with R 3.6.3 (R Core Team, 2020), JMP 14.0.0 (SAS Institute, 2018) and PRIMER 6.1.12 (Clarke & Gorley, 2006).

5.2.5. Accounting for the effects of latitude and specimens’ size

Because isotope values of fish muscle tissue, as well as trophic level and resource use can be influenced by location and specimens’ size (see chapter 3), these parameters need to be accounted for prior analyses. To investigate the effects of year, location and fish size on bulk isotopes and trophic parameters, a general linear model was fitted for the corrected values of δ15N and δ13C, with total length, as well as the latitude and years of sampling as the explanatory variables. For species with significant relationships between bulk isotopes, trophic parameters and total length, large specimens from the modern group and small ones from museum collections were excluded before analyses, until both groups achieved statistically indistinguishable distributions of total length (Wilcoxon test p > 0.05). Similarly, modern and historical groups were only compared when samples came from the same latitudinal range.

146

5.3. Results

5.3.1. Changes in community niche width

The area occupied by the communities’ isotopic space varied between before and after the full expansion of industrialized fisheries in New Zealand. SEAc was significantly lower in the historical (1.82‰2) than in the modern community (3.56‰2, Figure 5.2). These differences were above the 95% credible interval calculated for the historical (1.51 to 2.18‰2) and modern (3.09 to 4.06‰2) communities. Other community metrics also presented an increase between periods, with increasing of δ13C range (0.68 to 1.27‰), δ15N range (1.4 to 2.43‰), CD (0.52 to 1.03‰), MNND (0.48 to 0.96‰) and SDNND (0.49 to 0.99‰). The increase in SEAc over time displayed different trends for different fish assemblages: while the inner shelf and mid slope assemblages had small changes over time, outer shelf (1.96 to 2 2 3.25‰ ) and slope (1.2 to 1.95‰ ) displayed large increases in SEAc (Figure 5.3). The observed differences in community-wide SEAc were mainly driven by changes in three species from the outer shelf: red cod (Pseudophycis bachus, Moridae), spiny dogfish (Squalus acanthias, Squalidae) and tarakihi (Nemadactylus macropterus, Cheilodactylidae) (Figure

5.4). Species-specific responses in SEAc were less clear from the slope assemblage, although ling (Genypterus blacodes, Ophidiidae) has had a large increase in SEAc between periods (Figure 5.4). δ13C range was the only metric to be significantly different between periods for the outer shelf assemblage (Figure 5.5a), i.e. with group differences above 95% credible intervals. Similarly, slope assemblage had an increase above the 75% credible intervals for δ13C range (1.16 to 2.12‰) and CD (0.72 to 1.04‰) (Figure 5.5b and 5.5c).

Figure 5.2. Isotopic niche space represented by corrected δ13C and δ15N values from fish communities of the east coast of the South Island sampled at different periods. Black and red circles represent specimens from historical and modern periods, respectively, with solid lines representing Bayesian 95% ellipses and dotted lines representing the minimum convex hull of each community 147

b BAR - Barracouta WAR - Common warehou STA - Giant stargazer RCO - Red cod SPD - Spiny dogfish TAR - Tarakihi HOK - Hoki LIN - Ling LDO - Lookdown dory SPE - Sea perch

148

Historical Modern

Figure 5.5. Bayesian inferred δ13C range for the outer shelf assemblage (a) and mean distance to the centroid (b) and δ13C range (c) for the slope assemblages of historical and modern samples. Black dots represent the mode value, with their respective 50, 75 and 95% credible intervals. Red crosses represent the calculated maximum likelihood value

5.3.2. Changes in species trophic level and resource use

Changes in %SPOM between periods were identified for species from the inner shelf and outer shelf, while differences in species TL were observed in all assemblages. A significant increase in %SPOM between periods was identified for blue cod, common warehou (Seriolella brama, Centrolophidae), spiny dogfish and tarakihi (Table 5.3 and Figure 5.6), with the largest differences displayed by tarakihi (average ± SE of 0.51 ± 0.05 to 0.67 ± 0.03). Similarly, TL also had significant increases between historical and modern samples for gurnard, red cod, tarakihi, hoki and orange roughy, with the largest observed differences for orange roughy (3.12 ± 0.04 to 3.49 ± 0.04). Sea perch was the only species with a decrease in TL between time periods (Table 5.3 and Figure 5.6). At least 4293 unique permutations were

149 compared for each species during PERMANOVA analysis of trophic parameters from bulk

isotope estimates.

Trophiclevel

Figure 5.6. Average of estimated trophic level vs contribution from pelagic production to a fish’s diet (%SPOM) for specimens collected from the east coast of the South Island of New Zealand in historical (H) and modern (M) periods, with respective standard errors. Arrows point from historical to modern species values. Inner shelf (a), outer shelf (b), slope and mid slope (c) assemblages were plotted separately

150

Table 5.3. Average ± standard error of contribution from pelagic production to a fish’s diet (%SPOM) and estimated trophic level (TL) of commercial species collected from the east coast of the South Island during historical and modern periods. t values from PERMANOVA comparison between periods are shown with its respective significance %SPOM Trophic level Species Assemblage Historical Modern t (perm) Historical Modern t (perm) Blue cod Inner shelf 0.5 ± 0.03 0.6 ± 0.01 3.126** 3.11 ± 0.07 3.07 ± 0.04 0.394 Elephant fish 0.43 ± 0.04 0.34 ± 0.04 1.414 2.7 ± 0.11 2.97 ± 0.01 1.528 Gurnard 0.41 ± 0.03 0.41 ± 0.01 0.035 3.12 ± 0.03 3.42 ± 0.03 6.955*** Leatherjacket 0.52 ± 0.03 0.51 ± 0.02 0.475 3.02 ± 0.04 3.1 ± 0.05 1.420 Barracouta Outer shelf 0.59 ± 0.06 0.52 ± 0.02 1.530 2.93 ± 0.11 2.76 ± 0.08 1.300 Common warehou 0.51 ± 0.02 0.56 ± 0.01 2.379* 3.19 ± 0.04 3.17 ± 0.05 0.250 Giant stargazer 0.46 ± 0.02 0.5 ± 0.01 2.000 4.03 ± 0.17 3.9 ± 0.06 0.898 Red cod 0.59 ± 0.05 0.55 ± 0.02 0.902 2.95 ± 0.04 3.26 ± 0.04 3.986** Spiny dogfish 0.51 ± 0.01 0.59 ± 0.03 2.151* 2.31 ± 0.03 2.87 ± 0.08 5.553*** Tarakihi 0.51 ± 0.05 0.67 ± 0.03 3.128** 2.96 ± 0.05 3.11 ± 0.04 2.2154* Ling Slope 0.46 ± 0.03 0.5 ± 0.03 0.957 4.06 ± 0.1 4.15 ± 0.1 0.600 Sea perch 0.52 ± 0.02 0.54 ± 0.02 0.651 2.98 ± 0.06 2.64 ± 0.05 4.267*** Hoki 0.69 ± 0.05 0.71 ± 0.01 0.210 3.33 ± 0.07 3.68 ± 0.03 4.029** Lookdown dory 0.61 ± 0.06 0.53 ± 0.01 1.572 3.01 ± 0.04 2.93 ± 0.03 1.613 Orange roughy Mid slope 0.72 ± 0.01 0.69 ± 0.01 1.706 3.12 ± 0.04 3.49 ± 0.04 6.413*** 0.05 (*), 0.005(**) and 0.0005 (***) levels of significance

13 A significant decrease in δ CEAA between periods was identified for red cod, tarakihi, 13 hapuka and orange roughy, and decrease for hoki (Table 5.4), although δ CEAA displayed no relationship with %SPOM (Figure 5.7a). Between 126 and 2891 unique permutations were 13 compared for each species during PERMANOVA analysis of δ CEAA and TLGlx-Phe. TLBulk

estimates agreed with TLGlx-Phe, indicating the accuracy of both techniques to estimate TL

(Figure 5.7b), but TLGlx-Phe differences between periods differed for some species compared to

TLBulk results. TLGlx-Phe only differed significantly between periods for tarakihi and orange roughy, with large differences between historical and modern samples (3.34 ± 0.08 to 4.25 ± 0.04 for orange roughy) (Table 5.4).

Table 5.4. Average ± standard error of δ13C of essential amino acids and trophic level (TL) estimated 15 15 from δ NGlx - δ NPhe of commercial species collected from the east coast of the South Island during historical and modern periods. t values from PERMANOVA comparison between periods are shown with its respective significance Essential AA Trophic level (Glx-Phe) Species Assemblage Historical Modern t (perm) Historical Modern t (perm) Blue cod Inner shelf -21.19 ± 0.62 -21.71 ± 0.55 0.627 3.53 ± 0.11 3.5 ± 0.08 0.274 Barracouta Outer shelf -24.97 ± 0.3 -26.04 ± 0.88 1.247 3.6 ± 0.05 3.41 ± 0.11 1.69 Red cod -22.67 ± 0.68 -25.64 ± 0.37 3.412* 3.53 ± 0.17 3.71 ± 0.14 0.78 Tarakihi -19.46 ± 0.25 -22.79 ± 0.37 7.633*** 3.12 ± 0.05 3.53 ± 0.03 5.7409*** Hapuka Slope -20.18 ± 0.35 -23.75 ± 0.45 5.956*** 4.53 ± 0.04 4.75 ± 0.08 2.172 Ling -23.11 ± 0.57 -22.89 ± 0.34 0.303 4.62 ± 0.22 5.26 ± 0.28 1.866 Hoki -25.1 ± 0.76 -21.72 ± 0.44 3.326* 4.18 ± 0.1 4.15 ± 0.07 0.267 Orange roughy Mid slope -21.35 ± 1.22 -24.88 ± 0.24 2.841* 3.34 ± 0.08 4.25 ± 0.04 10.39* 0.05 (*), 0.005(**) and 0.0005 (***) levels of significance

151

Figure 5.7. Linear regressions between the average contribution from pelagic production to a fish’s diet (%SPOM) and δ13C values of essential amino acids (a) and trophic level estimates from bulk isotopes and compound-specific isotopic analysis of amino acids (CSIA-AA) (b). Grey and black symbols represent historical and modern specimens, respectively, with their linear regression lines and standard error bars

5.3.3. The effects of temperature, prey abundance and fisheries

SOI, PML SST, Munida sightings and MTI varied in different ways between 1953 and 2018. While SOI and PML SST displayed cyclic periods of high and low values, expected from climate data, Munida sightings had a general decreasing and MTI an increasing trend with time (Figure 5.8). When all species were pooled together, SOI had a significant contribution to the general linear model explaining %SPOM (slope ± SE of 0.05 ± 0.02, p = 0.011). Similarly, variation of TL estimated from bulk isotopes had a relationship with both PML SST (0.17 ± 0.05, p = 0.0006) and MTI (-0.39 ± 0.18, p = 0.032). SOI (0.22 ± 0.05, p < 0.0001), MTI (0.56 ± 0.18, p = 0.002) and Munida sightings (0.01 ± 0.001, p = 0.0001) also

152

15 contributed significantly to the linear model explaining the δ NSource values. Because of the variability in the ecological parameters among species and assemblages, linear models did not provide a good fit to the whole dataset (r2 between 0.02 and 0.05). To account for species variability, a linear mixed effect model was run with the same variables but with species as a random effect. Results were similar then previous models, but with a higher proportion of the 2 15 variability explained (r = 0.28 for %SPOM, 0.44 for δ NSource and 0.69 for TL) and significant species random effect at p < 0.02 for all variables. When species were considered 15 as random effect, MTI had a positive relationship to %SPOM (p = 0.011), δ NSource (p = 0.029) and a negative to TL (p < 0.0001), but the mixed effects model residuals from the last two variables were skewed from the normal distribution and were not improved with transformation, hampering our comparisons.

Figure 5.8. Yearly deviation of the average of Munida sightings, Southern Oscillation Index (SOI), the temperature recorded from University of Otago Portobello Marine lab (PML SST) and New Zealand Marine Trophic Index (NZ MTI) from 1953 to 2018

Relationships between trophic parameters had a higher proportion of the variance explained when each assemblage was analyzed separately, especially for species from the slope. %SPOM of inner shelf species had a significant relationship with Munida sightings, PML SST and MTI, although the regression only explained a small proportion of %SPOM variation (r2 = 0.12, Table 5.5). MTI had no effect on %SPOM of the outer shelf and slope assemblages, but SOI and PML SST had contribution explaining the variability of that variable, but similar to inner shelf species it explained a small proportion of its variability (r2 153

= 0.07 for outer shelf and 0.23 for slope, Table 5.5). TL of inner shelf species had a linear relationship with Munida sightings and MTI (r2 = 0.13, Table 5.5). Variation of TL from the outer shelf assemblage was only explained by SOI (r2 = 0.19), while TL from the slope assemblage had a relationship with all variables at p < 0.005, except for MTI (r2 = 0.24). All 15 2 variables influenced δ NSource values for the inner shelf assemblage at p < 0.05 (r = 0.1). Similar results were obtained from the slope assemblage (r2 = 0.35), but no relationship was 15 found for PML SST and MTI (Table 5.5). δ NSource estimated from the outer shelf assemblage did not have a significant relationship with any variable analyzed. Trophic parameters from mid slope assemblage, comprising orange roughy only, were not significantly explained by any variables tested. Inner and outer shelf assemblages had a better fit between TL and the explanatory variables (r2 = 0.13 and 0.18, respectively), mostly due to the contributions from elephant fish (Callorhinchus milii, Callorhinchidae) (species-specific general linear model, r2= 0.69), gurnard (r2 = 0.71), barracouta (r2 = 0.41) and spiny dogfish 2 15 (r = 0.56). Slope assemblages had a better relationship between those variables and δ NSource (r2 = 0.35), mainly driven by ling (r2 = 0.51) and hoki (r2 = 0.98) responses, although the small number of sampling years for hoki limited the regression analysis (n = 5). Linear mixed effects models were not run for each specific assemblage due to the small amount of species levels (< 5 for all assemblages except outer shelf) for the random effect analysis.

Table 5.5. Results of general linear model between estimated trophic level (TL), the contribution from 15 15 pelagic production to a fish’s diet (%SPOM), estimated δ N of the basal organic matter (δ NSource); and Southern Oscillation Index (SOI), water temperature recorded from the University of Otago Portobello Marine Laboratory (PML SST), Munida sightings and New Zealand Marine Trophic Index (NZ MTI), for each fish assemblage

%SPOM Munida sightings PML SST SOI NZ MTI Assemblages Estimate SE p value Estimate SE p value Estimate SE p value Estimate SE p value Inner shelf -0.003 0.001 0.003 -0.098 0.025 0.000 0.019 0.044 0.666 0.209 0.079 0.009 Outer shelf 0.000 0.001 0.589 0.094 0.028 0.001 -0.118 0.047 0.012 -0.227 0.125 0.071 Slope 0.002 0.001 0.029 0.013 0.027 0.644 0.093 0.021 <.0001 0.076 0.114 0.507 Mid slope 0.004 0.006 0.566 0.142 0.251 0.581 -0.144 0.196 0.472 0.176 0.685 0.801 δ15NSource Munida sightings PML SST SOI NZ MTI Assemblages Estimate SE p value Estimate SE p value Estimate SE p value Estimate SE p value Inner shelf 0.006 0.002 0.015 -0.165 0.072 0.022 0.340 0.125 0.007 0.535 0.225 0.018 Outer shelf 0.003 0.003 0.218 -0.099 0.081 0.221 0.051 0.134 0.702 0.364 0.358 0.311 Slope 0.005 0.002 0.061 0.406 0.085 <.0001 0.247 0.064 0.000 0.382 0.355 0.285 Mid slope -0.003 0.005 0.566 -0.117 0.208 0.581 0.120 0.162 0.472 -0.146 0.569 0.801 TL Munida sightings PML SST SOI NZ MTI Assemblages Estimate SE p value Estimate SE p value Estimate SE p value Estimate SE p value Inner shelf -0.005 0.002 0.012 0.048 0.056 0.398 0.128 0.098 0.194 -0.468 0.177 0.009 Outer shelf -0.001 0.002 0.629 -0.087 0.075 0.246 0.558 0.124 <.0001 0.368 0.332 0.269 Slope 0.013 0.004 0.001 0.701 0.131 <.0001 -0.301 0.099 0.003 -0.431 0.547 0.432 Mid slope 0.003 0.024 0.915 0.477 0.987 0.636 -0.144 0.769 0.854 0.803 2.696 0.770

154

5.3.4. Accounting for the effects of latitude and specimens’ size

Latitude and total length had significant effects on bulk isotope and trophic parameter values for several species in the present study (Appendices A.5.1 and A.5.2). A total of 658 specimens were sampled and analyzed, 389 from modern and 269 from historical groups. After accounting for the effects of latitude and total length, 540 samples (337 modern and 203 historical) were analyzed for bulk isotopes and 510 for the results of the isotope mixing model (325 modern and 185 historical). A breakdown of sample sizes by species can be seen in Appendix A.5.3. Because historical samples of hapuka were significantly smaller than individuals collected for the modern group, comparisons between TLBulk and %SPOM between periods were not possible for that species.

155

5.4. Discussion

When investigating the trophic structure variability of fish communities composed of large omnivores, the fundamental niche of each species needs to be considered, since their resource use will be affected by how each species interact with the environment (Deudero et al., 2004). These interactions are usually linked to individual size and habitat use, and ultimately to the ecology of each species (Dalponti et al., 2018). Bulk isotopes and trophic parameters estimated for the exploited community varied with latitude along the coast, as well as with total length for most species. The magnitude and direction of those changes differed 13 15 between species, although δ C and %SPOM; δ N and TLBulk tended to increase with latitude and total length, respectively. These results corroborate studies that suggest that fish TL is related to body size, but TL variation is mainly driven by ontogenetic shifts and habitat use within populations (Dalponti et al., 2018; Ladds et al., 2020).

The results indicated a wider niche occupied by modern communities, mainly driven by outer shelf and slope assemblages. There was a general trend towards a larger reliance on pelagic production when compared to macroalgae, as well as a shift to species feeding on higher trophic levels in the modern period. Although the analysis between historical and modern communities was designed to investigate differences in trophic parameters before and after the expansion of industrialized fisheries in New Zealand, a linear relationship between these parameters and environmental variables was also observed for different species and assemblages. These results are a strong evidence that different species and assemblages have been affected differently by environmental changes and fisheries activities, which should be accounted for in the management of these resources.

Although not all species displayed significant changes in %SPOM and TL between periods, several changes occurred in the direction of values previously occupied by other species. For example, spiny dogfish specimens from the modern group occupied a higher trophic level relying more heavily on pelagic production, similar to red cod values from past samples. Furthermore, lookdown dory and orange roughy shifted to similar values in modern samples previously occupied by sea perch and hoki in the region, respectively. Small shifts of leatherjacket towards %SPOM and TL values previously occupied by blue cod was also observed in the inner shelf assemblage. Although I was unable in the present study to confirm changes in fish diets, the nature of omnivore fishes, including their large array of prey species, indicate that the species analyzed here were likely to use resources abundant enough to sustain their populations, relying on different resources over time due to productivity levels, competition or habitat degradation (Jack & Wing, 2011).

156

5.4.1. Mid slope assemblage

The mid slope assemblage comprised of orange roughy has only demonstrated a significant increase in TLBulk between periods. The differences observed did not have a linear relationship with Munida occurrence, SOI, PML SST or MTI. Orange roughy is a long-lived (~130 years), bathypelagic species that started to be exploited in New Zealand in the 1970s (Andrews & Tracey, 2003; Fisheries New Zealand, 2019). Because of its large biomass and appreciation in the international market, linked to the large decrease in landings of Atlantic cod (Gadus morhua, Gadidae) in the North Hemisphere, it quickly became an important export product for New Zealand (Johnson & Haworth, 2004). Spawning stock biomass for commercial fish species in the study region decreased to 20% between 1980 and 2005, increasing to around 40% in 2017, within the current management target of 30 to 50% (Fisheries New Zealand, 2019). The removal of large numbers of individuals from fish populations by fisheries since 1970 is likely reflected in the observed increase in TLBulk seen in the present study. Populations with decreased biomass can in some cases have a predation release effect on prey species, allowing them to feed at higher TL (Augé et al., 2012;

Pinkerton et al., 2015b). The lack of relationship between TLBulk shifts and MTI values could be a result of the increase in the national MTI between 1980 and 2000, followed by a decrease in modern times. Therefore, MTI values from the 1980s are very similar to modern values, impairing any linear relationships between these values and the TLBulk shifts recorded for orange roughy.

5.4.2. Slope assemblage

The slope assemblage comprised sea perch, ling, hoki and lookdown dory. An expansion of the community niche breadth between periods was observed, mostly driven by an increase in niche breadth of ling. The change was related to an increase in resource use and trophic diversity (δ13C range and CD) in that community. Although CD is also a function of trophic packing, no significant change in MNND between periods was found, suggesting that the increase in CD values was mainly a result of increasing diversity in resource use as measured by the range of δ13C. Therefore, species were observed to be accessing a wider array of basal resources through food web interactions compared to those used in the historical periods. Although ling was observed to have had a significant increase in its niche breadth, it was the only species from the slope to demonstrate no shifts in %SPOM and TLBulk between periods. The result suggests that ling population was still feeding at a similar average TL and its associated food web was relying on the same average combination of basal 157 organic matter sources, but specialization among individuals had increased for the modern samples. Like the mid slope assemblage, sea perch, hoki and lookdown dory did not demonstrate significant shifts in %SPOM between periods, but TLBulk of fishes sampled in the present, except for sea perch, had higher values than the historical group. Biomass estimates demonstrate no strong variability with time for lookdown dory and sea perch populations in the study region, but those are not available for long periods (1990s to present). On the other hand, hoki stocks had the same pattern observed for orange roughy, with steep declines in biomass after the start of the deepwater fisheries, reaching 30% of the spawning biomass of 1970 in 2005 (Fisheries New Zealand, 2019). This was followed by a recovery in 2014 to 50- 60% of the spawning biomass estimated for 1970.

Among the species with significant shifts in TLBulk between periods, sea perch was the only one to display smaller values in modern samples. Sea perch is a benthic forager with diet dominated by fish, crustaceans and salps (pelagic tunicata) (Horn et al., 2012). Horn et al. (2012) demonstrated that the importance of salps to the diet of sea perch and other salp specialist fishes varied together among years in the Chatham Rise. This result suggests that the diet of sea perch is dependent on prey availability and salp abundance, which could have been higher in modern periods, especially due ocean warming (Pakhomov et al., 2006). Since salps are microphages, i.e. feeds at small particles in the water column and therefore at small TL, increase in salp consumption by sea perch could have driven the decrease in TL observed in the present study.

SOI was the only variable to contribute significantly to the temporal trends of 15 %SPOM, δ NSoure and TLBulk for the slope assemblage. The linear model slopes suggest that during years with higher SOI values (towards La Niña anomaly) and large Munida occurrences, the community relied more heavily on pelagic producers supported by a higher 15 δ NSource and lowering their estimated TLBulk. Interestingly, years with higher temperatures recorded at PML were consistent with higher TL estimated for species from the slope assemblage, suggesting that the decreases in the individual’s TLBulk during years of high SOI (and usually warmer temperatures) could have a smaller influence from temperature changes and larger influences from another environmental or biological variables. Warmer years could decrease pelagic productivity by reducing water mixing and transport at surface and subsurface, especially in the north of the east coast of the South Island (Chiswell & Sutton, 2020). Reduced new production would increase particle recycling by bacteria and nutrient reutilization by phytoplankton, especially during sinking, increasing the δ15N values of the basal organic matter pool used by the slope assemblage, consequently increasing their TLBulk

158

(Altabet, 1988). Although the latter is true, PML SST values did not demonstrate a relationship with %SPOM for this assemblage, likely resulting from the lack of resolution of the environmental variables when attempting to understand this complex system (Winder & Schindler, 2004).

5.4.3. Outer shelf assemblage

The increase in niche breadth between periods was driven by tarakihi, spiny dogfish and red cod values and was related to a higher diversity in resource use, inferred by the significantly larger δ13C range in the modern community. Giant stargazer and barracouta were the only species that did not demonstrate a significant change in %SPOM or TLBulk between periods. Other species presented either an increase in the reliance of pelagic primary production (common warehou), TLBulk (red cod) or both (spiny dogfish and tarakihi) between periods. Compared to the other species, barracouta and giant stargazer feed on a more limited prey pool that excludes most of the benthic invertebrates, such as worms, echinoderms, bivalves and cephalochordates, commonly found in the diet of the other species (Table 5.1). The reduced diet diversity of these species could have led to an inability to exploit different prey types supported by different basal organic matter sources. This observation might explain their steady %SPOM and TLBulk values, as well as the lack of a significant difference in SEAc between periods. The results from the present study suggest that the trophic interactions of species with large prey fields have likely changed in the outer shelf and slope before and after the full expansion of New Zealand fisheries. Likewise, changes in diet composition caused by shifts in community structure due to fisheries activities were previously identified in important fishing grounds in George Banks (Garrison, 2000; Link & Garrison, 2002). Although these shifts likely affected species interactions and energy flow in that ecosystem, piscivorous species kept feeding at similar guilds in those studies, feeding on different prey species but at a similar TL supported by a comparable basal organic matter source. Differently, prey consumption and food web interactions resulted from changes in community composition can present more dramatic changes for opportunistic species, as seen for spiny dogfish in Argentina (Alonso et al., 2002), which corroborate the findings of the present study.

SOI had relationships with the variation in %SPOM and TLBulk through time for the outer shelf assemblage, but regression slope values had reverse signals and equations r2 were smaller than the ones seen from fishes from the slope. The outer shelf community was comprised of highly mobile species that can exploit a large array of habitats and resources, 159 including individuals feeding exclusively on the ocean floor (tarakihi), to pelagic ones (barracouta). This high variability in niches likely resulted in a lack of agreement between trophic parameters and the independent variables among species. For example, while trophic parameters for common warehou had no relationship with Munida sightings, PML SST, SOI and MTI; %SPOM of red cod and TLBulk of barracouta had significant correlations with all variables analyzed. Although the sample sizes limited the analysis of separate species, these results indicated that the direction and magnitude of the effects of climate, anthropogenic and natural processes depended on the niche of each species.

5.4.4. Inner shelf assemblage

Fishes from the inner shelf assemblages did not demonstrate large shifts in niche breadth before and after the full expansion of fisheries in New Zealand, with only blue cod

(increase in %SPOM) and gurnard (increase in TLBulk) having a significant shift in the trophic parameters between periods. Fishes from the inner shelf assemblages are known to have been exploited for longer periods than offshore and deep-sea communities. There is evidence of the depletion of large fish inhabiting New Zealand coastal waters since the middle of the 19th and 20th centuries (Macdiarmid et al., 2018; Parsons et al., 2009), both in size and biomass. For example, a fisheries report written by A.E. Hefford in 1937 indicated a decrease in fisheries productivity in coastal trawling grounds off the east coast of the South Island (Johnson & Haworth, 2004), while the reduced occurrence of hapuka in coastal waters was noted by the Marine Department decades earlier: “At Nugget Bay the fishermen report that fishing on the inshore grounds has been rather poor, but further off in the deeper water the catches were quite equal to other years. Hapuku [hapuka] is the principal fish taken” (Fisher, 1913). After the removal of large, high trophic level individuals through fisheries, the niche previously occupied by species such as hapuka was likely filled by other species, which could then impede the recovery of hapuka through competition and predation of juveniles. A similar effect of competitive exclusion was indicated by modeling Atlantic cod populations in the Scotian Shelf (Bundy & Fanning, 2005). Therefore, the largest effects of fisheries on the trophic interactions of the inner shelf food web have likely occurred before the time frame of the present study, explaining the relatively steady state of the current food web.

In contrast with the comparisons between periods, continuous linear models of community trophic structure had significant relationships from changes in climate, anthropogenic and biological processes. Higher occurrences of Munida resulted in decreases in reliance on pelagic production and TL by the inner shelf assemblage, with a subsequent 160

15 increase in δ NSource. Years with colder temperatures also corresponded to decreased 15 %SPOM and δ NSource values, which could be linked to the larger number of Munida inshore (Meerhoff et al., 2013), although those variables were not highly correlated. SOI values had 15 the reverse correlation with δ NSource, indicating again that the effects of SOI on the trophic parameters analyzed could be a result of different environmental processes (e.g. occurrence of draughts) then increased sea surface temperature. The relationships observed between Munida sightings and the other parameters for the inner shelf assemblage presented an inverse trend of the ones found for fishes from the slope. Munida is known to have both pelagic and benthic morphotypes that feed on different basal organic matter pools (Funes et al., 2018). Occurrence and biomass of macroalgae were higher in the diet of Munida collected in shallower waters compared to deeper water specimens in the Beagle Channel, Argentina (Romero et al., 2004). Munida close to shore was likely utilizing organic matter from kelp and reducing the reliance on pelagic organic matter to the inner shelf food web. The same process was observed for the fish from the other assemblages, where invertebrates would rely more heavily on pelagic production, increasing the assemblage’s %SPOM values.

The inner shelf was the only assemblage to demonstrate a significant relationship between trophic parameters and MTI. These results suggest that, within the timeframe of the present study, fisheries activities only had a continuous effect on the trophic parameters of the coastal community analyzed, when compared to the species inhabiting the outer shelf, slope and mid slope. Nevertheless, there was a general trend of increasing fish reliance on basal organic matter from pelagic producers and occupying a lower TL with the continuous 15 expansion of fisheries. δ NSource also increased significantly with MTI, which was in line with the decreased TLBulk values observed. The relationship between trophic parameters, prey abundance, PML SST and MTI were investigated between 1953 to 2018. During that period MTI rose from 3.62 to 4.14 in 1997, falling to 4 in the most recent years. While inner shelf species were exploited since the 19th century, other assemblages remained largely off-target until the 1970s and 1980s, which could explain the lack of contribution observed from MTI to the linear models with trophic parameters.

5.4.5. CSIA-AA estimates

The use of both bulk isotopes and CSIA-AA provided an opportunity for comparing and validating the trophic parameter estimates using two different techniques. This is especially important when analyzing samples from past environments with unavailable isotope baseline from primary producers. In the present study, although some samples had 161

large differences between bulk and AA estimates, in general, species average values of TLBulk agreed with TLGlx-Phe, indicating the accuracy of both estimates. Differences in TLGlx-Phe between periods were also generally consistent with TL from bulk isotopes. Conversely, 13 %SPOM displayed no relationship with δ CEAA, indicating the poor performance of EAA for tracking organic matter sources between macroalgae and SPOM, which can be related to the smaller carbon pool in EAA compared to bulk tissues (Whiteman et al., 2019).

Although historical and modern samples analyzed for CSIA-AA were sourced from similar locations, sample sizes hampered the ability to account for the effects of total length on the trophic parameters estimated by CSIA-AA. Hapuka, red cod and orange roughy had a significant positive relationship between bulk TL and total length, with larger specimens comprising the modern group (Wilcoxon’s test, p < 0.02). These differences in individual size likely influenced TLGlx-Phe values observed for these species, since individuals from the historical group were significantly smaller than from modern samples. Interestingly, no significant increase in TLGlx-Phe was observed for red cod and hapuka between historical and modern samples. These results suggest that smaller individuals of these species have currently a higher TL, occupied by larger individuals in the historical food web. Although the mechanisms of these shifts are still unknown, reduction in the abundance of larger, high TL fishes have the potential to release species and/or individuals from predation, which can occupy the available niche due to species removal (Pace et al., 1999; Stortini et al., 2018). Like for orange roughy and hoki, these findings are in line with the temporal reduction in the abundance of these species (Fisher, 1913; Fisheries New Zealand, 2019).

5.4.6. Caveats

While all samples were arithmetically corrected for different lipid concentrations, the accumulation of urea is also known to affect isotope values in muscle tissue of cartilaginous fishes (Li et al., 2016; Takagi et al., 2014). These effects were not corrected in the present study, since there is no correction equation for these effects for fishes that have been under fixation and preservation (Burgess & Bennett, 2017). Because we are comparing the same species between sites and periods, errors from the lack of ammonia correction are expected not to hinder the analysis of spiny dogfish and elephant fish, the only cartilaginous fish species analyzed in the present study.

Temporal shifts in δ15N of animal tissue can be related to nutrient enrichment from coastal runoff, due to fertilizers, sewage, and other anthropogenic activities (Baker et al., 2010). Although this process has been shown to occur along the Otago coast (Sabadel et al., 162 in press), δ15N enrichment was only seen for herbivorous species feeding directly on kelp, therefore this effect was not considered in the present study.

Because a relationship between length and isotopic values used to calculate the trophic level of fish was observed, the results presented here were dependent on the specimen’s size. While restricting the sizes of specimens compared between periods may be appropriate due to collection biases, size-TL relationship as well as the size structure of the populations at each period are needed to proper investigate trophic structure changes at the community level.

Because specimens from museum collections were only available in small quantities with low level of replication at the year level, grouping them into assemblages was necessary for the analyses between fishes’ trophic parameters, prey abundance, environmental and fisheries data. Bias could have arisen from high number of specimens collected from single years, and lack of samples in others. Analysis of the distribution of samples among years has shown that samples were well distributed along the historical periods for most species, with an increase in sample size for the modern period. Although most datasets presented a normal 15 distribution of residuals, %SPOM and δ NSource of outer shelf assemblage showed lack of agreement in the normal quantile plot. This result was probably originated from the fact that fishes from this assemblage present contrasting life histories, hindering the comparison of trophic parameters when grouped, as discussed previously.

5.4.7. Implications for the ecology and management of fish communities

By removing top predators, exploitation can reduce trophic redundancy, increase trophic diversity and create longer food webs (Saporiti et al., 2014). Removal of large individuals from a community can also have a cascade effect and result in mesopredator release, when smaller predators start preying on resources previously used by higher trophic level species (Estes, 1998; Pace et al., 1999). While New Zealand coastal waters have been fished for centuries, even before the arrival of Europeans (Johnson & Haworth, 2004), offshore and deeper environments would only be largely affected by exploitation after the industrialization of fisheries around 1970 (Durante et al., 2020a). An increase in the variability of resource use and niche breadth of consumers has been linked to impacted and less productive marine systems (Donázar-Aramendía et al., 2019; Jack & Wing, 2011), as well as to reduced growth rates in fish populations (Jacquemin & Doll, 2013). While there was an increase in niche breadth throughout the whole ecosystem, four species (red cod, tarakihi, spiny dogfish and ling) were responsible for the increases between the periods analyzed, which indicate the specific impact of fisheries activities (Mancinelli & Vizzini, 163

2015). Although inner shelf assemblages had little change between the periods analyzed, trophic parameters presented a linear correlation with MTI through time. I speculate that the trophic structure of inner shelf species has been relatively stable in the past decades, evidenced by comparisons between years, while variation in trophic parameters correlated to MTI indicates a much earlier impact of fisheries in that community.

Changes in primary production driven by climate variability could have affected the marine food webs, with consequences to fish abundance and fisheries yields (Brown et al., 2010). The impact of these changes on the commercial fish communities is complex and depends on intra- and interspecific interactions, resonating to the entire food web (Casini et al., 2008). For example, climate forcing has been demonstrated to affect recruitment and abundance of young of the year of important commercial fish species in New Zealand, including hoki (Bull & Livingston, 2001) and red cod (Beentjes & Renwick, 2001). Negative SOI and colder water temperatures were identified as better conditions for the recruitment of both species and were likely related to increased pelagic productivity and food availability.

There is a prolific body of literature describing the effects of climate change and fisheries on marine communities worldwide (Crowder et al., 2008; Doney et al., 2012; Harley et al., 2006; Ward & Myers, 2005). These relationships not only indicate that fish communities have endured large shifts in species composition, distribution, size frequency and trophic relationships over the past but are also likely to be affected by these stressors in the future. I have demonstrated in the present study that the trophic structure of different exploited fish assemblages have been affected since the expansion of New Zealand industrialized fisheries and had different continuous relationships with fishery activities, climate forcing and shifts in prey abundance. Trophic structure is an important aspect of a fish’s niche and a key concept that needs to be understood before applying ecosystem approaches to fisheries management (Crowder et al., 2008; Thrush & Dayton, 2010). Application of ecosystem-based approaches requires a shift from stock-population to community-habitat management unit, focused on species interactions and overall productivity of specific habitats (Mangel & Levin, 2005). Productivity and trophic structure can be affected by both climate and anthropogenic stressors, so the study of their individual and synergistic effects on exploited communities is necessary for the proper management of the seascape and these resources (Hare, 2014). While we can manage the impacts of fishing on those communities, the same is not true for climate-driven and natural stressors. While sea surface temperature is increasing in the study region (Shears & Bowen, 2017), Munida sightings have been decreasing with time (Figure 5.8). Although we still do not know the

164 future effects of these changes in the ecosystem, the present study is the first to demonstrate how these shifts may alter the trophic structure of important fisheries resources in New Zealand, and how they differ between different habitats. While more investigations are necessary to uncover the specific mechanisms of those changes, current ecological knowledge must be applied to the management of these communities with an ecosystem-based approach.

165

Chapter 6 Long-term shifts in the structure and length-frequency of fish communities and the relationship with temperature, fisheries and life history

Abstract

Fish communities, especially species inhabiting the continental shelf, have been exploited for centuries in most marine systems. The composition and size structure of these communities can be affected by both fisheries and environmental influences, but long-term studies with fisheries-independent data are still rare. In the present study, data from research bottom trawl surveys conducted between 1991 and 2018 were analyzed to identify long-term changes in species composition and size distributions of demersal fish species. The community analyzed inhabits the continental shelf and upper slope of the east coast of the South Island of New Zealand. Analysis of variance, hierarchical clustering and MDS plots demonstrated that community composition has changed over the years, with a tendency towards larger overlap among species in space in most recent periods. Statistical differences in depth use for the whole community were not found, although, several species have changed their depth distributions over time, mainly related to shifts in abundance at different depths. Total catch rate (kg.km-2) of the whole community combined has increased significantly, while the average trophic level decreased over time for all depth groups analyzed, primarily explained by an increase in the abundance of intermediate trophic level species. In general, data from fishing pressure, temperature, and sediment type did not covary with changes in species abundance. However, fishing pressure presented a significantly higher correlation with the variation in length-frequency of species at intermediate trophic levels, suggesting the impacts of fisheries are heterogeneous throughout the community. The present study demonstrated how the structure of an important fish community has changed through time and how these changes affected species at different trophic levels, an important step towards the multi-species management of fisheries resources. Although the community presented compositional changes, the community can be considered stable over the periods analyzed, with only few species showing signs of long-term decline in abundance.

166

6.1. Introduction

Marine fishes have been exploited for centuries worldwide, but there is still a lack of knowledge on long-term changes in abundance, distribution, and length-frequency of communities from important fishing grounds (Chrysafi & Kuparinen, 2016). Most of the data available on these communities come from commercial fisheries landings data. However, because landing data systematically bias sampling by actively seeking large aggregations, unreporting discards and by presenting differences in fishing power, long-term analysis of these parameters should be undertaken with fisheries-independent data to accurately resolve community changes (Durante et al., 2020a). Accordingly, long-term analysis of abundance trends from fisheries-independent surveys is important to detect the impact of anthropogenic activities on marine biodiversity, especially communities known to be affected by exploitation (Branch et al., 2010).

Fish abundance and community composition can be influenced by recruitment, mortality and migration of individuals, but are also dependent on ocean productivity, prey availability and environmental factors, such as water temperature (Hixon et al., 2012; Mutshinda et al., 2009). In exploited environments, fisheries also play an important role in shaping the abundance of these populations, with potential synergetic effects coupled with ecological and environmental processes (Beukhof et al., 2019; Lima et al., 2020; Verba et al., 2020). For example, although the collapse of sardines in the California Current ecosystem during the 1950s was mainly a result of weak recruitment, exploitation rates were linked to the range and rates of those changes (Lindegren et al., 2013). Further, the dynamics between sardines and anchovies’ populations since 1661 were mainly driven by climate forcing and species-specific life history traits. While fisheries research and stock assessments often focus on single populations, or even parts of a population, long-term changes in the composition of fish communities have been reported worldwide, with shifts in species abundance and dominance within each community (Moyes & Magurran, 2019; Safiq et al., 2018).

When studying the effects of different stressors on the structure of complex aquatic communities, the use of species traits instead of species identity can simplify analysis and facilitate the resolution of generalized impacts on different habitats (Beukhof et al., 2019; Jennings & Polunin, 1997). Species traits such as longevity, sedentary lifestyle and high trophic level can also reflect enhanced vulnerability to environmental and anthropogenic stressors (Henriques et al., 2014). Because of these differences, changes in species composition can affect the distribution of trophic levels within a community (Shannon et al., 2014). These community shifts are usually linked to the selective exploitation of high trophic

167 level species and can alter the energy flux and fisheries yields of specific stocks (Sosa-López et al., 2005; Udy et al., 2019c). Because of this selectivity, trophodynamic indicators, such as mean trophic level of the commercial fisheries catch, have been demonstrated to be a good tool for the evaluation of the impact of fisheries at an ecosystem level (Cury et al., 2005; Kleisner et al., 2014). Nevertheless, it is challenging to identify the drivers of change in the trophic level composition of large communities (Sosa-López et al., 2005).

Although shifts in fish assemblage composition, trophic structure and species traits can be good indicators of environmental and anthropogenic impacts on marine communities (Garrison, 2000), length-frequency variation of a population will likely respond faster to those effects than compositional shifts (Beukhof et al., 2019). Shifts in length-frequency of a population can have a large impact on trophic interactions, since a small decrease in length at age (reduced growth rates resulting in increased numbers of small individuals) can exponentially increase a species’ mortality due to predation (Audzijonyte et al., 2013). Size- based indicators such as mean and maximum length of individuals within populations and communities have been used to resolve the effects of fishing on exploited populations, but no single indicator was found to respond effectively to exploitation and environmental effects (Shin et al., 2005). By using information from complete size distributions within fish populations, Tu et al. (2018) demonstrated that exploitation rates and temperature changes have long-term effects on the size structure of commercial species, although these effects were different for species from contrasting habitats.

The fisheries resources of New Zealand have been exploited commercially since 1860, with an important expansion of fishing grounds after 1970 (Johnson & Haworth, 2004). This was followed by a reduction of both landings and mean trophic level composition of the catch after the full expansion of industrialized fisheries in the year 2000 (Durante et al., 2020a), indicating a change in the trophic architecture of those communities (Pauly et al., 1998). During the development of New Zealand fisheries, exploitation intensity was not homogeneous within its waters, with fishing pressure initially focused on the inshore and demersal communities with a more recent history of deep-sea and offshore activities (Durante et al., 2020a). Because of the selective nature of fisheries and spatial heterogeneity of fishing effort, these activities likely had distinct effects on species with contrasting life history strategies. Further, trawling became the prevalent fishing method (Fisheries New Zealand, 2019), with the potential to modify the structure of fish assemblages over time (Watson et al., 2006). The amount of fisheries research undertaken has also been heterogeneous across marine systems in New Zealand. For example, inshore habitats have had relatively low

168 coverage of fisheries observers and discard/bycatch assessments compared to other deepwater and large volume fisheries, which has increased the predominance of ‘unknown stock status’ assigned to many inshore species (Fisheries New Zealand, 2019; Mace et al., 2014).

Understanding if and how community assemblages of important commercial species have changed over time is an important step towards multispecies management (Mahon et al., 1998). Species abundance, distribution and length-frequency data are routinely collected during research trawl surveys in New Zealand, and provide a reliable data source for long- term studies on the structure of fish communities (MacGibbon et al., 2019; O’Driscoll et al., 2018; Stevenson & MacGibbon, 2018). Regions surveyed include the historically important east coast of the South Island, which includes two of the largest fisheries ports in New Zealand (Lyttelton and Timaru). The region is important for commercial and recreational fisheries and is managed as a single fisheries management area (FMA 3) for most exploited species (Fisheries New Zealand, 2019). The continental shelf along the east coast is heterogeneous, comprised of rocky reefs in the south (Otago, ~ 46˚ S) and north (Kaikoura, ~ 42˚S), and a sandy gently sloping bottom in between, extending offshore along the Canterbury Bight (Figure 6.1). This geomorphology facilitated the development of an extensive inshore trawl fishery, resulting in one of the most intensively trawled areas in New Zealand waters since the expansion of industrialized fishing, measured on Catch Effort Landing Returns (CELRs) (Baird et al., 2011). Seawater temperature has also been demonstrated to vary greatly along the region, both at short (seasonal and yearly, see Chapter 2) and long (decades) time scales (Hopkins et al., 2010; Shears & Bowen, 2017), but the long- term effects of exploitation and temperature on the composition and size frequency of the fish community have not been previously studied. Nevertheless, Beentjes et al., (2002) has demonstrated the existence of different fish assemblages with contrasting depth distribution between winter and summer surveys in the region, but temporal investigations were not made in that study.

The National Institute of Water and Atmospheric Research of New Zealand (NIWA) has published several reports with the results of Fisheries New Zealand trawl surveys where they have analyzed the main trends in the abundance, length-frequency and distribution of the commercial fish species along the east coast of the South Island (MacGibbon et al., 2019). Instead of analyzing the shifts in abundance and size composition of the commercial fish community between years, the aim of the present study was to investigate long-term variations of these parameters. Here surveys were pooled in year groups to reduce between- year variability and to answer the following questions: (1) has the exploited fish assemblage

169 composition and size structure changed between periods? (2) Has the composition of trophic level within the exploited community changed? (3) How much of this variation could be explained by fisheries and environmental variables? (4) Did species with contrasting life history, ecological traits and habitat use have different relationships with fisheries and environmental variables?

Figure 6.1. Map with the location of trawling stations (30–400 m) inside the Canterbury Bight used in the present study divided by time periods. White represents land, while shades of grey represent depth of 0–100, 100–200 200–400 and deeper than 400 meters

170

6.2. Methods

6.2.1. Data collection

Twelve bottom trawl surveys of the Canterbury Bight and Pegasus Bay were undertaken by NIWA on behalf of Fisheries New Zealand between 1991 and 2018 (MacGibbon et al., 2019). The trawl surveys are also known as the east coast South Island trawl survey time series. The main objective of the surveys was to determine the relative distribution and abundance of important commercial species of fish, including red cod (Pseudophycis bachus, Moridae), giant stargazer (Kathetostoma giganteum, Uranoscopidae), sea perch (Helicolenus percoides, Sebastidae), tarakihi (Nemadactylus macropterus, Cheilodactylidae), spiny dogfish (Squalus acanthias, Squalidae), elephant fish (Callorhinchus milii, Callorhinchidae), red gurnard (Chelidonichthys kumu, Triglidae) and dark ghost shark (Hydrolagus novaezelandiae, Chimaeridae). All surveys were conducted from the R.V. Kaharoa, using standardized bottom trawl nets with a headline height of 4 to 5 m. One-hour tows were conducted during daylight hours at a speed of 3 knots over the ground. The catch from each tow was sorted by species and weighed to the nearest 0.1 kg. Relative abundance in the form of catch rates (kg.km-2) was then calculated for each species based on the area swept by the trawl doors and distance towed. Length, sex and individual weights were recorded for Quota Management System (QMS) species as well as other selected species (Fisheries New Zealand, 2019), either for the whole catch or, for larger catches, from a sub- sample of about 100 randomly selected fish. The twelve surveys were conducted in 1991– 1994, 1996, 2007–2009, 2012, 2014, 2016 and 2018 during late autumn-winter (April–June) and used a two-phase random stratified survey design (Francis, 1989). Between 1996 and 2000 five surveys were undertaken during summer (December–January) but were discontinued due to extreme fluctuations in catchability between surveys (Beentjes & Stevenson, 2001; Francis et al., 2001). The survey area of approximately 23 000 km2 was initially stratified by depth ranges and these were arbitrarily divided into nine smaller strata along the coast (1991–1993), and into 17 strata starting in 1994 (MacGibbon et al., 2019). Strata depth categories represented the inshore (30–100 m), shelf edge (100–200 m) and continental slope (200–400 m). Simulations were carried out using catch data from previous surveys to determine the allocation of trawl stations within each stratum, with the aim of achieving specified coefficients of variation. Trawl stations were then randomly allocated inside each stratum with a minimum separation distance of 5 km and a minimum of three tows per stratum. Surveys after 2007 included an extra depth range, between 10 and 30

171 meters, to improve the sampling of coastal species such as elephant fish and red gurnard, but these data are not part of the present study.

Relative abundance data (catch rates - kg.km-2), of 32 fish (eight cartilaginous fishes and 24 teleosts) and one squid species, as well as length frequency of 25 fish species (seven cartilaginous fishes and 18 teleosts) were analyzed from these twelve surveys. Length frequency data of rattails (Macrouridae), crested bellowfish (Notopogon lilliei, Centriscidae), javelinfish (Lepidorhynchus denticulatus, Macrouridae), silverside (Argentina elongate, Argentinidae), carpet shark (Cephaloscyllium isabellum, Scyliorhinidae), witch (Arnoglossus scapha, Bothidae) and scaly gurnard (Lepidotrigla brachyoptera, Triglidae) were not analyzed.

6.2.2. Data processing

Data acquired from the east coast South Island trawl surveys were extracted from the Fisheries New Zealand trawl database and processed to ensure the accurate comparison between years and regions of interest. Because differences in community composition and spatial distribution were found between summer and winter surveys (Beentjes et al., 2002), only surveys that were conducted during winter were analyzed in the present study. Similarly, data from stations shallower than 30 meters were excluded from the present study since the shallow strata (10–30 m) were not included in surveys before 2007. To avoid analyzing catch rates influenced by poor gear performance, only tows with gear performance of 1 and 2, considered satisfactory for abundance estimates, were kept for analysis (Beentjes & Stevenson, 2001; MacGibbon et al., 2019). Due to the lack of environmental data and identification of some species, data from the 1991 survey was not used in the community structure analysis. To account for interannual variability on the data, survey years were aggregated into three discrete year groups (1992–96, 2007–09 and 2012–18), with analysis undertaken between groups and not between single years. 1992–96 period also included length-frequency data from 1991.

6.2.3. Species assemblage composition and size structure

To investigate changes in species composition between year groups, resemblance matrices of species catch rates were calculated using a Bray-Curtis similarity index. Bray- Curtis similarity is suitable for analyses of community data, because unlike Euclidian distance 172 it preserves species identity (Bray & Curtis, 1957), giving more weighing to abundant than rare species. Catch rates were fourth root transformed before analysis to avoid similarity inflation due to differences in natural abundance between species. A “dummy” species present in every sample with a value of 1 was also included before resemblance calculation, which improves Bray Curtis similarity comparison for samples with few or zero individuals (Clarke et al., 2006). Changes in species associations between periods were investigated with hierarchical cluster analysis and non-metric multi-dimensional scaling (MDS) plots, using between-species Bray-Curtis similarity, which grouped species with similar occurrences and abundances in each period. Species can be visually identified and grouped in both hierarchical clusters and MDS plots (Beentjes et al., 2002), enabling this study to investigate changes in species associations between periods.

Resemblance matrices of Bray-Curtis similarity between tows were analyzed using pairwise permutational analysis of variance (PERMANOVA) to investigate changes in the community structure between periods. PERMANOVA is a nonlinear, non-normal analysis of variance that tests between and within group similarities in the multivariate space, using permutation to compute statistical tests, instead of statistical tables (Anderson, 2001). Because statistics calculated during PERMANOVA take into account the distances of the multivariate cloud within and between the group centroid, it is sensitive to both location and dispersion effects (Clarke & Gorley, 2006). A test of homogeneity of dispersions (PERMDISP) was then applied, which can help to explain significant PERMANOVA results that are not linked to multivariate location effects. PERMDISP can help to identify between- period differences due to species abundance variability in the multivariate space, which could indicate a community under stress (Chapman et al., 1995; Warwick & Clarke, 1993). PERMANOVA and PERMDISP were run for the whole dataset, as well as for each depth group, which represents the survey division of strata (30–100, 100–200 and 200–400 meters). To elucidate changes in community composition, differences in the average catch rates of each species between periods were investigated with Student’s t-test for each depth group separately.

Because assemblage composition is usually well described by depth in the study region (Beentjes et al., 2002), preferred depth for each species was calculated and compared between periods. A locally estimated scatterplot smoothing regression (LOESS) was fitted between catch rates and station depths for each species in each period, while 5% and 95% of the cumulative distribution of catch rates were set as the deeper and shallower distribution of each species between the depths sampled (depth range use), preferred depths were

173 characterized by 50% of their cumulative abundance. LOESS is a non-parametric approach that fits small models to subsets of the data, being suitable for complex and non-linear analysis (Cleveland et al., 1992).

Length frequency data were scaled by the percentage of the catch sampled, area swept by the trawl doors, and stratum area using the NIWA program SURVCALC (Francis & Fu, 2012). This was output as the sum of individuals per centimeter size class by strata and overall for all strata combined, resulting in the estimated population length-frequency distributions of each species in a given survey year (MacGibbon et al., 2019). The scaled number of individuals per centimeter size class for each survey was then averaged within each period per species. Differences in the average length of the whole community between periods were analyzed with an ANOVA, while the distribution of size classes was visually compared for each species and for all species combined.

6.2.4. Trophic level composition of the fish community

To investigate temporal changes in the trophic structure of the community, a weighted average trophic level of each tow was calculated and plotted for the three depth groups. The average trophic level of the catch was then compared between year groups at each depth category using a Tukey-Kramer post hoc test. Trophic level estimates were retrieved from Fishbase (Fishbase.org). Trophic level is known to increase with fish length, because larger fish can feed on larger, higher trophic level prey (Dalponti et al., 2018), but can also show changes with time (Fanelli & Cartes, 2010). Since length measurements were not taken from every species, a static value of trophic level was assigned for each species independent of size or potential temporal changes in species trophic level.

6.2.5. The effect of fisheries and environmental variables

Because the datasets used in the community composition and length-frequency analysis were spatially explicit and non-explicit, respectively, different datasets were used to investigate the relationship of fisheries and environmental variables to changes in the species composition and size structure of individual species in the community. After depth and latitude, bottom temperature and sediment were the two environmental variables known to influence the composition of fish assemblages in the region (Beentjes et al., 2002). Therefore, bottom temperature recorded from every tow and sediment data were used to describe the 174 abundance variability of each species. While the temperature was recorded with a CTD (an oceanographic instrument that records conductivity, temperature and depth) attached to the trawl gear, sediment data were obtained from NIWA maps, which describe the distribution of sand, mud, gravel, as well as sediments comprising more than 50% of calcium carbonate (Leathwick et al., 2012). For the few tows missing temperature values, an expectation maximum likelihood algorithm was used to estimate a value for that location, assuming a multi-normal distribution model for the data (Clarke & Gorley, 2006). Because the bottom temperature is correlated with depth, bottom temperature anomaly was used in the analysis of community composition. A LOESS was fitted between bottom temperature and depth from every tow from 1992 to 2018, the equation was then used to calculate the anomaly of temperature recorded in each station for that specific depth (Beentjes et al., 2002). To investigate the effects of commercial fishing, the distribution of average commercial catches (kg.ha) between 2007 and 2017, grouped into ten catch intensity categories, was used as an index of fishing intensity at every tow location (Fisheries New Zealand unpublished data, Appendix A.6.1). Catch values in this dataset are likely to be a function of fish abundance, and were used here as a proxy of fish removal due to fisheries activities, being the only high resolution, spatially explicit dataset of fisheries activities in the region (see Appendix A.6.2). The 1993 survey was excluded from the environmental analysis due to the high number of missing temperature values.

The yearly average surface temperature recorded at the University of Otago Portobello Marine Lab (PML) was used as an independent variable in the length-frequency analysis. Surface temperature recorded at PML is a localized environmental proxy, but has been shown to follow the overall fluctuation in the temperature of boundary currents in the region, associated with wind regimes (Shears & Bowen, 2017). Total reported commercial landings of each species in this region, reported by fisheries statistical area (018, 020, 022 and 024)(Fisheries New Zealand, 2019), were provided by Fisheries New Zealand and used as a general proxy for fish removal from the population between different years in the length- frequency analysis. These data corresponded to the total estimated catch for five and eight most abundant species caught, depending on data type: Catch Effort Landing Returns (CELR, 5), Trawl Catch Effort Return (TCER, 8), Lining Catch Effort Return (LCER, 8), Net Catch Effort Landing Returns (NCELR, 8), Trawl Catch Effort and Processing Return (TCEPR, 5) and Lining Trip Catch Effort Return (LTCER, 8). Catches were combined for each fishery general statistical area along the east coast of the South Island (see Chapter 4, Figure 4.1). Due to confidentiality, only data from an area in which three or more fishers reported catch were combined. Analogous to the catch intensity map based on kg.ha (Appendix A.6.1), 175 landings reported by statistical area are also a function of species abundance in this region, not only fishing effort (see above paragraph). Because fisheries data from 2018 were not available at the time of the analyses, data from 2017 were also used in the analysis of 2018 length-frequency data. This approximation is justified by the small changes in both total allowable commercial catch and the total catch in the region between 2017 and 2018 (Fisheries New Zealand, 2019).

Distance based linear models (DISTLM) were used to estimate the contribution of each environmental variable and fisheries data to the variation in the catch rate of each species separately. DISTLM is a statistical tool that tests the explanatory power and significance of the community resemblance matrices by independent variables (Mcardle & Anderson, 2001). The adjusted r2 (adjusted for sample size effect) of each model was compared to investigate the contributions of different variables to the catch rate variability of each species (Peres-Neto et al., 2006). Similarly, redundancy analysis was used to investigate the effect of exploitation and temperature on the length-frequency variation of each species. In this analysis, data matrices of species length composition across years were used as the dependent variable, while sea surface temperature and commercial fishing landings were used as explanatory variables. Length composition and year-class strength of some fish species in the region have been shown to respond to environmental variables with temporal lags sometimes larger than one year (Beentjes & Renwick, 2001; Bull & Livingston, 2001). Therefore, 0, 1- and 2- years lag in the temperature data was used in the redundancy analysis, with the explanatory fraction from fishing, temperature and its interaction to changes in the length-frequency of each species calculated, after Tu et al., (2018). The models with the largest total adjusted r2 explaining the most variability in the length-frequency data were then compared between species groups with contrasting life history and ecological traits.

6.2.6. The effect of species life history traits

Differences in depth use and catch rates between the earlier (1992–96) and later periods (2012–18) for species in different trophic level groups (smaller than 3.5, 3.5 to 4 and larger than 4) were investigated with Wilcoxon’s each pair comparison tests. Differences in the average and distribution of explanation power of fisheries and environmental variables to changes in catch rates and length-frequency were tested between species with different life history traits with Student’s t and Wilcoxon’s test. Species were combined in three groups according to the value distribution of each trait, resulting in the same number of species in

176 each group. Traits analyzed in the present study included species trophic level, maximum total length and age, as well as mean preferred temperature and its range (Kaschner et al., 2016). For consistency, all life history traits were retrieved from Fishbase (Appendix A.6.3, Fishbase.org). When only parametric tests were used to compare groups, assumptions were checked by histogram visualization, Shapiro Wilk’s and Levene’s tests. Statistical analyses were undertaken with R 3.6.3 (R Core Team, 2020), JMP 14.0.0 (SAS Institute, 2018) and PRIMER 6.1.12 (Clarke & Gorley, 2006).

177

6.3. Results

6.3.1. Species assemblage composition and size structure

A total of 1011 research survey tows were analyzed, 372 from 1992–96, 276 from 2007–09 and 363 from 2012–18. Pooling and analyzing the data into three time periods reduced the variability in between year differences and allowed the present study to look at long-term trends of catch rates, preferred depth and average total length. Total catch rates of the whole community combined had a significant increase among periods, with average ± SD of 990 ± 842, 13 356 ± 977 and 15 250 ± 852 kg.km-2, for 1992–96, 2007–09 and 2012–18, respectively (p < 0.0001, Wilcoxon’s test). Averages of all variables were calculated for each species for each period analyzed (Table 6.1). Dendrograms from the cluster analysis showed different relationships among species between time periods. While seven assemblages were identified in 1992–96 and 2007–09, only five could be visualized from the 2012–18 dendrogram (Figures 6.2 and 6.3). In general, these groups can be separated into shallow (B, D, F and G in 1992–96, A and G in 2007–09 and A in 2012–18), intermediate (E in 1992–96, E and F in 2007–09 and D in 2012–18) and deepwater species (A and C in 1992–96, B in 2007–09 and C in 2012–18), as well as a group with jack mackerels (C in 2007–09 and B in 2012–18) and groups with blue cod and crested bellow fish (D in 2007–09 and E in 2012–18) (Figure 6.2). While the dendrograms showed shifts in species associations with time, the largest difference was the fewer number of clusters in 2012–18 data than in the previous periods (Figure 6.2). Except for blue cod (Parapercis colias, Pinguipedidae), silverside, crested bellowfish, hoki (Macruronus novaezelandiae, Merlucciidae), javelinfish, rattails, ling (Genypterus blacodes, Ophidiidae), jack mackerels (Trachurus spp, Carangidae), leatherjacket (Meuschenia scaber, Monacanthidae), rig (Mustelus lenticulatus, Triakidae), common warehou (Seriolella brama, Centrolophidae) and New Zealand sole (Peltorhamphus novaezeelandiae, Rhombosoleidae), the other 19 species could be classified as one single fish assemblage in 2012–18. MDS plots were in agreement with the pattern observed in the dendrograms, except for the assemblage comprised of blue cod, silverside and crested bellowfish in 2012‒18, which would be better classified as species with no assemblage, like in the 1992‒96 and 2007‒09 periods (Figure 6.3).

178

Table 6.1. Summary of frequency of occurrence (FO, %), average catch rate (CR, kg.km-2), preferred depth (PD, meters) and average length (AL, cm) for each species in each one of the three periods. AL from 1992-1996 also included values taken in 1991

1992-1996 2007-2009 2012-2018 Species code Common name FO CR PD AL FO CR PD AL FO CR PD AL BAR Barracouta 88 480.3 148 53 88 1006.8 95 52 91 1233.1 87 52 NMP Tarakihi 60 68.5 184 22 57 82.2 79 22 63 93.5 67 23 GIZ Giant stargazer 79 24.0 180 38 79 29.2 129 39 77 30.6 191 37 GUR Gurnard 28 13.8 49 36 52 53.6 42 34 60 61.2 42 33 SPD Spiny dogfish 98 904.1 154 60 100 1245.5 178 55 98 919.3 247 51 RAT Rattails 34 132.9 179 - 26 52.3 228 - 23 139.9 194 - GSH Ghost shark 35 159.7 301 46 51 247.6 271 43 44 611.4 288 49 SPE Sea perch 66 107.5 104 24 65 100.3 154 24 63 171.4 117 25 ELE Elephant fish 31 20.1 60 51 41 37.8 61 43 37 167.5 62 55 RCO Red cod 79 383.1 195 39 63 149.0 235 40 58 179.7 195 36 LIN Ling 66 35.5 329 60 46 19.2 339 52 38 19.3 347 58 JMM Slender jack mackerel 45 31.4 125 48 9 2.5 110 50 7 0.7 192 47 JMD Greenback jack mackerel 30 6.1 37 44 18 2.1 106 45 22 2.6 140 40 JMN Yellowtail jack mackerel 10 5.0 69 43 3 0.2 41 37 1 0.0 65 40 BCO Blue cod 13 4.7 84 35 21 4.0 108 33 24 9.1 86 31 CBE Crested bellowsfish 13 19.7 133 - 20 14.9 146 - 28 77.2 130 - HAP Hapuku 30 4.8 194 58 41 7.7 180 57 35 9.1 195 60 LSO Lemon sole 41 3.3 130 27 39 3.7 67 26 50 3.8 217 26 SPO Rig 23 2.2 102 66 21 6.7 76 74 23 5.7 68 66 JAV Javelin fish 4 1.1 367 - 7 6.7 362 - 7 12.1 319 - ESO New Zealand sole 5 0.3 53 27 8 0.2 43 25 5 0.3 39 23 SSI Silverside 6 0.2 236 - 21 1.1 224 - 35 2.3 318 - HOK Hoki 11 17.7 362 59 9 38.0 323 52 11 18.0 272 44 WAR Common warehou 15 7.1 106 30 21 32.9 123 31 14 4.2 219 24 LEA Leatherjacket 5 1.0 77 25 11 4.9 37 21 15 11.3 34 23 CAR Carpet shark 54 28.8 124 - 84 46.1 129 - 84 49.7 121 - SSK Smooth skate 39 19.6 164 93 53 30.3 198 46 48 33.0 190 58 RSK Rough skate 35 14.1 201 52 50 34.3 87 48 58 41.6 129 49 NOS Arrow squid 93 72.7 283 26 91 60.6 273 - 89 65.7 323 19 SWA Silver warehou 45 11.2 283 24 57 27.7 259 26 48 20.2 230 25 WIT Witch 75 20.2 235 - 79 25.9 164 - 85 54.9 218 - SCH School shark 36 7.9 103 72 42 14.9 79 74 52 15.7 146 69 SCG Scaly gurnard 49 7.8 123 - 68 39.2 124 - 69 27.8 121 -

Significant differences in the community structure between periods were indicated from the PERMANOVA analyses, but not from the PERMDISP results (Table 6.2). The same results were obtained when analyses were undertaken for each depth category separately, but PERMDISP analysis only showed significant differences between 1992–96 and 2007–09 for the 100–200 m and between 1992–96 and 2012–18 for the 30–100 m depth category (Table 6.2). The average distance to centroid in the Euclidean space (Bray-Curtis similarity data) of the significantly different PERMDISP comparisons indicate a reduction in variability of catch rates with time for both 30–100 m (31.5 ± 0.45 to 29.9 ± 0.45, average ± SE) and 100–200 m categories (30.6 ± 0.6 to 27.6 ± 0.6).

179

similarity

Curtis

node average of Bray average node

- Between

Figure 6.2. Dendrograms from hierarchical cluster analysis of species catch rates in the whole study region for the different year periods of 1992-1996, 2007-2009 and 2012-2018. Clusters were based on Bray-Curtis similarities of catch rates (kg.km-2) between species. Species codes can be found in Table 6.1 and Appendix A.6.3

180

Figure 6.3. Non-metric multi-dimensional scaling plots of Bray-Curtis similarity resemblance matrices between species for each period analyzed. Groups identified in Figure 6.2 are depicted inside shapes with their respective letter names. Species codes can be found in Table 6.1 and Appendix A.6.3

181

Table 6.2. Results from between-periods pairwise PERMANOVA and PERMDISP analysis of the community composition for the whole region and at different depth categories

PERMANOVA PERMDISP Depth category Pairwise comparison average similarity t-values All 1992-1996 - 2007-2009 48.98*** 0.82 1992-1996 - 2012-2018 48.18*** 0.94 2007-2009 - 2012-2018 50.92** 0.08 30–100 1992-1996 - 2007-2009 54.35*** 1.30 1992-1996 - 2012-2018 53.73*** 2.47* 2007-2009 - 2012-2018 56.97** 1.03 100–200 1992-1996 - 2007-2009 55.08*** 3.33** 1992-1996 - 2012-2018 53.07*** 1.75 2007-2009 - 2012-2018 59.24** 1.96 200–400 1992-1996 - 2007-2009 56.1** 0.42 1992-1996 - 2012-2018 55.50*** 1.47 2007-2009 - 2012-2018 57.68** 0.98 * indicates 0.05 (*), 0.005 (**) and 0.0005 (***) levels of significance

Preferred depth (161.7 ± 91.6, 147.4 ± 90.7 and 164.3 ± 89.4 m, average ± SD) and depth range (170.1 ± 117, 156.5 ± 111.8 and 187.4 ± 98.1 m) of the whole community were not significantly different between the three time periods (1992–96, 2007–09 and 2012–18) at p < 0.05 (Wilcoxon’s test), respectively. On the other hand, several fishes had changes in preferred depth between the periods analyzed, but because of the nature of the data (LOESS regression estimates), they were not statistically tested. While greenback jack mackerel (Trachurus declivis), school shark (Galeorhinus galeus, Triakidae), common warehou, slender jack mackerel (Trachurus murphyi) and spiny dogfish showed an increase in occurrence in deeper habitats, New Zealand sole, leatherjacket, rig, barracouta (Thyrsites atun, Gempylidae) and tarakihi displayed a shift to shallower waters (Figure 6.4). Species showed a wide range in minimum and maximum depth of occurrence, which varied among periods. Greenback jack mackerel, school shark, common warehou, crested bellowfish, silverside and hoki had expanded their depth range, while leatherjacket and tarakihi showed a reduced depth range between periods within the region (Figure 6.4).

182

Figure 6.4. Range and preferred depth (horizontal line) values between periods analyzed for 15 fish species with the largest differences between periods. Crosses mark the average between minimum (5% of cumulative catch), maximum (95% of cumulative catch) and preferred depth (50% of cumulative catch). Based on catch rates (kg.km-2)

183

To investigate if changes in preferred depth were related to shifts in the distribution or in the relative abundance of each species at depth, differences in average catch rates between periods were tested for each species at each depth group. Common warehou, New Zealand sole, crested bellowfish and hoki had no significant increase in catch rates between periods for all depths analyzed. All jack mackerel species had significant declines in catch rates with time, in the 30–100 m (greenback, slender and yellowtail jack mackerels), 100–200 m (slender and yellowtail jack mackerels) and 200–400 m categories (slender jack mackerel) (Table 6.3). In contrast, school shark and tarakihi displayed an increase in catch rates in the 100–200 m depth category, with no long-term changes in other depths. Catch rates at 30–100 m depth for rig, leatherjacket, barracouta and silverside showed an increase between the periods analyzed, with barracouta and silverside also showing significant increases in relative abundance at 100–200 m and 200–400 m (silverside only). Spiny dogfish displayed an increase in relative abundance between 1992–96 and 2007–09 at 30–100 m but had a significant decrease in catch rates at 100–200 m and an increase at 200–400 m (Table 6.3). Red cod relative abundance was estimated to have decreased by 10-fold at 30–100 m between the periods analyzed, from an average of 460 kg.km-2 in 1992–96 to 45 kg.km-2 in 2012–18. Both skate species, rough (Zearaja nasuta, Rajidae) and smooth skate (Dipturus innominatus, Rajidae) showed an increase in relative abundance between periods at 30–100 m (rough skate only), 100–200 m and 200–400 m (smooth skate only). A similar pattern was also displayed by dark ghost shark, with a five thousand percent and a 4-fold increase in relative abundance between periods at 100–200 m and 200–400 m, respectively (Table 6.3).

The average total length of the whole fish community (25 species combined) did not vary significantly between periods, with averages ± standard deviation of 45.3 ± 17.6, 42 ± 14.65 and 41.73 ± 14.72 cm, for 1992-06, 2007–09 and 2012–18, respectively (ANOVA, df (2,74) = 0.4, p = 0.69). Most species analyzed had similar size distributions among periods, with only changes in relative abundance and hence population numbers. In contrast, elephant fish showed an increase in abundance of specimens smaller than 45 cm in 2007–09, while individuals over 55 cm dominated the catches in 2012–18 (Figure 6.5). The increase in the dominance of larger individuals in 2012–18 was also seen for ghost shark (Figure 6.5), while hoki and common warehou had a decrease in abundance of individuals larger than 40 and 30 cm in 2012–18, respectively, compared to larger numbers recorded in 2007–09 (not shown).

184

Table 6.3. Average catch rate (kg.km-2) ± SE for each species at specific depth groups (30–100, 100– 200 and 200–400 m) for each period analyzed (1992–96, 2007–09 and 2012–18). Periods separated by different letters represent significant different average catch rates within the same depth category (Student’s t-test)

Species Depth group 92-96 07-09 12-18 Leatherjacket 30‒100 2 ± 4 A*** 10 ± 4 A* 22 ± 4 B Gurnard 28 ± 13 A 103 ± 15 B*** 116 ± 13 B*** Slender jack mackerel 10 ± 3 A 0.4 ± 4 AB 0.1 ± 3 B* Rig 4 ± 2 A 12 ± 2 B* 10 ± 2 B* Scaly gurnard¹ 4 ± 2 A 22 ± 3 B*** 22 ± 2 B*** School shark 10 ± 2 A** 21 ± 2 B 12 ± 2 A* Yellowtail jack mackerel 20 ± 5 A 0.4 ± 6 B* 1 ± 5 B* Greenback jack mackerel 10 ± 2 A 2 ± 3 B* 3 ± 2 B* Silverside¹ <0.1 ± 0.1 A 0.1 ± 0.1 AB 0.4 ± 0.1 B* Rough skate 24 ± 7 A 60 ± 8 B** 66 ± 7 B*** Barracouta 600 ± 159 A 1406 ± 183 B*** 1725 ± 159 B*** Silver warehou 1 ± 1 A 3 ± 1 B* 3 ± 1 B** Red cod 460 ± 105 A 69 ± 121 B* 45 ± 105 B* Spiny dogfish 868 ± 212 A 1531 ± 245 B* 1106 ± 212 AB Gurnard 100‒200 0.1 ± 1 A 2 ± 1 AB 4 ± 1 B** Elephant fish 1 ± 7 A 6 ± 8 AB 22 ± 7 B* Slender jack mackerel 1 ± 0.2 A <0.1 ± 0.3 AB <0.1 ± 0.2 B* Blue cod 1.4 ± 2 A 7 ± 2 AB 9 ± 2 B* Scaly gurnard 18 ± 11 A 84 ± 12 B*** 48 ± 11 C* School shark 9 ± 5 A 13 ± 6 AB 27 ± 5 B* Tarakihi 18 ± 29 A 40 ± 34 AB 110 ± 29 B* Yellowtail jack mackerel 62 ± 16 A 7 ± 19 B* 1 ± 16 B* Silverside¹ 0.3 ± 0.4 A 2 ± 1 B* 4 ± 0.4 C** Giant Stargazer 31 ± 5 A 50 ± 5 B* 41 ± 5 AB Smooth skate 14 ± 7 A 45 ± 8 B** 40 ± 7 B* Rough skate 6 ± 3 A** 9 ± 4 A** 20 ± 3 B Barracouta 510 ± 103 A 851 ± 119 B* 974 ± 102 B** Witch 25 ± 17 A*** 57 ± 20 A* 112 ± 17 B Spiny dogfish 1268 ± 200 A 973 ± 232 AB 484 ± 199 B* Sea perch 177 ± 73 A 224 ± 85 AB 391 ± 73 B* Hapuka 5 ± 2 A 9 ± 2 AB 12 ± 2 B** Ghost shark 18 ± 110 A*** 393 ± 128 A** 893 ± 109 B Yellowtail jack mackerel 200‒400 4 ± 1 A <0.1 ± 2 B* 0.1 ± 2 B** Lemon sole 1 ± 1 A* 0.4 ± 1 A* 4 ± 1 B Silverside¹ 0.3 ± 1 A 3 ± 1 AB 5 ± 1 B*** Smooth skate 9 ± 7 A 25 ± 9 AB 32 ± 8 B* Rough skate 1 ± 1 A 4 ± 1 AB 5 ± 1 B* Silver warehou 54 ± 28 AB 140 ± 33 A 43 ± 31 B* Witch 39 ± 12 A* 13 ± 14 A*** 86 ± 13 B Spiny dogfish 275 ± 235 A 892 ± 280 AB 1305 ± 259 B** Ghost shark 578 ± 264 A*** 689 ± 314 A** 2149 ± 291 B Hoki 107 ± 36 A 244 ± 43 B* 130 ± 40 AB Javelinfish 7 ± 18 A 43 ± 21 AB 88 ± 19 B** ¹Significance in B calculated between A and B or in C calculated between B and C * indicates 0.05 (*), 0.005 (**) and 0.0005 (***) levels of significance

185

Figure 6.5. Average scaled population numbers of elephant fish (a) and dark ghost shark (b) for the three time periods analyzed, 1991-1996, 2007-2009 and 2012-2018

6.3.2. Trophic level composition of the fish community

The average trophic level of the community caught in 1992–96 was higher compared to 2012–18 for all the depths analyzed. Turkey-Kramer tests showed significant differences in the average trophic level of the community among all periods in the 100–200 m depth

186 category, between 1992–96 and the two most recent periods in the 30–100 m and between 1992–96 and 2012–18 in the 200–400 m category (p < 0.05, Figure 6.6).

Figure 6.6. Weighted average trophic level of each tow between 1992 and 2018 for different depth categories. Size of the dots represent total catch rate in that trawl station for 31 species of fish combined, except rattails, ranging from 11 to 36885 kg.km-2. Lines are linear fits with shaded 95% confidence interval. The averages ± standard error of the weighted average trophic levels for 1992– 96, 2007–09 and 2012–18 were of 4.07 ± 0.01, 3.97 ± 0.01 and 3.93 ± 0.01 for 30–100 m, 4.09 ± 0.01, 3.94 ± 0.02 and 3.88 ± 0.01 for 100–200 m and 4.02 ± 0.03, 4 ± 0.04 and 3.91 ± 0.03 for 200–400 m, respectively

6.3.3. The effect of fisheries and environmental variables

Fisheries and environmental variables used in the present study did not present large correlations with catch rates as estimated from the adjusted r2. Temperature, sediment type and fishing accounted together for only 0.2 to 7% of the variability in catch rates of the species analyzed (average ± SE of 2.6 ± 0.3%). Temperature and fishing pressure explained most of this variation (0.9 ± 0.1% for both), followed by sediment (0.8 ± 0.1%). In contrast, fishing pressure explained nearly 10% of the variation in the length-frequency of the species analyzed (9.6 ± 0.03%), compared to 8% for temperature (8.2 ± 1.1%) and their interaction (6 ± 1.4%) (Figure 6.7). 187

Figure 6.7. Boxplot of adjusted r2 values from distance based linear models and redundancy analysis of fishing and environmental variables explaining the long-term variability in catch rates (a) and length-frequency (b) of commercial fish species along the east coast of the South Island. Outliers represented as dots. Note the different y-axes scales

6.3.4. The effect of species life history traits and habitat use

When species were combined into trophic level groups, species with trophic level between 3.5 and 4 showed a significantly higher increase in average transformed catch rates in the region between 1992–96 and 2012–18, compared to fishes with a trophic level lower than 3.5 (average ± SE of 0.7 ± 0.18 versus 0.04 ± 0.23, p = 0.045, Wilcoxon’s test) and a trophic level higher than 4 (0.07 ± 0.12, p = 0.015, Wilcoxon’s test). Significant differences in depth use for fishes at different trophic level groups were not found.

188

The explanatory power (adjusted r2) of fisheries, temperature, and sediment on catch rate variability did not vary significantly with species trophic level. On the other hand, when fishes were grouped by preferred temperature (lower than 11.2, 11.2 to 13.5 and higher than 13.5˚C), fishes with an average preferred temperature higher than 13.5˚C had a higher average proportion of their catch rate variability explained by fisheries (average ± SE of 1.3 ± 0.3%) than fishes in the lower preferred temperature group (0.2 ± 0.3%, p = 0.034, Student’s t each pair comparison test) (Figure 6.8), although adjusted r2 distributions were not significantly different among groups (Wilcoxon’s test, p = 0.1). Species grouped according to maximum length (smaller than 55, 55 to 100 and larger than 100 cm), age groups (less than 15, between 15 and 25 and more than 25 years) or range of preferred temperature (lower than 6.5, 6.5 to 8 and higher than 8˚C) did not present significant differences in catch rate variability explained by fisheries, temperature or sediment type.

Figure 6.8. Explanatory power of temperature, sediment and fishing on the variation of relative abundance of species in different mean preferred temperature groups, estimated through the adjusted r2 from distance based linear models

189

Although species with higher trophic level showed a larger average decrease in total length between 1992–96 and 2012–18, differences between groups were not significant (average ± SD of -1.01 ± 2.08 cm for trophic levels lower than 3.5, -1.85 ± 3.54 cm for trophic levels between 3.5 and 4, and -7.32 ± 11.5 cm for trophic levels higher than 4, p = 0.55, Wilcoxon’s test). These differences were mainly due to a large decrease in average length of spiny dogfish, hoki and smooth skate. Length-frequency of species with a trophic level between 3.5 and 4 presented a higher mean adjusted r2 from fishing than lower and higher trophic level species, with an average ± SE of 21.3 ± 4% compared to 3.5 ± 4.2 and 4.8 ± 3.9%, respectively (p < 0.007, Student’s t-test) (Figure 6.9). Significant results were also found for the distribution of adjusted r2 from fishing between trophic level groups of 3.5-4 and lower than 3.5 (p = 0.049). Significant differences between the contribution of temperature, fishing or their interaction and any other life history trait were not found.

Figure 6.9. Explanatory power of fishing, temperature and their interaction on the variation in the length-frequency of species in different trophic level categories, estimated through the adjusted r2 from redundancy analysis

190

6.4. Discussion

The results of the present study demonstrate that length-frequency distributions of individual species and the structure of the fish community off the east coast of the South Island have gone through some changes between the periods analyzed. Overall shifts in depth use and length-frequency were not expressed at the community level, while the observed trends were largely not correlated with fishing or the environmental variables tested.

Cluster analyses from a previous study in the region over the 1992–96 period corroborate results of the present study, with species grouped in assemblages and separated mainly by depth use (Beentjes et al., 2002). In that study, the fish assemblages between 30 and 400 meters of depth were found to have high distributional overlap, which has increased over time based on the present study. While the identification of species associations can be useful information for the management of multispecies fisheries, the lack of structure and stability through time can make both single and multispecies management difficult in the region (Mahon et al., 1998). Many species previously categorized in shallow and deepwater groups were classified in the “intermediate depth” group in 2012–18. PERMANOVA results indicated a significant change in the community structure between periods, for the whole region and at specific depth groups (see Figures 6.2 and 6.3). From the lack of significance from PERMDISP results, the differences observed between periods were due to location changes in the multivariate space. This result can be attributed to changes in absolute catch rates of each species, instead of shifts in the variability of catch rates between tows at each period. In fact, 17 species displayed significant changes in catch rates in at least one depth group (Table 6.3). Conversely, PERMDISP showed a significant decrease in catch rate variability at 30–100 and 100–200 m between 1992-06 and the periods of 2012–18 and 2007– 09, respectively. The decrease in catch rate variability could indicate a more stable and less impacted community with a more homogeneous distribution throughout the seascape (Hsieh et al., 2006; Warwick & Clarke, 1993), which is related to the large increase in the relative abundance of several species in recent periods, such as ghost shark, barracouta and spiny dogfish. Although the ecological consequences of these changes are still unclear, homogenization of marine fish communities has been previously linked to human activities and ocean warming, with consequences to ecosystem function (Magurran et al., 2015; Safiq et al., 2018). Therefore, the results highlight the importance of ongoing, long-term, fisheries- independent surveys to study the composition of exploited communities.

Common warehou, New Zealand sole, crested bellowfish and hoki were the only species with identified changes in depth use that were not related to changes in relative

191 abundance at different depths. While New Zealand sole and hoki were found in shallower habitats, common warehou and crested bellowfish were distributed along a larger range of depth in the most recent periods. The present study was not designed to investigate the mechanisms driving those changes, but because New Zealand sole and hoki inhabit coastal and deep waters (Roberts et al., 2015), these results suggest a gradual expansion and retraction of their distribution, respectively, inside the study region over time.

Trophic level composition of the fish community was observed to be decreasing over time in all depth categories analyzed, indicating that higher trophic level species are becoming relatively less abundant in the region. Although temporal changes in the trophic architecture of fish communities have been demonstrated worldwide (Caddy & Garibaldi, 2000; Durante et al., 2020a; Pauly et al., 1998), this is the first time that fisheries-independent data, which are relatively free of the biases of commercial catch reporting records, have been used to resolve these patterns in New Zealand. Using reported commercial landings data, Durante et al., (2020a) demonstrated that New Zealand exploited fish communities have undergone a decrease in mean trophic level of species composition after the full expansion of New Zealand fisheries around the year 2000 (see Chapter 1). The results from the present study corroborate those findings, indicating that the decrease in the mean trophic level of the fish community is not just a function of the expansion of fishery grounds or market demands, but changes in community composition. By selectively removing large predators through fisheries (Pauly et al., 1998), smaller species could have been released from predation, which in turn could control the abundance of small individuals of the predator species, both through predation and competition, hampering its recovery (Bundy & Fanning, 2005). Ultimately, reduced trophic level composition of the community could lead to an increase in overall abundance and catches of lower trophic level species (Pinnegar et al., 2002), as seen in the present study.

In open systems, such as most of the marine environments, it is hard to pinpoint the specific stressors driving changes in the species composition of communities. Matching temporal events, such as the start of commercial fisheries, is one way to estimate the effects of specific stressors (Ward & Myers, 2005). The first period of sampling in 1992–96 occurred just before the full expansion of New Zealand industrialized fisheries, around the year 2000 (Durante et al., 2020a). The results indicated that the relative abundance of species feeding at intermediate trophic level (3.5 to 4) have increased since 1992–96, and there were no long- term decreases in average catch rates for any of the species analyzed, except for jack mackerels and red cod. Red cod is an important predator and prey species in the region, preying on 28 other fishes including juveniles of high trophic level species such as skates and

192 sharks, and is characterized as the most generalist predator of this community (Graham, 1939). Conversely, ghost shark (100-400 m), spiny dogfish (200–400 m) and barracouta (30- 200 m), which feed mainly on crustaceans, bivalves, salps, echinoderms and a small percentage of teleosts (Hanchet, 1991; Stevens et al., 2011) had large increases in catch rates between periods. Red cod are a short-lived, fast-growing species prone to high mortality and recruitment variably with highly fluctuating interannual abundance (Beentjes & Renwick, 2001). The decrease in abundance of red cod evidenced in the present study could have contributed to reduced predation on mid trophic level species in the region, especially juveniles, resulting in an increased abundance of intermediate trophic level species (Myers et al., 2007). Durant & Hjermann, (2017) have demonstrated that fish populations in high latitudes depend relatively more on age structure for recruitment in warmer conditions, which can be easily impacted by fisheries activities. The average temperature in the study region is known to have increased in the past century (Shears & Bowen, 2017) and this may have affected the recruitment of red cod (Beentjes & Renwick, 2001), contributing to its reduced abundance observed in the present study. The large increase in abundance of spiny dogfish, barracouta and ghost shark over the years evidenced in the present study and by MacGibbon et al., (2019), could also have contributed to the changes in the predator/prey dynamic in the region, impairing the recruitment of other species, such as red cod. Although it is challenging to identify the drivers and mechanisms of the changes observed here, environmental and ecological processes are likely to play important synergetic roles.

NIWA’s trawl survey dataset limits the analyses to the past 28 years. The fish community could have experienced more dramatic changes in composition and abundance in previous years, especially in the period between 1970 and 1992 when industrialized fishing was expanding rapidly across different environments (Durante et al., 2020a). Earlier trawl surveys and historical fisheries data from the east coast of the South Island indicate that fish communities were more heavily exploited during the 1970s and 1980s when compared to the present, especially by international fishing fleets (Fenaughty & Bagley, 1981). Both total biomass and individual mean mass of high trophic level species were found to have decreased to 13 and 50 percent respectively after industrialized fisheries activities in the tropical Pacific Ocean, followed by an increase in abundance of smaller and macropredator species (Ward & Myers, 2005). The predicted changes in community composition off east coast South Island before 1992 would likely be related to the reduction in the abundance of coastal top predators, with more dramatic effects on the community dynamic than the changes reported in the present study (Hill et al., 2020; Thurstan et al., 2010).

193

Data from the temperature, sediment type and fishing pressure did not covary with the catch rates of the fish species, in line with previous investigations (Beentjes et al., 2002). Fishes with higher preferred temperatures had a significantly higher proportion of their abundance explained by fisheries activities, estimated from the adjusted r2 (Figure 6.8). Several species such as gurnard, tarakihi, New Zealand sole, mackerels and rig present high preferred temperature and are heavily targeted by fisheries, explaining the trends observed. These results could also be related to the higher commercial catches of fish in the northern, warmer region of the study area (Appendix A.6.1). The results did not corroborate findings of Perry, (2005) that larger shifts in latitudinal and depth distribution due to temperature occurred for species with reduced maximum length. These differences could be related to a larger number of species analyzed from a larger area (North Sea) with fewer gaps in the time series data in the study by Perry, (2005) compared to the present one.

The average fish length among all species showed a small but statistically non- significant decrease with time, with species at higher trophic levels displaying the largest average changes. Size variation was better explained by fisheries activities, especially for species at intermediate trophic levels. Truncated length distributions, resulting in reduced numbers of large fish, can have a strong impact on the abundance of fish populations. For example, computational models have shown that reduced size-at-age for ling and tarakihi resulted in reduced relative biomass, fisheries catches, and increased predation mortality, with larger effects during scenarios of higher mortality rates due to fisheries (Audzijonyte et al., 2013). This pattern was not observed in the studied community, since most species presented an increased abundance over time, coupled with a non-significant decrease in average length. In the present study, data from fisheries activities showed a higher correlation with the length- frequency variation of species at intermediate trophic levels. With the decline of large predators, lower trophic level species have become the focus of industrialized fisheries worldwide (Jackson, 2001; Pauly et al., 1998; Wing & Wing, 2001), which could explain the patterns observed in the present study.

Fisheries New Zealand provided commercial landings from inside the study region for all species analyzed in the present study. The dataset has an inherently low resolution of catches, but these are the only available data with which to compare commercial catches inside the east coast of the South Island for all species analyzed in the present study, since the size of Fishery Management Areas vary significantly between species (Fisheries New Zealand, 2019).

194

Although previous studies have shown that fishing and temperature effects have different impacts on species from different habitats (Tu et al., 2018), the community analyzed in the present study was mainly comprised of demersal species, therefore, comparisons of the results between species from distinct habitats were not possible. Furthermore, as stated by Tu et al., (2018), the analytical methods used in the present study still assumes linear responses of composition and size structure to the independent variables (fishing and environmental), which is also acknowledged here. Although this might have affected some of the present results, this is an important step in understanding the effects of fishing and environmental variables on the local exploited fish community. Furthermore, although the trophic level of fish species have likely varied between the periods analyzed here (see Chapter 5), and are probably not well represented by values from Fishbase (Pinkerton et al., 2015a), the use of average trophic level was only used as a tool to compare changes in community composition through time.

It was demonstrated in the present study that changes in the community composition and size structure of a commercially important fish community has experienced shifts in the past 28 years. Although these changes were species-specific, they resulted in overall larger catch rates and reduced trophic level composition of the community. Assemblages were observed to be more homogenized in space in the most recent periods, but this might be the result of increased abundance, and therefore larger distributions of several species. Small changes in species average length, as well as a generally stable or increasing catch rates through time for most species, could indicate a current stable community, although compositional and length-frequency shifts before fisheries industrialization (in the 1970s) could have been more dramatic. The explanatory power of fishing pressure and environmental variables was small but varied for species at different trophic levels and those species more heavily exploited. The present analysis is the first to investigate long-term changes in the composition and length-frequency of the fish community in the east coast of the South Island of New Zealand. How these shifts can be species-specific and affect species feeding at different trophic levels was highlighted. Understanding long-term shifts in the composition of important exploited communities is an important step towards the multispecies management of fisheries and the implementation of ecosystem-based management in New Zealand.

195

General discussion

The management of fisheries resources is moving towards an ecosystem-based approach worldwide, but this process has been slow and with few successful case studies (Pitcher et al., 2009; Wing & Jack, 2014). While most managers and researchers agree on the potential benefits of this management approach, practical baseline knowledge for its implementation is still missing in many places (Edwards et al., 2010). Although trophic structure and food web stability are considered important to the resilience of healthy marine ecosystems (Gutiérrez, 2007; Thrush & Dayton, 2010), there is still a lack of long-term studies investigating the variability of trophic ecology in marine communities, including exploited ones. Furthermore, the shifting baseline syndrome of fisheries has been shown to contribute to the lack of sustainability of commercial fisheries, even under standardized approaches such as the maximum sustainable yield of single species/stocks (Pauly, 1995). Long-term information on the impacts of fisheries activities and the trophic structure of fish communities is important for understanding the current ecological status and effective implementation of ecosystem-based management of fishery resources. The present thesis has uncovered some of the variability in the trophic structure of exploited fishes along the east coast of the South Island of New Zealand. Resource use and trophic levels of fish species presented regional and temporal variability, which in some cases corresponded to contrasting commercial landings and catch per unit of effort (CPUE) in the most recent years. The niche breadth of the exploited community also varied in space (between Otago and Kaikoura) and time (before and after the expansion of New Zealand fisheries), but only a few species were responsible for those changes. These ecological shifts were correlated with both natural and anthropogenic factors, that had different effects on communities from distinct marine habitats.

Throughout the present study, literature reviews, meta-analyses, oceanographic surveys, and lab experiments were undertaken to generate baseline information on the ecology of the exploited fish communities. It was imperative to understand the history of fisheries in New Zealand and how they interacted with different fish communities throughout the history of exploitation. Understanding the distribution of isotope values of different basal organic matter sources was also essential for the estimation of trophic parameters used in the current thesis. Similarly, to study long-term changes in the ecology of fishes, the development of a method to access the isotopic values from muscle tissues stored in museum collections was necessary to resolve trophic structure changes over time.

196

By retrieving values of reported commercial fisheries landings and combining these data by species it was possible to demonstrate how fishing activities have expanded in the New Zealand Exclusive Economic Zone. While inshore species have been exploited since the start of the datasets (1931), offshore and deep-water species only started to be extensively fished between 1970 and 1980. It was demonstrated that the Marine Trophic Index (MTI) of the catch can represent differences in species composition and the type of habitat exploited. Exploitation impacted different habitats and species in contrasting ways, which was further investigated using an isotopic ecology approach. After the full expansion of fisheries in New Zealand, MTI and landing values decreased and stabilized, a trend also observed in other fishing grounds with less adaptable management approaches.

The analysis of the natural distribution of isotopes is a powerful tool for investigating ecological interactions in aquatic systems, but isotope values representing the basal organic matter source supporting the food web must be known for the estimation of trophic parameters. In the study region, δ15N and δ13C values of suspended particulate organic matter (SPOM) were related to processes such as continental runoff, water mass mixing and the northward transport of Subtropical Water along the coast. High variability in isotope values was identified at surface and maximum fluorescence depth, especially south of Akaroa. Using carbon to nitrogen (C:N) and nitrate to phosphate ratios, a subset of samples representing new phytoplankton production south of Akaroa was identified. Averages of δ13C and δ15N of SPOM were calculated for important fishery management areas and at depth. δ13C and δ15N values of the most abundant macroalgae species were also calculated from samples taken from sites in Otago and Kaikoura, showing almost no variation between sites compared to SPOM. SPOM and macroalgae represented the two most important basal organic matter sources in the region, used as sources of basal organic matter for the study of the local food-webs.

Using these basal organic matter sources and applying an isotopic mixing model approach, it was demonstrated that the resource use and trophic positions occupied by each fish species varied along the coast and between different habitats. Community-wide metrics also differed between the two most important regions sampled, Otago and Kaikoura. The fish community from Kaikoura had less trophic overlap and higher resource use variability, with species relying more heavily on pelagic production and occupying higher trophic positions. These regional differences were linked to smaller fishery yields and CPUE in Kaikoura when compared to Otago. While these results could correspond to natural differences in the isotope values and trophic structure of the exploited community, some of the results could also be

197 linked to habitat degradation and loss of connectivity due to fisheries activities and the Kaikoura earthquake in 2016. The observed patterns were most pronounced for tarakihi (Nemadactylus macropterus, Cheilodactylidae), a commercially important fish that feeds on benthic invertebrates and have shown small spawning biomass and high fishing mortality in the study region since the year 2000 (Fisheries New Zealand, 2019).

To investigate long-term changes in the trophic structure of the fish community, fishes collected in the past century and preserved by the Otago Museum and the Museum of New Zealand Te Papa Tongarewa were used. Using meta-analysis and laboratory experiments, it was demonstrated that fixation by formalin, followed by preservation in ethanol or isopropanol (standard procedure in museum collections) predictably affected the bulk isotope 13 15 values (δ CBulk and δ NBulk) in fish muscle tissue. These effects could be corrected with an arithmetic equation as a function of C:N and the proportion of nitrogen in the muscle tissue, making it possible to retrieve isotopic data from fishes in museums’ wet collections. Furthermore, isotopes of specific amino acids were also shown not to vary with preservation 13 15 (δ CAA) or to vary in an expected way (δ NAA), allowing for estimation of trophic level. Trophic level estimated from bulk muscle tissue and specific amino acids generally agreed, suggesting accuracy from both methodologies.

By correcting the effects of preservation, long-term changes in the trophic structure of the fish communities could be investigated based on isotopic ecology. Specimens collected after the expansion of industrialized fisheries in New Zealand tended to be relying more heavily on pelagic production and feeding at a higher trophic level than museum samples collected before 1996. These shifts occurred mainly for species with broader diets inhabiting the outer shelf and slope. Community niche breadth was observed to be larger for modern samples, mainly due to higher resource use diversity seen for tarakihi, red cod (Pseudophycis bachus, Moridae), ling (Genypterus blacodes, Ophidiidae) and spiny dogfish (Squalus acanthias, Squalidae). Estimating the absolute values of trophic position for each species was not an aim of the present study, alternatively, estimates were used to compare the trophic structure of the fish communities along the coast and over time, with specific methods attributed to each chapter. Therefore, although the results and implications of different chapters can be discussed altogether, specific values of trophic parameters varied, mainly due to different values of trophic fractionation applied. These differences were necessary due to the contrasting amount of data, species with different diets and the inclusion of isotope analysis of specific amino acids in the long-term analysis of trophic parameters.

198

Results from regional and temporal shifts in the trophic parameters of the fish community were very similar, with most of the variation in trophic structure and niche breadth observed between Kaikoura and Otago also identified between historical and modern samples. While the drivers of those changes could not be indicated in the regional comparison, present and modern samples were collected before and after the full expansion of New Zealand’s industrialized fisheries. Therefore, the observed pattern suggests that differences in the trophic structure of the exploited fish community between Kaikoura and Otago were likely linked to fisheries activities, as seen for the temporal analysis, although other environmental drivers cannot be ruled out. To incorporate samples along the whole study region, modern specimens used in the temporal analysis of trophic parameters included those used in the regional investigations. This could have favored the similarity between the results, indicating that there were no regional differences in trophic structure identified from historical samples, only from modern ones. Although the latter is true, the Bayesian approach used in the present study generates credible intervals related to the variability in the δ13C and δ15N values of populations and communities, therefore it is not biased towards extreme values, being suitable for this type of analysis (Jackson et al., 2011). Furthermore, specimens analyzed in the long-term investigations of trophic parameters were collected throughout the same range of latitude for both historical and modern groups, limiting latitudinal effects.

The comparisons made here indicated that fisheries activities had a large impact on the trophic parameters of the fish community, but linear models suggested relationships between fisheries activities and trophic parameters throughout the years were only significant for inner shelf species. Environmental variation and prey abundance were the factors influencing the same parameters for the outer shelf and slope assemblages. The contrasting results obtained by between-period comparison and by linear models represent the complexity of the system studied, as well as the different scales of each approach. Linear models allowed the present study to evaluate higher-frequency changes in trophic parameters, at the cost of grouping species into assemblages. For example, while outer shelf and slope assemblages showed the largest differences in resource use, trophic level and niche breadth after the full expansion of fisheries, high-frequency changes of the same factors were only attributed to prey abundance and environmental variability. These variables could also have influenced the between-year differences in the trophic parameters of blue cod and sea perch found in the most recent sampling period. This is the first time that the variability in the abundance of Munida (Munida gregaria, Munididae) was observed to influence the trophic parameters of fish assemblages, highlighting the importance of the understanding of ecosystem processes to the management of fisheries resources. In contrast, inner shelf species did not show any major 199 changes in their trophic structure between periods, but it was the only assemblage to show significant linear relationships with MTI values. The observed pattern is interesting, but not surprising, since coastal fish communities have been under the largest amount of pressure from fisheries activities, evidenced from the analyses of fisheries data. Previous studies have also demonstrated that coastal fish communities have shown clear changes in the composition, abundance and trophic relationships worldwide (Jackson, 2001; Myers et al., 2007; Myers & Worm, 2003).

With well resolved isotopic values of basal organic matter, trophic parameters can be calculated and used in long-term studies to support management decisions. While fisheries effort can be reduced or increased with new policies and market demand, the effects of prey abundance and environmental conditions on the trophic structure of important assemblages can only be mitigated. Reduction in the reliance of organic matter derived from macroalgae by the fish community throughout the years could indicate a decrease in macroalgae productivity in the region, which is in line with global declines of macroalgae abundance in the past century (Krumhansl et al., 2016). These changes can ultimately affect fisheries yields and need to be considered in management decisions. Therefore, to protect the niche of exploited species, these natural processes, as well as the importance of prey abundance and macroalgae habitat should be studied and considered during management.

Long-term changes in the composition of the broader fish community were also observed from the analysis of data from trawl surveys in the region. Data from the most recent surveys showed reduced assemblage segregation and a reduction in the average trophic level of the community. These changes were mainly driven by an increase in abundance of intermediate trophic level species and a reduction in the abundance of red cod and jack mackerels (Carangidae). These results corroborate the temporal decrease in MTI values after the full expansion of fisheries. Both results demonstrate that the decrease in the average trophic level of the community is not only a consequence of the expansion of fishing grounds or market demands but is also a result of shifts in community composition. Although shifts in the trophic level composition of the exploited community can represent a concern to the sustainability of these resources, they were mainly linked to the increase in abundance of several species in the present study. Because of the temporal mismatch between data from commercial fisheries and trawl surveys, it was not possible to evaluate shifts in the community composition before the expansion of industrialized fisheries in the 1970s. Therefore, although the present fish community presents a relatively stable overall abundance and length-frequency, the effects of earlier fisheries activities to the community composition

200

(e.g. removal of coastal predators) is still not clear. Nevertheless, the general decrease in average trophic level of the community was followed by an increase in trophic level of specific species, indicating that a decrease in the trophic level composition of the community can support higher fish abundance, but could also allow few species to feed at higher trophic levels, corroborating earlier findings from mass-balance model studies (Pinkerton et al., 2015b).

Some of the species that presented shifts in resource use and trophic level after the expansion of fisheries also showed trends in their average relative abundance. For example, red cod presented a decrease in abundance in coastal environments, as well as a significant increase in trophic level with time, while spiny dogfish had its abundance increased in the same depth strata, and its resource use and trophic level both shifted towards values occupied by red cod before 1996. Both species showed an increase in niche breadth between periods, which suggest increases in the number of basal sources of organic matter supporting the species diet and/or the number of feeding pathways, which can be linked to increased competition for resources. While the analysis of the present thesis was not designed to uncover the drivers of those changes, the use of isotopic ecology proved to be a powerful tool to uncover past ecological states, with the potential to be used in research for the conservation of these fisheries resources. The importance of considering specimen size, habitat use and the latitude of sample collection in isotopic ecology was also demonstrated and highlighted throughout the thesis. Some analyses undertaken during the present study were limited by small sample sizes of some species, especially those retrieved from museum collections. Future research could focus on collecting fish tissue, as well as samples from primary producers, from specific management areas throughout the years. This would allow more complete investigations of temporal changes in the trophic structure of important fish stocks to inform future management decisions. This approach requires small amount of time, space, and personnel, allowing sampling to be undertaken during regular stock assessment surveys.

The present study has demonstrated that fish communities and their niches are not just changing with exploitation and climate effects, but they have been changing for at least as long as we were able to get samples to analyze. Although the study region is relatively small, the observed patterns were prevalent inside an important fishing ground in New Zealand, a country that has proudly claimed to have one of the best fishery management systems in the world. These results highlight the importance of studying the trophic structure of exploited communities as a conservation tool beyond the maximum sustainable yield mindset. With every new generation of scientists and managers, there is potential for new ecological

201 baselines to be accepted, normalizing the presence of impacted environments, and bringing management goals incrementally towards a degraded system. We have shown here the importance of uncovering those ecological baselines by developing new methodologies and accessing previously collected samples. We emphasize the need for considering ecological baselines from less impacted or pristine systems when implementing ecosystem-based management, because looking back sometimes is the only way forward.

202

Appendices

Appendix A.1.1. Publication associated with this thesis

Durante, L., Beentjes, M., & Wing, S. (2020). Shifting trophic architecture of marine fisheries in New Zealand: Implications for guiding effective ecosystem‐based management. Fish and Fisheries. https://doi.org/10.1111/faf.12463

203

Appendix A.1.2. Data on all Quota Management System fish species used in chapter one. Source: Fisheries New Zealand and Fishbase

Common name QMS code Scientific name Habitat Trophic level Alfonsino BYX Beryx splendens, Beryx decadactylus Benthopelagic 4.12 Anchovy ANC Engraulis australis Pelagic-neritic 3 Barracouta BAR Thyrsites atun Benthopelagic 3.85 Bigeye tuna BIG Thunnus obesus Pelagic-oceanic 4.41 Blue cod BCO Parapercis colias Demersal 3.82 Blue mackerel EMA Scomber australasicus Pelagic-neritic 4.09 Blue moki MOK Latridopsis ciliaris Demersal 3.22 Blue shark BWS Prionace glauca Pelagic-oceanic 4.31 Bluenose BNS Hyperoglyphe antarctica Benthopelagic 3.95 Butterfish BUT Odax pullus Demersal 2.35 Black cardinalfish CDL Epigonus telescopus Bathydemersal 3.59 Blue warehou WAR Seriolella brama Benthopelagic 3.51 Elephant fish ELE Callorhinchus milii Demersal 3.6 Colistium nudipinnis, Peltorhamphus novaezelandiae, Colistium guntheri, Rhombosolea retiaria, Rhombosolea plebeia, Flatfish FLA Demersal 3.3 Rhombosolea leporina, Rhombosolea tapirina, Pelotretis flavilatus Freshwater eels SFE, LFE, ANG Anguilla australis, Anguilla dieffenbachii, Anguilla reinhardtii Demersal 3.82 Frostfish FRO Lepidopus caudatus Benthopelagic 3.95 Garfish GAR Hyporhamphus ihi Pelagic-neritic 3.19 Gemfish SKI Rexea solandri Benthopelagic 4.31 Dark ghost shark GSH Hydrolagus novaezealandiae Bathydemersal 3.52 Stargazer GIZ Kathetostoma giganteum Demersal 4.7 Grey mullet GMU Mugil cephalus Benthopelagic 2.48 Red gurnard GUR Chelidonichthys kumu Demersal 3.68 Hake HAK Merluccius australis Benthopelagic 4.45 Groper HPB Polyprion oxygeneios, Polyprion americanus Demersal 4.3 Hok HOK Macruronus novaezelandiae Benthopelagic 4.34 Trachurus declivis, Trachurus novaezelandiae, Trachurus Jack mackerels JMA Pelagic-oceanic 3.52 murphyi John dory JDO Zeus faber Benthopelagic 4.38 Kahawai KAH Arripis trutta and Arripis xylabion Pelagic-neritic 3.9 Kingfish KIN Seriola lalandi Benthopelagic 4.04 Leatherjacket LEA Meuschenia scaber Demersal 3.04 Ling LIN Genypterus blacodes Bathydemersal 4.2 Lookdown dory LDO Cyttus traversi Bathydemersal 4.25 Mako shark MAK Isurus oxyrinchus Pelagic-oceanic 4.43 Moonfish MOO Lampris guttatus Bathypelagic 4.22 Orange roughy ORH Hoplostethus atlanticus Bathypelagic 4.25 Pseudocyttus maculatus, Allocyttus niger, Neocyttus Oreos OEO Bathypelagic 4.19 rhomboidalis and Allocyttus verucosus Pacific bluefin tuna TOR Thunnus orientalis Pelagic-oceanic 4.05 Pale ghost shark GSP Hydrolagus bemisi Bathydemersal 3.8 Parore PAR Girella tricuspidata Benthopelagic 2.19 Pilchard PIL Sardinops sagax Pelagic-neritic 2.64 Porae POR Nemadactylus douglasii Demersal 3.39 Porbeagle shark POS Lamna nasus Pelagic-oceanic 4.47 Ray’s bream RBM Brama brama Pelagic-neritic 4.08 Red cod RCO Pseudophycis bachus Demersal 4.5 Red snapper RSN Centroberyx affinis Benthopelagic 3.81 Redbait RBT Emmelichthys nitidus Bathydemersal 3.43 Ribaldo RIB Mora moro Bathypelagic 3.75 Rig SPO Mustelus lenticulatus Demersal 3.49 Rough skate RSK Zearaja nasuta Demersal 3.68 Rubyfish RBY Plagiogeneion rubiginosum Bathydemersal 3.4 School shark SCH Galeorhinus galeus Benthopelagic 4.26 Sea perch SPE Helicolenus percoides Demersal 4.01 Silver warehou SWA Seriolella punctata Benthopelagic 3.4 Smooth skate SSK Dipturus innominata Demersal 4.1 Snapper SNA Pagrus auratus Reef-associated 3.46 Southern blue whiting SBW Micromesistius australis Benthopelagic 3.55 Southern bluefin tuna STN Thunnus maccoyii Pelagic-oceanic 3.93 Spiny dogfish SPD Squalus acanthias Benthopelagic 4.33 Sprat SPR Sprattus antipodum, S. muelleri Pelagic-neritic 3.2 Swordfish SWO Xiphias gladius Pelagic-oceanic 4.5 Tarakihi NMP Nemadactylus macropterus Demersal 3.41 Trevally TRE Pseudocaranx dentex Reef-associated 3.9 Trumpeter TRU Latris lineata Pelagic 3.7 White warehou WWA Seriolella caerulea Pelagic-neritic 3.2 Yellow-eyed mullet YEM Aldrichetta forsteri) Demersal 2.51 Yellowfin tuna YFN Thunnus albacares Pelagic-oceanic 4.38

204

Appendix A.2.1. Details on the global multimodel approach

A multimodel information theoretic approach was chosen to model isotope values of SPOM (δ13C and δ15N) and seawater (δ18O) along the ECSI. This approach allows one to run all possible models with suitable variables and average the best in predicting the dependent variable, resulting in better predicting power and a more parsimonious model (Grueber et al., 2011).

A.2.1.1. Standardization of variables To allow proper estimates and interpretation of both directions and magnitude of parameter estimates from model averages, all variables have been centralized using equation A.2.1 after Gelman (2008), where 푋𝑖 is the value 푋 of variable 𝑖. 푋̅𝑖 is the average of the variable 𝑖 and SD is the standard deviation of variable 𝑖. Variable month refers to each sampling season, which were grouped into 2017/18 (January, February and March) and 2018/19 cruises (November).

푋푖−푋̅푖 A.2.1) Standardized 푋𝑖 = 2∗푆퐷

A.2.1.2. Fitting, evaluating and averaging models

Linear models were fitted using the function lm in the R package lme4 (Bates et al.,

2015). The models with lowest AICc values (delta < 2) were identified and obtained using the dredge and get.model functions in the MuMIn package in R (Barto, 2015). Models were then averaged using the function model.avg also from MuMIn package. The full R code can be seen underneath (R Core Team, 2020):

“Dataframe <- as.data.frame(Data) d13C_model <- lm(d13CVPDB ~ Lon + Lat + Month + Fluor + S + T + Ammonia + SiO2 + Nox + rdepth, Dataframe, na.action=na.fail)

mode_set_13C <- dredge(d13C_model, trace=FALSE, rank="AICc") top_models_13C <- get.models(mode_set_13C, subset=delta<2) model.avg(top_models_13C) summary(model.avg(top_models_13C)) top_models_13C

Data2 <- read.csv("For_N.csv",header = TRUE,sep = ',') Dataframe2 <- as.data.frame(Data2)

205 d15N_model <- lm(d15NAIR ~ Lon + Lat + Month + Fluor + S + T + Ammonia + SiO2 + Nox + rdepth, Dataframe2, na.action=na.fail) mode_set_15N <- dredge(d15N_model, trace=FALSE, rank="AICc") top_models_15N <- get.models(mode_set_15N, subset=delta<2) model.avg(top_models_15N) summary(model.avg(top_models_15N)) top_models_15N

Data3 <- read.csv("For_O.csv",header = TRUE,sep = ',') Dataframe3 <- as.data.frame(Data3) d18O_model <- lm(d18OVSMOW ~ Lon + Lat + Month + Fluor + S + T + Ammonia + SiO2 + Nox + rdepth, Dataframe3, na.action=na.fail) mode_set_18O <- dredge(d18O_model, trace=FALSE, rank="AICc") top_models_18O <- get.models(mode_set_18O, subset=delta<2) model.avg(top_models_18O) summary(model.avg(top_models_18O)) top_models_18O”

206

Appendix A.2.2. Isosurfaces of oceanographic variables along the east coast of the South Island during austral summer of 2017/18 at maximum fluorescence depth. Stations are represented by black dots

207

Appendix A.2.3. Isosurfaces of oceanographic variables along the east coast of the South Island during austral summer of 2018/19 at maximum fluorescence depth. Stations are represented by black dots

208

Appendix A.2.4. Linear relationship between NOx and phosphate (orthophosphate) values from surface water samples collected along the east coast of the South Island during the austral Summers of 2017/18 and 2018/19. Linear equation: NOx = -14.55 + 7.38 * Phosphate, n = 72, r2 = 0.94 and p < 0.0001. Concentrations are given in μg.L-1

209

Appendix A.2.5. Maximum fluorescence depth (meters) measured during the 2017/18 (a) and 2018/19 (b) sampling seasons along the east coast of the South Island. Data was displayed using the DIVA interpolation. Stations are represented by black dots

210

13 δ C Surface Estimate Adjusted SE Relative importance p-values (Intercept) -23.42 0.25 - <0.0001 Latitude -1.62 1.52 0.61 0.29 Longitude 1.50 1.42 0.58 0.29 NOx 0.46 0.69 0.47 0.51 SiO2 -1.26 0.69 0.89 <0.1

Temperature 0.11 0.51 0.05 0.83 Sampling season -0.28 0.54 0.32 0.61

Fluorescence 0.58 0.59 0.66 0.32 Salinity -0.15 0.38 0.26 0.69 Ammonia -0.23 0.45 0.34 0.60 δ15N Surface (Intercept) 4.86 0.12 - <0.0001 Latitude 0.04 0.16 0.25 0.78 Longitude 0.09 0.21 0.25 0.67

NOx -5.59 0.28 1.00 <0.0001 Salinity 0.02 0.11 0.25 0.84 δ18O Surface (Intercept) 0.28 0.02 - < 0.0001 Latitude 0.01 0.02 0.20 0.75 NOx -0.20 0.05 1.00 <0.0001

SiO2 0.02 0.04 0.40 0.53

Temperature -0.08 0.06 0.80 0.19 Sampling season 0.00 0.03 0.20 0.96 Salinity 0.05 0.04 0.80 0.20 Ammonia 0.00 0.02 0.20 0.81

211

Appendix A.2.7. Nonparametric comparisons of δ15N of SPOM at surface for each pair of statistical area using Wilcoxon’s test. * represents p-values lower than 0.05

Stat Stat area Score Mean Std Err Dif Z p-Value area Difference 18 17 9.8750 3.679900 2.68350 0.0073* 23 22 6.1750 3.166667 1.95000 0.0512 38 22 5.3333 4.917090 1.08465 0.2781 38 24 5.2381 6.646398 0.78811 0.4306 20 17 4.8000 1.914854 2.50672 0.0122* 39 24 3.0119 5.019011 0.60010 0.5484 38 17 2.4000 2.049390 1.17108 0.2416 23 17 2.0250 1.837117 1.10227 0.2703 39 22 1.9833 3.801316 0.52175 0.6018 23 18 1.6500 3.872983 0.42603 0.6701 20 18 0.3750 3.679900 0.10190 0.9188 23 20 -0.2250 1.837117 -0.12247 0.9025 38 23 -0.6250 1.767767 -0.35355 0.7237 39 38 -0.7500 1.224745 -0.61237 0.5403 39 17 -1.0500 1.807392 -0.58095 0.5613 39 23 -1.1250 1.620185 -0.69437 0.4875 24 22 -2.2286 3.561472 -0.62574 0.5315 38 20 -2.4000 2.049390 -1.17108 0.2416 22 17 -2.6667 3.055050 -0.87287 0.3827 39 20 -3.1500 1.807392 -1.74284 0.0814 24 17 -4.8286 3.805360 -1.26889 0.2045 38 18 -6.8250 6.358066 -1.07344 0.2831 22 20 -8.2667 3.055050 -2.70590 0.0068* 24 23 -9.0774 4.015100 -2.26081 0.0238* 39 18 -9.6250 4.815773 -1.99864 0.0456* 24 20 -11.8857 3.806010 -3.12288 0.0018* 22 18 -13.7083 3.499510 -3.91722 <.0001* 24 18 -17.2298 3.742771 -4.60348 <.0001*

212

Appendix A.2.8. Nonparametric comparisons of δ15N of SPOM at MF depth for each pair of statistical area using Wilcoxon’s test. * represents p-values lower than 0.05

Stat area Stat Score Mean Std Err Z p-Value area Difference Dif

39 24 7.6500 3.801316 2.01246 0.0442* 18 17 6.5313 3.540441 1.84476 0.0651 38 24 6.4000 4.917090 1.30158 0.1931 23 22 5.7212 2.887306 1.98148 0.0475* 20 18 4.5521 3.540441 1.28574 0.1985 23 18 3.5938 3.307189 1.08665 0.2772 20 17 2.6667 1.527525 1.74574 0.0809 39 22 2.5962 3.396831 0.76429 0.4447 38 22 2.1538 4.341216 0.49614 0.6198 23 17 1.4583 1.649916 0.88388 0.3768 39 17 0.4167 1.443376 0.28868 0.7728 23 20 0.2917 1.649916 0.17678 0.8597 38 17 0.0000 1.490712 0.00000 1.0000 39 38 0.0000 1.224745 0.00000 1.0000 38 23 -0.6250 1.767767 -0.35355 0.7237 39 23 -1.1250 1.620185 -0.69437 0.4875 38 20 -1.3333 1.490712 -0.89443 0.3711 22 17 -1.6410 3.049450 -0.53814 0.5905 39 20 -2.0833 1.443376 -1.44338 0.1489 38 18 -3.7188 5.205166 -0.71443 0.4750 24 22 -4.0205 3.117088 -1.28983 0.1971 39 18 -4.7813 4.003904 -1.19415 0.2324

22 20 -6.5641 3.049450 -2.15255 0.0314* 24 17 -7.6000 3.376389 -2.25093 0.0244* 24 23 -8.3917 3.166667 -2.65000 0.0080* 24 20 -8.8000 3.376389 -2.60634 0.0092* 22 18 -9.2716 3.179335 -2.91622 0.0035* 24 18 -14.6604 3.267687 -4.48648 <.0001*

213

Appendix A.2.9. Nonparametric comparisons of δ18O of seawater from the whole east coast of the South Island for each pair of depth sampled using Wilcoxon’s test. * represents p-values lower than 0.05

Depth Depth Score Mean Std Err Z p-Value Difference Dif 1000 500 -2.0313 3.30470 -0.61465 0.5388 150 0 -6.5399 6.23579 -1.04877 0.2943 1000 150 -11.6827 4.72608 -2.47196 0.0134* 500 150 -12.8221 3.89725 -3.29004 0.0010* 500 0 -24.5826 6.70646 -3.66550 0.0002* 1000 0 -25.5653 10.62205 -2.40681 0.0161*

214

Appendix A.4.1. Publication associated with this thesis

Durante, L., Sabadel, A., Frew, R., Ingram, T., & Wing, S. (2020). Effects of fixatives on stable isotopes of fish muscle tissue: implications for trophic studies on preserved specimens. Ecological Applications. https://doi.org/10.1002/eap.2080

215

Appendix A.4.2. CSIA-AA sample preparation extracted and modified from Sabadel et al., (2016)

A.4.2.1. Reagents and solvents

All reagents and solvents used were of analytical grade: hydrochloric acid (HCl), ethyl acetate (Fisher chemicals), acetone, triethylamine (Scharlau), acetic acid, dichloromethane, isopropanol and acetic anhydride (Merck Chemicals). All AAs were purchased individually in crystalline form from: Sigman-Aldrich (alanine (Ala), aspartic acid (Asp), Glutamic acid (Glu), glycine (Gly), norleucine (Nle) and phenylalanine (Phe)) or Mann Assayed chemicals (isoleucine (Ile), leucine (Leu), proline (Pro), serine (Ser), threonine (Thr) and valine (Val)) for use as a high-purity calibration standard for GC-C-IRMS.

A.4.2.2. Amino acid extraction

The AA extraction occurred by hydrolyzing the samples. Initially, the samples (finely grinded fat-free fish muscle tissues) were placed in borosilicate glass kimax® culture tubes.

An internal standard mixture (50 µl of 1 mg/ml Nle in 0.1 M HCl) was added to the sample to monitor any possible fractionation differences due to the derivatization. Norleucine was chosen as it makes an excellent an internal standard for three main reasons: (i) it is a synthetic AA that does not naturally occur in proteins, (ii) it is stable to acid hydrolysis and (iii) it can be chromatographically separated from other protein AA.

The samples were then covered with 2 ml of 6 M HCl and heated at 110 oC for 24 hours in a N2 atmosphere. After 24 hours, the tubes were removed from the oven and allowed to cool down until reaching room temperature. They were then centrifuged (3000 rpm, 10 min) and each sample’s supernatant was transferred into a clean kimax® tube, preferably avoiding as much remaining tissue as possible. This was done a second time by adding 2 ml of 0.1 M HCl to the samples, centrifuging and extracting the new supernatant. Combined o supernatants were dried at 60 C under a stream of N2.

A.4.2.3. Amino acid derivatization

Samples were all derivatized at the same time, along with four samples containing the AAs standard solution.

216

Esterification - The esterification of the carboxylic acid group was completed by adding 0.5 ml of a solution of isopropanol, previously acidified with acetyl chloride (4:1, v/v), to each kimax® tube. It is very important to carefully seal the tubes with PTFE before proceeding, to ensure that no solvents will leave the tube during the reaction. These tubes were then heated in an oven at 100 oC for 1 hour. The esterification of the carboxylic acid was conducted under acidic conditions (because HCl is generated when forming isopropyl acetate from acetyl chloride and isopropanol), so that the OH- group could be protonated to generate water as a leaving group. The carboxylic acids were not dissociated and therefore they will react with the isopropyl acetate to form an ester (see Figure A.4.2.1). After 1 hour, the reaction was stopped by placing the tubes in a freezer (-4 oC). After a couple of minutes, when the samples had cooled down to below room temperature, they were placed under a o gentle stream of N2 at 40 C to remove excess solvent, until dryness.

Acetylation - Step two, the acetylation of the amine group was performed by adding 1 ml of a mixture of acetone:triethylamine:acetic anhydride (5:2:1, v/v/v) to each sample followed by heating at 60 oC for 10 min. During this time, the reaction was carried out in a basic environment (triethylamine), so that free amine nucleophiles were present in the reaction system. The nucleophiles will react with the acetic anhydride to form a secondary amide. At this stage, it is important to note that threonine and serine also acetylate at their OH group on the side chain.

Figure A.4.2.1. NAIP ester formation. This happens in two distinct reactions: Step 1 - Esterification and Step 2 - Acylation.

Once the derivatization was completed, the reagents were evaporated at room temperature under a stream of N2 and then the AAs derivatives were re-dissolved in 2 ml of ethyl acetate. The extraction was performed by adding 1 ml of saturated NaCl in Milli-Q

217 water (Millipore Corporation: (MQ), 18.2 MΩ•cm at 25 °C), vortexing, and then drawing off the organic phase. This was done twice to ensure quantitative extraction. The combined organic phases were then dried under a gentle N2 stream at room temperature. Any residual water was removed by adding dichloromethane (1 ml x 3) and drying it with gentle stream of

N2 in an ice bath. It is important to have a final solution containing as few water molecules as possible, as they would interfere with the GC-FID or GC-C-IRMS analysis.

Finally, the derivatives were re-dissolved in a minimum volume of ethyl acetate (i.e. 40 µl). The final solution was sufficiently concentrated to provide reasonable signal strength when analyzed by a GC (> 100 mV).

A.4.2.4. Amino acid GC-IRMS measurements

13 15 δ CAA and δ NAA were measured by gas chromatography/combustion/isotope ratio mass spectrometer (GC-IRMS), using a Thermo Trace gas chromatograph, the GC combustion III interface, and a Deltaplus XP isotope ratio mass spectrometer (Thermo Fisher Scientific). 200 nl aliquots of derivatized AA were injected, inlet at 270 °C in splitless mode, carried by helium at 1.4 ml min−1 and separated on a VF-35ms column (30 m long, 0.32 mm ID and a 1.0 μm film thickness). The GC temperature program is summarized in Table A.4.2.1 and an example of a chromatogram is presented in Fig A.4.2.2.

Table A.4.2.1. GC-FID and GC-C-IRMS program, modified from Styring et al. 2012.

Rate (oC/min) Temp (oC) Hold time (min) Initial - 40 5 Ramp 1 15 120 0 Ramp 2 3 180 0 Ramp 3 1.5 210 0 Ramp 4 5 270 7

13 For δ CAA measurements, the oxidation reactor was set at 950 °C and the reduction o 15 reactor at 200 C; while for δ NAA measurements, the oxidation rector was set at 980 °C and the reduction reactor at 650 °C and a liquid nitrogen cold trap employed after the reduction reactor. Samples were analyzed in duplicates or triplicates – when needed - along with amino acid standards of known isotopic composition (STD; measured by EA-IRMS). Each run contained no more than 10 samples.

218

15 Fig A.4.2.2. Example of a GC-IRMS chromatogram from the measurement of δ NAA. Compounds names are displayed for the 15 AAs from the standard solution. Note that Lys and Tyr co-elute, hence why they were not used in this thesis.

A.4.2.5. Data correction

Correction of raw δ15N data - Correction was done by plotting the mean δ15N value of each individual amino acid standard measured on GC-IRMS vs. their ‘real’ value measured on EA-IRMS (Fig A.4.2.3 – EA-IRMS measurements done as described in the thesis). The r2 is always > 0.99. Raw δ15N values for amino acids in the samples were then corrected using 15 the fitting equation and the value further adjusted by adjusting to the difference in δ NNle (internal standard) value in the sample and in the standards.

8.0

6.0

4.0

2.0 y = 0.8815x - 1.801 R² = 0.9947

0.0 IRMS - -10.0 -5.0 0.0 5.0 10.0 15.0

GC -2.0

N 15

δ -4.0

-6.0

-8.0 15 δ NEA-IRMS Fig A.4.2.3. Example of standard calibration for AA corrections using EA-IRMS values vs GC-IRMS values of the STD.

219

Correction of raw δ13C data used in chapter 4 - Similar preliminary assessment than for raw δ15N data: no drift and SD < 1‰. The raw δ13C data as corrected individually as follow, based on mean values of the STD:

13 13 ∆Corr= 훿 CSTD IRMS − 훿 CSTD EA A.4.2.1)

13 13C − ∆ 훿 CSample = 훿 Sample IRMS Corr A.4.2.2)

Correction of raw δ13C data used in chapter 5 - Similar preliminary assessment than for raw δ15N data: no drift and SD < 1‰. The raw δ13C data as corrected individually as follow, based following (O’Brien et al., 2002):

13 13 13 훿 C − 훿 C + 훿 C ∗ 휌 13 Sample IRMS STD IRMS STD EA 푆푇퐷 훿 CSample = A.4.2.3) 휌푆푇퐷

13 Where δ CSample-IRMS is the value of the derivatized AA in the sample measured by the 13 GC-IRMS; δ CSTD-IRMS is the value of the derivatized AA in the STD measured by GC- 13 IRMS; δ CSample-EA is the value of the underivatized AA in the STD measured by EA-IRMS; and ρSTD is the proportion of C in the derivative from the AA. While equations A.4.2.1 and A.4.2.2 were used to evaluate proportional differences in δ13C due to fixatives, equation A.4.2.3 is more suitable for the correction of added carbons when absolute values of δ13C from samples are needed, i.e. in chapter 5 (O’Brien et al., 2002).

A.4.2.6. Propagation of error during TP calculations

When using the equation below for TP calculation:

(∂15N −∂15N +β) TP = Glu Phe + 1 A.4.2.4) Glu−Phe TEF

Where β represents the isotopic difference between Glu and Phe in the primary 15 15 producers: δ NGlu - δ NPhe = 3.4‰ for aquatic cyanobacteria and algae (Chikaraishi et al.,

220

2009). The trophic enrichment factor (TEF) value of 7.6‰ is commonly used in aquatic food web studies for low TP (McMahon & McCarthy, 2016; Ohkouchi et al., 2017). The TEF is the average 15N enrichment in one or more trophic AAs relative to source AAs per TP and is key to estimate the TP accurately. Recent studies have highlighted the issues with the use of a fixed TEF across a food web and more specifically at higher TPs (Bradley et al., 2015). Indeed, as pointed out by the meta-analysis of Nielsen et al., (2015), TEF decreases with higher TP. Change in diet with higher trophic level species was found to be the main cause of this TEF decrease (McMahon & McCarthy, 2016). Thus, a TEF = 7.6 was used (Chikaraishi et al., 2009; McMahon et al., 2015).

The propagation of the error σTL can be calculated, following (Dale et al., 2011), by using the following equation:

σ σ 2 σ 2 TL = √( numerator) + ( TEF) A.4.2.5) |TL| numerator TEF

In which σnumerator can be calculated as follow:

2 2 2 √ 15 15 σnumerator = (σδ NGlu) + (σδ NPhe) + (σβ) A.4.2.6)

With σβ equals to 0.9‰ and σTEF equals to 1.1‰ (Chikaraishi et al., 2009).

221

Appendix A.4.3. Meta-data from published literature on the effects of formalin fixation and formalin preservation on bulk carbon and nitrogen isotopes

δ¹³C shifts (‰) δ¹⁵N shifts (‰) Period Source reference Fish species Formalin Formalin (weeks) Formalin Ethanol Formalin Ethanol +ethanol +ethanol (Bosley & Pleuronectes 16 -0.74 - -2.17 1.21 - -1.41 Wainright, 1999) americanus (Ogawa et al., Hemibarbus barbus 9 - - - -0.25 -0.18 - 2001) 62 - - - 0.17 - - 117 - - - -0.1 - - Lepomis macrochirus 9 - - - 0.1 0.12 - 62 - - - -0.3 - - 117 - - - 0 - - Micropterus salmoides 9 - - - 0.1 0.07 - 62 - - - -0.1 - - 117 - - - -0.2 - - Zacco platyput 9 - - - 0.3 -0.21 - 62 - - - 0.4 - - 117 - - - 0.45 - - (Kaehler & Argyrosomus 1 -0.27 0.77 - -0.3 -0.71 - Pakhomov, 2001) hololepidotus 4 -0.59 0.64 - 1.04 0.99 - 12 -0.65 0.59 - 0.71 0.81 - (Arrington & Arius felis 6 -1.11 -0.16 -1.12 0.36 0.2 0.62 Winemiller, 2002) Cynoscion nebulosus Dorosoma cepedianum Mugil cephalus (Edwards et al., Rhinichthys cataractae 27 - - -2 - - 0.4 2002) Percina caprodes 700 - - -0.8 - - 0.5 Percina roanoka Etheostoma tippecanoe Rhinichthys cataractae 4 - - -1.5 - - - (Sarakinos et al., Catostomus occidentalis 26 1.33 0.21 - 0.16 0.37 - 2002) (Sweeting et al., Gadus morhua 84 -1.73 0.48 - 0.8 1.05 - 2004) (Kelly et al., Salvelinus alpinus 10 -2.21 -0.78 - 0.66 0.35 - 2006) Salvelinus alpinus 10 -2.78 -0.2 - 0.03 0.06 - (Fanelli & Cartes, Hoplostethus 24 -0.06 - - 0.16 - - 2010) mediterraneus 48 -0.71 - - -0.54 - - 96 -2.08 - - 0.16 - - Hymenocephalus 24 -1 - - -0.05 - - italicus 48 -0.1 - - -0.22 - - 96 -1.57 - - 0.35 - - Nezumia aequalis 24 0.78 - - 0.09 - - 48 -0.15 - - -0.06 - - 96 -0.88 - - 0.31 - - (Akın et al., 2011) Perca fluviatilis 12 -1.2 0.48 - 2.05 2.19 - 24 -1.14 0.3 - 0.34 0.7 - Blicca bjoerkna 12 -1.09 0.26 - 0.71 0.8 - 24 -1.48 0.03 - 0.31 0.35 - Continues in the next page

222

δ¹³C shifts (‰) δ¹⁵N shifts (‰) Source Period Fish species Formalin Formalin reference (weeks) Formalin Ethanol Formalin Ethanol +ethanol +ethanol (Xu et al., Siniperca chuatsi 89 -0.89 0.68 - 0.24 0.52 - 2011) Hypophthalmichthys nobilis 89 -0.75 0.52 - 0.08 0.35 - Ctenopharyngodon idellus 89 -1.38 1.02 - 0.08 0.38 - (Correa, 2012) Salmo trutta 9 - 0.91 - - 0.44 - Oncorhynchus mykiss 9 - 0.94 - - 0.71 - Galaxias platei 9 - 0.23 - - 0.45 - (Lau et al., Pseudogastromyzon myersi 4 -0.79 -0.84 -1.2 0.44 0.3 0.26 2012) 8 -1.78 -1.01 -2.24 0.28 0.05 0.08 12 -1.29 -0.26 -0.9 0.45 0.26 -0.03 24 -1.81 -1.21 -1.09 -0.43 0.42 0.28 48 -2.43 -1.86 -1.39 -0.83 -0.9 -0.98 Liniparhomaloptera disparis 4 -1.6 -1.07 -2.45 0.54 0.05 0.38 8 -2.98 -2.23 -1.56 -0.49 0.21 0.63 12 -1.05 -0.75 -0.82 0.88 1.4 1.85 24 -1.45 -0.75 -1.04 0.82 0.65 1.05 48 -1.18 -0.5 -0.33 -0.4 0.26 -0.21 Ctenogobius duospilus 4 -1.18 0.88 -0.92 0.25 0.75 0.3 8 -1 0.3 -0.06 -0.02 0.3 0.39 12 -0.3 0.68 -0.84 0.18 0.31 0.36 24 -1.19 0.18 -0.13 0.59 0.86 0.27 48 -0.83 0.34 0.43 0.07 0.06 0.29 (González- Vanmanenia sp. 8.6 -0.74 - - 0.32 - - Bergonzoni et Lepidocephalichthys 8.6 -0.64 - - 0.46 - - al., 2015) berdmorei Garra cambodgiensis 8.6 -0.7 - - 0.39 - - Rhodeus ocellatus 8.6 -1.5 - - 0.01 - - Puntius semifasciolatus 8.6 -1.2 - - 0.29 - - Barilius pulchellus 8.6 -0.77 - - 0.37 - - Danio chrysotaeniatus 8.6 -0.89 - - 0.34 - - Scaphiodonichthys 8.6 -0.78 - - 0.31 - - acanthopterus Pteronemacheilusmeridionalis 8.6 -0.6 - - 0.31 - - Schistura kengtungensis 8.6 -0.85 - - 0.32 - - Schistura kloetzliae 8.6 -1.06 - - 0.38 - - Channa gachua 8.6 -1.2 - - -0.34 - - Rhinogobius brunneus 8.6 -0.97 - - 0.13 - - Rhinogobius giurinus 8.6 -1.4 - - 0.48 - - Creteuchiloglanis 8.6 -0.76 - - 0.4 - - longipectoralis Glyptothorax laosensis 8.6 -1.1 - - 0.55 - - Mastacembelus armatus 8.6 -0.73 - - 0.4 - - (Kishe- Haplochromine cichlid 500 - - -0.66 - - 0.34 Machumu et species al., 2017) (Hetherington Thunnus albacares 1 -1.5 0.1 - 1.3 1.1 - et al., 2019) Thunnus albacares 96 - - -0.8 - - 1.4

223

Appendix A.5.1. Results from general linear models between bulk isotope values, year of collection, specimens’ total length and latitude from chapter 5 data

δ13C Intercept Year Total length Latitude Species Estimate SE p value Estimate SE p value Estimate SE p value Estimate SE p value Barracouta -36.64 8.29 0.0003*** 0.010 0.005 0.053 -0.001 0.001 0.193 0.033 0.05 0.520 Blue cod 20.00 12.96 0.128 -0.011 0.007 0.092 0.002 0.002 0.286 0.372 0.05 <.0001*** Common warehou -4.40 6.86 0.532 -0.007 0.004 0.103 -0.001 0.002 0.583 0.011 0.07 0.877 Elephant fish -13.66 26.40 0.612 -0.007 0.014 0.644 0.002 0.002 0.377 -0.181 0.27 0.513 Giant stargazer 19.37 6.80 0.008* -0.020 0.004 <.0001*** 0.001 0.001 0.062 -0.022 0.06 0.706 Gurnard 23.64 12.23 0.060 -0.014 0.005 0.017 0.004 0.001 0.001** 0.349 0.08 <.0001*** Hapuka -32.58 15.19 0.051 0.003 0.010 0.774 0.003 0.001 0.014* -0.160 0.10 0.124 Hoki -46.02 12.97 0.003** 0.019 0.004 0.0005** 0.001 0.001 0.173 0.234 0.16 0.168 Leatherjacket -18.80 10.83 0.089 0.001 0.005 0.815 -0.002 0.003 0.550 0.037 0.06 0.571 Ling -68.93 13.65 <.0001*** 0.032 0.007 <.0001*** 0.000 0.001 0.428 0.292 0.08 0.0005** Lookdown dory -29.48 19.01 0.133 0.027 0.007 0.0004*** 0.006 0.001 <.0001*** 1.013 0.24 0.0002*** Orange roughy 1.01 17.99 0.956 -0.004 0.009 0.691 0.003 0.002 0.091 0.289 0.14 0.044* Red cod -24.08 12.66 0.063 0.011 0.006 0.077 0.000 0.001 0.670 0.367 0.08 <.0001*** Sea perch -9.78 4.36 0.028* -0.001 0.002 0.563 0.007 0.001 <.0001*** 0.185 0.03 <.0001*** Spiny dogfish 17.95 15.68 0.262 -0.015 0.009 0.122 -0.003 0.001 0.026* 0.132 0.18 0.481 Tarakihi 19.37 12.08 0.114 -0.013 0.006 0.026* 0.000 0.002 0.861 0.261 0.09 0.006* δ15N Intercept Year Total length Latitude Species Estimate SE p value Estimate SE p value Estimate SE p value Estimate SE p value Barracouta 8.52 15.73 0.594 0.009 0.009 0.347 0.000 0.001 0.646 0.291 0.10 0.007* Blue cod 47.82 14.98 0.002** -0.013 0.008 0.098 0.003 0.002 0.070 0.213 0.06 0.001** Common warehou 26.46 11.96 0.044* -0.007 0.007 0.310 0.000 0.003 0.943 -0.048 0.12 0.692 Elephant fish 25.26 33.50 0.462 -0.013 0.018 0.498 0.003 0.002 0.209 -0.251 0.34 0.475 Giant stargazer 65.43 8.21 <.0001*** -0.028 0.005 <.0001*** 0.002 0.001 0.008* -0.065 0.07 0.357 Gurnard -19.80 12.20 0.112 0.019 0.005 0.001** 0.003 0.001 0.0086* 0.098 0.08 0.219 Hapuka 16.25 20.29 0.438 -0.017 0.013 0.201 0.009 0.001 <.0001*** -0.594 0.13 0.0005** Hoki -25.15 13.97 0.094 0.025 0.005 <.0001*** 0.001 0.001 0.484 0.263 0.17 0.152 Leatherjacket -31.02 11.92 0.013* 0.016 0.006 0.006* -0.003 0.003 0.380 -0.317 0.07 <.0001*** Ling -7.43 16.73 0.659 0.020 0.009 0.028* 0.000 0.001 0.724 0.411 0.09 <.0001*** Lookdown dory 87.85 37.03 0.025* 0.002 0.013 0.906 0.010 0.002 <.0001*** 1.855 0.46 0.0004*** Orange roughy 26.38 21.27 0.226 0.011 0.011 0.301 0.007 0.002 0.003** 0.821 0.16 <.0001*** Red cod -10.76 11.20 0.341 0.018 0.005 0.002** 0.006 0.001 <.0001*** 0.293 0.07 0.0002*** Sea perch 61.73 8.29 <.0001*** -0.020 0.004 <.0001*** 0.010 0.002 <.0001*** 0.248 0.06 <.0001*** Spiny dogfish -36.50 24.24 0.143 0.045 0.014 0.004** -0.001 0.002 0.643 0.902 0.29 0.004** Tarakihi 1.40 11.23 0.901 0.007 0.005 0.181 -0.004 0.001 0.013* 0.020 0.08 0.815

224

Appendix A.5.2. Results from general linear models between proportion of pelagic basal organic matter supporting a fish’s det (%SPOM), trophic level (TL) and year of collection, specimens’ total length and latitude from chapter 5 data

%SPOM Intercept Year Total length Latitude Species Estimate SE p value Estimate SE p value Estimate SE p value Estimate SE p value Barracouta 3.98 1.60 0.022* -0.0011 0.0009 0.268 0.0002 0.0001 0.014* 0.034 0.010 0.003** Blue cod -2.84 1.56 0.073 0.0018 0.0008 0.028* 0.0001 0.0002 0.505 0.006 0.006 0.389 Common warehou 0.18 0.75 0.811 0.0009 0.0004 0.070 0.0002 0.0002 0.382 0.031 0.007 0.001* Elephant fish 3.90 2.91 0.199 0.0004 0.0016 0.813 -0.0003 0.0002 0.110 0.090 0.030 0.008* Giant stargazer -1.06 1.28 0.414 0.0019 0.0007 0.013* -0.0002 0.0001 0.126 0.048 0.011 0.0001*** Gurnard 1.66 2.31 0.476 -0.0002 0.0010 0.824 -0.0003 0.0002 0.240 0.016 0.015 0.297 Hapuka 5.07 3.38 0.158 -0.0009 0.0021 0.690 -0.0003 0.0002 0.155 0.058 0.022 0.0191* Hoki 14.91 4.01 0.002** -0.0024 0.0013 0.082 0.0000 0.0002 0.945 0.217 0.050 0.001** Leatherjacket 5.96 1.73 0.001** -0.0019 0.0008 0.021* 0.0007 0.0005 0.141 0.041 0.010 0.0003*** Ling -0.72 1.74 0.680 0.0012 0.0009 0.180 -0.0001 0.0001 0.04* 0.026 0.010 0.01* Lookdown dory 13.17 3.56 0.001** -0.0030 0.0013 0.026* -0.0003 0.0002 0.168 0.149 0.044 0.002** Orange roughy 2.44 3.71 0.516 0.0012 0.0018 0.524 -0.0001 0.0004 0.725 0.095 0.028 0.002** Red cod 4.25 1.67 0.014* -0.0004 0.0008 0.650 -0.0003 0.0001 0.02* 0.064 0.011 <.0001*** Sea perch -0.31 1.01 0.763 0.0014 0.0005 0.009* -0.0008 0.0002 <.0001*** 0.040 0.007 <.0001*** Spiny dogfish -3.25 1.79 0.080 0.0023 0.0010 0.037* 0.0005 0.0002 0.004** 0.025 0.021 0.251 Tarakihi -0.11 1.64 0.946 0.0009 0.0008 0.236 0.0007 0.0002 0.001** 0.031 0.012 0.014* TL Intercept Year Total length Latitude Species Estimate SE p value Estimate SE p value Estimate SE p value Estimate SE p value Barracouta -0.84 5.23 0.874 0.004 0.003 0.171 -0.0001 0.0003 0.647 0.11 0.03 0.003** Blue cod 11.17 4.00 0.007* -0.003 0.002 0.147 0.0006 0.0004 0.224 0.05 0.02 0.003** Common warehou 6.19 3.27 0.080 -0.002 0.002 0.449 0.0002 0.0009 0.835 0.00 0.03 0.955 Elephant fish 4.41 7.66 0.573 -0.003 0.004 0.488 0.0010 0.0005 0.078 -0.08 0.08 0.305 Giant stargazer 24.03 3.64 <.0001*** -0.010 0.002 <.0001*** 0.0010 0.0003 0.005* 0.00 0.03 0.998 Gurnard -10.23 3.49 0.006* 0.007 0.002 <.0001*** 0.0003 0.0003 0.406 0.03 0.02 0.265 Hapuka 6.24 7.96 0.448 -0.008 0.005 0.148 0.0040 0.0005 <.0001** -0.24 0.05 0.0004*** Hoki -23.02 7.02 0.006* 0.011 0.002 0.0004*** 0.0001 0.0003 0.798 -0.13 0.09 0.167 Leatherjacket -11.67 3.11 0.001** 0.006 0.001 0.0003*** -0.0012 0.0009 0.174 -0.09 0.02 <.0001*** Ling 10.26 5.97 0.093 0.001 0.003 0.733 0.0004 0.0002 0.048* 0.20 0.03 <.0001*** Lookdown dory 18.23 10.74 0.102 -0.001 0.004 0.902 0.0027 0.0007 0.0004*** 0.35 0.13 0.0135* Orange roughy 2.24 6.04 0.714 0.004 0.003 0.231 0.0019 0.0006 0.0043** 0.16 0.05 0.002** Red cod -5.22 3.85 0.181 0.004 0.002 0.026* 0.0019 0.0003 <.0001*** 0.03 0.03 0.272 Sea perch 19.07 2.70 <.0001*** -0.007 0.001 <.0001*** 0.0024 0.0005 <.0001*** 0.06 0.02 0.001** Spiny dogfish -15.12 6.51 0.028* 0.015 0.004 0.001** 0.0001 0.0006 0.906 0.28 0.08 0.001** Tarakihi -6.63 3.67 0.077 0.005 0.002 0.01** -0.0017 0.0004 0.001** -0.02 0.03 0.530

225

Appendix A.5.3. Sample sizes by species, period and assemblage used in the analyses of bulk isotope data (δ13C and δ15N), trophic parameters (%SPOM and trophic level) and compound specific isotopic analysis of amino acids (CSIA-AA) in chapter 5

Species Period Assemblage All specimens Bulk isotopes Trophic parameters CSIA-AA Leatherjacket Historical Inner shelf 23 23 23 - Blue cod 12 12 12 5 Gurnard 20 17 17 - Elephant fish 15 15 13 - Common warehou Outer shelf 8 8 6 - Barracouta 8 8 8 6 Tarakihi 21 20 20 10 Spiny dogfish 15 12 12 - Giant stargazer 16 10 5 - Red cod 24 10 10 7 Sea perch Slope 32 23 23 - Lookdown dory 18 6 6 - Hoki 11 11 11 6 Hapuka 8 - - 6 Ling 20 20 11 7 Orange roughy Mid slope 18 8 8 5 Leatherjacket Modern Inner shelf 26 26 26 - Blue cod 58 58 58 6 Gurnard 25 20 20 - Elephant fish 5 5 5 - Common warehou Outer shelf 11 11 11 - Barracouta 15 15 15 5 Tarakihi 62 48 48 5 Spiny dogfish 17 16 16 - Giant stargazer 17 17 17 - Red cod 42 33 33 5 Sea perch Slope 39 33 33 - Lookdown dory 12 9 9 - Hoki 7 7 7 4 Hapuka 12 - - 8 Ling 27 27 15 5 Orange roughy Mid slope 14 12 12 5

226

Appendix A.6.1. Distribution of annually averaged commercial catch (kg/ha) from all species and fishing methods between 2007 and 2017 reported to Fisheries New Zealand. Values were combined in ten catch intensity groups, with values from 1 (low intensity) to 10 (high intensity). A detailed description of the methods for the mapping of this dataset can be found in appendix A.6.2. Geographic limit of the east coast of the South Island trawl survey is depicted as the white dashed line. Source: Fisheries New Zealand

MPI Catch Intensity map (2007 to

227

Appendix A.6.2. Details on the creation of the Ministry of Primary Industries (MPI) fishing intensity map between 2007 and 2017, extracted from the metadata of the Geodatabase Raster Dataset

Tags

Fisheries New Zealand, MPI, Commercial, Fishing, Catch, Ministry, Primary, Industries

Summary

Annually averaged commercial catch (kg/ha) from all fishing methods (except freshwater fishing) reported to the Ministry for Primary Industries (MPI) over ten years.

Description

The distribution of total commercial catch is estimated for all fishing events reported in statutory catch and effort returns for the period 1 October 2007 to 30 September 2017.The location of fishing events is reported by either start (or start and end) coordinates (precise to 1 nautical mile) or by large statistical areas. The total catch of all species from each fishing event is spread uniformly over a polygon of space estimated to be occupied by that fishing. Trawl fishing polygons are derived from the length and width of the door-spread for the duration of the tow. The path of each tow is taken as a straight line between start and end coordinates where these are reported, or between start and estimated end coordinates. Where not required to report end coordinates, (as is the case for most inshore trawling) tow end points are derived using the direction of the next tow start position or the direction of the landing point for the last tow of the day. This has proven to be a reasonably good predictor of trawl direction. Line fishing is attributed to a circle with the center at the reported start position and a radius of the reported length of line set. Set net fishing is attributed to a circle with the center at the reported start position and radius of 2 nm in accordance with the definition of a single set netting event prescribed in reporting regulations. Jig fishing reports a single nightly position and is assumed to occur within 5 nm of that position. Hand and Pot fishing reports by statistical area, and where available, information on habitat and depth or information supplied by fishers is used to define the parts of each statistical area where each

228 type of fishing is likely to have occurred. In the case of lobster potting and paua diving, an informal map of reef area supplied by the Department of Conservation is used to estimate where this fishing may have occurred. Catch intensity (kg/ha) is mapped to a square kilometer grid for all fishing events and summed over gear types. The data is aggregated into grid squares of between 1 and 2500 km2 as required to give 9-year annual averages of data from at least three permit holders. Catch per unit area values are classified into ten intensity classes and displayed using heatmap colors. MPI has high confidence in the data on catch quantities used to create this dataset but the spatial distributions of those catches are only approximate and should be used with caution especially at large map scales (maps of small spatial extent). Nevertheless, the aggregation of a large number of fishing events tends to provide consistent patterns that have passed scrutiny when tested with groups of fishers.

Credits

Ministry for Primary Industries (MPI) - Manatū Ahu Matua

Use limitations

Data was created to provide a generalized context and is not guaranteed to be 100% accurate. All values and locations displayed within these datasets are based on the best available information at the time of creation. Where accurate location information is lacking, fishing events have been aggregated to broader areas such as a fishery statistical area, or where an informed assumption can be made, to a subset of a statistical area (such as rocky areas within a statistical area for paua fishing events).See Legal Constraints for any licenses attached to this dataset.

Extent

West -180.000000 East 180.000000

North -25.542249 South -56.077018

Scale Range

Maximum (zoomed in) 1:5,000

Minimum (zoomed out) 1:150,000,000

229

Appendix A.6.3. Life history traits of species used in the present study. Predominantly species of RATs: two saddle rattail (Coelorinchus biclinozonalis), Oblique banded rattail (Coelorinchus aspercephalus) and Bollons rattail (Coelorinchus bollonsi). Source: Fishbase.org

Maximum Maximum Range of preferred Mean preferred Trophic Code Common name Scientific name length (cm) age (years) temperature (˚C) temperature (˚C) level BAR Barracouta Thyrsites atun 200 10 9.8 11.2 3.85 NMP Tarakihi Nemadactylus macropterus 70 50 4.9 14 3.41 GIZ Giant Stargazer Kathetostoma giganteum 90 20 6.5 11.2 4.7 GUR Gurnard Chelidonichthys kumu 60 15 11.7 19.3 3.68 SPD Spiny dogfish Squalus acanthias 95 26 14.5 9.9 4.33 RAT Rattails ------GSH Ghost shark Hydrolagus novaezealandiae 96 - 5.7 10.8 3.52 SPE Sea perch Helicolenus percoides 47 42 6.1 13 4.01 ELE Elephant fish Callorhinchus milii 125 15 5.9 15.2 3.6 RCO Red cod Pseudophycis bachus 90 6 5.6 11 4.5 LIN Ling Genypterus blacodes 200 30 9.4 7.2 4.2 JMM Yellowtail jack mackerel Trachurus novaezelandiae 50 25 8.1 15.3 3.2 JMD Greenback jack mackerel Trachurus declivis 64 25 9.1 14.4 3.9 JMN Slender jack mackerel Trachurus murphyi 70 16 8.5 16.5 3.3 BCO Blue cod Parapercis colias 45 17 8 12.9 3.82 CBE Crested bellowsfish Notopogon lilliei 27 - 6.6 11.7 3.5 HAP Hapuka Polyprion oxygeneios 160 60 7.4 11.7 4.5 LSO Lemon sole Pelotretis flavilatus 40 - 7.5 13.3 3.3 SPO Rig Mustelus lenticulatus 150 20 9.7 13.7 3.49 JAV Javelinfish Thorntooth grenadier 55 - 5.6 9.2 3.8 ESO New Zealand sole Peltorhamphus novaezeelandiae 55 7 7 16.6 3.1 SSI Silverside Argentina elongata 25 - 7 11.2 3.4 HOK Hoki Macruronus novaezelandiae 120 25 6.7 8 4.34 WAR Common warehou Seriolella brama 76 15 4.8 14.4 3.51 LEA Leatherjacket Meuschenia scaber 31 - 4.3 15.1 3.04 CAR Carpet shark Cephaloscyllium isabellum 100 - 7 11.8 4.2 SSK Smooth skate Dipturus innominata 240 24 6.5 8.2 4.1 RSK Rough skate Zearaja nasuta 118 9 5.2 9.2 3.68 NOS Arrow squid Doryteuthis plei - - - - - SWA Silver warehou Seriolella punctata 66 15 5.7 13 3.4 WIT Witch Arnoglossus scapha 28.1 - 7.9 13.3 3.7 SCH School shark Galeorhinus galeus 193 55 16.5 12.3 4.26 SCG Scaly gurnard Lepidotrigla brachyoptera 14 - 8 13.5 3.4

230

References

Abrantes, K. G., Barnett, A., & Bouillon, S. (2014). Stable isotope-based community metrics as a tool to identify patterns in food web structure in east African estuaries. Functional Ecology, 28(1), 270–282. https://doi.org/10.1111/1365-2435.12155 Adams, S. M., McLean, R. B., & Huffman, M. M. (1982). Structuring of a Predator Population Through Temperature-Mediated Effects on Prey Availability. Canadian Journal of Fisheries and Aquatic Sciences, 39(8), 1175–1184. https://doi.org/10.1139/f82-155 Akaike, H. (1973). Information theory and an extension of the maximum likelihood principle. In B. N. Petrov & F. Csak (Eds.), Second international symposium on information theory (pp. 267–281). Akademai Kiado. Akın, Ş., Şahin, C., Turan, D., Gözler, A. M., Verep, B., Bozkurt, A., Çelik, K., Çetin, E., & Aracı, A. (2011). Effects of preservation methods on the δ13C and δ15N signatures in muscle tissues of two freshwater fish species. Fresenius Environmental Bulletin, 20(9a), 2419–2426. Alberts, B., Johnson, A., Lewis, J., Morgan, D., Raff, M., Roberts, K., Walter, P., Wilson, J., & Hunt, T. (2015). Molecular biology of the cell (Sixth edit). Garland Science, Taylor and Francis Group. Alder, J., Cullis-Suzuki, S., Karpouzi, V., Kaschner, K., Mondoux, S., Swartz, W., Trujillo, P., Watson, R., & Pauly, D. (2010). Aggregate performance in managing marine ecosystems of 53 maritime countries. Marine Policy, 34(3), 468–476. https://doi.org/10.1016/j.marpol.2009.10.001 Alestra, T., Gerrity, S., Dunmore, R., Marsden, I., Pirker, J., & Schiel, D. (2019). Rocky reef impacts of the Kaikōura earthquake: quantification and monitoring of nearshore habitats and communities. New Zealand Aquatic Environment and Biodiversity Report No. 212. Alonso, M. K., Crespo, E. A., García, N. A., Pedraza, S. N., Mariotti, P. A., & Mora, N. J. (2002). Fishery and ontogenetic driven changes in the diet of the spiny dogfish, Squalus acanthias, in Patagonian waters, Argentina. Environmental Biology of Fishes, 63(1997), 193–202. https://doi.org/10.1023/A:1014229432375 Altabet, M. A. (1988). Variations in nitrogen isotopic composition between sinking and suspended particles: implications for nitrogen cycling and particle transformation in the open ocean. Deep Sea Research Part A. Oceanographic Research Papers, 35(4), 535– 554. https://doi.org/10.1016/0198-0149(88)90130-6

Altabet, M. A. (2001). Nitrogen isotopic evidence for micronutrient control of fractional NO3 − utilization in the equatorial Pacific. Limnology and Oceanography, 46(2), 368–380. https://doi.org/10.4319/lo.2001.46.2.0368 Amundsen, P. A. (1995). Feeding strategy of Arctic charr (Salvelinus alpinus): general opportunist, but individual specialist. Nordic Journal of Freshwater Research, 71, 150– 156. Anderson, M., Gorley, R. N., & Clarke, K. R. (2008). PERMANOVA + for PRIMER: Guide to Software and Statistical Methods (Vol. 1, p. 1:218). PRIMERr-e Ltd. Anderson, M. J. (2001). A new method for non-parametric multivariate analysis of variance. Austral Ecology, 26(1), 32–46. https://doi.org/10.1111/j.1442-9993.2001.01070.pp.x

231

Anderson, O. F., Bagley, N. ., Hurst, R. J., Francis, M. P., Clark, M. R., & McMillan, P. . (1998). Atlas of New Zealand Fish and Squid Distributions From Research Bottom Trawls. NIWA Technical Report 42. Andrews, A. H., & Tracey, D. M. (2003). Age validation of deepwater fish species, with particular reference to New Zealand orange roughy, black oreo, smooth oreo, and black cardinalfish. Final research report for Ministry of Fisheries Research Project DEE2000/02. Annala, J. H. (1988). New Zealand Fisheries Assessment Research Document 88/28. Archer, W., Braun, D. R., Harris, J. W. K., McCoy, J. T., & Richmond, B. G. (2014). Early Pleistocene aquatic resource use in the Turkana Basin. Journal of Human Evolution, 77, 74–87. https://doi.org/10.1016/j.jhevol.2014.02.012 Arkema, K., Abramson, S. C., & Dewsbury, B. M. (2014). Marine ecosystem-based management : from characterization to implementation. Frontiers in Ecology and the Environment, 4(10), 525–532. https://doi.org/10.1890/1540- 9295(2006)4[525:MEMFCT]2.0.CO;2 Arrington, D. A., & Winemiller, K. O. (2002). Preservation effects on stable isotope analysis of fish muscle. Transactions of the American Fisheries Society, 131(2), 337–342. https://doi.org/10.1577/1548-8659(2002)131<0337:PEOSIA>2.0.CO;2 Auber, A., Travers-Trolet, M., Villanueva, M. C., & Ernande, B. (2015). Regime Shift in an Exploited Fish Community Related to Natural Climate Oscillations. PLOS ONE, 10(7), 1–18. https://doi.org/10.1371/journal.pone.0129883 Audzijonyte, A., Kuparinen, A., Gorton, R., & Fulton, E. A. (2013). Ecological consequences of body size decline in harvested fish species: positive feedback loops in trophic interactions amplify human impact. Biology Letters, 9(2), 1–5. https://doi.org/10.1098/rsbl.2012.1103 Augé, A., Lalas, C., Davis, L., & Chilvers, B. (2012). Autumn diet of recolonising female New Zealand sea lions based at Otago Peninsula, South Island, New Zealand. New Zealand Journal of Marine and Freshwater Research, 46(1), 97–110. https://doi.org/10.1080/00288330.2011.606326 Bagley, N. W., Ballara, S. L., Horn, P. L., & Hurst, R. J. (1998). A summary of commercial landings and a validated ageing method for blue warehou, Seriolella brama (Centrolophidae), in New Zealand waters, and a stock assessment of the Southern (WAR 3) Fishstock. New Zealand Fisheries Assessment Research Document . Bailey, H., & Secor, D. H. (2016). Coastal evacuations by fish during extreme weather events. Scientific Reports, 6(1), 1–9. https://doi.org/10.1038/srep30280 Baird, S. J., Wood, B. A., & Bagley, N. W. (2011). Nature and extent of commercial fishing effort on or near the seafloor within the New Zealand 200 n. mile Exclusive Economic Zone, 1989–90 to 2004–05. New Zealand Aquatic Environment and Biodiversity Report No. 73. https://fs.fish.govt.nz/Doc/22814/AEBR_73.pdf.ashx Baisden, W. T., Keller, E. D., Van Hale, R., Frew, R. D., & Wassenaar, L. I. (2016). Precipitation isoscapes for New Zealand: enhanced temporal detail using precipitation- weighted daily climatology. Isotopes in Environmental and Health Studies, 52(4–5), 343–352. https://doi.org/10.1080/10256016.2016.1153472 Baker, D. M., Webster, K. L., & Kim, K. (2010). Caribbean octocorals record changing carbon and nitrogen sources from 1862 to 2005. Global Change Biology, 16(10), 2701– 2710. https://doi.org/10.1111/j.1365-2486.2010.02167.x

232

Bargh, B. (2016). The struggle for Maori FIshing Rights: Te Ika a Māori. Huia Publishers. Barnes, C., Jennings, S., & Barry, J. T. (2009). Environmental correlates of large-scale spatial variation in the δ13C of marine animals. Estuarine, Coastal and Shelf Science, 81(3), 368–374. https://doi.org/10.1016/j.ecss.2008.11.011 Barto, K. (2015). MuMIn: multi-model inference (Version 1.13.4). R Centre for Statistical Computing. Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting Linear Mixed-Effects Models Using lme4. Journal of Statistical Software, 67(1), 1–48. https://doi.org/10.18637/jss.v067.i01 Battista, W., Kelly, R. P., Erickson, A., & Fujita, R. (2018). Fisheries Governance Affecting Conservation Outcomes in the United States and European Union. Coastal Management, 46(5), 388–452. https://doi.org/10.1080/08920753.2018.1498711 Bearhop, S., Adams, C. E., Waldron, S., Fuller, R. A., & Macleod, H. (2004). Determining trophic niche width: a novel approach using stable isotope analysis. Journal of Animal Ecology, 73(5), 1007–1012. https://doi.org/10.1111/j.0021-8790.2004.00861.x Beentjes, M. P., Bull, B., Hurst, R. J., & Bagley, N. W. (2002). Demersal fish assemblages along the continental shelf and upper slope of the east coast of the South Island, New Zealand. New Zealand Journal of Marine and Freshwater Research, 36(1), 197–223. https://doi.org/10.1080/00288330.2002.9517080 Beentjes, M. P., & Page, M. (2018). Relative abundance, size and age structure, and stock status of blue cod off Kaikoura in 2017. New Zealand Fisheries Assessment Report 2018/37. Ministry for Primary Industries. Beentjes, M. P., & Renwick, J. A. (2001). The relationship between red cod, Pseudophycis bachus, recruitment and environmental variables in New Zealand. Environmental Biology of Fishes, 61(3), 315–328. https://doi.org/10.1023/A:1010943906264 Beentjes, M. P., & Stevenson, M. L. (2001). Review of east coast South Island summer trawl survey time series, 1996–97 to 1999–2000. NIWA Technical Report 108. Begossi, A., & Caires, R. (2015). Art, fisheries and ethnobiology. Journal of Ethnobiology and Ethnomedicine, 11(1), 1–8. https://doi.org/10.1186/1746-4269-11-17 Bentley, N., Kendrick, T. H., & MacGibbon, D. J. (2014). Fishery characterisation and catch-per-unit-effort analyses for sea perch (Helicolenus spp.) in New Zealand, 1989-90 to 2009-10. New Zealand Fisheries Assessment Report 2014/27. Bess, R. (2005). Expanding New Zealand’s quota management system. Marine Policy, 29(4), 339–347. https://doi.org/10.1016/j.marpol.2004.04.008 Best, E. (2005). Fishing Methods and Devices of the Maori. Te Papa Press. Beukhof, E., Dencker, T., Pecuchet, L., & Lindegren, M. (2019). Spatio-temporal variation in marine fish traits reveals community-wide responses to environmental change. Marine Ecology Progress Series, 610, 205–222. https://doi.org/10.3354/meps12826 Beverton, R. J. H., & Holt, S. J. (1993). On the Dynamics of Exploited Fish Populations (Third edit). Springer-Science+Business Media B.V. Bigg, G. R., & Rohling, E. J. (2000). An oxygen isotope data set for marine waters. Journal of Geophysical Research: Oceans, 105(C4), 8527–8535. https://doi.org/10.1029/2000JC900005 Bishop, J. (2006). Standardizing fishery-dependent catch and effort data in complex fisheries with technology change. Reviews in Fish Biology and Fisheries, 16(1), 21–38. 233

https://doi.org/10.1007/s11160-006-0004-9 Blaber, S. J. M., & Bulman, C. M. (1987). Diets of fishes of the upper continental slope of eastern Tasmania: content, calorific values, dietary overlap and trophic relationships. Marine Biology, 95(3), 345–356. https://doi.org/10.1007/BF00409564 Blanchard, J. L., Law, R., Castle, M. D., & Jennings, S. (2011). Coupled energy pathways and the resilience of size-structured food webs. Theoretical Ecology, 4(3), 289–300. https://doi.org/10.1007/s12080-010-0078-9 Blum, F. (1893). Der formaldehyd als antisepticum. Munch Med Wochensch, 8, 601. Bolnick, D. I., Svanbäck, R., Fordyce, J. A., Yang, L. H., Davis, J. M., Hulsey, C. D., & Forister, M. L. (2003). The ecology of individuals: Incidence and implications of individual specialization. American Naturalist, 161(1), 1–28. https://doi.org/10.1086/343878 Bosley, K. L., & Wainright, S. C. (1999). Effects of preservatives and acidification on the stable isotope ratios (15N:14N, 13C:12C) of two species of marine animals. Canadian Journal of Fisheries and Aquatic Sciences, 56(11), 2181–2185. https://doi.org/10.1139/f99-153 Bowes, R. E., Thorp, J. H., & Reuman, D. C. (2017). Multidimensional metrics of niche space for use with diverse analytical techniques. Scientific Reports, 7(1), 1–11. https://doi.org/10.1038/srep41599 Boyd, P., LaRoche, J., Gall, M., Frew, R., & McKay, R. M. L. (1999). Role of iron, light, and silicate in controlling algal biomass in subantarctic waters SE of New Zealand. Journal of Geophysical Research: Oceans, 104(C6), 13395–13408. https://doi.org/10.1029/1999JC900009 Bradford-Grieve, J., Livingston, M., Sutton, P., & Hadfield, M. (2007). Ocean variability and declining hoki stocks: an hypothesis as yet untested. New Zealand Science Review, 63(3– 4), 76–81. http://www.seaaroundus.org/NewsletterF.htm, Bradford, J. M., & Roberts, P. E. (1978). Distribution of reactive phosphorus and plankton in relation to upwelling and surface circulation around New Zealand. New Zealand Journal of Marine and Freshwater Research, 12(1), 1–15. https://doi.org/10.1080/00288330.1978.9515717 Bradley, C. J., Wallsgrove, N. J., Choy, C. A., Drazen, J. C., Hetherington, E. D., Hoen, D. K., & Popp, B. N. (2015). Trophic position estimates of marine teleosts using amino acid compound specific isotopic analysis. Limnology and Oceanography: Methods, 13(9), 476–493. https://doi.org/10.1002/lom3.10041 Branch, T. A., Watson, R., Fulton, E. A., Jennings, S., McGilliard, C. R., Pablico, G. T., Ricard, D., & Tracey, S. R. (2010). The trophic fingerprint of marine fisheries. Nature, 468(7322), 431–435. https://doi.org/10.1038/nature09528 Bray, J. R., & Curtis, J. T. (1957). An ordination of the upland forest communities of Southern Wisconsin. Ecological Monographs, 27(4), 325–349. https://doi.org/10.2307/1942268 Brazner, J. C., Tanner, D. K., & Morrice, J. A. (2011). Fish-Mediated Nutrient and Energy Exchange between a Lake Superior Coastal Wetland and its Adjacent Bay. Journal of Great Lakes Research, 27(1), 98–111. https://doi.org/10.1016/S0380-1330(01)70625-9 Brown, C. J., Fulton, E. A., Hobday, A. J., Matear, R. J., Possingham, H. P., Bulman, C., Christensen, V., Forrest, R. E., Gehrke, P. C., Gribble, N. A., Griffiths, S. P., Lozano- montes, H., Martin, J. M., Metcalf, S., Okey, T. A., Watson, R., & Richardson, A. J.

234

(2010). Effects of climate-driven primary production change on marine food webs: implications for fisheries and conservation. Global Change Biology, 16(4), 1194–1212. https://doi.org/10.1111/j.1365-2486.2009.02046.x Bugoni, L., McGill, R. A. R., & Furness, R. W. (2008). Effects of preservation methods on stable isotope signatures in bird tissues. Rapid Communications in Mass Spectrometry, 22(16), 2457–2462. https://doi.org/10.1002/rcm.3633 Bull, B., & Livingston, M. E. (2001). Links between climate variation and year class strength of New Zealand hoki (Macruronus novaezelandiae) : An update. New Zealand Journal of Marine and Freshwater Research, 35(5), 871–880. https://doi.org/10.1080/00288330.2001.9517049 Bulman, C., & Blaber, S. (1986). Feeding ecology of Macruronus novaezelandiae (Hector) (Teleostei : Merlucciidae) in south-eastern Australia. Marine and Freshwater Research, 37(5), 621. https://doi.org/10.1071/MF9860621 Bulman, C., & Koslow, J. (1992). Diet and food consumption of a deep-sea fish orange roughy Hoplostethus atlanticus (Pisces Trachichthyidae), off southeastern Australia. Marine Ecology Progress Series, 82(June), 115–129. https://doi.org/10.3354/meps082115 Bundy, A., & Fanning, L. P. (2005). Can Atlantic cod (Gadus morhua) recover? Exploring trophic explanations for the non-recovery of the cod stock on the eastern Scotian Shelf, Canada. Canadian Journal of Fisheries and Aquatic Sciences, 62(7), 1474–1489. https://doi.org/10.1139/f05-086 Burgess, K. B., & Bennett, M. B. (2017). Effects of ethanol storage and lipid and urea extraction on δ15N and δ13C isotope ratios in a benthic elasmobranch, the bluespotted maskray Neotrygon kuhlii. Journal of Fish Biology, 90(1), 417–423. https://doi.org/10.1111/jfb.13164 Butler, E. C. V., Butt, J. A., Lindstrom, E. J., Teldesley, P. C., Pickmere, S., & Vincent, W. F. (1992). Oceanography of the Subtropical Convergence Zone around southern New Zealand. New Zealand Journal of Marine and Freshwater Research, 26(2), 131–154. https://doi.org/10.1080/00288330.1992.9516509 Caddy, J. F. (1998). How Pervasive is “Fishing Down Marine Food Webs”? Science, 282(5393), 1383a. https://doi.org/10.1126/science.282.5393.1383a Caddy, J. F., & Garibaldi, L. (2000). Apparent changes in the trophic composition of world marine harvests: the perspective from the FAO capture database. Ocean & Coastal Management, 43, 615–655. https://doi.org/10.1016/S0964-5691(00)00052-1 Caldwell, L. K. (1970). The Ecosystem as a Criterion for Public Land Policy. Natural Resources Journal, 10(2), 203–221. https://digitalrepository.unm.edu/nrj/vol10/iss2/1 Canals, M., Puig, P., de Madron, X. D., Heussner, S., Palanques, A., & Fabres, J. (2006). Flushing submarine canyons. Nature, 444(7117), 354–357. https://doi.org/10.1038/nature05271 Carbines, G., & Mckenzie, J. (2001). Movement patterns and stock mixing of blue cod in Southland (BC05). Ministry of Fisheries Research Project BCO9702. Cardinale, M., Bartolino, V., Svedäng, H., Sundelöf, A., Poulsen, R. T., & Casini, M. (2015). A centurial development of the North Sea fish megafauna as reflected by the historical Swedish longlining fisheries. Fish and Fisheries, 16(3), 522–533. https://doi.org/10.1111/faf.12074 Carrascal, A. (2016). Digitize it (Version 4.2.0). EZT wain v. 1.12.

235

Casini, M., Bartolino, V., Molinero, J., & Kornilovs, G. (2010). Linking fisheries, trophic interactions and climate: threshold dynamics drive herring Clupea harengus growth in the central Baltic Sea. Marine Ecology Progress Series, 413, 241–252. https://doi.org/10.3354/meps08592 Casini, M., Lövgren, J., Hjelm, J., Cardinale, M., Molinero, J.-C., & Kornilovs, G. (2008). Multi-level trophic cascades in a heavily exploited open marine ecosystem. Proceedings of the Royal Society B: Biological Sciences, 275(1644), 1793–1801. https://doi.org/10.1098/rspb.2007.1752 Chapman, M. G., Underwood, A. J., & Skilleter, G. A. (1995). Variability at different spatial scales between a subtidal assemblage exposed to the discharge of sewage and two control assemblages. Journal of Experimental Marine Biology and Ecology, 189(1–2), 103–122. https://doi.org/10.1016/0022-0981(95)00017-L Chen, H., Yin, K., Wang, H., Zhong, S., Wu, N., Shi, F., Zhu, D., Zhu, Q., Wang, W., Ma, Z., Fang, X., Li, W., Zhao, P., & Peng, C. (2011). Detecting one-hundred-year environmental changes in western China using seven-year repeat photography. PLoS ONE, 6(9), 1–13. https://doi.org/10.1371/journal.pone.0025008 Cherel, Y., Corre, M., Jaquemet, S., Ménard, F., Richard, P., & Weimerskirch, H. (2008). Resource partitioning within a tropical seabird community: new information from stable isotopes. Marine Ecology Progress Series, 366, 281–291. https://doi.org/10.3354/meps07587 Cherel, Y., & Hobson, K. (2007). Geographical variation in carbon stable isotope signatures of marine predators: a tool to investigate their foraging areas in the Southern Ocean. Marine Ecology Progress Series, 329, 281–287. https://doi.org/10.3354/meps329281 Chikaraishi, Y., Kashiyama, Y., Ogawa, N., Kitazato, H., & Ohkouchi, N. (2007). Metabolic control of nitrogen isotope composition of amino acids in macroalgae and gastropods: implications for aquatic food web studies. Marine Ecology Progress Series, 342, 85–90. https://doi.org/10.3354/meps342085 Chikaraishi, Y., Ogawa, N. O., Kashiyama, Y., Takano, Y., Suga, H., Tomitani, A., Miyashita, H., Kitazato, H., & Ohkouchi, N. (2009). Determination of aquatic food-web structure based on compound-specific nitrogen isotopic composition of amino acids. Limnology and Oceanography: Methods, 7(11), 740–750. https://doi.org/10.4319/lom.2009.7.740 Chiswell, S. M., Bostock, H. C., Sutton, P. J., & Williams, M. J. (2015). Physical oceanography of the deep seas around New Zealand: a review. New Zealand Journal of Marine and Freshwater Research, 49(2), 286–317. https://doi.org/10.1080/00288330.2014.992918 Chiswell, S. M., & Schiel, D. R. (2001). Influence of along‐shore advection and upwelling on coastal temperature at Kaikoura Peninsula, New Zealand. New Zealand Journal of Marine and Freshwater Research, 35(2), 307–317. https://doi.org/10.1080/00288330.2001.9517000 Chiswell, S. M., & Sutton, P. J. H. (2020). Relationships between long-term ocean warming, marine heat waves and primary production in the New Zealand region. New Zealand Journal of Marine and Freshwater Research, 1–22. https://doi.org/10.1080/00288330.2020.1713181 Chrysafi, A., & Kuparinen, A. (2016). Assessing abundance of populations with limited data: Lessons learned from data-poor fisheries stock assessment. Environmental Reviews, 24(1), 25–38. https://doi.org/10.1139/er-2015-0044

236

Clark, I. N., Major, P. J., & Mollet, N. (1988). Development and Implementation of New Zealand’s ITQ Management System. Marine Resource Economics, 5(4), 325–349. https://doi.org/10.1086/mre.5.4.42628934 Clark, M. R. (1985). Feeding relationships of seven fish species from the Campbell Plateau, New Zealand. New Zealand Journal of Marine and Freshwater Research, 19(3), 365– 374. https://doi.org/10.1080/00288330.1985.9516101 Clark, M. R., Althaus, F., Schlacher, T. A., Williams, A., Bowden, D. A., & Rowden, A. A. (2016). The impacts of deep-sea fisheries on benthic communities: a review. ICES Journal of Marine Science, 73(suppl_1), 51–69. https://doi.org/10.1093/icesjms/fsv123 Clarke, K. R., & Gorley, R. N. (2006). PRIMER v6: User Manual/Tutorial (6.1.12). PRIMER-E. Clarke, K. R., Somerfield, P. J., & Chapman, M. G. (2006). On resemblance measures for ecological studies, including taxonomic dissimilarities and a zero-adjusted Bray-Curtis coefficient for denuded assemblages. Journal of Experimental Marine Biology and Ecology, 330(1), 55–80. https://doi.org/10.1016/j.jembe.2005.12.017 Cleveland, W. S., Grosse, E., & Shyu, W. M. (1992). Local regression models. In J. M. Chambers & T. J. Hastie (Eds.), Statistical Models in S. Wadsworth & Brooks/Cole. Cohen, D. M., Inada, T., Iwamoto, T., & Scialabba, N. (1990). FAO species catalogue. Gadiform fishes of the world (Order Gadiformes). In An annotated and illustrated catalogue of cods, hakes, grenadiers and other gadiform fishes known to date. (p. 442). FAO Fish. Synop. Coleman, N., & Mobley, M. (1984). Diets of commercially exploited fish from Bass Strait and adjacent Victorian Waters, south-eastern Australia. Marine and Freshwater Research, 35(5), 549–560. https://doi.org/10.1071/MF9840549 Cook, J. (1769). Cook’s Journal: Daily Entries. In Voyaging Accounts. National Library of Australia. http://southseas.nla.gov.au/journals/cook/17691007.html Cornelissen, I. J. M., Vijverberg, J., van den Beld, A. M., Helmsing, N. R., Verreth, J. A. J., & Nagelkerke, L. A. J. (2018). Heterogeneity in food-web interactions of fish in the Mwanza Gulf, Lake Victoria: a quantitative stable isotope study. Hydrobiologia, 805(1), 113–130. https://doi.org/10.1007/s10750-017-3297-x Correa, C. (2012). Tissue preservation biases in stable isotopes of fishes and molluscs from Patagonian lakes. Journal of Fish Biology, 81(6), 2064–2073. https://doi.org/10.1111/j.1095-8649.2012.03453.x Cortés, E. (1999). Standardized diet compositions and trophic levels of sharks. ICES Journal of Marine Science, 56(5), 707–717. https://doi.org/10.1006/jmsc.1999.0489 Costello, C., Ovando, D., Hilborn, R., Gaines, S. D., Deschenes, O., & Lester, S. E. (2012). Status and solutions for the world’s unassessed fisheries. Science, 338(6106), 517–520. https://doi.org/10.1126/science.1223389 Cotter, A. J. ., Course, G., Buckland, S. ., & Garrod, C. (2002). A PPS sample survey of English fishing vessels to estimate discarding and retention of North Sea cod, haddock, and whiting. Fisheries Research, 55(1–3), 25–35. https://doi.org/10.1016/S0165- 7836(01)00306-X Croot, P. L., & Hunter, K. A. (1998). Trace metal distributions across the continental shelf near Otago Peninsula, New Zealand. Marine Chemistry, 62(3–4), 185–201. https://doi.org/10.1016/S0304-4203(98)00036-X Crowder, L. B., Hazen, E. L., Avissar, N., Bjorkland, R., Latanich, C., & Ogburn, M. B. 237

(2008). The impacts of fisheries on marine ecosystems and the transition to ecosystem- based management. Annual Review of Ecology, Evolution, and Systematics, 39(1), 259– 278. https://doi.org/10.1146/annurev.ecolsys.39.110707.173406 Cryer, M., Mace, P. M., & Sullivan, K. J. (2016). New Zealand’s ecosystem approach to fisheries management. Fisheries Oceanography, 25(1), 57–70. https://doi.org/10.1111/fog.12088 Cury, P. M., Shannon, L. J., Roux, J.-P., Daskalov, G. M., Jarre, A., Moloney, C. L., & Pauly, D. (2005). Trophodynamic indicators for an ecosystem approach to fisheries. ICES Journal of Marine Science, 62(3), 430–442. https://doi.org/10.1016/j.icesjms.2004.12.006 Da Silva, R., Mazumdar, A., Mapder, T., Peketi, A., Joshi, R. K., Shaji, A., Mahalakshmi, P., Sawant, B., Naik, B. G., Carvalho, M. A., & Molletti, S. K. (2017). Salinity stratification controlled productivity variation over 300 ky in the Bay of Bengal. Scientific Reports, 7(1), 14439. https://doi.org/10.1038/s41598-017-14781-3 Dahl, K. (1907). The scales of the herring as a mean of determining age, growth and migration. Report on Nonekian Fishery and Marine-Investigations Vol. 11. No. 6. https://brage.bibsys.no/xmlui/bitstream/handle/11250/114860/sh_vol02_06_1907.pdf?se quence=1&isAllowed=y Dale, J., Wallsgrove, N., Popp, B., & Holland, K. (2011). Nursery habitat use and foraging ecology of the brown stingray Dasyatis lata determined from stomach contents, bulk and amino acid stable isotopes. Marine Ecology Progress Series, 433, 221–236. https://doi.org/10.3354/meps09171 Dalerum, F., & Angerbjörn, A. (2005). Resolving temporal variation in vertebrate diets using naturally occurring stable isotopes. Oecologia, 144(4), 647–658. https://doi.org/10.1007/s00442-005-0118-0 Dalponti, G., Guariento, R., & Caliman, A. (2018). Hunting high or low: body size drives trophic position among and within marine predators. Marine Ecology Progress Series, 597(April), 39–46. https://doi.org/10.3354/meps12584 Dalton, M., Holland, D. S., Squires, D., Terry, J., & Tomberlin, D. (2018). An Economic Perspective on National Standard 1. NOAA Technical Memorandum. NMFS-F/SPO-180. Dammhahn, M., & Kappeler, P. M. (2014). Stable isotope analyses reveal dense trophic species packing and clear niche differentiation in a malagasy primate community. American Journal of Physical Anthropology, 153(2), 249–259. https://doi.org/10.1002/ajpa.22426 Dansgaard, W. (1964). Stable isotopes in precipitation. Tellus, 16(4), 436–468. https://doi.org/10.1111/j.2153-3490.1964.tb00181.x Darimont, C. T., Carlson, S. M., Kinnison, M. T., Paquet, P. C., Reimchen, T. E., & Wilmers, C. C. (2009). Human predators outpace other agents of trait change in the wild. Proceedings of the National Academy of Sciences, 106(3), 952–954. https://doi.org/10.1073/pnas.0809235106 Davis, J. P., & Wing, S. R. (2012). Niche partitioning in the Fiordland wrasse guild. Marine Ecology Progress Series, 446, 207–220. https://doi.org/10.3354/meps09452 De Leo, F. C., Smith, C. R., Rowden, A. A., Bowden, D. A., & Clark, M. R. (2010). Submarine canyons: hotspots of benthic biomass and productivity in the deep sea. Proceedings of the Royal Society B: Biological Sciences, 277(1695), 2783–2792. https://doi.org/10.1098/rspb.2010.0462

238

Demirhan, S. A., & Seyhan, K. (2007). Life history of spiny dogfish, Squalus acanthias (L. 1758), in the southern Black Sea. Fisheries Research, 85(1–2), 210–216. https://doi.org/10.1016/j.fishres.2007.02.009 Deudero, S., Pinnegar, J. K., Polunin, N. V. C., Morey, G., & Morales-Nin, B. (2004). Spatial variation and ontogenic shifts in the isotopic composition of Mediterranean littoral fishes. Marine Biology, 145(5), 971–981. https://doi.org/10.1007/s00227-004-1374-y DiFiore, P. J., Sigman, D. M., & Dunbar, R. B. (2009). Upper ocean nitrogen fluxes in the Polar Antarctic Zone: Constraints from the nitrogen and oxygen isotopes of nitrate. Geochemistry, Geophysics, Geosystems, 10(11), 1–22. https://doi.org/10.1029/2009GC002468 Donázar-Aramendía, I., Sánchez-Moyano, J. E., García-Asencio, I., Miró, J. M., Megina, C., & García-Gómez, J. C. (2019). Human pressures on two estuaries of the Iberian Peninsula are reflected in food web structure. Scientific Reports, 9(1), 1–10. https://doi.org/10.1038/s41598-019-47793-2 Doney, S. C., Ruckelshaus, M., Emmett Duffy, J., Barry, J. P., Chan, F., English, C. A., Galindo, H. M., Grebmeier, J. M., Hollowed, A. B., Knowlton, N., Polovina, J., Rabalais, N. N., Sydeman, W. J., & Talley, L. D. (2012). Climate change impacts on marine ecosystems. Annual Review of Marine Science, 4(1), 11–37. https://doi.org/10.1146/annurev-marine-041911-111611 Dunn, M. R., Connell, A. M., Forman, J., Stevens, D. W., & Horn, P. L. (2010). Diet of Two Large Sympatric Teleosts, the Ling (Genypterus blacodes) and Hake (Merluccius australis). PLoS ONE, 5(10), 1–11. https://doi.org/10.1371/journal.pone.0013647 Durant, J., & Hjermann, D. (2017). Age-structure, harvesting and climate effects on population growth of Arcto-boreal fish stocks. Marine Ecology Progress Series, 577, 177–188. https://doi.org/10.3354/meps12210 Durante, L. M., Beentjes, M. P., & Wing, S. R. (2020a). Shifting trophic architecture of marine fisheries in New Zealand: Implications for guiding effective ecosystem‐based management. Fish and Fisheries, 21(4), 813–830. https://doi.org/10.1111/faf.12463 Durante, L. M., Cruz, I. C. S., & Lotufo, T. M. C. (2018). The effect of climate change on the distribution of a tropical zoanthid (Palythoa caribaeorum) and its ecological implications. PeerJ, 6, 1–26. https://doi.org/10.7717/peerj.4777 Durante, L. M., Sabadel, A. J. M., Frew, R. D., Ingram, T., & Wing, S. R. (2020b). Effects of fixatives on stable isotopes of fish muscle tissue: implications for trophic studies on preserved specimens. Ecological Applications, 30(4), 1–16. https://doi.org/10.1002/eap.2080 Dürr, H. H., Meybeck, M., Hartmann, J., Laruelle, G. G., & Roubeix, V. (2011). Global spatial distribution of natural riverine silica inputs to the coastal zone. Biogeosciences, 8(3), 597–620. https://doi.org/10.5194/bg-8-597-2011 Dziak, J. J., Coffman, D. L., Lanza, S. T., Li, R., & Dziak, J. (2012). Sensitivity and specificity of information criteria. https://methodology.psu.edu/media/techreports/12- 119.pdf Ebert, D. A., Compagno, L. J. V., & Cowley, P. D. (1992). A preliminary investigation of the feeding ecology of squaloid sharks off the west coast of southern Africa. South African Journal of Marine Science, 12(1), 601–609. https://doi.org/10.2989/02577619209504727 Eddy, T. D., Araújo, J. N., Bundy, A., Fulton, E. A., & Lotze, H. K. (2016). Effectiveness of lobster fisheries management in New Zealand and Nova Scotia from multi-species and ecosystem perspectives. ICES Journal of Marine Science: Journal Du Conseil, 74(1), 239

146–157. https://doi.org/10.1093/icesjms/fsw127 Edgar, G. J., & Shaw, C. (1995). The production and trophic ecology of shallow-water fish assemblages in southern Australia II. Diets of fishes and trophic relationships between fishes and benthos at Western Port, Victoria. Journal of Experimental Marine Biology and Ecology, 194(1), 83–106. https://doi.org/10.1016/0022-0981(95)00084-4 Edwards, M., Beaugrand, G., Hays, G. C., Koslow, J. A., & Richardson, A. J. (2010). Multi- decadal oceanic ecological datasets and their application in marine policy and management. Trends in Ecology & Evolution, 25(10), 602–610. https://doi.org/10.1016/j.tree.2010.07.007 Edwards, M. S., Turner, T. F., & Sharp, Z. D. (2002). Short- and long-term effects of fixation and preservation on stable isotope values (δ13C, δ15N, δ34S) of fluid-preserved museum specimens. Copeia, 2002(4), 1106–1112. https://doi.org/10.1643/0045- 8511(2002)002[1106:SALTEO]2.0.CO Eide, M., Olsen, A., Ninnemann, U. S., & Eldevik, T. (2017). A global estimate of the full oceanic 13C Suess effect since the preindustrial. Global Biogeochemical Cycles, 31(3), 492–514. https://doi.org/10.1002/2016GB005472 Ellis, J. R., Pawson, M. G., & Shackley, S. E. (1996). The comparative feeding ecology of six species of shark and four species of ray (Elasmobranchii) in the North-East Atlantic. Journal of the Marine Biological Association of the United Kingdom, 76(1), 89–106. https://doi.org/10.1017/S0025315400029039 Elton, C. S. (1927). Animal ecology. Sidgwick and Jackson. Epstein, S., & Mayeda, T. (1953). Variation of O18 content of waters from natural sources. Geochimica et Cosmochimica Acta, 4(5), 213–224. https://doi.org/10.1016/0016- 7037(53)90051-9 Espinasse, B., Pakhomov, E., Hunt, B., & Bury, S. (2019). Latitudinal gradient consistency in carbon and nitrogen stable isotopes of particulate organic matter in the Southern Ocean. Marine Ecology Progress Series, 631, 19–30. https://doi.org/10.3354/meps13137 Estes, J. A. (1998). Killer whale predation on sea otters linking oceanic and nearshore ecosystems. Science, 282(5388), 473–476. https://doi.org/10.1126/science.282.5388.473 Fagan, K.-A., Koops, M. A., Arts, M. T., & Power, M. (2011). Assessing the utility of C:N ratios for predicting lipid content in fishes. Canadian Journal of Fisheries and Aquatic Sciences, 68(2), 374–385. https://doi.org/10.1139/F10-119 Fanelli, E., & Cartes, J. (2010). Temporal variations in the feeding habits and trophic levels of three deep-sea demersal fishes from the western Mediterranean Sea, based on stomach contents and stable isotope analyses. Marine Ecology Progress Series, 402, 213–232. https://doi.org/10.3354/meps08421 Fantle, M. S., Dittel, A. I., Schwalm, S. M., Epifanio, C. E., & Fogel, M. L. (1999). A food web analysis of the juvenile blue crab, Callinectes sapidus, using stable isotopes in whole animals and individual amino acids. Oecologia, 120(3), 416–426. https://doi.org/10.1007/s004420050874 FAO. (1994). National Research Council Nutrient Requirements of Poultry – Ninth Revised Edition (1994). Journal of Applied Poultry Research, 3(1), 101. https://doi.org/10.1093/japr/3.1.101 FAO. (2008). Fisheries management. 2. The ecosystem approach to fisheries. 2.1 Best practices in ecosystem modelling for informing an ecosystem approach to fisheries (p. 78p). FAO. http://www.fao.org/3/i0151e/i0151e00.pdf

240

Fenaughty, J. M., & Bagley, N. M. (1981). WJ Scott New Zealand Trawling Survey - South Island East Coast. Technical Report 157. Fenberg, P. B., & Roy, K. (2008). Ecological and evolutionary consequences of size-selective harvesting: how much do we know? Molecular Ecology, 17(1), 209–220. https://doi.org/10.1111/j.1365-294X.2007.03522.x Fiedler, F. N., Port, D., Giffoni, B. B., Sales, G., & Fisch, F. (2017). Pelagic longline fisheries in southeastern/south Brazil. Who cares about the law? Marine Policy, 77, 56–64. https://doi.org/10.1016/j.marpol.2016.12.011 Fisher, F. (1913). Marine Department: Annual Report for 1912-1913 (Vol. 53). https://doi.org/10.1017/CBO9781107415324.004 Fisheries New Zealand. (2016). Region - South-East Coast (FMA 3). Fisheries Infosite. https://fs.fish.govt.nz/Page.aspx?pk=41&fyk=39 Fisheries New Zealand. (2019). Fisheries Assessment Plenary, May 2019: stock assessments and stock status. www.fisheries.govt.nz/news-and-resources/science-and- research/fisheries-research/ Fluharty, D., Aparicio, P., Blackburn Alaska Groundfish Boehlert, G., Coleman, F., Conkling, P., Costanza, R., Dayton, P., Francis, R., Hanan, D., Hinman, K., Houde, E., Kitchell, J., Langton, R., Lubchenco, J., Mangel, M., Nelson, R., O’Connel, V., Orbach, M., Sissenwine, M., … Morgan, A. (1999). Ecosystem-Based Fishery Management. A report to Congress by the Ecossystem Principles Advisoty Panel. http://www.nmfs.noaa.gov/sfa/EPAPrpt.pdf Forman, J., & Dunn, M. (2010). The influence of ontogeny and environment on the diet of lookdown dory, Cyttus traversi. New Zealand Journal of Marine and Freshwater Research, 44(4), 329–342. https://doi.org/10.1080/00288330.2010.523080 Fox, C. H., Johnson, F. B., Whiting, J., & Roller, P. P. (1985). Formaldehyde fixation. Journal of Histochemistry & Cytochemistry, 33(8), 845–853. https://doi.org/10.1177/33.8.3894502 Fox, D. S., & Starr, R. M. (1996). Comparison of commercial fishery and research catch data. Canadian Journal of Fisheries and Aquatic Sciences, 53(12), 2681–2694. https://doi.org/10.1139/f96-230 Francis, M. P., Hurst, R. J., McArdle, B. H., Bagley, N. W., & Anderson, O. F. (2002). New Zealand demersal fish assemblages. Environmental Biology of Fishes, 65(2), 215–234. https://doi.org/10.1023/A:1020046713411 Francis, R. I. C. C. (1984). An adaptive strategy for stratified random trawl surveys. New Zealand Journal of Marine and Freshwater Research, 18(1), 59–71. https://doi.org/10.1080/00288330.1984.9516030 Francis, R. I. C. C. (1989). A standard approach to biomass estimation from bottom trawl surveys. New Zealand Fisheries Assessment Research Document. (Unpublished report held in NIWA library, Wellington.) Research Document 89/3. Francis, R. I. C. C., & Fu, D. (2012). SurvCalc User Manual v1.2-2011-09-28. NIWA Technical Report 134. Francis, R. I. C. C., Hurst, R. J., & Renwick, J. A. (2001). An evaluation of catchability assumptions in New Zealand stock assessments. New Zealand Fisheries Assessment Report 2001/1. Francois, R., Altabet, M. A., Goericke, R., McCorkle, D. C., Brunet, C., & Poisson, A. (1993). Changes in the δ13C of surface water particulate organic matter across the 241

subtropical convergence in the SW Indian Ocean. Global Biogeochemical Cycles, 7(3), 627–644. https://doi.org/10.1029/93GB01277 Frank, K. T., & Leggett, W. C. (1994). Fisheries ecology in the context of ecological and evolutionary theory. Annu. Rev. Ecol. Syst, 25, 401–422. Frew, R. D., Heywood, K. J., & Dennis, P. F. (1995). Oxygen isotope study of water masses in the Princess Elizabeth Trough, Antarctica. Marine Chemistry, 49(2–3), 141–153. https://doi.org/10.1016/0304-4203(95)00003-A Froese, R., & Pauly, D. (2017). Fishbase. World Wide Web electronic publication. www.fishbase.org Fry, B. (2006). Stable Isotopic Ecology (First Edit). Springer Science+Business Media, LLC. Fry, B., Joern, A., & Parker, P. L. (1978). Grasshopper Food Web Analysis: Use of Carbon Isotope Ratios to Examine Feeding Relationships Among Terrestrial Herbivores. Ecology, 59(3), 498–506. https://doi.org/10.2307/1936580 Fry, B., Mumford, P. L., Tam, F., Fox, D. D., Warren, G. L., Havens, K. E., & Steinman, A. D. (1999). Trophic position and individual feeding histories of fish from Lake Okeechobee, Florida. Canadian Journal of Fisheries and Aquatic Sciences, 56(4), 590– 600. https://doi.org/10.1139/cjfas-56-4-590 Fuita, T., Kitagawa, D., Okuyama, Y., Ishito, Y., Inada, T., & Jin, Y. (1995). Diets of the demersal fishes on the shelf off Iwate, northern Japan. Marine Biology, 123(2), 219–233. https://doi.org/10.1007/BF00353613 Funes, M., Irigoyen, A. J., Trobbiani, G. A., & Galván, D. E. (2018). Stable isotopes reveal different dependencies on benthic and pelagic pathways between Munida gregaria ecotypes. Food Webs, 17, 1–9. https://doi.org/10.1016/j.fooweb.2018.e00101 Fung, T., Farnsworth, K. D., Reid, D. G., & Rossberg, A. G. (2015). Impact of biodiversity loss on production in complex marine food webs mitigated by prey-release. Nature Communications, 6(1), 1–8. https://doi.org/10.1038/ncomms7657 Furlan, M., Castilho, A. L., Fernandes-goes, L. C., Fransozo, V., Bertini, G., & Costa, R. C. Da. (2013). Effect of environmental factors on the abundance of decapod crustaceans from soft bottoms off southeastern Brazil. Anais Da Academia Brasileira de Ciências, 85(4), 1345–1356. https://doi.org/10.1590/0001-3765201394812 Garcia, S. M., Zerbi, A., Aliaume, C., Do Chi, T., & Lasserre, G. (2003). The ecosystem approach to fisheries. Issues, terminology, principles, institutional foundations, implementation and outlook. In FAO Fisheries Technical Paper. No. 443. https://doi.org/10.1007/978-3-319-03041-8_82 Garrison, L. (2000). Fishing effects on spatial distribution and trophic guild structure of the fish community in the Georges Bank region. ICES Journal of Marine Science, 57(3), 723–730. https://doi.org/10.1006/jmsc.2000.0713 Gelman, A. (2008). Scaling regression inputs by dividing by two standard deviations. Statistics in Medicine, 27(15), 2865–2873. https://doi.org/10.1002/sim.3107 Godfriaux, B. L. (1970). Food of predatory demersal fish in Hauraki Gulf. New Zealand Journal of Marine and Freshwater Research, 4(4), 325–336. https://doi.org/10.1080/00288330.1970.9515351 Godfriaux, B. L. (1974). Food of Tarakihi in Western bay of plenty and Tasman bay, New Zealand. New Zealand Journal of Marine and Freshwater Research, 8(1), 111–153. https://doi.org/10.1080/00288330.1974.9515494

242

Goldman, C. R., Elser, J. J., Richards, R. C., Reuters, J. E., Priscu, J. C., & Levin, A. L. (1996). Thermal stratification, nutrient dynamics, and phytoplankton productivity during the onset of spring phytoplankton growth in Lake Baikal, Russia. Hydrobiologia, 331(1– 3), 9–24. https://doi.org/10.1007/BF00025403 Gomes, M. C., Haedrich, R. L., & Villagarcia, M. G. (1995). Spatial and temporal changes in the groundfish assemblages on the north-east Newfoundland/Labrador Shelf, north-west Atlantic, 1978–1991. Fisheries Oceanography, 4(2), 85–101. https://doi.org/10.1111/j.1365-2419.1995.tb00065.x González-Bergonzoni, I., Vidal, N., Wang, B., Ning, D., Liu, Z., Jeppesen, E., & Meerhoff, M. (2015). General validation of formalin-preserved fish samples in food web studies using stable isotopes. Methods in Ecology and Evolution, 6(3), 307–314. https://doi.org/10.1111/2041-210X.12313 Gordon, J. D. M., & Duncan, J. A. R. (1987). Aspects of the biology of Hoplostethus atlanticus and H. mediterraneus (Pisces: Berycomorphi) from the slopes of the rockall Trough and the Porcupine Sea Bight (north-eastern Atlantic). Journal of the Marine Biological Association of the United Kingdom, 67(1), 119–133. https://doi.org/10.1017/S0025315400026400 Goring, D. G., & Bell, R. G. (1999). El Niño and decadal effects on sea‐level variability in northern New Zealand: A wavelet analysis. New Zealand Journal of Marine and Freshwater Research, 33(4), 587–598. https://doi.org/10.1080/00288330.1999.9516902 Graham, B., & Bury, S. (2019). Marine Isoscapes for trophic and animal movement studies in the southwest Pacific Ocean New Zealand Aquatic Environment and Biodiversity Report No. 218. http://www.mpi.govt.nz/news-and-resources/publications Graham, D. H. (1939). Food of fishes of Otago Harbour and Adjacent Sea. Royal Society of New Zealand, 421–436. Graham, M. (1934). Modern theory of exploiting a fishery, and application to North Sea trawling. ICES Journal of Marine Science, 10(1), 264–274. https://doi.org/10.1093/icesjms/10.3.264 Greenstreet, S. P. R., Spence, F. B., Shanks, A. M., & McMillan, J. A. (1999). Fishing effects in northeast Atlantic shelf seas: Patterns in fishing effort, diversity and community structure. II. Trends in fishing effort in the North Sea by UK registered vessels landing in Scotland. Fisheries Research, 40(2), 107–124. https://doi.org/10.1016/S0165- 7836(98)00207-0 Greig, M. J., & Gilmour, A. E. (1992). Flow through the Mernoo Saddle, New Zealand. New Zealand Journal of Marine and Freshwater Research, 26(2), 155–165. https://doi.org/10.1080/00288330.1992.9516510 Grinnell, J. (1904). The origin and distribution of the chest-nut-backed chickadee. The Auk, 21(3), 364–382. https://doi.org/10.2307/4070199 Grueber, C. E., Nakagawa, S., Laws, R. J., & Jamieson, I. G. (2011). Multimodel inference in ecology and evolution: challenges and solutions. Journal of Evolutionary Biology, 24(4), 699–711. https://doi.org/10.1111/j.1420-9101.2010.02210.x Guerra, M. B. (2018). Foraging ecology of sperm whales at Kaikōura. A thesis submitted in partial fulfilment of the requirements for the Doctor of Philosophy diploma in Marine Sciences. University of Otago. Guerra, M., Dawson, S., Sabadel, A., Slooten, E., Somerford, T., Williams, R., Wing, L., & Rayment, W. (2020). Changes in habitat use by a deep-diving predator in response to a coastal earthquake. Deep Sea Research Part I: Oceanographic Research Papers, 243

158(January), 1–10. https://doi.org/10.1016/j.dsr.2020.103226 Gulland, J. A. (1987). The management of North Sea fisheries. Marine Policy, 11(4), 259– 272. https://doi.org/10.1016/0308-597X(87)90017-0 Gutiérrez, C. M. (2007). Magnuson-Stevens Fishery Conservation and Management Act. https://www.nefmc.org/files/msa_amended_2007.pdf Haami, B. (2008). “Kiore – Pacific rats - Hunting and eating kiore.” Te Ara - the Encyclopedia of New Zealand. http://www.teara.govt.nz/en/kiore-pacific-rats/page-3 Halpern, B. S., Walbridge, S., Selkoe, K. A., Kappel, C. V., Micheli, F., D’Agrosa, C., Bruno, J. F., Casey, K. S., Ebert, C., Fox, H. E., Fujita, R., Heinemann, D., Lenihan, H. S., Madin, E. M. P., Perry, M. T., Selig, E. R., Spalding, M., Steneck, R., & Watson, R. (2008). A Global Map of Human Impact on Marine Ecosystems. Science, 319(5865), 948–952. https://doi.org/10.1126/science.1149345 Hamilton, L. J. (2006). Structure of the Subtropical Front in the Tasman Sea. Deep Sea Research Part I: Oceanographic Research Papers, 53(12), 1989–2009. https://doi.org/10.1016/j.dsr.2006.08.013 Hamling, I. J., Hreinsdóttir, S., Clark, K., Elliott, J., Liang, C., Fielding, E., Litchfield, N., Villamor, P., Wallace, L., Wright, T. J., D’Anastasio, E., Bannister, S., Burbidge, D., Denys, P., Gentle, P., Howarth, J., Mueller, C., Palmer, N., Pearson, C., … Stirling, M. (2017). Complex multifault rupture during the 2016 Mw 7.8 Kaikōura earthquake, New Zealand. Science, 356(6334), 10p. https://doi.org/10.1126/science.aam7194 Hanchet, S. (1991). Diet of spiny dogfish, Squalus acanthias Linnaeus, on the east coast, South Island, New Zealand. Journal of Fish Biology, 39(3), 313–323. https://doi.org/10.1111/j.1095-8649.1991.tb04365.x Handel, E. Van. (1985). Rapid determination of total lipids in mosquitoes. Journal of the American Mosquito Control Association, 1(3), 302–304. http://www.biodiversitylibrary.org/content/part/JAMCA/JAMCA_V01_N3_P302- 304.pdf Hannides, C. C. S., Popp, B. N., Landry, M. R., & Graham, B. S. (2009). Quantification of zooplankton trophic position in the North Pacific Subtropical Gyre using stable nitrogen isotopes. Limnology and Oceanography, 54(1), 50–61. https://doi.org/10.4319/lo.2009.54.1.0050 Harding, J., & Jellyman, P. (2015). Earthquakes, catastrophic sediment additions and the response of urban stream communities. New Zealand Journal of Marine and Freshwater Research, 49(3), 346–355. https://doi.org/10.1080/00288330.2015.1013969 Hare, J. A. (2014). The future of fisheries oceanography lies in the pursuit of multiple hypotheses. ICES Journal of Marine Science, 71(8), 2343–2356. https://doi.org/10.1093/icesjms/fsu018 Harley, C. D. G., Hughes, A. R., Kristin, M., Miner, B. G., Sorte, C. J. B., Carol, S., Randall Hughes, A., Hultgren, K. M., Thornber, C. S., Rodriguez, L. F., Tomanek, L., & Williams, S. L. (2006). The impacts of climate change in coastal marine systems. Ecology Letters, 9(2), 228–241. https://doi.org/10.1111/j.1461-0248.2005.00871.x Harmelin-Vivien, M., Loizeau, V., Mellon, C., Beker, B., Arlhac, D., Bodiguel, X., Ferraton, F., Hermand, R., Philippon, X., & Salen-Picard, C. (2008). Comparison of C and N stable isotope ratios between surface particulate organic matter and microphytoplankton in the Gulf of Lions (NW Mediterranean). Continental Shelf Research, 28(15), 1911– 1919. https://doi.org/10.1016/j.csr.2008.03.002

244

Hasson, A., Koch-Larrouy, A., Morrow, R., Juza, M., & Penduff, T. (2012). The origin and fate of mode water in the southern Pacific Ocean. Ocean Dynamics, 62(3), 335–354. https://doi.org/10.1007/s10236-011-0507-3 Hawke, D. J. (1992). Salinity variability in shelf waters near Otago Peninsula, New Zealand, on a time scale of hours. New Zealand Journal of Marine and Freshwater Research, 26(2), 167–173. https://doi.org/10.1080/00288330.1992.9516511 Hawke, D. J., & Hunter, K. A. (1992). Reactive P distribution near Otago Peninsula, New Zealand: An advection-dominated shelf system containing a headland eddy. Estuarine, Coastal and Shelf Science, 34(2), 141–155. https://doi.org/10.1016/S0272- 7714(05)80101-5 Heath, R. A. (1981). Oceanic fronts around southern New Zealand. Deep Sea Research Part A. Oceanographic Research Papers, 28(6), 547–560. https://doi.org/10.1016/0198- 0149(81)90116-3 Heincke, F. (1898). Naturgeschichte des Herings I. Die Lokal? formen und die Wanderungen des Herings in den euro? Päischen Meeren. Abh. D. Seef. Ver., 2, 128. Hendry, A. P., Farrugia, T. J., & Kinninson, M. T. (2008). Human influences on rates of phenotypic change in wild animal populations. Molecular Ecology, 17(1), 20–29. https://doi.org/10.1111/j.1365-294X.2007.03428.x Henriques, S., Pais, M. P., Vasconcelos, R. P., Murta, A., Azevedo, M., Costa, M. J., & Cabral, H. N. (2014). Structural and functional trends indicate fishing pressure on marine fish assemblages. Journal of Applied Ecology, 51(3), 623–631. https://doi.org/10.1111/1365-2664.12235 Hesslein, R. H., Hallard, K. A., & Ramlal, P. (1993). Replacement of sulfur, carbon, and nitrogen in tissue of growing broad whitefish (Coregonus nasus) in response to a change in diet traced by δ34S, δ13C, and δ15N. Canadian Journal of Fisheries and Aquatic Sciences, 50(10), 2071–2076. https://doi.org/10.1139/f93-230 Hetherington, E. D., Kurle, C. M., Ohman, M. D., & Popp, B. N. (2019). Effects of chemical preservation on bulk and amino acid isotope ratios of zooplankton, fish, and squid tissues. Rapid Communications in Mass Spectrometry, 33(10), 935–945. https://doi.org/10.1002/rcm.8408 Hill, S. L., Hinke, J., Bertrand, S., Fritz, L., Furness, R. W., Ianelli, J. N., Murphy, M., Oliveros‐Ramos, R., Pichegru, L., Sharp, R., Stillman, R. A., Wright, P. J., & Ratcliffe, N. (2020). Reference points for predators will progress ecosystem‐based management of fisheries. Fish and Fisheries, 21(2), 368–378. https://doi.org/10.1111/faf.12434 Hilton, G. M., Thompson, D. R., Sagar, P. M., Cuthbert, R. J., Cherel, Y., & Bury, S. J. (2006). A stable isotopic investigation into the causes of decline in a sub-Antarctic predator, the rockhopper penguin. Global Change Biology, 12(4), 611–625. https://doi.org/10.1111/j.1365-2486.2006.01130.x Hinz, H., Moranta, J., Balestrini, S., Sciberras, M., Pantin, J. R., Monnington, J., Zalewski, A., Kaiser, M. J., Sköld, M., Jonsson, P., Bastardie, F., & Hiddink, J. G. (2017). Stable isotopes reveal the effect of trawl fisheries on the diet of commercially exploited species. Scientific Reports, 7(1), 12p. https://doi.org/10.1038/s41598-017-06379-6 Hixon, M. A., Anderson, T. W., Buch, K. L., Johnson, D. W., McLeod, J. B., & Stallings, C. D. (2012). Density dependence and population regulation in marine fish: a large-scale, long-term field manipulation. Ecological Monographs, 82(4), 467–489. https://doi.org/10.1890/11-1525.1 Hjort, J. (1914). Rapports Et Procrs-Verbaux. Rapports et Procès-Verbaux, 20, 237. 245

Hobson, K. A., Piatt, J. F., & Pitocchelli, J. (1994). Using Stable Isotopes to Determine Seabird Trophic Relationships. The Journal of Animal Ecology, 63(4), 786–798. https://doi.org/10.2307/5256 Hobson, K., Ambrose, W., & Renaud, P. (1995). Sources of primary production, benthic- pelagic coupling, and trophic relationships within the Northeast Water Polynya: insights from δ13C and δ15N analysis. Marine Ecology Progress Series, 128(1–3), 1–10. https://doi.org/10.3354/meps128001 Hopkins, J., Shaw, A. G. P., & Challenor, P. (2010). The Southland Front, New Zealand: Variability and ENSO correlations. Continental Shelf Research, 30(14), 1535–1548. https://doi.org/10.1016/j.csr.2010.05.016 Horn, P. L., & Ballara, S. L. (2018). A comparison of a trawl survey index with CPUE series for hake (Merluccius australis) off the west coast of South Island (HAK 7). New Zealand Fisheries Assessment Report 2018/13. http://www.mpi.govt.nz/news-and- resources/publications Horn, P. L., & Sullivan, K. J. (1996). Validated aging methodology using otoliths, and growth parameters for hoki (Macruronus novaezelandiae) in New Zealand waters. New Zealand Journal of Marine and Freshwater Research, 30(2), 161–174. https://doi.org/10.1080/00288330.1996.9516705 Hsieh, C. H., Reiss, C. S., Hunter, J. R., Beddington, J. R., May, R. M., & Sugihara, G. (2006). Fishing elevates variability in the abundance of exploited species. Nature, 443(7113), 859–862. https://doi.org/10.1038/nature05232 Hu, Y., Shang, H., Tong, H., Nehlich, O., Liu, W., Zhao, C., Yu, J., Wang, C., Trinkaus, E., & Richards, M. P. (2009). Stable isotope dietary analysis of the Tianyuan 1 early modern human. Proceedings of the National Academy of Sciences, 106(27), 10971–10974. https://doi.org/10.1073/pnas.0904826106 Hussey, N. E., MacNeil, M. A., McMeans, B. C., Olin, J. A., Dudley, S. F. J., Cliff, G., Wintner, S. P., Fennessy, S. T., & Fisk, A. T. (2014). Rescaling the trophic structure of marine food webs. Ecology Letters, 17(2), 239–250. https://doi.org/10.1111/ele.12226 Huxley, T. H. (1881). The Herring. Nature, 23, 607–613. https://www.nature.com/articles/023607a0.pdf ICES. (2018). Greater North Sea Ecoregion – Ecosystem overview. ICES Fisheries Overviews, 1(Novemeber), 1–22. https://doi.org/10.17895/ices.pub.4647 Ingram, T., & Shurin, J. B. (2009). Trait-based assembly and phylogenetic structure in northeast Pacific rockfish assemblages. Ecology, 90(9), 2444–2453. https://doi.org/10.1890/08-1841.1 Inouye, L. S., & Lotufo, G. R. (2006). Comparison of macro-gravimetric and micro- colorimetric lipid determination methods. Talanta, 70(3), 584–587. https://doi.org/10.1016/j.talanta.2006.01.024 Irwin, G., & Walrond, C. (2016). “When was New Zealand first settled? - Extinction and decline.” Te Ara - the Encyclopedia of New Zealand. http://www.teara.govt.nz/en/when- was-new-zealand-first-settled/page-7 Jack, L., & Wing, S. R. (2010). Maintenance of old-growth size structure and fecundity of the red rock lobster Jasus edwardsii among marine protected areas in Fiordland, New Zealand. Marine Ecology Progress Series, 404, 161–172. https://doi.org/10.3354/meps08499 Jack, L., & Wing, S. R. (2011). Individual variability in trophic position and diet of a marine

246

omnivore is linked to kelp bed habitat. Marine Ecology Progress Series, 443, 129–139. https://doi.org/10.3354/meps09468 Jack, L., & Wing, S. R. (2013). A safety network against regional population collapse: mature subpopulations in refuges distributed across the landscape. Ecosphere, 4(5), 1–16. https://doi.org/10.1890/ES12-00221.1 Jack, L., Wing, S. R., & McLeod, R. R. J. (2009). Prey base shifts in red rock lobster Jasus edwardsii in response to habitat conversion in Fiordland marine reserves: implications for effective spatial management. Marine Ecology Progress Series, 381(April 2009), 213–222. https://doi.org/10.3354/meps07971 Jackson, A. L., Inger, R., Parnell, A. C., & Bearhop, S. (2011). Comparing isotopic niche widths among and within communities: SIBER - Stable Isotope Bayesian Ellipses in R. Journal of Animal Ecology, 80(3), 595–602. https://doi.org/10.1111/j.1365- 2656.2011.01806.x Jackson, J. B. C. (2001). Historical overfishing and the recent collapse of coastal ecosystems. Science, 293(5530), 629–637. https://doi.org/10.1126/science.1059199 Jackson, M. C., Donohue, I., Jackson, A. L., Britton, J. R., Harper, D. M., & Grey, J. (2012). Population-level metrics of trophic structure based on stable isotopes and their application to invasion ecology. PLoS ONE, 7(2), 1–12. https://doi.org/10.1371/journal.pone.0031757 Jacquemin, S. J., & Doll, J. C. (2013). Long-term fish assemblages respond to habitat and niche breadth in the West Fork White River, Indiana. Ecology of Freshwater Fish, 22(2), 280–294. https://doi.org/10.1111/eff.12025 Jennings, S., Greenstreet, S. P. R., & Reynolds, J. D. (1999). Structural change in an exploited fish community: A consequence of differential fishing effects on species with contrasting life histories. Journal of Animal Ecology, 68(3), 617–627. https://doi.org/10.1046/j.1365- 2656.1999.00312.x Jennings, S., Pinnegar, J. K., Polunin, N. V. C., & Warr, K. J. (2001). Impacts of trawling disturbance on the trophic structure of benthic invertebrate communities. Marine Ecology Progress Series, 213, 127–142. https://doi.org/10.3354/meps213127 Jennings, S., & Polunin, N. V. C. (1997). Impacts of predator depletion by fishing on the biomass and diversity of non-target reef fish communities. Coral Reefs, 16(2), 71–82. https://doi.org/10.1007/s003380050061 Jennings, S., & Warr, K. J. (2003). Environmental correlates of large-scale spatial variation in the δ15N of marine animals. Marine Biology, 142(6), 1131–1140. https://doi.org/10.1007/s00227-003-1020-0 Jeppesen, E., Søndergaard, M., Jensen, J. P., Mortensen, E., & Hansen, A.-M. (1998). Cascading trophic interactions from fish to bacteria and nutrients after reduced sewage loading: an 18-year study of a shallow hypertrophic lake. Ecosystems, 1(3), 250–267. https://doi.org/10.1007/s100219900020 Jillett, J. B. (1969). Seasonal hydrology of waters off the Otago peninsula, South‐Eastern New Zealand. New Zealand Journal of Marine and Freshwater Research, 3(3), 349–375. https://doi.org/10.1080/00288330.1969.9515303 Johnson, D., & Haworth, J. (2004). Hooked - The Sory of New Zealand Fishing Industry. Hazard Press. Jones, K. N., Currie, K. I., McGraw, C. M., & Hunter, K. A. (2013). The effect of coastal processes on phytoplankton biomass and primary production within the near-shore

247

Subtropical Frontal Zone. Estuarine, Coastal and Shelf Science, 124, 44–55. https://doi.org/10.1016/j.ecss.2013.03.003 Kadoya, T., & McCann, K. S. (2015). Weak Interactions and Instability Cascades. Scientific Reports, 5(1), 1–7. https://doi.org/10.1038/srep12652 Kaehler, S., & Pakhomov, E. (2001). Effects of storage and preservation on the δ13C and δ15N signatures of selected marine organisms. Marine Ecology Progress Series, 219, 299–304. https://doi.org/10.3354/meps219299 Kailola, P. J., Williams, M. J., Stewart, P. C., Reichelt, R. E., McNee, A., & Grieve, C. (1993). Australian Fisheries Resources. Bureau of Resource Sciences, Department of Primary Industries and Energy, Australia. Karlsson, O., Costa, D., Goebel, M., Hårding, K., Hückstädt, L., Brault, E., Teilmann, J., McCarthy, M., Koch, P., Härkönen, T., McMahon, K., & Goetz, K. (2019). Trophic position and foraging ecology of Ross, Weddell, and crabeater seals revealed by compound-specific isotope analysis. Marine Ecology Progress Series, 611, 1–18. https://doi.org/10.3354/meps12856 Kaschner, K., Kesner-Reyes, K., Garilao, C., Rius-Barile, J., Rees, T., & Froese, R. (2016). AquaMaps: Predicted range maps for aquatic species. Version 10/2019. www.aquamaps.org Keeling, C. D. (1979). The Suess effect: 13Carbon-14Carbon interrelations. Environment International, 2(4–6), 229–300. https://doi.org/10.1016/0160-4120(79)90005-9 Kelly, B., Dempson, J. B., & Power, M. (2006). The effects of preservation on fish tissue stable isotope signatures. Journal of Fish Biology, 69(6), 1595–1611. https://doi.org/10.1111/j.1095-8649.2006.01226.x Killingley, J. S. (1980). Migrations of california gray whales tracked by oxygen-18 variations in their epizoic barnacles. Science, 207(4432), 759–760. https://doi.org/10.1126/science.207.4432.759 Kishe-Machumu, M. A., van Rijssel, J. C., Poste, A., Hecky, R. E., & Witte, F. (2017). Stable isotope evidence from formalin-ethanol-preserved specimens indicates dietary shifts and increasing diet overlap in Lake Victoria cichlids. Hydrobiologia, 791(1), 155–173. https://doi.org/10.1007/s10750-016-2925-1 Klein, R. G., Avery, G., Cruz-Uribe, K., Halkett, D., Parkington, J. E., Steele, T., Volman, T. P., & Yates, R. (2004). The Ysterfontein 1 Middle Stone Age site, South Africa, and early human exploitation of coastal resources. PNSA, 101(16), 5708–5715. https://doi.org/10.1073/pnas.0400528101 Kleisner, K., Mansour, H., & Pauly, D. (2014). Region-based MTI: resolving geographic expansion in the Marine Trophic Index. Marine Ecology Progress Series, 512, 185–199. https://doi.org/10.3354/meps10949 Koenigs, C., Miller, R., & Page, H. (2015). Top predators rely on carbon derived from giant kelp Macrocystis pyrifera. Marine Ecology Progress Series, 537, 1–8. https://doi.org/10.3354/meps11467 Konchina, Y. V. (1992). Trophic status of Peruvian pseudoneritic fish in oceanic epipelagic water. J. Ichthyol., 32(8), 20–39. Krumhansl, K. A., Okamoto, D. K., Rassweiler, A., Novak, M., Bolton, J. J., Cavanaugh, K. C., Connell, S. D., Johnson, C. R., Konar, B., Ling, S. D., Micheli, F., Norderhaug, K. M., Pérez-Matus, A., Sousa-Pinto, I., Reed, D. C., Salomon, A. K., Shears, N. T., Wernberg, T., Anderson, R. J., … Byrnes, J. E. K. (2016). Global patterns of kelp forest

248

change over the past half-century. Proceedings of the National Academy of Sciences, 113(48), 13785–13790. https://doi.org/10.1073/pnas.1606102113 Kurle, C., & McWhorter, J. (2017). Spatial and temporal variability within marine isoscapes: implications for interpreting stable isotope data from marine systems. Marine Ecology Progress Series, 568(March), 31–45. https://doi.org/10.3354/meps12045 Ladds, M., Pinkerton, M. H., Jones, E., Durante, L., & Dunn, M. (2020). Relationship between morphometrics and trophic levels in deep-sea fishes. Marine Ecology Progress Series, 637(1978), 225–235. https://doi.org/10.3354/meps13243 Lamare, M. D., & Wing, S. R. (2001). Calorific content of New Zealand marine macrophytes. New Zealand Journal of Marine and Freshwater Research, 35(2), 335–341. https://doi.org/10.1080/00288330.2001.9517004 Larkin, P. A. (1977). An epitaph for the concept of maximum sustained yield. Transactions of the American Fisheries Society, 106(1), 1–11. https://doi.org/10.1577/1548- 8659(1977)106<1:AEFTCO>2.0.CO;2 Larsen, T., Ventura, M., Andersen, N., O’Brien, D. M., Piatkowski, U., & McCarthy, M. D. (2013). Tracing carbon sources through aquatic and terrestrial food webs using amino acid stable isotope fingerprinting. PLoS ONE, 8(9), 1–9. https://doi.org/10.1371/journal.pone.0073441 Lassalle, G., Lobry, J., Le Loc’h, F., Bustamante, P., Certain, G., Delmas, D., Dupuy, C., Hily, C., Labry, C., Le Pape, O., Marquis, E., Petitgas, P., Pusineri, C., Ridoux, V., Spitz, J., & Niquil, N. (2011). Lower trophic levels and detrital biomass control the Bay of Biscay continental shelf food web: Implications for ecosystem management. Progress in Oceanography, 91(4), 561–575. https://doi.org/10.1016/j.pocean.2011.09.002 Lau, D. C. P., Leung, K. M. Y., & Dudgeon, D. (2012). Preservation effects on C/N ratios and stable isotope signatures of freshwater fishes and benthic macroinvertebrates. Limnology and Oceanography: Methods, 10(2), 75–89. https://doi.org/10.4319/lom.2012.10.75 Lawton, R. J., Wing, S. R., & Lewis, A. M. (2010). Evidence for discrete subpopulations of sea perch (Helicolenus ercoides) across four fjords in Fiordland, New Zealand. New Zealand Journal of Marine and Freshwater Research, 44(4), 309–322. https://doi.org/10.1080/00288330.2010.519777 Layman, C. A., Araujo, M. S., Boucek, R., Hammerschlag-Peyer, C. M., Harrison, E., Jud, Z. R., Matich, P., Rosenblatt, A. E., Vaudo, J. J., Yeager, L. A., Post, D. M., & Bearhop, S. (2012). Applying stable isotopes to examine food-web structure: an overview of analytical tools. Biological Reviews, 87(3), 545–562. https://doi.org/10.1111/j.1469- 185X.2011.00208.x Layman, C. A., Arrington, D. A., Montaña, C. G., & Post, D. M. (2007). Can stable isotope ratios provide for community-wide measures of trophic structure? Ecology, 88(1), 42– 48. https://doi.org/10.1890/0012-9658(2007)88[42:CSIRPF]2.0.CO;2 Leathwick, J. R., Rowden, A., Nodder, S., Gorman, R., Bardsley, S., Pinkerton, M., Baird, S. J., Hadfield, M., Currie, K., & Goh, A. (2012). A Benthic-optimised Marine Environment Classification (BOMEC) for New Zealand waters. In New Zealand Aquatic Environment and Biodiversity Report No. 88. http://deepwater.hosting.outwide.net/wp- content/uploads/2013/08/Leathwick-et-al-2012-BOMEC-AEBR-88.pdf Leduc, D., Nodder, S. D., Rowden, A. A., Gibbs, M., Berkenbusch, K., Wood, A., De Leo, F., Smith, C., Brown, J., Bury, S. J., & Pallentin, A. (2020). Structure of infaunal communities in New Zealand submarine canyons is linked to origins of sediment organic matter. Limnology and Oceanography, April, 1–25. https://doi.org/10.1002/lno.11454

249

Leduc, D., Rowden, A. A., Nodder, S. D., Berkenbusch, K., Probert, P. K., & Hadfield, M. G. (2014). Unusually high food availability in Kaikoura Canyon linked to distinct deep-sea nematode community. Deep Sea Research Part II: Topical Studies in Oceanography, 104(June), 310–318. https://doi.org/10.1016/j.dsr2.2013.06.003 Levin, P. S., Holmes, E. E., Piner, K. R., & Harvey, C. J. (2006). Shifts in a Pacific Ocean fish assemblage: the potential influence of exploitation. Conservation Biology, 20(4), 1181–1190. https://doi.org/10.1111/j.1523-1739.2006.00400.x Li, Y., Zhang, Y., Hussey, N. E., & Dai, X. (2016). Urea and lipid extraction treatment effects on δ15N and δ13C values in pelagic sharks. Rapid Communications in Mass Spectrometry, 30(1), 1–8. https://doi.org/10.1002/rcm.7396 Lima, M., Canales, T. M., Wiff, R., & Montero, J. (2020). The Interaction Between Stock Dynamics, Fishing and Climate Caused the Collapse of the Jack Mackerel Stock at Humboldt Current Ecosystem. Frontiers in Marine Science, 7(March), 1–9. https://doi.org/10.3389/fmars.2020.00123 Lindegren, M., Checkley, D. M., Rouyer, T., MacCall, A. D., & Stenseth, N. C. (2013). Climate, fishing, and fluctuations of sardine and anchovy in the California Current. Proceedings of the National Academy of Sciences, 110(33), 13672–13677. https://doi.org/10.1073/pnas.1305733110 Link, J. S., & Garrison, L. P. (2002). Changes in piscivory associated with fishing induced changes to the finfish community on Georges Bank. Fisheries Research, 55(1–3), 71–86. https://doi.org/10.1016/S0165-7836(01)00300-9 Link, J. S., & Marshak, A. R. (2019). Characterizing and comparing marine fisheries ecosystems in the United States: determinants of success in moving toward ecosystem- based fisheries management. Reviews in Fish Biology and Fisheries, 29(1), 23–70. https://doi.org/10.1007/s11160-018-9544-z Lock, K., & Leslie, S. (2007). New Zealand’s Quota Management System: a history of the first 20 Years. In Motu Economic and Public Policy Research. http://deepwatergroup.org//wp-content/uploads/2013/03/QMS.pdf Lockerbie, L. (1940). Excavations at Kings Rock, Otago, with a discussion of the fish-hook barb as an ancient feature of Polynesian culture. The Journal of the Polynesian Society, 49(3), 393–446. https://www.jstor.org/stable/pdf/20702824.pdf?refreqid=excelsior:dcd948d071f9bb16d0 a0b7494d85c04f Logan, J. M., Pethybridge, H., Lorrain, A., Somes, C. J., Allain, V., Bodin, N., Choy, C. A., Duffy, L., Goñi, N., Graham, B., Langlais, C., Ménard, F., Olson, R., & Young, J. (2020). Global patterns and inferences of tuna movements and trophodynamics from stable isotope analysis. Deep Sea Research Part II: Topical Studies in Oceanography. https://doi.org/10.1016/j.dsr2.2020.104775 Lorrain, A., Graham, B., Ménard, F., Popp, B., Bouillon, S., van Breugel, P., & Cherel, Y. (2009). Nitrogen and carbon isotope values of individual amino acids: a tool to study foraging ecology of penguins in the Southern Ocean. Marine Ecology Progress Series, 391, 293–306. https://doi.org/10.3354/meps08215 Lotze, H. K., Coll, M., Magera, A. M., Ward-Paige, C., & Airoldi, L. (2011). Recovery of marine animal populations and ecosystems. Trends in Ecology & Evolution, 26(11), 595–605. https://doi.org/10.1016/j.tree.2011.07.008 Lusseau, S. M. S., & Wing, S. R. (2006). Importance of local production versus pelagic subsidies in the diet of an isolated population of bottlenose dolphins Tursiops sp. Marine

250

Ecology Progress Series, 321(September), 283–293. https://doi.org/10.3354/meps321283 Macara, G. R. (2015). The Climate and Weather of Otago. In NIWA Science and Technology Series Number 67 (Second Edi, p. 42). NIWA. https://doi.org/10.31826/9781463208868- 002 MacDiarmid, A. B., Abraham, E., Baker, C. S., Carroll, E., Chagué-Goff, C., Cleaver, P., Francis, M. P., Goff, J., Horn, P., Jackson, J., Lalas, C., Lorrey, A., Marriot, P., Maxwell, K., McFadgen, B., McKenzie, A., Neil, H., Parsons, D., Patenaude, N., … Stirling, B. (2016). Taking Stock – The Changes to New Zealand Marine Ecosystems Since First Human Settlement: Synthesis of Major Findings, and Policy and Management Implications. New Zealand Aquatic Environment and Biodiversity Report No. 170. Macdiarmid, A. B., Cleaver, P., & Stirling, B. (2018). Historical evidence for the state and exploitation of the marine fish and invertebrate resources in the Hauraki Gulf and along the Otago-Catlins shelf 1769 – 1950. New Zealand Aquatic Environment and Biodiversity Report No. 194. Mace, P. M. (2001). A new role for MSY in single-species and ecosystem approaches to fisheries stock assessment and management. Fish and Fisheries, 2(1), 2–32. https://doi.org/10.1046/j.1467-2979.2001.00033.x Mace, P. M., Sullivan, K. J., & Cryer, M. (2014). The evolution of New Zealand’s fisheries science and management systems under ITQs. ICES Journal of Marine Science, 71(2), 204–215. https://doi.org/10.1093/icesjms/fst159 MacGibbon, D. J., Beentjes, M. P., Lyon, W. L., & Ladroit, Y. (2019). Inshore trawl survey of Canterbury Bight and Pegasus Bay, April–June 2018 (KAH1803). New Zealand Fisheries Assessment Report 2019/03. http://www.mpi.govt.nz/news-and- resources/publications MacKenzie, K. M., Longmore, C., Preece, C., Lucas, C. H., & Trueman, C. N. (2014). Testing the long-term stability of marine isoscapes in shelf seas using jellyfish tissues. Biogeochemistry, 121(2), 441–454. https://doi.org/10.1007/s10533-014-0011-1 Mackinson, S., Deas, B., Beveridge, D., & Casey, J. (2009). Mixed-fishery or ecosystem conundrum? Multispecies considerations inform thinking on long-term management of North Sea demersal stocks. Canadian Journal of Fisheries and Aquatic Sciences, 66(7), 1107–1129. https://doi.org/10.1139/F09-057 Magozzi, S., Yool, A., Vander Zanden, H. B., Wunder, M. B., & Trueman, C. N. (2017). Using ocean models to predict spatial and temporal variation in marine carbon isotopes. Ecosphere, 8(5), 1–20. https://doi.org/10.1002/ecs2.1763 Magurran, A. E., Dornelas, M., Moyes, F., Gotelli, N. J., & McGill, B. (2015). Rapid biotic homogenization of marine fish assemblages. Nature Communications, 6(1), 1–5. https://doi.org/10.1038/ncomms9405 Mahon, R., Brown, S. K., Zwanenburg, K. C. T., Atkinson, D. B., Buja, K. R., Claflin, L., Howell, G. D., Monaco, M. E., O’Boyle, R. N., & Sinclair, M. (1998). Assemblages and biogeography of demersal fishes of the east coast of North America. Canadian Journal of Fisheries and Aquatic Sciences, 55(7), 1704–1738. https://doi.org/10.1139/f98-065 Mancinelli, G., & Vizzini, S. (2015). Assessing anthropogenic pressures on coastal marine ecosystems using stable CNS isotopes: State of the art, knowledge gaps, and community- scale perspectives. Estuarine, Coastal and Shelf Science, 156(1), 195–204. https://doi.org/10.1016/j.ecss.2014.11.030 Mangel, M., & Levin, P. S. (2005). Regime, phase and paradigm shifts: making community 251

ecology the basic science for fisheries. Philosophical Transactions of the Royal Society B: Biological Sciences, 360(1453), 95–105. https://doi.org/10.1098/rstb.2004.1571 Manning, M. J., Hanchet, S. M., & Stevenson, M. L. (2004). A description and analysis of New Zealand’s spiny dogfish (Squalus acanthias) fisheries and recommendations on appropriate methods to monitor the status of the stocks. Final Research Report for Ministry of Fisheries Research Project SPD2002-01. https://fs.fish.govt.nz/Doc/22514/SPD2002-01 Spiny Dogfish methods monitor stocks Objective 1 final.pdf.ashx Marchal, P., Andersen, J. L., Aranda, M., Fitzpatrick, M., Goti, L., Guyader, O., Haraldsson, G., Hatcher, A., Hegland, T. J., Le Floc’h, P., Macher, C., Malvarosa, L., Maravelias, C. D., Mardle, S., Murillas, A., Nielsen, J. R., Sabatella, R., Smith, A. D. M., Stokes, K., … Ulrich, C. (2016). A comparative review of fisheries management experiences in the European Union and in other countries worldwide: Iceland, Australia, and New Zealand. Fish and Fisheries, 17(3), 803–824. https://doi.org/10.1111/faf.12147 Marine Stewardship Council. (2013). New Zealand’s top three fisheries celebrate sustainable management. Press Release. https://www.msc.org/media-centre/press-releases/new- zealand’s-top-three-fisheries-celebrate-sustainable-management Marine Stewardship Council. (2018). New Zealanders Choose Sustainable Seafood For Future Generations. Press Release. https://www.msc.org/media-centre/press- releases/new-zealanders-choose-sustainable-seafood-for-future-generations Marshall, K. N., Levin, P. S., Essington, T. E., Koehn, L. E., Anderson, L. G., Bundy, A., Carothers, C., Coleman, F., Gerber, L. R., Grabowski, J. H., Houde, E., Jensen, O. P., Möllmann, C., Rose, K., Sanchirico, J. N., & Smith, A. D. M. (2018). Ecosystem-based fisheries management for social-ecological systems: renewing the focus in the United States with next generation fishery ecosystem plans. Conservation Letters, 11(1), 1–7. https://doi.org/10.1111/conl.12367 Matich, P., Kiszka, J. J., Gastrich, K. R., & Heithaus, M. R. (2017). Trophic redundancy among fishes in an East African nearshore seagrass community inferred from stable- isotope analysis. Journal of Fish Biology, 91(2), 490–509. https://doi.org/10.1111/jfb.13354 Matthews, W. J., Marsh-Matthews, E., Adams, G. L., & Adams, S. R. (2014). Two Catastrophic Floods: Similarities and Differences in Effects on an Ozark Stream Fish Community. Copeia, 2014(4), 682–693. https://doi.org/10.1643/CE-14-041 MBIE. (2014). Sustainable Seas - Ko ngā moana whakauka. Ministry of Business, Innovation & Employment. http://www.mbie.govt.nz/info-services/science-innovation/national- science-challenges/sustainable-seas McAfee, D., Alleway, H. K., & Connell, S. D. (2020). Environmental solutions sparked by environmental history. Conservation Biology, 34(2), 386–394. https://doi.org/10.1111/cobi.13403 Mcardle, B. H., & Anderson, M. J. (2001). Fitting multivariate models to community data : a comment on distance-based redundancy analysis. Ecology, 82(1), 290–297. https://doi.org/10.1890/0012-9658(2001)082[0290:FMMTCD]2.0.CO;2 McClatchie, S., Pinkerton, M. H., & Livingston, M. E. (2005). Relating the distribution of a semi-demersal fish, Macruronus novaezelandiae, to their pelagic food supply. Deep Sea Research Part I: Oceanographic Research Papers, 52(8), 1489–1501. https://doi.org/10.1016/j.dsr.2005.02.007 McCutchan, J. H., Lewis, W. M., Kendall, C., & McGrath, C. C. (2003). Variation in trophic

252

shift for stable isotope ratios of carbon, nitrogen, and sulfur. Oikos, 102(2), 378–390. https://doi.org/10.1034/j.1600-0706.2003.12098.x McDonald-Madden, E., Sabbadin, R., Game, E. T., Baxter, P. W. J., Chadès, I., & Possingham, H. P. (2016). Using food-web theory to conserve ecosystems. Nature Communications, 7(1), 1–8. https://doi.org/10.1038/ncomms10245 McKenzie, J. R., Beentjes, M. P., Parker, S., Parsons, D. M., Armiger, H., O., W., Middleton, D., Langley, A., Buckthought, D., Walsh, C., Bian, R., Maolagáin, C. Ó., Stevenson, M., Sutton, C., Spong, K., Rush, N., & Smith, M. (2017). Fishery characterisation and age composition of tarakihi in TAR 1, 2 and 3 for 2013/14 and 2014/15. New Zealand Fisheries Assessment Report 2017/36. McKinney, C. R., McCrea, J. M., Epstein, S., Allen, H. A., & Urey, H. C. (1950). Improvements in mass spectrometers for the measurement of small differences in isotope abundance ratios. Review of Scientific Instruments, 21(8), 724–730. https://doi.org/10.1063/1.1745698 McLeod, R. J., & Wing, S. R. (2007). Hagfish in the new zealand fjords are supported by chemoautotrophy of forest carbon. Ecology, 88(4), 809–816. https://doi.org/10.1890/06- 1342 McLeod, R. J., Wing, S. R., & Skilton, J. E. (2010a). High incidence of invertebrate- chemoautotroph symbioses in benthic communities of the New Zealand fjords. Limnology and Oceanography, 55(5), 2097–2106. https://doi.org/10.4319/lo.2010.55.5.2097 McLeod, R., Wing, S. R., & Davis, J. (2010b). Habitat conversion and species loss alters the composition of carbon sources to benthic communities. Marine Ecology Progress Series, 411, 127–136. https://doi.org/10.3354/meps08677 McMahon, K. W., Hamady, L. L., & Thorrold, S. R. (2013). A review of ecogeochemistry approaches to estimating movements of marine animals. Limnology and Oceanography, 58(2), 697–714. https://doi.org/10.4319/lo.2013.58.2.0697 McMahon, K. W., & McCarthy, M. D. (2016). Embracing variability in amino acid δ15N fractionation: mechanisms, implications, and applications for trophic ecology. Ecosphere, 7(12), 1–26. https://doi.org/10.1002/ecs2.1511 McMahon, K. W., Thorrold, S. R., Elsdon, T. S., & Mccarthy, M. D. (2015). Trophic discrimination of nitrogen stable isotopes in amino acids varies with diet quality in a marine fish. Limnology and Oceanography, 60(3), 1076–1087. https://doi.org/10.1002/lno.10081 McMahon, K. W., Thorrold, S. R., Houghton, L. A., & Berumen, M. L. (2016). Tracing carbon flow through coral reef food webs using a compound-specific stable isotope approach. Oecologia, 180(3), 809–821. https://doi.org/10.1007/s00442-015-3475-3 McMullin, R. M., Wing, S. R., Wing, L., & Shatova, O. (2017). Trophic position of Antarctic ice fishes reflects food web structure along a gradient in sea ice persistence. Marine Ecology Progress Series, 564, 87–98. https://doi.org/10.3354/meps12031 Meerhoff, E., Castro, L., & Tapia, F. (2013). Influence of freshwater discharges and tides on the abundance and distribution of larval and juvenile Munida gregaria in the Baker river estuary, Chilean Patagonia. Continental Shelf Research, 61–62, 1–11. https://doi.org/10.1016/j.csr.2013.04.025 Migiro, G. (2018). Countries With The Largest Exclusive Economic Zones. World Atlas. https://www.worldatlas.com/articles/countries-with-the-largest-exclusive-economic- zones.html 253

Mill, A. C., Pinnegar, J. K., & Polunin, N. V. C. (2007). Explaining isotope trophic-step fractionation: Why herbivorous fish are different. Functional Ecology, 21(6), 1137– 1145. https://doi.org/10.1111/j.1365-2435.2007.01330.x Miller, S. M., Wing, S. R., & Hurd, C. L. (2006). Photoacclimation of Ecklonia radiata (Laminariales, Heterokontophyta) in Doubtful Sound, Fjordland, Southern New Zealand. Phycologia, 45(1), 44–52. https://doi.org/10.2216/04-98.1 Milligan, B., & Holt, L. A. (1977). The complete enzymic hydrolysis of crosslinked proteins. In M. Friedman (Ed.), Protein Crosslinking - Advances in Experimental Medicine and Biology book series (pp. 277–284). Springer Science+Business Media New York. Mills, J. A., Yarrall, J. W., Bradford-Grieve, J. M., Uddstrom, M. J., Renwick, J. A., & Merilä, J. (2008). The impact of climate fluctuation on food availability and reproductive performance of the planktivorous red-billed gull Larus novaehollandiae scopulinus. Journal of Animal Ecology, 77(6), 1129–1142. https://doi.org/10.1111/j.1365- 2656.2008.01383.x Ministry of Primary Industries. (2017). Aquatic Environment and Biodiversity Annual Review 2017. Compiled by the Fisheries Science Team, Ministry for Primary Industries. Mont’Alverne, R., Jardine, T. D., Pereyra, P. E. R., Oliveira, M. C. L. M., Medeiros, R. S., Sampaio, L. A., Tesser, M. B., & Garcia, A. M. (2016). Elemental turnover rates and isotopic discrimination in a euryhaline fish reared under different salinities: Implications for movement studies. Journal of Experimental Marine Biology and Ecology, 480, 36– 44. https://doi.org/10.1016/j.jembe.2016.03.021 Monteiro, P. V. (2017). The purse seine fishing of sardine in Portuguese waters: a difficult compromise between fish stock sustainability and fishing effort. Reviews in Fisheries Science & Aquaculture, 25(3), 218–229. https://doi.org/10.1080/23308249.2016.1269720 Morris, M., Stanton, B., & Neil, H. (2001). Subantarctic oceanography around New Zealand: Preliminary results from an ongoing survey. New Zealand Journal of Marine and Freshwater Research, 35(3), 499–519. https://doi.org/10.1080/00288330.2001.9517018 Mountjoy, J. J., Howarth, J. D., Orpin, A. R., Barnes, P. M., Bowden, D. A., Rowden, A. A., Schimel, A. C. G., Holden, C., Horgan, H. J., Nodder, S. D., Patton, J. R., Lamarche, G., Gerstenberger, M., Micallef, A., Pallentin, A., & Kane, T. (2018). Earthquakes drive large-scale submarine canyon development and sediment supply to deep-ocean basins. Science Advances, 4(3), 1–8. https://doi.org/10.1126/sciadv.aar3748 Moyes, F., & Magurran, A. E. (2019). Change in the dominance structure of two marine-fish assemblages over three decades. Journal of Fish Biology, 94(1), 96–102. https://doi.org/10.1111/jfb.13868 MPI. (2018). Status of stock as at December 2018 or last assessment date. https://www.mpi.govt.nz/dmsdocument/17653-stock-status-table-for-fish-stocks MPI. (2020). Public feedback sought on marine protection for southeastern South Island. News & Resources. https://www.mpi.govt.nz/news-and-resources/media-releases/public- feedback-sought-on-marine-protection-for-south-eastern-south- island/?utm_source=notification-email MRAG, & IEEP. (2007). European Commission Studies and Pilot Projects for Carrying Out the Common Fisheries Policy No FISH/2006/17. Impact assessment of discard policy for specific fisheries. https://ec.europa.eu/fisheries/sites/fisheries/files/docs/publications/impact_assessment_di scard_policy_2007_en.pdf

254

Mullan, A. (1996). Influence of Southern Oscillation on New Zealand Weather. In Proceedings of Western Pacific International Meeting and Workshop on TOGA-COARE. http://horizon.documentation.ird.fr/exl-doc/pleins_textes/doc34-08/31756.pdf Mullin, M. M., Rau, G. H., & Eppley, R. W. (1984). Stable nitrogen isotopes in zooplankton: some geographic and temporal variations in the North Pacific. Limnology and Oceanography, 29(6), 1267–1273. https://doi.org/10.4319/lo.1984.29.6.1267 Muñoz, M., & Casadevall, M. (1997). Fish remains from Arbreda Cave (Serinya Girona), northeast Spain, and their palaeoecological significance. Journal of Quaternary Science, 12(2), 111–115. https://doi.org/10.1002/(SICI)1099-1417(199703/04)12:2<111::AID- JQS294>3.0.CO;2-P Murdoch, R. C., Proctor, R., Jillett, J. B., & Zeldis, J. R. (1990). Evidence for an eddy over the continental shelf in the downstream lee of Otago Peninsula, New Zealand. Estuarine, Coastal and Shelf Science, 30(5), 489–507. https://doi.org/10.1016/0272- 7714(90)90069-4 Mutshinda, C. M., O’Hara, R. B., & Woiwod, I. P. (2009). What drives community dynamics? Proceedings of the Royal Society B: Biological Sciences, 276(1669), 2923– 2929. https://doi.org/10.1098/rspb.2009.0523 Myers, R. A., Baum, J. K., Shepherd, T. D., Powers, S. P., & Peterson, C. H. (2007). Cascading effects of the loss of apex predatory sharks from a coastal ocean. Science, 315(5820), 1846–1850. https://doi.org/10.1126/science.1138657 Myers, R. A., & Worm, B. (2003). Rapid worldwide depletion of predatory fish communities. Nature, 423(6937), 280–283. https://doi.org/10.1038/nature01610 Nature Publishing Group. (1943). Trawling and the stocks of fish. Nature, 151(3829), 323– 323. https://doi.org/10.1038/151323a0 Nelson, C. S., Northcote, T. G., & Hendy, C. H. (1989). Potential use of oxygen and carbon isotopic composition of otoliths to identify migratory and non‐migratory stocks of the New Zealand common smelt: A pilot study. New Zealand Journal of Marine and Freshwater Research, 23(3), 337–344. https://doi.org/10.1080/00288330.1989.9516370 Newell, R. G. (2004). Maximising Value in Multi-species Fisheries (p. 79). Fulbright New Zealand. https://www.fulbright.org.nz/publications/2004-newell/ Newland, K. (1999). The Zulu Herring Drifter: A Case Study of Historic Vessel preservation at the Scottish Fisheries Museum. Stavanger Museum Årbok, 108(1998), 83–104. Newsome, S. D., Bentall, G. B., Tinker, M. T., Oftedal, O. T., Ralls, K., Estes, J. A., & Fogel, M. L. (2010). Variation in δ13C and δ15N diet–vibrissae trophic discrimination factors in a wild population of California sea otters. Ecological Applications, 20(6), 1744–1752. https://doi.org/10.1890/09-1502.1 Nicoll, D. (2018). Southland crayfish industry set to benefit from commercial catch increase. Stuff.Co.Nz. https://www.stuff.co.nz/southland-times/news/100677321/southland- crayfish-industry-set-to-benefit-from-commercial-catch-increase Nielsen, J. G., Cohen, D. M., Markle, D. F., & Robins, C. R. (1999). Ophidiiform fishes of the world (Order Ophidiiformes). An annotated and illustrated catalogue of pearlfishes, cusk-eels, brotulas and other ophidiiform fishes known to date. In FAO Fish. Synop. The Food and Agriculture Organization of the United Nations. Nielsen, J. M., Popp, B. N., & Winder, M. (2015). Meta-analysis of amino acid stable nitrogen isotope ratios for estimating trophic position in marine organisms. Oecologia, 178(3), 631–642. https://doi.org/10.1007/s00442-015-3305-7

255

NOAA. (2019). NMFS Headquarters Ecosystem Based Fisheries Management Implementation Plan. https://www.fisheries.noaa.gov/national/ecosystems/ecosystem- based-fishery-management-implementation-plans Nodder, S. D., Duineveld, G. C. A., Pilditch, C. A., Sutton, P. J., Probert, P. K., Lavaleye, M. S. S., Witbaard, R., Chang, F. H., Hall, J. A., & Richardson, K. M. (2007). Focusing of phytodetritus deposition beneath a deep-ocean front, Chatham Rise, New Zealand. Limnology and Oceanography, 52(1), 299–314. https://doi.org/10.4319/lo.2007.52.1.0299 O’Brien, D. M., Fogel, M. L., & Boggs, C. L. (2002). Renewable and nonrenewable resources: Amino acid turnover and allocation to reproduction in Lepidoptera. Proceedings of the National Academy of Sciences, 99(7), 4413–4418. https://doi.org/10.1073/pnas.072346699 O’Connell-Milne, S. A., Wing, S. R., Suanda, S. H., Udy, J. A., Durante, L. M., Salmond, N., & C., W. L. (in review). Bivalve filter feeding and oceanographic forcing drive the fluxes of organic matter and nutrients at an estuarine-coastal interface. Marine Ecology Progress Series. O’Connor, S., Ono, R., & Clarkson, C. (2011). Pelagic fishing at 42,000 years before the present and the maritime skills of modern humans. Science, 334(6059), 1117–1121. https://doi.org/10.1126/science.1207703 O’Driscoll, R. L. (1998). Feeding and schooling behaviour of barracouta (Thyrsites atun) off Otago, New Zealand. Marine and Freshwater Research, 49(1), 21–24. https://doi.org/10.1071/MF97074 O’Driscoll, R. L., Ballara, S. L., Macgibbon, D. J., & Schimel, A. C. G. (2018). Trawl survey of hoki and middle depth species in the Southland and Sub-Antarctic, November– December 2016 (TAN1614). New Zealand Fisheries Assessment Report 2018/39. http://www.mpi.govt.nz/news-and-resources/publications O’Leary, M. H. (1981). Carbon isotope fractionation in plants. Phytochemistry, 20(4), 553– 567. https://doi.org/10.1016/0031-9422(81)85134-5 Ogawa, N. O., Chikaraishi, Y., & Ohkouchi, N. (2013). Trophic position estimates of formalin-fixed samples with nitrogen isotopic compositions of amino acids: an application to gobiid fish (Isaza) in Lake Biwa, Japan. Ecological Research, 28(5), 697– 702. https://doi.org/10.1007/s11284-012-0967-z Ogawa, N. O., Koitabashi, T., Oda, H., Nakamura, T., Ohkouchi, N., & Wada, E. (2001). Fluctuations of nitrogen isotope ratio of gobiid fish (Isaza) specimens and sediments in Lake Biwa, Japan, during the 20th century. Limnology and Oceanography, 46(5), 1228– 1236. https://doi.org/10.4319/lo.2001.46.5.1228 Ohkouchi, N., Chikaraishi, Y., Close, H. G., Fry, B., Larsen, T., Madigan, D. J., McCarthy, M. D., McMahon, K. W., Nagata, T., Naito, Y. I., Ogawa, N. O., Popp, B. N., Steffan, S., Takano, Y., Tayasu, I., Wyatt, A. S. J., Yamaguchi, Y. T., & Yokoyama, Y. (2017). Advances in the application of amino acid nitrogen isotopic analysis in ecological and biogeochemical studies. Organic Geochemistry, 113, 150–174. https://doi.org/10.1016/j.orggeochem.2017.07.009 Otago Regional Council. (2020). Managing Our Environment - Water. Online Maps and Data. https://www.orc.govt.nz/managing-our- environment/wathttps://www.orc.govt.nz/managing-our-environment/water/water- monitoring-and-alerts Pace, M. L., Cole, J. J., Carpenter, S. R., & Kitchell, J. F. (1999). Trophic cascades revealed

256

in diverse ecosystems. Trends in Ecology & Evolution, 14(12), 483–488. https://doi.org/10.1016/S0169-5347(99)01723-1 Palkovacs, E. P., Kinnison, M. T., Correa, C., Dalton, C. M., & Hendry, A. P. (2012). Fates beyond traits: ecological consequences of human-induced trait change. Evolutionary Applications, 5(2), 183–191. https://doi.org/10.1111/j.1752-4571.2011.00212.x Pantoja, S., Repeta, D. J., Sachs, J. P., & Sigman, D. M. (2002). Stable isotope constraints on the nitrogen cycle of the Mediterranean Sea water column. Deep Sea Research Part I: Oceanographic Research Papers, 49(9), 1609–1621. https://doi.org/10.1016/S0967- 0637(02)00066-3 Parsons, D. M., Morrison, M. A., MacDiarmid, A. B., Stirling, B., Cleaver, P., Smith, I. W. G., & Butcher, M. (2009). Risks of shifting baselines highlighted by anecdotal accounts of New Zealand’s snapper (Pagrus auratus) fishery. New Zealand Journal of Marine and Freshwater Research, 43(4), 965–983. https://doi.org/10.1080/00288330909510054 Patrick, W. S., & Cope, J. (2014). Examining the 10-year rebuilding dilemma for U.S. Fish stocks. PLoS ONE, 9(11), 1–10. https://doi.org/10.1371/journal.pone.0112232 Paulin, C. D. (1989). Redescription of Helicolenus percoides (Richardson) and H. barathri (Hector) from New Zealand (Pisces, Scorpaenidae). Journal of the Royal Society of New Zealand, 19(3), 319–325. https://doi.org/10.1080/03036758.1989.10427185 Paulin, C. D. (2007). Perspectives of Mäori fishing history and techniques. Ngä ähua me ngä püräkau me ngä hangarau ika o te Mäori. Te Papa Museum of New Zealand, 18, 11–47. www.tepapa.govt.nz/sites/default/files/tuhinga.18.2007.pt2_.p11-47.paulin.pdf Pauly, D. (1995). Anecdotes and the shifting baseline syndrome of fisheries. Trends in Ecology & Evolution, 10(10), 430. https://doi.org/10.1016/S0169-5347(00)89171-5 Pauly, D., Alder, J., Bennett, E., Christensen, V., Tyedmers, P., & Watson, R. (2003). The Future for Fisheries. Science, 302(5649), 1359–1361. https://doi.org/10.1126/science.1088667 Pauly, D., Christensen, V., Dalsgaard, J., Froese, R., & Torres Jr., F. (1998). Fishing Down Marine Food Webs. Science, 279(5352), 860–863. https://doi.org/10.1126/science.279.5352.860 Pauly, D., & Watson, R. (2005). Background and interpretation of the ‘Marine Trophic Index’ as a measure of biodiversity. Philosophical Transactions of the Royal Society B: Biological Sciences, 360(1454), 415–423. https://doi.org/10.1098/rstb.2004.1597 Pauly, D., Zeller, D., & Palomares, M. L. D. (2020). Sea Around Us Marine Trophic Index. Sea Around Us Data. seaaroundus.org Paxton, J. R., Hoese, D. F., Allen, G. R., & Hanley, J. E. (1989). Zoological Catalogue of Australia: Pisces: Petromyzontidae to Carangidae (Zoological Catalogue of Australia). CSIRO Publishing. Peres-Neto, P., Legendre, P., Dray, S., & Borcard, D. (2006). Variation partitioning of species data matrices: estimation and comparison of fraction. Ecology, 87(10), 2614–2625. https://doi.org/10.1890/0012-9658(2006)87[2614:VPOSDM]2.0.CO;2 Perez, J. A. A., Wahrlich, R., Pezzuto, P. R., Schwingel, P. R., Lopes, F. R. A., & Rodrigues- Ribeiro, M. (2003). Deep-sea Fishery off Southern Brail: Recent Trends of the Brazilian Fishing Industry. Journal of Northwest Atlantic Fishery Science, 31(October), 1–18. https://doi.org/10.2960/J.v31.a13 Perry, A. L. (2005). Climate Change and Distribution Shifts in Marine Fishes. Science, 308(5730), 1912–1915. https://doi.org/10.1126/science.1111322 257

Persson, A., & Hansson, L. (1999). Diet shift in fish following competitive release. Canadian Journal of Fisheries and Aquatic Sciences, 56(1), 70–78. https://doi.org/10.1139/f98-141 Petchey, O. L., Eklöf, A., Borrvall, C., & Ebenman, B. (2008). Trophically unique species are vulnerable to cascading extinction. The American Naturalist, 171(5), 568–579. https://doi.org/10.1086/587068 Peterson, G., Allen, C. R., & Holling, C. S. (1998). Ecological resilience, biodiversity, and scale. Ecosystems, 1(1), 6–18. https://doi.org/10.1007/s100219900002 Phillips, D. L., & Gregg, J. W. J. W. (2001). Uncertainty in source partitioning using stable isotopes. Oecologia, 127(2), 171–179. https://doi.org/10.1007/s004420000578 Phillips, D. L., Inger, R., Bearhop, S., Jackson, A. L., Moore, J. W., Parnell, A. C., Semmens, B. X., & Ward, E. J. (2014). Best practices for use of stable isotope mixing models in food-web studies. Canadian Journal of Zoology, 92(10), 823–835. https://doi.org/10.1139/cjz-2014-0127 Phillips, D. L., Newsome, S. D., & Gregg, J. W. (2005). Combining sources in stable isotope mixing models: Alternative methods. Oecologia, 144(4), 520–527. https://doi.org/10.1007/s00442-004-1816-8 Pikitch, E. K., Santora, C., Babcock, E. A., Bakun, A., Bonfil, R., Conover, D. O., Dayton, P., Doukakis, P., Fluharty, D., Heneman, B., Houde, E. D., Link, J., Livingston, P. A., Mangel, M., McAllister, M. K., Pope, J., & Sainsbury, K. J. (2004). Ecosystem-based fishery management. Science, 305(5682), 346–347. https://doi.org/10.1126/science.1098222 Pincinato, R. B. M., & Gasalla, M. A. (2019). Exploring simple ecological indicators on landings and market trends in the South Brazil Shelf Large Marine Ecosystem. Fisheries Management and Ecology, 26(3), 200–210. https://doi.org/10.1111/fme.12340 Pinkerton, M. H. (2010). Headline indicators for the New Zealand ocean. Unpublished NIWA report for Coasts & Oceans IO2 project. https://www.niwa.co.nz/sites/niwa.co.nz/files/headline_indicators_59.pdf Pinkerton, M. H. (2011). A balanced trophic model of the Chatham Rise, New Zealand. Unpublished NIWA report. https://niwa.co.nz/sites/niwa.co.nz/files/chatham- model_32.pdf Pinkerton, M. H., Anderson, O. F., Bury, S., Brown, J., Nodder, S., & Forman, J. (2015a). Marine Trophic Index based on research trawl surveys of the Chatham Rise, 1992–2014. Pinkerton, M. H., Anderson, O. F., Dunn, M., Edwards, C., & Roux, M. (2017). Ocean ecosystem indicators for New Zealand : review of the Marine Trophic Index. Report prepared for Ministry for the Environment. Pinkerton, M. H., Lundquist, C. J., Duffy, C. A. J., & Freeman, D. J. (2008). Trophic modelling of a New Zealand rocky reef ecosystem using simultaneous adjustment of diet, biomass and energetic parameters. Journal of Experimental Marine Biology and Ecology, 367(2), 189–203. https://doi.org/10.1016/j.jembe.2008.09.022 Pinkerton, M. H., MacDiarmid, A., Beaumont, J., Bradford-Grieve, J., Francis, M. P., Jones, E., Lalas, C., Lundquist, C. J., Mckenzie, A., Nodder, S. D., Paul, L., Stenton-Dozey, J., Thompson, D., & Zeldis, J. (2015b). Changes to the food-web of the Hauraki Gulf during the period of human occupation: a mass-balance model approach. New Zealand Aquatic Environment and Biodiversity Report No. 160. www.mpi.govt.nz/news- resources/publications.aspx Pinnegar, J. K., Jennings, S., O’Brien, C. M., & Polunin, N. V. C. (2002). Long-term changes

258

in the trophic level of the Celtic Sea fish community and fish market price distribution. Journal of Applied Ecology, 39(3), 377–390. https://doi.org/10.1046/j.1365- 2664.2002.00723.x Pinnegar, J. K., & Polunin, N. V. C. (1999). Differential fractionation of δ13C and δ15N among fish tissues: implications for the study of trophic interactions. Functional Ecology, 13(2), 225–231. https://doi.org/10.1046/j.1365-2435.1999.00301.x Pitcher, T. J., Kalikoski, D., Short, K., Varkey, D., & Pramod, G. (2009). An evaluation of progress in implementing ecosystem-based management of fisheries in 33 countries. Marine Policy, 33(2), 223–232. https://doi.org/10.1016/j.marpol.2008.06.002 Pocheville, A. (2015). The ecological niche: history and recent controversies. In Handbook of Evolutionary Thinking in the Sciences (pp. 547–586). Springer Netherlands. https://doi.org/10.1007/978-94-017-9014-7_26 Polechová, J., & Storch, D. (2019). Ecological niche. In Encyclopedia of Ecology (2nd ed., pp. 72–80). Elsevier. https://doi.org/10.1016/B978-0-12-409548-9.11113-3 Polis, G. A., Anderson, W. B., & Holt, R. D. (1997). Towards an integration of landscape and food web ecology: the dynamics of spatially subsidized food webs. Annu. Rev. Ecol. Syst, 28, 289–316. Popp, B. N., Graham, B. S., Olson, R. J., Hannides, C. C. S., Lott, M. J., López‐Ibarra, G. A., Galván‐Magaña, F., & Fry, B. (2007). Insight into the trophic ecology of yellowfin tuna, Thunnus albacares, from compound‐specific nitrogen isotope analysis of proteinaceous amino acids. In T. E. Dawson & R. T. W. Siegwolf (Eds.), Terrestrial Ecology (First edit, pp. 173–190). Elsevier Ltd. https://doi.org/10.1016/S1936-7961(07)01012-3 Post, D. M. (2002). Using stable isotopes to estimate trophic position: models, methods, and assumptions. Ecology, 83(3), 703–718. https://doi.org/10.2307/3071875 Post, D. M., Layman, C. A., Arrington, D. A., Takimoto, G., Quattrochi, J., & Montaña, C. G. (2007). Getting to the fat of the matter: models, methods and assumptions for dealing with lipids in stable isotope analyses. Oecologia, 152(1), 179–189. https://doi.org/10.1007/s00442-006-0630-x Quevedo, M., Svanbäck, R., & Eklöv, P. (2009). Intrapopulation niche partitioning in a generalist predator limits food web connectivity. Ecology, 90(8), 2263–2274. https://doi.org/10.1890/07-1580.1 Quillfeldt, P., & Masello, J. F. (2020). Compound-specific stable isotope analyses in Falkland Islands seabirds reveal seasonal changes in trophic positions. BMC Ecology, 20(1), 1–12. https://doi.org/10.1186/s12898-020-00288-5 R Core Team. (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing. Ramos, R., & González-Solís, J. (2012). Trace me if you can: the use of intrinsic biogeochemical markers in marine top predators. Frontiers in Ecology and the Environment, 10(5), 258–266. https://doi.org/10.1890/110140 Reed, D. C., Rassweiler, A. R., Miller, R. J., Page, H. M., & Holbrook, S. J. (2016). The value of a broad temporal and spatial perspective in understanding dynamics of kelp forest ecosystems. Marine and Freshwater Research, 67(1), 14–24. https://doi.org/10.1071/MF14158 Rijnsdorp, A. D., Peck, M. A., Engelhard, G. H., Möllmann, C., & Pinnegar, J. K. (2009). Resolving the effect of climate change on fish populations. ICES Journal of Marine Science, 66(7), 1570–1583. https://doi.org/10.1093/icesjms/fsp056

259

Roberts, C. D., Stewart, A. L., & Struthers, C. D. (2015). The Fishes of New Zealand (First edit). Te Papa Press. Rodgers, K., & Wing, S. R. (2008). Spatial structure and movement of blue cod Parapercis colias in Doubtful Sound, New Zealand, inferred from δ13C and δ15N. Marine Ecology Progress Series, 359, 239–248. https://doi.org/10.3354/meps07349 Rojas, P., Flores, H., & Sepúlveda, J. (1985). Alimentación del Bacalao de Juan Fernández Polyprion oxygeneios (Bloch y Scneider , 1801) (Pisces : Percichthyidae). Investigaciones Marinas En El Archipiélago de Juan Fernández, January, 305–309. Romero, M. C., Lovrich, G. A., Tapella, F., & Thatje, S. (2004). Feeding ecology of the crab Munida subrugosa (Decapoda: Anomura: Galatheidae) in the Beagle Channel, Argentina. Journal of the Marine Biological Association of the United Kingdom, 84(2), 359–365. https://doi.org/10.1017/S0025315404009282h Rosenberg, A. A., Bolster, W. J., Alexander, K. E., Leavenworth, W. B., Cooper, A. B., & McKenzie, M. G. (2005). The history of ocean resources: modelling cod biomass using historical records. Frontiers in Ecology and the Environment, 3(2003), 84–90. https://doi.org/10.1890/1540-9295(2005)003[0078:THOORM]2.0.CO;2 Ruiz-Cooley, R. I. I., Garcia, K. Y. Y., & Hetherington, E. D. D. (2011). Effects of lipid removal and preservatives on carbon and nitrogen stable isotope ratios of squid tissues: Implications for ecological studies. Journal of Experimental Marine Biology and Ecology, 407(1), 101–107. https://doi.org/10.1016/j.jembe.2011.07.002 Russell, B. C. (1983). The food and feeding habits of rocky reef fish of north‐eastern New Zealand. New Zealand Journal of Marine and Freshwater Research, 17(2), 121–145. https://doi.org/10.1080/00288330.1983.9515991 Russell, E. S. (1931). Some theoretical considerations on the “overfishing” problem. ICES Journal of Marine Science, 6(1), 3–20. https://doi.org/10.1093/icesjms/6.1.3 Sabadel, A. J. M., Durante, L. M., & Wing, S. R. (in press). Amino acid stable isotopes of reef fish species uncover Suess and nitrogen enrichment effects on local kelp ecosystems. Marine Ecology Progress Series. Sabadel, A. J. M., Woodward, E. M. S., Van Hale, R., & Frew, R. D. (2016). Compound- specific isotope analysis of amino acids: A tool to unravel complex symbiotic trophic relationships. Food Webs, 6, 9–18. https://doi.org/10.1016/j.fooweb.2015.12.003 Sae-Bird Scientific. (2017). Software Manual Seasoft V2 : SBE Data Processing (p. 177). Sea-Brid Scientific. [email protected] Safiq, A. D., Lockwood, J. L., & Brown, J. A. (2018). Homogenization of Fish Assemblages Off the Coast of Florida. In R. Rozzi (Ed.), Biocultural Homogenization to Biocultural Conservation (pp. 289–300). Springer Nature. https://doi.org/10.1007/978-3-319-99513- 7_18 Saldanha, L. (1997). Fauna submarina atlântica: Portugal continental, Açores, Madeira (6th editio). Mem Martins: Publicac̜ ões Europa-América. Salinger, M. J., Renwick, J., Behrens, E., Mullan, A. B., Diamond, H. J., Sirguey, P., Smith, R. O., Trought, M. C. T., Alexander, L., Cullen, N. J., Fitzharris, B. B., Hepburn, C. D., Parker, A. K., & Sutton, P. J. (2019). The unprecedented coupled ocean-atmosphere summer heatwave in the New Zealand region 2017/18: drivers, mechanisms and impacts. Environmental Research Letters, 14(4), 1–24. https://doi.org/10.1088/1748-9326/ab012a Saporiti, F., Bearhop, S., Silva, L., Vales, D. G., Zenteno, L., Crespo, E. A., Aguilar, A., & Cardona, L. (2014). Longer and less overlapping food webs in anthropogenically

260

disturbed marine ecosystems: confirmations from the past. PLoS ONE, 9(7), 1–13. https://doi.org/10.1371/journal.pone.0103132 Sarakinos, H. C. H., Johnson, M. L., & Zanden, M. J. Vander. (2002). A synthesis of tissue- preservation effects on carbon and nitrogen stable isotope signatures. Canadian Journal of Zoology, 80(2), 381–387. https://doi.org/10.1139/z02-007 SAS Institute. (2018). JMP (14.2.0). https://www.jmp.com/en_us/software/data-analysis- software.html# Savoye, N., Aminot, A., Tréguer, P., Fontugne, M., Naulet, N., & Kérouel, R. (2003). Dynamics of particulate organic matter δ15N and δ13C during spring phytoplankton blooms in a macrotidal ecosystem (Bay of Seine, France). Marine Ecology Progress Series, 255, 27–41. https://doi.org/10.3354/meps255027 Schaefer, M. B. (1991). Some aspects of the dynamics of populations important to the management of the commercial marine fisheries. Society for Mathematical Biology, 53(1), 253–279. https://doi.org/10.1007/BF02464432 Schiel, D. R., Alestra, T., Gerrity, S., Orchard, S., Dunmore, R., Pirker, J., Lilley, S., Tait, L., Hickford, M., & Thomsen, M. (2019). The Kaikōura earthquake in southern New Zealand: Loss of connectivity of marine communities and the necessity of a cross‐ ecosystem perspective. Aquatic Conservation: Marine and Freshwater Ecosystems, 29(9), 1520–1534. https://doi.org/10.1002/aqc.3122 Schlitzer, R. (2018). Ocean Data View. Schluter, D., & McPhail, J. D. (1992). Ecological character displacement and speciation in sticklebacks. The American Naturalist, 140(1), 85–108. https://doi.org/10.1086/285404 Schmidt, S. N., Vander Zanden, M. J., & Kitchell, J. F. (2009). Long-term food web change in Lake Superior. Canadian Journal of Fisheries and Aquatic Sciences, 66(12), 2118– 2129. https://doi.org/10.1139/F09-151 Schwarz, G. (1978). Estimating the dimension of a model. The Annals of Statistics, 6(2), 461– 464. https://www.jstor.org/stable/pdf/2958889.pdf?refreqid=excelsior%3A105e808a331bc3cb c728e472d5ffabc2 Shackell, N. L., Frank, K. T., Fisher, J. A. D., Petrie, B., & Leggett, W. C. (2010). Decline in top predator body size and changing climate alter trophic structure in an oceanic ecosystem. Proceedings of the Royal Society B: Biological Sciences, 277(1686), 1353– 1360. https://doi.org/10.1098/rspb.2009.1020 Shanahan, E. (2019). Latest EU fisheries management plan threatens sustainability targets and long-term ocean health. WWF Communication Post. https://wwf.panda.org/our_work/oceans/smart_fishing/latest_fishing_news/?342991/Late st-EU-fisheries-management-plan-threatens-sustainability-targets-and-long-term-ocean- health Shannon, L., Coll, M., Bundy, A., Gascuel, D., Heymans, J., Kleisner, K., Lynam, C., Piroddi, C., Tam, J., Travers-Trolet, M., & Shin, Y. (2014). Trophic level-based indicators to track fishing impacts across marine ecosystems. Marine Ecology Progress Series, 512, 115–140. https://doi.org/10.3354/meps10821 Shaw, A. (2001). Measurements of an oceanic front using a front-following algorithm for AVHRR SST imagery. Remote Sensing of Environment, 75(1), 47–62. https://doi.org/10.1016/S0034-4257(00)00155-3 Shaw, A. G. P., & Vennell, R. (2000). Variability of water masses through the Mernoo

261

Saddle, South Island, New Zealand. New Zealand Journal of Marine and Freshwater Research, 34(1), 103–116. https://doi.org/10.1080/00288330.2000.9516918 Shears, N. T., & Bowen, M. M. (2017). Half a century of coastal temperature records reveal complex warming trends in western boundary currents. Scientific Reports, 7(1), 1–9. https://doi.org/10.1038/s41598-017-14944-2 Shin, Y.-J., Rochet, M.-J., Jennings, S., Field, J. G., & Gislason, H. (2005). Using size-based indicators to evaluate the ecosystem effects of fishing. ICES Journal of Marine Science, 62(3), 384–396. https://doi.org/10.1016/j.icesjms.2005.01.004 Sigman, D. M., & Casciotti, K. L. (2001). Nitrogen isotopes in the ocean. In Encyclopedia of Ocean Sciences (pp. 1884–1894). Elsevier. https://doi.org/10.1006/rwos.2001.0172 Simmons, G., Bremner, G., Whittaker, H., Clarke, P., Teh, L., Zylich, K., Zeller, D., Pauly, D., Stringer, C., Torkington, B., & Haworth, N. (2016). Reconstruction of marine fisheries catches for New Zealand (1950-2010) (Working Paper Series). http://www.seaaroundus.org/doc/PageContent/OtherWPContent/Simmons+et+al+2016+- +NZ+Catch+Reconstruction+-+May+11.pdf Simmons, J. E. (2014). Fluid Preservation: a comprehensive reference. Rowman & Littlefield. Singh, G. G., Sinner, J., Ellis, J., Kandlikar, M., Halpern, B. S., Satterfield, T., & Chan, K. M. A. (2017). Mechanisms and risk of cumulative impacts to coastal ecosystem services: An expert elicitation approach. Journal of Environmental Management, 199, 229–241. https://doi.org/10.1016/j.jenvman.2017.05.032 Sissenwine, M. P., & Mace, P. M. (1992). ITQs in New Zealand: the era of fixed quota in perpetuity. Fishery Bulletin, 90(January), 147–160. Skinner, M. M., Cross, B. K., & Moore, B. C. (2017). Estimating in situ isotopic turnover in Rainbow Trout (Oncorhynchus mykiss) muscle and liver tissue. Journal of Freshwater Ecology, 32(1), 209–217. https://doi.org/10.1080/02705060.2016.1259127 Slooten, E., Simmons, G., Dawson, S. M., Bremner, G., Thrush, S. F., Whittaker, H., McCormack, F., Robertson, B. C., Haworth, N., Clarke, P. J., Pauly, D., & Zeller, D. (2017). Evidence of bias in assessment of fisheries management impacts. Proceedings of the National Academy of Sciences, 114(25), E4901–E4902. https://doi.org/10.1073/pnas.1706544114 Smith, I. (2013). Pre-European Maori exploitation of marine resources in two New Zealand case study areas: species range and temporal change. Journal of the Royal Society of New Zealand, 43(1), 1–37. https://doi.org/10.1080/03036758.2011.574709 Smith, P. (1998). Molecular identification of sea perch species. Final Report to the Ministry of Fisheries for Project MOF706 - Unpublished report. Smith, R. O., Vennell, R., Bostock, H. C., & Williams, M. J. M. (2013). Interaction of the subtropical front with topography around southern New Zealand. Deep Sea Research Part I: Oceanographic Research Papers, 76, 13–26. https://doi.org/10.1016/j.dsr.2013.02.007 Solemdal, P., & Sinclair, M. (1989). Johan Hjort -founder of modern Norwegian fishery research and pioneer in recruitment thinking. Cons. Int. Explor. Mer, 191, 339–344. http://www.ices.dk/sites/pub/Publication Reports/Marine Science Symposia/Phase 2/Rapport et Proces-Verbaux des Reunions - Volume 191 - 1989 - Partie 46 de 161.pdf Somes, C. J., Schmittner, A., Galbraith, E. D., Lehmann, M. F., Altabet, M. A., Montoya, J. P., Letelier, R. M., Mix, A. C., Bourbonnais, A., & Eby, M. (2010). Simulating the

262

global distribution of nitrogen isotopes in the ocean. Global Biogeochemical Cycles, 24(4), 1–16. https://doi.org/10.1029/2009GB003767 Sosa-López, A., Mouillot, D., Do Chi, T., & Ramos-Miranda, J. (2005). Ecological indicators based on fish biomass distribution along trophic levels: an application to the Terminos coastal lagoon, Mexico. ICES Journal of Marine Science, 62(3), 453–458. https://doi.org/10.1016/j.icesjms.2004.12.004 Sponaugle, S. (2010). Otolith microstructure reveals ecological and oceanographic processes important to ecosystem-based management. Environmental Biology of Fishes, 89(3–4), 221–238. https://doi.org/10.1007/s10641-010-9676-z Stats NZ. (2019). Fish monetary stock account, year ended September 1996–2018. www.stats.govt.nz%09%0A Stenseth, N. C., & Dunlop, E. S. (2009). Evolution: unnatural selection. Nature, 457(February), 803–804. https://doi.org/10.1038/457803a Stevens, D. W., Hurst, R. J., & Bagley, N. W. (2011). Feeding habits of New Zealand fishes : a literature review and summary of research trawl database records 1960 to 2000. New Zealand Aquatic Environment and Biodiversity Report No. 85. Stevens, W. D., O’Driscoll, R. L., Ballara, S. L., & Schimel, A. C. G. (2018). Trawl survey of hoki and middle-depth species on the Chatham Rise, January 2018 (TAN1801). New Zealand Fisheries Assessment Report 2018/41. http://www.mpi.govt.nz/news-and- resources/publications Stevenson, M. L. (2004). Trawl survey of the west coast of the South Island and Tasman and Golden Bays, March-April 2003 (KAHO304) - New Zealand Fisheries Assessment Report 200414. Stevenson, M. L., & MacGibbon, D. J. (2018). Inshore trawl survey of the west coast South Island and Tasman and Golden Bays, March-April 2017 (KAH1703). New Zealand Fisheries Assessment Report 2018/18. http://www.mpi.govt.nz/news-and- resources/publications Stortini, C., Frank, K., Leggett, W., Shackell, N., & Boyce, D. (2018). Support for the trophic theory of island biogeography across submarine banks in a predator-depleted large marine ecosystem. Marine Ecology Progress Series, 607, 155–169. https://doi.org/10.3354/meps12799 Styring, A. K., Kuhl, A., Knowles, T. D. J., Fraser, R. A., Bogaard, A., & Evershed, R. P. (2012). Practical considerations in the determination of compound-specific amino acid δ15N values in animal and plant tissues by gas chromatography-combustion-isotope ratio mass spectrometry, following derivatisation to their N-acetylisopropyl e. Rapid Communications in Mass Spectrometry, 26(19), 2328–2334. https://doi.org/10.1002/rcm.6322 Suring, E., & Wing, S. R. (2009). Isotopic turnover rate and fractionation in multiple tissues of red rock lobster (Jasus edwardsii) and blue cod (Parapercis colias): Consequences for ecological studies. Journal of Experimental Marine Biology and Ecology, 370(1–2), 56– 63. https://doi.org/10.1016/j.jembe.2008.11.014 Sutton, C. P. (2004). Estimation of age, growth, and mortality of giant stargazer (Kathetostoma giganteurn) from Southland trawl surveys between 1993 and 1996. New Zealand Fisheries Assessment Report 2004/38. Sutton, P. J. H. (2003). The Southland Current: A subantarctic current. New Zealand Journal of Marine and Freshwater Research, 37(3), 645–652. https://doi.org/10.1080/00288330.2003.9517195 263

Swartz, W., Sala, E., Tracey, S., Watson, R., & Pauly, D. (2010). The Spatial Expansion and Ecological Footprint of Fisheries (1950 to Present). PLoS ONE, 5(12), 1–6. https://doi.org/10.1371/journal.pone.0015143 Sweeting, C. J., Polunin, N. V. C., & Jennings, S. (2004). Tissue and fixative dependent shifts of δ13C and δ15N in preserved ecological material. Rapid Communications in Mass Spectrometry, 18(21), 2587–2592. https://doi.org/10.1002/rcm.1661 Takagi, W., Kajimura, M., Tanaka, H., Hasegawa, K., Bell, J. D., Toop, T., Donald, J. A., & Hyodo, S. (2014). Urea-based osmoregulation in the developing embryo of oviparous cartilaginous fish (Callorhinchus milii): contribution of the extraembryonic yolk sac during the early developmental period. Journal of Experimental Biology, 217(8), 1353– 1362. https://doi.org/10.1242/jeb.094649 Tallis, H. M., Wing, S. R., & Frew, R. D. (2004). Historical evidence for habitat conversion and local population decline in a New Zealand Fjord. Ecological Applications, 14(2), 546–554. https://doi.org/10.1890/03-5104 Thavarajah, R., Mudimbaimannar, V., Rao, U., Ranganathan, K., & Elizabeth, J. (2012). Chemical and physical basics of routine formaldehyde fixation. Journal of Oral and Maxillofacial Pathology, 16(3), 400–405. https://doi.org/10.4103/0973-029X.102496 Thrush, S. F., & Dayton, P. K. (2010). What can ecology contribute to ecosystem-based management? Annual Review of Marine Science, 2(1), 419–441. https://doi.org/10.1146/annurev-marine-120308-081129 Thurstan, R. H., Brockington, S., & Roberts, C. M. (2010). The effects of 118 years of industrial fishing on UK bottom trawl fisheries. Nature Communications, 1(1), 1–6. https://doi.org/10.1038/ncomms1013 Tohse, H., & Mugiya, Y. (2008). Sources of otolith carbonate: experimental determination of carbon incorporation rates from water and metabolic CO2, and their diel variations. Aquatic Biology, 1, 259–268. https://doi.org/10.3354/ab00029 .s quota management system – incoherent and conflicted׳Torkington, B. (2016). New Zealand Marine Policy, 63, 180–183. https://doi.org/10.1016/j.marpol.2015.03.017 Tracey, D. M., Horn, P. L., Andrews, A. H., Marriott, P. M., & Dunn, M. R. (2007). Age and growth, and an investigation of age validation of lookdown dory (Cyttus traversi). Final Research Report for Ministry of Fisheries Research Project LDO2004-01. Treguer, P., Nelson, D. M., Van Bennekom, A. J., DeMaster, D. J., Leynaert, A., & Queguiner, B. (1995). The silica balance in the world ocean: a reestimate. Science, 268(5209), 375–379. https://doi.org/10.1126/science.268.5209.375 Troupin, C., Barth, A., Sirjacobs, D., Ouberdous, M., Brankart, J.-M., Brasseur, P., Rixen, M., Alvera-Azcárate, A., Belounis, M., Capet, A., Lenartz, F., Toussaint, M.-E., & Beckers, J.-M. (2012). Generation of analysis and consistent error fields using the Data Interpolating Variational Analysis (DIVA). Ocean Modelling, 52–53(August), 90–101. https://doi.org/10.1016/j.ocemod.2012.05.002 Trueman, C. N., Johnston, G., O’Hea, B., & MacKenzie, K. M. (2014). Trophic interactions of fish communities at midwater depths enhance long-term carbon storage and benthic production on continental slopes. Proceedings of the Royal Society B: Biological Sciences, 281(1787), 1–10. https://doi.org/10.1098/rspb.2014.0669 Trueman, C. N., Mackenzie, K. M., & Palmer, M. R. (2012). Identifying migrations in marine fishes through stable-isotope analysis. Journal of Fish Biology, 81(2), 826–847. https://doi.org/10.1111/j.1095-8649.2012.03361.x

264

Trueman, C. N., MacKenzie, K. M., & St John Glew, K. (2017). Stable isotope-based location in a shelf sea setting: accuracy and precision are comparable to light-based location methods. Methods in Ecology and Evolution, 8(2), 232–240. https://doi.org/10.1111/2041-210X.12651 Tsuchiya, M., & Talley, L. D. (1996). Water-property distributions along an eastern Pacific hydrographic section at 135W. Journal of Marine Research, 54(3), 541–564. https://doi.org/10.1357/0022240963213583 Tu, C. Y., Chen, K. T., & Hsieh, C. H. (2018). Fishing and temperature effects on the size structure of exploited fish stocks. Scientific Reports, 8(1), 1–10. https://doi.org/10.1038/s41598-018-25403-x Tuck, I., Cole, R., & Devine, J. A. (2009). Ecosystem indicators for New Zealand fisheries. New Zealand Aquatic Environment and Biodiversity Report No. 42. Tuck, I. D., Hewitt, J. E., Handley, S. J., & Lundquist, C. J. (2017). Assessing the effects of fishing on soft sediment habitat, fauna and process. New Zealand Aquatic Environment and Biodiversity Report No. 178. Udy, J. A., Wing, S. R., Jowett, T., O’Connell‐Milne, S. A., Durante, L. M., McMullin, R. M., & Kolodzey, S. (2019a). Regional differences in kelp forest interaction chains are influenced by both diffuse and localized stressors. Ecosphere, 10(10), 1–20. https://doi.org/10.1002/ecs2.2894 Udy, J. A., Wing, S. R., O’Connell-Milne, S. A., Durante, L. M., McMullin, R. M., Kolodzey, S., & Frew, R. D. (2019b). 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 Udy, J., Wing, S., O’Connell‐Milne, S., Kolodzey, S., McMullin, R., Durante, L., & Frew, R. (2019c). Organic matter derived from kelp supports a large proportion of biomass in temperate rocky reef fish communities: Implications for ecosystem‐based management. Aquatic Conservation: Marine and Freshwater Ecosystems, 29(9), 1503–1519. https://doi.org/10.1002/aqc.3101 Valdés, J., Ortlieb, L., Gutierrez, D., Marinovic, L., Vargas, G., & Sifeddine, A. (2008). 250 years of sardine and anchovy scale deposition record in Mejillones Bay, northern Chile. Progress in Oceanography, 79(2–4), 198–207. https://doi.org/10.1016/j.pocean.2008.10.002 Van Hale, R. (2003). Stable isotope biogeochemistry of the Otago continental shelf. A thesis submitted in partial fulfilment of the requirements for the Doctor of Philosophy diploma. University of Otago. Van Hale, R. A., Frew, R. D., & Frew A, R. D. (2010). Rayleigh distillation equations applied to isotopic evolution of organic nitrogen across a continental shelf. Marine and Freshwater Research, 61(3), 369–378. https://doi.org/10.1071/MF09041 Vander Zanden, M. J., Chandra, S., Allen, B. C., Reuter, J. E., & Goldman, C. R. (2003). Historical food web structure and restoration of native aquatic communities in the Lake Tahoe (California–Nevada) basin. Ecosystems, 6(3), 274–288. https://doi.org/10.1007/s10021-002-0204-7 Vanderklift, M., & Bearham, D. (2014). Variation in δ13C and δ15N of kelp is explained by light and productivity. Marine Ecology Progress Series, 515(November), 111–121. https://doi.org/10.3354/meps10967 Venugopal, V., & Shahidi, F. (1996). Structure and composition of fish muscle. Food Reviews International, 12(2), 175–197. https://doi.org/10.1080/87559129609541074 265

Verba, J. T., Pennino, M. G., Coll, M., & Lopes, P. F. M. (2020). Assessing drivers of tropical and subtropical marine fish collapses of Brazilian Exclusive Economic Zone. Science of The Total Environment, 702, 1–37. https://doi.org/10.1016/j.scitotenv.2019.134940 Verburg, P. (2007). The need to correct for the Suess effect in the application of δ13C in sediment of autotrophic Lake Tanganyika, as a productivity proxy in the Anthropocene. Journal of Paleolimnology, 37(4), 591–602. https://doi.org/10.1007/s10933-006-9056-z Vincent, W. F., Howard‐Williams, C., Tildesley, P., & Butler, E. (1991). Distribution and biological properties of oceanic water masses around the South Island, New Zealand. New Zealand Journal of Marine and Freshwater Research, 25(1), 21–42. https://doi.org/10.1080/00288330.1991.9516451 Vlieg, P. (1988). Proximate composition of New Zealand marine finfish and shellfish. http://www.fao.org/3/ae581e/ae581e00.htm Vlieg, P., & Body, D. R. (1988). Lipid contents and fatty acid composition of some New Zealand freshwater finfish and marine finfish, shellfish, and roes. New Zealand Journal of Marine and Freshwater Research, 22(2), 151–162. https://doi.org/10.1080/00288330.1988.9516287 Wagner, A., Hülsmann, S., Horn, W., Schiller, T., Schulze, T., Volkmann, S., & Benndorf, J. (2013). Food-web-mediated effects of climate warming: consequences for the seasonal Daphnia dynamics. Freshwater Biology, 58(3), 573–587. https://doi.org/10.1111/j.1365- 2427.2012.02809.x Wainright, S. C., Fogarty, M. J., Greenfield, R. C., & Fry, B. (1993). Long-term changes in the Georges Bank food web: trends in stable isotopic compositions of fish scales. Marine Biology, 115(3), 481–493. https://doi.org/10.1007/BF00349847 Walrond, C. (2006). “Fishing industry”, Te Ara - the Encyclopedia of New Zealand. Story. http://www.teara.govt.nz/en/fishing-industry/print Walters, C. J., Christensen, V., Martell, S. J., & Kitchell, J. F. (2005). Possible ecosystem impacts of applying MSY policies from single-species assessment. ICES Journal of Marine Science, 62(3), 558–568. https://doi.org/10.1016/j.icesjms.2004.12.005 Wang, Y., Huan, Z., YuWei, C., Lu, Z., & Guangchun, L. (2019). Trophic niche width and overlap of three benthic living fish species in poyang lake: a stable isotope approach. Wetlands, 39(S1), 17–23. https://doi.org/10.1007/s13157-018-0995-8 Ward, P., & Myers, R. A. (2005). Shifts in open ocean fish communities coinciding with the commencement of commercial fishing. Ecology, 86(4), 835–847. https://doi.org/10.1890/03-0746 Ward, T. D., Algera, D. A., Gallagher, A. J., Hawkins, E., Horodysky, A., Jørgensen, C., Killen, S. S., McKenzie, D. J., Metcalfe, J. D., Peck, M. A., Vu, M., & Cooke, S. J. (2016). Understanding the individual to implement the ecosystem approach to fisheries management. Conservation Physiology, 4(1), 1–18. https://doi.org/10.1093/conphys/cow005 Warwick, R. M., & Clarke, K. R. (1993). Increased variability as a symptom of stress in marine communities. Journal of Experimental Marine Biology and Ecology, 172(1–2), 215–226. https://doi.org/10.1016/0022-0981(93)90098-9 Waterman, J. J. (2001). Measures, Stowage Rates and Yields of Fishery Products - Torry Advisory Note No. 17. http://www.fao.org/3/x5898e/x5898e00.htm#Contents Watson, R., Revenga, C., & Kura, Y. (2006). Fishing gear associated with global marine catches. Fisheries Research, 79(1–2), 103–111.

266

https://doi.org/10.1016/j.fishres.2006.01.013 West, J. B., Bowen, G. J., Dawson, T. E., & Tu, K. P. (2010). Isoscapes: understanding movement, pattern, and process on Earth through isotope mapping (Vol. 1). Springer. https://doi.org/10.1016/j.soilbio.2006.01.002 Whiteman, J. P., Smith, E. A. E., Besser, A. C., & Newsome, S. D. (2019). A guide to using compound-specific stable isotope analysis to study the fates of molecules in organisms and ecosystems. Diversity, 11(1), 1–18. https://doi.org/10.3390/d11010008 Williams, B. G. (1973). The effect of the environment on the morphology of Munida Gregaria (Fabricius) (Decapoda, Anomura). Crustaceana, 24(2), 197–210. https://doi.org/10.1163/156854073X00362 Williams, B. G. (1980). The pelagic and benthic phases of post-metamorphic Munida gregaria (Fabricius) (Decapoda, Anomura). Journal of Experimental Marine Biology and Ecology, 42(2), 125–141. https://doi.org/10.1016/0022-0981(80)90171-9 Wilson, J. (2005). European discovery of New Zealand - Abel Tasman. Te Ara - the Encyclopedia of New Zealand. http://www.teara.govt.nz/en/european-discovery-of-new- zealand/page-2 Winder, M., & Schindler, D. E. (2004). Climate change uncouples trophic interactions in an aquatic ecosystem. Ecology, 85(8), 2100–2106. https://doi.org/10.1890/04-0151 Wing, S., & Jack, L. (2012). Resource specialisation among suspension-feeding invertebrates on rock walls in Fiordland, New Zealand, is driven by water column structure and feeding mode. Marine Ecology Progress Series, 452, 109–118. https://doi.org/10.3354/meps09588 Wing, S. R., Beer, N., & Jack, L. (2012). Resource base of blue cod Parapercis colias subpopulations in marginal fjordic habitats is linked to chemoautotrophic production. Marine Ecology Progress Series, 466, 205–214. https://doi.org/10.3354/meps09929 Wing, S. R., & Jack, L. (2013). Marine reserve networks conserve biodiversity by stabilizing communities and maintaining food web structure. Ecosphere, 4(11), 1–14. https://doi.org/10.1890/ES13-00257.1 Wing, S. R., & Jack, L. (2014). Fiordland: the ecological basis for ecosystem management. New Zealand Journal of Marine and Freshwater Research, 48(4), 577–593. https://doi.org/10.1080/00288330.2014.897636 Wing, S. R., Leichter, J. J., Perrin, C., Rutger, S. M., Bowman, M. H., & Cornelisen, C. D. (2007). Topographic shading and wave exposure influence morphology and ecophysiology of Ecklonia radiata (C. Agardh 1817) in Fiordland, New Zealand. Limnology and Oceanography, 52(5), 1853–1864. https://doi.org/10.4319/lo.2007.52.5.1853 Wing, S. R., McLeod, R., Clark, K., & Frew, R. (2008). Plasticity in the diet of two echinoderm species across an ecotone: microbial recycling of forest litter and bottom-up forcing of population structure. Marine Ecology Progress Series, 360, 115–123. https://doi.org/10.3354/meps07434 Wing, S. R., & Wing, E. (2001). Prehistoric fisheries in the Caribbean. Coral Reefs, 20(1), 1– 8. https://doi.org/10.1007/s003380100142 Wold, S., Sjöström, M., & Eriksson, L. (2001). PLS-regression: a basic tool of chemometrics. Chemometrics and Intelligent Laboratory Systems, 58(2), 109–130. https://doi.org/10.1016/S0169-7439(01)00155-1 Worm, B. (2005). Global patterns of predator diversity in the open oceans. Science, 267

309(5739), 1365–1369. https://doi.org/10.1126/science.1113399 Worm, B., Hilborn, R., Baum, J. K., Branch, T. A., Collie, J. S., Costello, C., Fogarty, M. J., Fulton, E. A., Hutchings, J. A., Jennings, S., Jensen, O. P., Lotze, H. K., Mace, P. M., McClanahan, T. R., Minto, C., Palumbi, S. R., Parma, A. M., Ricard, D., Rosenberg, A. A., … Zeller, D. (2009). Rebuilding global fisheries. Science, 325(5940), 578–585. https://doi.org/10.1126/science.1173146 Xu, J., Yang, Q., Zhang, M. M., Zhang, M. M., Xie, P., & Hansson, L.-A. A. (2011). Preservation effects on stable isotope ratios and consequences for the reconstruction of energetic pathways. Aquatic Ecology, 45(4), 483–492. https://doi.org/10.1007/s10452- 011-9369-5 Yorke, C., Hanns, B., Shears, N., Page, H., & Miller, R. (2019). Living kelp versus plankton as food sources for suspension feeders. Marine Ecology Progress Series, 614, 21–33. https://doi.org/10.3354/meps12906 Young, M. (1925). Notes on the habits of whalefeed (Munida gregaria). N.Z. JI Sci. Technol., 7, 318–319. Zeldis, J. (1985). Ecology of Munida gregaria (Decapoda, Anomura) distribution and abundance, population dynamics and fisheries. Marine Ecology Progress Series, 22, 77– 99. https://doi.org/10.3354/meps022077 Zeldis, J. R., & Jillett, J. B. (1982). Aggregation of pelagic Munida gregaria (Fabricius) (Decapoda, Anomura) by coastal fronts and internal waves. Journal of Plankton Research, 4(4), 839–857. https://doi.org/10.1093/plankt/4.4.839

268