Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Stable isotope analysis of Rivers Inlet sockeye salmon (Oncorhynchus nerka) : investigating the contribution… Doson Coll, Yago 2015

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2015_september_dosoncoll_yago.pdf [ 2.13MB ]
Metadata
JSON: 24-1.0167721.json
JSON-LD: 24-1.0167721-ld.json
RDF/XML (Pretty): 24-1.0167721-rdf.xml
RDF/JSON: 24-1.0167721-rdf.json
Turtle: 24-1.0167721-turtle.txt
N-Triples: 24-1.0167721-rdf-ntriples.txt
Original Record: 24-1.0167721-source.json
Full Text
24-1.0167721-fulltext.txt
Citation
24-1.0167721.ris

Full Text

STABLE ISOTOPE ANALYSIS OF RIVERS INLET SOCKEYE SALMON (ONCORHYNCHUS NERKA): INVESTIGATING THE CONTRIBUTION OF ENVIRONMENTAL CONDITIONS IN THE HIGH SEAS TO BRITISH COLUMBIA POPULATION DECLINES by Yago Doson Coll  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in The Faculty of Graduate and Postdoctoral Studies (Resource Management and Environmental Studies)  THE UNIVERISTY OF BRITSH COLUMBIA (Vancouver) May 2015 © Yago Doson Coll, 2015 ii  Abstract Sockeye salmon (Oncorhynchus nerka) populations in BC have undergone varying degrees of decline coinciding with a shift to a warmer phase of the Pacific Decadal Oscillation (PDO) in 1977. The PDO, and other climate cycles, have been shown to significantly affect the physical and biological characteristics of the North East Pacific Ocean. Changes in ocean productivity have implications for pelagic food webs and may cause shifts in the abundance of potential prey for sockeye salmon, impacting their long-term production patterns. We investigated the coupling of ocean conditions and population fluctuations using Rivers Inlet as a case study, a system that suffered probably the most catastrophic sockeye stock collapse in BC history. Stable isotope analysis was used to access information on ocean conditions stored in the carbon and nitrogen isotope ratios of archived sockeye scales for the period 1915-2013. Our results indicated that Rivers Inlet sockeye salmon experienced highly variable open ocean conditions during this period. Both decadal scale shifts in North Pacific climate (e.g., PDO) and interannual scale shifts in climate (e.g., El Niño/La Niña events) were reflected in the physical and biological environment of the offshore Gulf of Alaska. Positive phases of the PDO and El Niño events were associated with a warmer and less productive ocean, while negative phases of the PDO and La Niña events were associated with a colder and more productive ocean. Moreover, the carbon and nitrogen stable isotope time-series indicated that the foraging habits of Rivers Inlet sockeye salmon were affected by these shifts of North Pacific climate. A lengthening (shortening) of the food web was associated to warm (cold) and less productive (more productive) periods. In addition, the isotope data also supports Rivers Inlet sockeye salmon shifting diet depending upon prey availability. We concluded that a combination of the iii  two factors was responsible for the changes in the feeding ecology of Rivers Inlet sockeye salmon during the period 1915-2013. Such variation in the feeding ecology of Rivers Inlet sockeye salmon could potentially have a negative effect in the overall survival rates of sockeye salmon.             iv  Preface This thesis is the result of a long journey that started four years ago. Both my supervisor Dr. Evgeny Pakhomov and co-supervisor Dr. Brian Hunt started the work for this thesis way before I did. They proposed the research topic, applied for funding and made all the arrangements needed in order to obtain all tissue samples and perform the stable isotope analysis. The collaboration of the Department of Fisheries and Oceans Canada (DFO) and the Wuikinuxv Nation was vital to obtain all the tissue samples used in this study. Dr. Sven Kaehler at the IsoEnvironmental lab in Rhodes University performed the stable isotope analysis, providing the isotope ratios of Rivers Inlet sockeye salmon scales for us to interpret.  The amount of hours that I spent gathering and reading scientific literature, running statistical analyses, and interpreting the resulting data were not as surprising as the amount of hours I spent trying to figure out how to structure every paragraph of this thesis. To anyone reading this, if you ever have trouble to figure out a problem and you cannot think of a good idea on how to approach it, go outside for a walk, a run or a cycle. I had the most brilliant ideas while riding my bike.       v  Table of contents Abstract ......................................................................................................................................................... ii Preface ......................................................................................................................................................... iv Table of contents .......................................................................................................................................... v List of tables ................................................................................................................................................. vii List of figures ............................................................................................................................................... viii Acknowledgements ....................................................................................................................................... x Dedication .................................................................................................................................................... xii  Introduction .......................................................................................................................................... 1 1 Materials and methods ....................................................................................................................... 15 22.1 Overview of study site and Owikeno sockeye stock. .................................................................. 15 2.2 Sample collection ........................................................................................................................ 17 2.2.1 Archived samples ................................................................................................................ 17 2.2.2 Recent samples ................................................................................................................... 18 2.3 Sample preparation .................................................................................................................... 20 2.4 Measurement of stable isotopes ................................................................................................ 21 2.4.1 13C correction for Suess Effect .......................................................................................... 21 2.4.2 13C correction for C:N ........................................................................................................ 22 2.4.3 Muscle and scale isotopic offset ......................................................................................... 26 2.5 Statistical analyses ...................................................................................................................... 26 2.6 Principal component analysis (PCA) ............................................................................................ 27 2.6.1 Data and methods. .............................................................................................................. 28 2.7 Additional data sets .................................................................................................................... 32 2.7.1 Baseline Isotope data. ......................................................................................................... 32 2.7.2 Nutrient data. ...................................................................................................................... 33  Results ................................................................................................................................................. 34 33.1 Comparison of 2013 pre-spawn and post-spawn samples ......................................................... 34 3.2 Isotopic relatioship between muscle and scale tissue ................................................................ 42 3.3 Carbon and nitrogen stable isotope time-series from rivers inlet sockeye salmon scales ......... 47 3.4 Principal component analysis (PCA) ............................................................................................ 52 3.4.1 PCA on the 33 environmental time-series (PCAenv) ............................................................ 52 vi  3.4.2 PCA on the 15 physical time-series (PCAphys) ...................................................................... 57 3.4.3 PCA on the 18 biological time-series (PCAbiol) ..................................................................... 60 3.5 Relationships between environmental conditions and stable isotope time-series.................... 64 3.6 Baseline isotopic data ................................................................................................................. 66 3.7 Nutrient data ............................................................................................................................... 68  Discussion ............................................................................................................................................ 70 44.1 Comparison of pre-spawn and post-spawn isotope data: physiological implications ............... 70 4.2 Isotopic relationship between muscle and scale tissue .............................................................. 72 4.3 Principal component analysis (PCA) ............................................................................................ 73 4.4 Stable isotope variability ............................................................................................................. 75 4.5 Stable isotopes and climate shifts .............................................................................................. 85 4.6 Environmental conditions and sockeye salmon returns ............................................................. 91 4.7 Conclusions ................................................................................................................................. 95  References ................................................................................................................................................. 97 Appendix  .................................................................................................................................................. 108      vii  List of tables Table 2.1. Numeric and alphabetic abbreviations for the 33 environmental time-series used in this analysis. A brief description and source of information are provided for each time-series in Appendix A. The time-series are plotted geographically in Figure 2.5. ...................................... 29 Table 3.1. Summary statistics of 13C, 13C’ and 15N obtained from muscle and scale tissue samples of 2011, 2012, 2013 pre-spawn and 2013 post-spawn Rivers Inlet sockeye salmon...................... 46 Table 3.2. Paired t-test results comparing stable isotope signatures from scales and muscle tissue samples of 2011, 2012, 2013 pre-spawn and 2013 post-spawn Rivers Inlet sockeye salmon. .. 46 Table 3.3. Observed range of 13C’ and 15N in the stable isotope analysis of three different populations of sockeye salmon. ...................................................................................................................... 49 Table 3.4. Eigenvalues, percentage of variance and cumulative percentage of variance for the first five principal components of the principal component analysis performed to the 33 environmental variables. ..................................................................................................................................... 52 Table 3.5. The 33 environmental time-series and their corresponding loadings of the first two principal components of the PCA. High loadings (r>|0.1|) are underlined. .............................................. 56 Table 3.6. Eigenvalues, percentage of variance and cumulative percentage of variance for the first five principal components of the principal component analysis performed to the 15 physical time-series. .......................................................................................................................................... 57 Table 3.7. The 15 physical time-series and their corresponding loadings of the first two principal components of the PCA. High loadings (r>|0.2|) are underlined. .............................................. 59 Table 3.8. Eigenvalues, percentage of variance and cumulative percentage of variance for the first five principal components of the principal component analysis performed to the 18 biological time-series. .......................................................................................................................................... 60 Table 3.9. The 18 biological time-series and their corresponding loadings of the first two principal components of the PCA. High loadings (r>|0.18|) are underlined. ............................................ 63 Table 3.10. Statistical results from simple linear regression of nitrogen and carbon stable isotope time-series with 0, 1 and 2 year lag PCA scores. Correlation coefficient (r), the probability (p-value) that the correlation was greater (or less) than 0, degrees of freedom (df), and number of data points compared (N) are given. Correlations significant at < 0.05 are shaded. .......................... 65 Table 3.11. Statistical results from simple linear regression of nitrogen stable isotope time-series for the period 1986-2010 with 0, 1 and 2 year lag NO3 time-series. Correlation coefficient (r), the probability (p-value) that the correlation was greater (or less) than 0, degrees of freedom (df), and number of data points compared (N) are given. .................................................................. 69  viii  List of figures  Figure 1.1. General circulation in the Northeast Pacific showing Ocean Station Papa (OSP) and the four major domains defined by Ware and McFarlane (1989). ............................................................. 3 Figure 1.2. Rivers Inlet (Statistical Area 9) sockeye salmon catch and escapement data for the period 1948-2012. .................................................................................................................................. 14 Figure 2.1. Location of Rivers Inlet on the central coast of British Columbia. ........................................... 15 Figure 2.2. Major sockeye salmon spawning areas in the Owikeno Lake watershed. ............................... 16 Figure 2.3. Relationship between C:N ratios and 13C of Rivers Inlet sockeye scales fitted using an exponential model for (a) all sampled years and, (b) 2013 pre-spawn and post-spawn samples. ..................................................................................................................................................... 23 Figure 2.4. Relationship between C:N ratios and 13C of Rivers Inlet sockeye muscle tissue fitted using an exponential model for 2013 pre-spawn and post-spawn samples. ............................................ 25 Figure 2.5. Geographic distribution of our 33 environmental time-series. Their location on the map indicates where each variable was measured or has influence. See Table 2.1 for a definition of each abbreviation. ....................................................................................................................... 30 Figure 3.1. Boxplot of 2013 non-lipid-normalized 13C stable isotope ratios for Rivers Inlet sockeye salmon (a) muscle of pre-spawn and post-spawn fish and, (b) scales of pre-spawn and post-spawn salmon. Median values (solid horizontal line), interquartile range (box outline), and 95% confidence intervals (whiskers) are shown. ................................................................................ 35 Figure 3.2. Boxplot of 2013 C:N ratios for Rivers Inlet sockeye salmon (a) muscle of pre-spawn and post-spawn fish and, (b) scales of pre-spawn and post-spawn salmon. Median values (solid horizontal line), interquartile range (box outline), and 95% confidence intervals (whiskers) are shown. ......................................................................................................................................... 37 Figure 3.3. Boxplot of 2013 lipid-normalized 13C (13C ‘) stable isotope ratios for Rivers Inlet sockeye salmon (a) muscle of pre-spawn and post-spawn fish and, (b) scales of pre-spawn and post-spawn salmon. Median values (solid horizontal line), interquartile range (box outline), and 95% confidence intervals (whiskers) are shown. ................................................................................ 39 Figure 3.4. Boxplot of 2013 15N stable isotope ratios for Rivers Inlet sockeye salmon (a) muscle of pre-spawn and post-spawn fish and, (b) scales of pre-spawn and post-spawn salmon. Median values (solid horizontal line), interquartile range (box outline), and 95% confidence intervals (whiskers) are shown. ................................................................................................................. 41 Figure 3.5. Boxplot of differences in non-lipid-normalized 13C stable isotope ratios between muscle and scales (13Cm-sc) of 2011, 2012, 2013 pre-spawn and 2013 post-spawn Rivers Inlet sockeye salmon. Median values (solid horizontal line), interquartile range (box outline), and 95% confidence intervals (whiskers) are shown. ................................................................................ 43 Figure 3.6. Boxplot of differences in lipid-normalized 13C (13C ‘) stable isotope ratios between muscle and scales (13C’m-sc) of 2011, 2012, 2013 pre-spawn and 2013 post-spawn Rivers Inlet sockeye salmon. Median values (solid horizontal line), interquartile range (box outline), and 95% confidence intervals (whiskers) are shown. ................................................................................ 44 ix  Figure 3.7. Boxplot of differences in 15N stable isotope ratios between muscle and scales (15Nm-sc) of 2011, 2012, 2013 pre-spawn and 2013 post-spawn Rivers Inlet sockeye salmon. Median values (solid horizontal line), interquartile range (box outline), and 95% confidence intervals (whiskers) are shown. ................................................................................................................. 45 Figure 3.8. Yearly C:N mean of Rivers Inlet sockeye salmon scales for the period from 1915-2013. Error bars represent standard deviation. ............................................................................................. 47 Figure 3.9. Yearly 13C (‰) mean of Rivers Inlet sockeye salmon scales for the period from 1915-2013. 48 Figure 3.10. Yearly (a) 13C’ (‰) and (b) 15N (‰) mean of Rivers Inlet sockeye salmon scales for the period from 1915-2013. Dashed vertical line indicates the 1977 regime shift. Error bars represent standard deviation. 13C data are lipid-normalized and corrected for Suess effect (see Materials and Methods section). ......................................................................................... 50 Figure 3.11. Relationship between the 13C’ and 15N time-series of RI sockeye salmon scales for the period 1915-2013. ....................................................................................................................... 51 Figure 3.12. The first two principal component scores from a principal component analysis of the 33 environmental time-series. Blue colored areas correspond to negative phases while red colored areas correspond to positive phases. Vertical dashed lines indicate the 1977 and 1989 regime shifts. ............................................................................................................................... 53 Figure 3.13. The first two principal component scores from a principal component analysis of the 15 physical time-series (PCAphys). . Blue colored areas correspond to negative phases while red colored areas correspond to positive phases. Vertical dashed lines indicate the 1977 and 1989 regime shifts. ............................................................................................................................... 58 Figure 3.14. The first two principal component scores from a principal component analysis of the 18 biological time-series. . Blue colored areas correspond to negative phases while red colored areas correspond to positive phases. Vertical dashed lines indicate the 1977 and 1989 regime shifts. ........................................................................................................................................... 61 Figure 3.15. Mean (± SD) nitrogen stable isotope ratios from sockeye salmon prey and suspended particulate organic matter (SPOM) collected between May 1991 and May 1993 along the Line P and at Ocean Station Papa (OSP) (Wu et al., 1996; 1998); squid samples taken during spring and summer of 1992 and 1994 from the continental shelf of the Gulf of Alaska (δ15N shelf signatures were converted to δ15N oceanic signatures based on the ∆δ15N between oceanic and shelf copepod signatures -Hobson et al., 1997); and from sockeye salmon scales collected for the period 1915-2013 (this study). Sample sizes given in parentheses. ............................... 67 Figure 3.16. Black dashed line shows average yearly nitrate concentration (M) in surface waters in the Gulf of Alaska between 50-57N and 145-160W for the period 1985-2010. Red line shows yearly 15N (‰) mean of Rivers Inlet sockeye salmon scales for the period 1986-2010............ 68 Figure 4.1. Summary of expected interactions between open ocean environmental conditions and Rivers Inlet sockeye salmon scale stable isotope signatures in response to climate variability. ........ 684      x  Acknowledgements  This research was made possible by funding provided by The Ambrose Monell Foundation and The G. Unger Vetlesen Foundation.  Also, I would like to express my most sincere gratitude to several individuals, who in one way or another contributed to the completion of this study. First, a very special thanks goes out to my supervisor Dr. Evgeny Pakhomov (UBC) and my co-supervisor Dr. Brian Hunt (UBC) for offering me the opportunity to lead this research project. Without their input, comments and advice this thesis would not be called a thesis. I am immensely grateful to Dr. Tony Pitcher (UBC) for giving me the opportunity to enroll as a UBC graduate student and for introducing me to Dr. Evgeny Pakhomov and Dr. Brian Hunt. I would also like to thank Dr. Scott Hinch (UBC) at the Faculty of Forestry for agreeing to be part of my examination committee and accepting to review my thesis in such a short period of time.  I am indebted to Darlene Gillespie of the Department of Fisheries and Oceans Canada (DFO), not only for providing access to archived sockeye salmon scales, but for her invaluable assistance through the entire research study. I am also indebted to Wayne Levesque (DFO) and Jason Slade (DFO) for accommodating me at Genesee camp, making the 2013 post-spawn samples collection possible. I am very grateful to the Wuikinuxv Nation for their collaboration in the project and the collection of 2011, 2012 and 2013 pre-spawn fish, with special thanks to Billie Johnson, Chris McConechy, and Jennifer Walkus. I thank Larysa Pakhomova (UBC) for laboratory assistance and Dr. Sven Kaehler (Rhodes University) for mass spectrometer isotopic analysis of scales and muscle tissue. I would like to thank Dr. Marc Trudel (DFO) and Dave Mackas (DFO, retired) for their input and valuable insights relevant to this study; and to Dr. Moira Galbraith (DFO), James Ingraham (NOAA, retired), Tracy Cone (DFO), Jennifer Shriver (Alaska Department xi  of Fish and Game), Steve Heinl (Alaska Department of Fish and Game), and Tamara Fraser (DFO) for providing important data sets needed for the completion of this research project. I thank Dr. Amanda Vincent (UBC), Dr. Rashid Sumaila (UBC), and Dr. Daniel Pauly (UBC) for their kind concern and consideration regarding my academic requirements. I also want to thank Dr. Michael Lipsen (UBC) at the Department of Earth and Ocean Sciences for accepting me as his Teacher Assistant during my academic years at UBC and for his helpful input and advice on this research project. I am forever grateful to my dear friend and colleague Anna Schuhbauer (UBC), not only for her help and input in writing, proof-reading and formatting this manuscript but for her unconditional support and friendship. I would also like to thank my fellow colleagues and friends at the Fisheries Centre (UBC) Catarina Wor, Roberto Licandeo, Dr. Ricardo Amoroso, Dr. Dana Miller and Shannon Obradovich for their input and advice in “R”, their moral support, and the overall help they offered me through my courses and research project.  I would also like to thank my lab colleagues and friends at the Department of Earth and Ocean Sciences (UBC), Kang Wang, Yulia Egorova, Nikita Sergeenko and Joanne Breckenridge for their friendship and support. Finally, I would like to thank with all my heart my dear friends Lara Marcus Zamora, Randy P. Morris, Jandries Van Aardt and Clara Calatayud Pavia for their financial and moral support, love, and friendship. I am here today because of them.      xii  Dedication I lovingly dedicate this thesis to my life companion Randy P. Morris and my best friend Lara Marcus Zamora, because without you none of this would have been possible. I also dedicate this thesis to my dear mother Genoveva Coll Castillo and my dear sister Cristina Doson Coll, because I miss you and I hope one day I will be able to help you as much as you helped me. Last but not least, I dedicate this thesis to the memory of Dr. Sven Kaehler, who was deeply involved in the project and ultimately made it happen.          1   Introduction 1Pacific salmon (Oncorhynchus spp.) are widely distributed in the North Pacific rim. Their range extends from southern California, north along the Canadian and Alaskan coasts to rivers draining into the Arctic Ocean, and then southward along Asian coastal areas of Russia, Japan, and Korea (Groot and Margolis, 1991). There are 8 species of Pacific salmon 6, of which (sockeye, pink, chum, chinook, coho, and steelhead salmon) are commonly found on the west coast of North America. Generally, all Pacific salmon are anadromous; they are born in rivers and streams, and migrate to the ocean, before returning to freshwater sources to spawn. In fact, two-thirds or more of the life history of Pacific salmon is normally spent in the pelagic environment of the open Pacific Ocean (Ware and McFarlane, 1989). Therefore, it is important to understand marine processes in the high seas and how they influence salmon growth and survival.  After entering the marine environment, British Columbia sockeye salmon (Oncorhynchus nerka) populations move north and west through the coastal waters of Canada and Alaska, and offshore into the pelagic Gulf of Alaska by the end of their first year at sea. They remain in the Gulf of Alasksa for approximately 1-3 years prior to their spawning migration back to their natal fresh waters (Burgner, 1991). During the pelagic phase of their life cycle, sockeye salmon move across two oceanic domains of the Gulf of Alaska, the Central Subarctic and the Transitional domains (Figure 1.1) (Pearcy, 1997). The Central Subarctic domain includes waters of the Alaska Gyre, whereas the Transitional Domain includes the easterly flowing Subarctic current (Figure 1.1) (Pearcy, 1997). Both oceanic domains lie in one of the major high-nitrate, low-chlorophyll 2  (HNLC) regions of the world’s oceans. Nutrients are replenished in surface waters by wind mixing from intense winter storms (Pearcy, 1997). Moreover, cyclonic circulation around the Gulf of Alaska and Ekman pumping from wind stress curl cause upwelling of nutrient-rich water in the Alaska Gyre (Pearcy, 1997). Even though nutrients are relatively plentiful, phytoplankton communities are dominated by small cells and a spring phytoplankton bloom does not develop, either because of intense grazing by microzooplankton or a lack of essential micronutrients such as iron (Pearcy, 1997). Nonetheless, the Central Subarctic and Transitional domains provide for offshore foraging for sockeye salmon, where they feed on a variety of prey items, including squid, fish, euphausiids, amphipods, copepods, pteropods, crustacean larvae and pelagic polychaetes (Brodeur, 1990). Such variation in their food habits is closely associated with spatial and temporal variation in prey abundance and availability (Aydin et al., 2000; Brodeur, 1990; Kaeriyama et al., 2000; Kaeriyama et al., 2004). For example, sockeye salmon in the Gulf of Alaska will switch their diets from highly caloric micronekton such as gonatid squid (Berryteuthis anonychus) to a lower trophic level diet based on zooplankton at times and places where squid are largely absent (Aydin et al., 2000; Brodeur, 1990; Kaeriyama et al., 2000; Kaeriyama et al., 2004).  3   Figure 1.1. General circulation in the Northeast Pacific showing Ocean Station Papa (OSP) and the four major domains defined by Ware and McFarlane (1989).   Large-scale shifts in climate affecting the Northeast Pacific have been correlated to salmon production (Hare and Francis, 1995; Mantua et al., 1997), salmon growth, age at maturity, and survival (Holt and Peterman, 2004; Pyper et al., 1999; Ruggerone et al., 2007). For example, it is now accepted that the North Pacific Ocean suffered a drastic climatic regime shift in the winter of 1976-1977 (Hare and Mantua, 2000). Mantua et al. (1997) termed this event the Pacific Decadal Oscillation (PDO) and it can be described as a pan-Pacific, recurring pattern of ocean-atmosphere variability that alternates between climate regimes every 20-30 years (Hare et al., 1999). Several studies have documented the climatic and ecosystem changes that took place during this regime shift and it is known that it had widespread consequences for the biota of 4  the North Pacific, including a major reorganization of the food web structure in the Gulf of Alaska (Hare and Francis, 1995; Hare and Mantua, 2000; Mantua et al., 1997). Coinciding with the regime shift, West Coast stocks (BC, Washington and Oregon) of Pacific salmon (Oncorhynchus spp.) experienced sharp declines and unstable returns, while in Alaska Pacific salmon production almost doubled after the late 1970’s (Hare et al., 1999, Mantua et al., 1997). Such fluctuations in Pacific salmon population trends, as they relate to large-scale climate variability, have been studied extensively (e.g. Francis and Hare, 1994; Hare and Francis, 1995; Mantua et al., 1997; Peterman et al., 1998). However, the underlying mechanisms linking climate forcing and salmon production remain poorly understood. West Coast stocks of sockeye salmon spend most of their lives foraging in the offshore waters of the Gulf of Alaska (Ware and McFarlane, 1989) where they put on up to 90% of their final adult body weight (Ishida et al., 1998). Thus, it is not unreasonable to think that climate processes affecting growth and survival of sockeye salmon stocks during the pelagic stage of their life cycle will impact long-term salmon production patterns. A brief review of some of the large-scale changes in climate affecting the Northeast Pacific ocean may help to understand when climate and oceanography have the opportunity to affect sockeye salmon abundance in the open ocean.  The 1977 regime shift was not the only occurrence of this phenomenon during the past century. There are indicators of two previous shifts in 1925 and 1947 (Mantua et al., 1997), as well as a fourth shift in 1989 (Hare and Mantua, 2000), and a possible fifth in 1999 (e.g. Batten and Welch, 2004; Litzow, 2006.). The last abrupt regime shift occurred in the winter of 2007/08 (Litzow and Mueter, 2014). Such climate shifts alternate between “warm” or positive phases and “cold” or negative phases of the PDO. In a similar way, the Northeast Pacific Ocean has 5  been affected by numerous El Niño/La Niña periods over the last century (e.g. Whitney and Welch, 2002), which are the equivalent to shorter positive and negative phase of the North Pacific climate regime respectively (Mantua et al., 1997). During positive phases (and El Niño events), there is an intensification and easterly shift in location of the Aleutian Low during the winter, resulting in a strong flow of warm, moist air into Alaska from the south, increased wind mixing, intensified Alaska current transporting warm waters from the south, and increased Ekman pumping in the cyclonic circulation (Trenberth and Hurrell, 1995). As a consequence, the Gulf of Alaska experiences increased sea surface temperatures (SSTs) and decreased salinity levels, leading to shallower mixed layers and increased stratification (Mantua and Hare, 2002). In contrast, negative phases (and La Niña events) correspond to opposite spatial patterns of North Pacific climate variation (Mantua and Hare, 2002). A weakening and retreat of the Aleutian Low toward the west-northwest occurs during cold phases (Rodionov et al., 2007), while upwelling-favorable winds strengthen over the California Current (Peterson and Schwing, 2003). A weaker Aleutian Low leads to less warm air being advected from the Northern Pacific, leading to below average SSTs in the Gulf of Alaska (Wendler, 2012), which in turn result in a less stratified ocean (Chhak and Di Lorenzo, 2007).   Such variation in the physical properties of the offshore waters of the Gulf of Alaska have been associated with changes in light availability and nutrient supply for phytoplankton, leading to large fluctuations in primary production, which in turn would impact zooplankton and fish populations that mature in the high seas (Whitney et al., 1998; Whitney and Freeland, 1999). Moreover, warmer SSTs in the Gulf of Alaska have been associated with changes in zooplankton communities (Mackas et al., 1998; Batten and Welch, 2004). Furthermore, North American 6  sockeye salmon stocks in the Gulf of Alaska appear to be distributed further north and west in warm winters than in cool winters (Blackbourn, 1987). Reduced overlap with highly caloric prey such as squid, largely absent in the northern latitudes of the Gulf of Alaska (Aydin et al., 2000; Kaeriyama et al., 2000; Kaeriyama et al., 2004), may result from sockeye being distributed further north. Therefore, climate shifts may affect the trophic ecology of sockeye salmon by limiting prey items, forcing a shift to a lower caloric diet, and/or resulting in a longer food chain and lower energy transfer efficiency, which in turn can negatively affect sockeye salmon growth and survival in the high seas.  Warmer SSTs in the high seas have been previously correlated to smaller adult sockeye (e.g., Cox and Hinch, 1996; Hinch et al., 1995b). Hinch et al. (1995b) developed empirical relationships between SSTs, zooplankton biomass, and terminal ocean weight for the early Stuart stock of Fraser River sockeye salmon. Their results indicated that warmer temperatures and lower food availability were associated with smaller sockeye (Hinch et al., 1995b). As this study may have been specific to a single stock, Cox and Hinch (1996) extended the analysis to 10 different stocks of Fraser River sockeye salmon. They concluded that size at maturity of both males and females in all stocks was smaller in years when SSTs were relatively warm (Cox and Hinch, 1996). Slower growth in warmer years may be caused indirectly by the changes in the feeding ecology of sockeye salmon influenced by oceanic processes listed above (e.g., reduced primary and secondary production associated to warm periods) and/or directly by increased metabolic demand associated to increased SSTs (Hinch et al., 1995b). This may be exacerbated by their more northerly distribution during warm years requiring a longer migration back to natal rivers for southern sockeye salmon stocks. Bioenergetics modeling supports a coupled 7  effect of higher metabolic costs and lower food availability and/or quality during warm periods in reduced adult size (Hinch et al., 1995b). Competitive interactions with other salmon species may also affect salmon populations, limiting prey and/or causing dietary shifts in response to increased inter- and intra-specific competition (Ruggerone et al., 2003; Tadokoro et al., 1996; Welch and Parsons, 1993).  The current understanding of how these environmentally driven changes relate to salmon trophic ecology and production in the high seas remains limited. Several studies on the high seas have given insight into migration patterns and ocean foraging behavior of salmon (e.g. French et al., 1976; Brodeur, 1990; Kaeriyama et al., 2000; Kaeriyama et al., 2004) but costs and logistical difficulties of working on the high seas have limited the research. Significant gaps remain in our understanding of salmon at sea, particularly after the initial post-smolt migration, and these knowledge gaps are a major obstacle to stock based management and conservation.  Stomach content analyses can be used to evaluate some aspects of the trophic ecology of sockeye salmon. However, such analyses emphasize what is present in the gut at the time of capture, but not necessarily the long-term accumulation of food. Furthermore, gut content analysis of salmon sampled in the open ocean is logistically a challenge. An alternative method is stable isotope analysis of a consumer tissue. This technique is advantageous because trophic level information is based on assimilated diet, not just recently ingested items. It is now well established that fish scales can provide the same information as soft tissues (Satterfield and Finney, 2002). The entire scale reflects the isotopic signature of the diet of that particular fish over its life, although biased toward the last growing season or period when most growth was 8  accumulated (last year at sea) (Trueman and Moore, 2007). Since fish scales are often archived for aging purposes, salmon scales present a powerful tool to retrospectively investigate the foraging ecology of sockeye salmon in the high seas.  Ratios of naturally occurring carbon (C) and nitrogen (N) stable isotopes of fish tissues reflect the signature of the base of the food chain and their position in the food web respectively. 15N increases by 1.3-5.3‰ (average = 3.4‰) per trophic transfer, while 13C trophic fractionation is lower,  increasing 0-1‰ per trophic transfer (Minagawa and Wada, 1984; Vander Zanden and Rasmussen, 2001). Therefore, the 15N  signature of fish tissues is often used to infer the trophic level at which a particular fish fed during the last months or year at sea depending on the metabolic turnover of the fish tissue used for the analysis  (Cabana and Rasmussen, 1996). However, 15N variation is also influenced by changes in nitrogen assimilation at the base of the food web. In fact, nitrogen isotopic fractionation associated with the uptake of dissolved nitrate by marine phytoplankton can vary up to 5‰, based on the availability of nitrate ions (Waser et al., 2000). Since there is an inverse relationship between surface NO3- and the 15N of particulate organic matter (POM) (Rau et al., 1998), a decrease in surface nitrate concentration is expected to confer an increase in 15N at the base of the food web, and this will subsequently be transferred up the food chain to salmon. Thus, there are two potential influences on 15N, one at the consumer level and one at the producer level. Information on surface nitrate concentration as well as detailed studies on sockeye salmon prey isotopic variability, and its controls in the offshore waters of the Gulf of Alaska are needed to distinguish between these factors. Very little data exist on the carbon and nitrogen isotopic composition of food web components in the vast area of the Gulf of Alaska, although sufficient enough to suggest 9  hypotheses regarding factors that may have been responsible for variations in the 15N of salmon during the period of this study.  Additional food web information is gained with 13C, which is used to infer the source of production in a food web (Peterson and Fry, 1987). The 13C values of top trophic level marine organisms are set by the composition of phytoplankton in the food web.  The isotopic composition of the phytoplankton is affected by the isotopic composition of the dissolved inorganic carbon (DIC) assimilated by primary producers (Burkhardt et al., 1999; Freeman and Hayes, 1992). During photosynthetic fixation of CO2 into organic material, algal cells discriminate against the heavier stable carbon isotope 13C (Burkhardt et al., 1999; Freeman and Hayes, 1992). Thus, 13C values have a strong inverse relationship with DIC concentration (Freeman and Hayes, 1992; Rau, 1994). Increased solubility of CO2 at lower temperatures increases the DIC concentration and availability of 12C for primary producers, resulting in lower baseline 13C values. Factors other than CO2 may also have a significant impact on the carbon isotopic composition in phytoplankton. Differences in algal growth rate as well as taxon-specific differences would have an effect on the carbon fractionation process (e.g., Rau et al., 1989). Phytoplankton growth rates have a strong negative relationship with carbon isotope fractionation, with rapidly growing populations having more enriched 13C values (Laws et al., 1997; Burkhardt et al., 1999). Also, large cells, and diatoms in particular, are typically enriched in 13C compared with other phytoplankton groups (Lara et al., 2010). The carbon isotope composition of pelagic consumers therefore varies in relation to temperature, primary productivity rates, and taxonomic composition of phytoplankton. Accurate climate and 10  oceanographic data of the Gulf of Alaska (e.g. SSTs) is needed to distinguish between these mechanisms driving the 13C signature. Schell et al. (1998) noted the existence of isotopic gradients in the 13C and 15N of zooplankton in the Gulf of Alaska. Regions of high primary productivity such as the continental shelf in the Gulf of Alaska contain zooplankton with higher 13C and 15N than areas further offshore (Schell et al., 1998). Thus, variations in movement patterns of sockeye salmon between coastal and open ocean environments can be inferred from stable isotope analysis. Stable isotope analysis of archived scales therefore presents the opportunity for time series analysis of trophic history, changes in production at the base of the food web, and the environmental drivers of these factors. Despite the advantages of such technique, there have been few stable isotope analyses of Pacific salmon in the ocean. Welch and Parsons (1993) were the first to study the trophic position and potential competitive interactions of Pacific salmon species using isotopic tracers. They were able to establish a trophic hierarchy of Pacific salmon species where sockeye salmon appear to be positioned between coho and pink salmon, with the potential for significant trophic overlap between these species (Welch and Parsons, 1993). These findings are consistent with empirical evidence for trophic competition in the high seas. For example, sockeye salmon growth rates were lower in years of large pink salmon populations in the western North Pacific (Krogius, 1960). Both Satterfield and Finney (2002), and Johnson and Schindler (2008) performed a meta-analysis of stable isotope ratios to examine the extent of trophic partitioning among Pacific salmon during their marine lives. Similar to Welch and Parsons (1993), they found a distinct pattern of trophic partitioning among species of Pacific salmon, with potential for competitive interaction between some groups. Such findings have significant management 11  implications since enhancement of salmon stocks with hatchery fish could reduce the availability of prey resources for wild fish in times of decreased forage production, with negative consequences for wild stocks (Beamish et al., 1997).  In addition to investigating intra-specific trophic ecology, Satterfield and Finney (2002) and Johnson and Schindler (2012) investigated the trophic ecology of Alaskan sockeye salmon stocks in relation to climate using time-series analysis of the stable isotope ratios of archived scale samples. Satterfield and Finney (2002) were unable to distinguish between the mechanisms driving the stable isotope signatures of Red Lake sockeye for the period 1966-1999. They did not observe any correlation between the carbon and nitrogen stable isotope time-series. Since both isotopes tend to be enriched up the food chain, they concluded that changes in the trophic ecology of Red Lake sockeye salmon were not the main factors driving the isotope variability observed. However, the inconsistency of their 15N time-series with surface nitrate data for the Northeast Pacific as well as isotopic data for sockeye salmon prey in the high seas indicated that the trophic ecology of Red Lake sockeye likely changed during the period of their study. In addition, their 13C time-series did not show any major variability around the mean or any long-term trend. Similar to Satterfield and Finney (2002), Johnson and Schindler (2012) did not observe a significant relationship between the carbon and nitrogen time-series of 8 different stocks of Bristol Bay sockeye salmon for the period 1964-2003. They did find a relationship between the isotope data and climate shifts, but the magnitude of variation was small. The strong relationship between their 13C time-series and climate indices suggested that physical forcing played a strong role in primary production rates during that period. The lack of correlation between their 15N time-series and climate indices indicated that the feeding 12  ecology of Bristol Bay sockeye changed coinciding with the 1977 regime shift. However, it was unclear if that change in the feeding ecology of sockeye salmon was caused by salmon shifting diet or by a fundamental change of the pelagic food web post-regime shift.  In this study we adopt a similar approach to Satterfield and Finney (2002) and Johnson and Schindler (2012), in examining the long term high seas trophic ecology of Rivers inlet sockeye salmon. Our research differs from these previous investigations in 3 ways: i) a uniquely large data set of sockeye salmon scales that spans 100 years; ii) a correction of scale 13C for C:N ratios; and iii) a Principal Component Analysis (PCA) on physical and biological indices relevant to the offshore waters of the Gulf of Alaska, which allowed for a more detailed interpretation of the stable isotope data. Rivers Inlet supported a major sockeye salmon commercial fishery through the 1900’s but suffered probably the most catastrophic sockeye stock collapse in BC history. After sustained fishing pressure for decades, the formerly third largest sockeye fishery in BC began to experience unstable returns in the 1970s before crashing in the 1990s (Rutherford et al. 1995; McKinnell et al., 2001) (Figure 1.2). In 1996, the fishery was closed yet stocks continued to decline, reaching a low of 5,500 adult returns in 1999, down from an 80 year average of 1 million (Rutherford et al. 1995) (Figure 1.2). During the 2000’s the stock recovered to 125,000 annual returns, but has remained below harvestable levels (200,000) despite the continued fishery closure (Rutherford et al. 1995) (Figure 1.2). Rivers Inlet sockeye salmon spawn in Owikeno Watershed, an area well known for its intense logging. One of the hypotheses for the declining catch in Rivers Inlet was that logging activities had reduced the production capacity of the spawning habitat to such an extent that Owikeno Lake was no longer capable of producing sockeye salmon juveniles at historic levels. However, assessments 13  conducted by the Department of Fisheries and Oceans Canada (DFO) showed no overall decline in freshwater survival or pre-smolt production suggesting that Rivers Inlet sockeye declined for reasons related to the marine environment (Rutherford et al. 1995; Rutherford and Wood, 2000; McKinnell et al., 2001).  The Department of Fisheries and Oceans Canada (DFO) has been collecting and storing Rivers Inlet sockeye salmon scales for determination of population age structure since the 1900s. Thus, the Rivers Inlet sockeye salmon scale collection offered the opportunity for a 100 year (1915-2013) retrospective study of carbon and nitrogen stable isotope ratios of the Rivers Inlet sockeye salmon stock. In this study we use these data to test the hypotheses that the expected long-term variability in the environmental conditions and feeding ecology of sockeye salmon in the high seas during the period 1915-2013, characterized by multiple interdecadal (e.g., PDO) and interannual (e.g., ENSO) climate shifts, will be recorded in the carbon and nitrogen stable isotope time-series of Rivers Inlet sockeye salmon scales. In addition, we attempted to assess any potential links between the identified environmental conditions that Rivers Inlet sockeye salmon experienced in the high seas and Rivers Inlet sockeye salmon population declines. Ultimately, this research aims to provide improved understanding of how large-scale changes in climate affecting the conditions experienced by sockeye salmon during their pelagic phase, and how this relates to salmon abundance. 14   Figure 1.2. Rivers Inlet (Statistical Area 9) sockeye salmon catch and escapement data for the period 1948-2012.             15   Materials and methods 22.1 Overview of study site and Owikeno sockeye stock. Situated in the central coast of BC, Canada (Figure 2.1), Rivers Inlet provides an ideal spawning and rearing habitat for sockeye salmon (Oncorhynchus nerka).  Figure 2.1. Location of Rivers Inlet on the central coast of British Columbia.  All sockeye production from Rivers Inlet is managed as a single stock (Area 9) and is composed of about 12 populations that spawn in the rivers, tributaries and shoreline of Owikeno Lake, including the 5km long Wannock River that drains Owikeno Lake to Rivers Inlet (Figure 2.2). Nelson et al. (2003) investigated the genetic divergence of sockeye salmon spawning in seven rivers of the Owikeno Lake watershed and concluded that there were no clear isolated populations of sockeye salmon spawning in Rivers Inlet, thus supporting management of Owikeno Lake sockeye salmon as a single stock. 16  Owikeno Lake is a deep (>300m at the deepest point), long (56Km) and turbid lake that comprises four distinct basins, each separated by shallows narrows (Foskett, 1958) (Figure 2.2). The two more seaward basins (Basin 1 and Basin 2) account for approximately 90% of the total lake area, and these are deep and highly turbid because of a strong input of glacial silt carried mostly by the two largest glacial streams that flow into Owikeno Lake, the Machmell and Sheemahant Rivers. In contrast, the two upper basins (Basin 3 and Basin 4) are much smaller, shallower and less turbid.  Figure 2.2. Major sockeye salmon spawning areas in the Owikeno Lake watershed.    17  2.2 Sample collection  2.2.1 Archived samples Rivers Inlet sockeye salmon scales have routinely been collected by DFO since 1912, for determination of stock age structure. Scales are collected in pairs for individual fish from each of the major spawning rivers, and stored dry on gummed paper cards which at the same time are catalogued and kept in filing cabinets within a temperature controlled, locked storage facility at the Pacific Biological Station (PBS), in the care of the Schlerochronology Lab. Scales archived for the period 1915-2010 were obtained from these gummed paper cards. Between 20 and 30 scales, each representing an individual fish, were selected for every sampled year.  We aimed to sample every 5 years within the period 1915-2010 but availability (or lack thereof) shaped the selection process.  As a result, for the period 1915-2010 we sampled the following years: 1915, 1920, 1925, 1930, 1935, 1940, 1946, 1960, 1965, 1970, 1986, 1990, 1992, 1995, 2000, 2004 and 2010. For years 1915, 1920, 1925, 1930, 1935, 1940, 1946, 1960, 1965, 1970 and 1990 scales came from pre-spawn fish caught by gillnets in the commercial fishery between the months of June and August. For the years 1986, 1992, 1995, 2000, 2004 and 2010 scales came from post-spawn dead/moribund fish sampled during routine stock assessments conducted by DFO during September-October. More details on how post-spawn fish were sampled can be found in section 2.2.2.   18  2.2.2 Recent samples We complemented our time-series of archived scales with Rivers Inlet sockeye salmon scales sampled for the years 2011, 2012 and 2013. Muscle tissue samples were also collected for these years to further investigate isotope relationships between muscle and scales (muscle and scale isotopic offset). Approximately 100 sockeye salmon were sampled every year as they entered the Wannock River, collected from the Wuikinuxv Nation food fishery. The food fishery started in June and continued throughout July to the month of August. Fish were caught using gill-nets. Fish heads were immediately cut off behind the gills and kept frozen at -20C before being transferred to the University of British Columbia (UBC) laboratory for processing. Heads were gradually thawed in a refrigerated chamber at 5C and finally exposed to room temperature for processing. Scales were removed from the skin covering the dorsal musculature just posterior to the head. Wedges of muscle measuring approximately 1cm2 were excised with a scalpel from the same area. Perhaps surprisingly, our time-series of sockeye salmon scales had a wide range of C:N ratios (ranging from values of 2 to 7). It is possible that the variability in C:N ratios observed in our collection of scales was caused by differences in scale tissue lipid content. Lipids are composed mainly of carbon and are therefore positively correlated with C:N ratios, to the extent that C:N ratios can be used as a proxy for lipid content (Bodin et al., 2007; Post et al., 2007; Schmidt et al., 2003). Our time-series of scales contained samples of both pre-spawn and post-spawn fish, two different groups that are expected to be in very different physical conditions. Since salmon cease feeding before starting their upstream migration, all energy requirements associated with the spawning anadromous migration need to be met by stored reserves (Jonsson et al., 19  1991). Lipids and proteins are the main storage products used by salmonids (Higgs et al. 1995). Doucett et al. (1999) observed significant effects of the spawning migration on the nutritional status of salmon. In fact, a 75% reduction in lipid content of red muscle on post-spawn fish relative to pre-spawn fish was observed (Doucett et al., 1999). Scales are reabsorbed by all pacific salmon species as they sexually mature and stop feeding as they migrate up freshwater river systems (Persson et al., 1999).  The need for migrating sockeye to extract stored nutrients from the body during their migration fast is the main cause as to why scales are reabsorbed by these fish (Kacem et al., 1998). A recent published study on lipid composition on fish scales of Atlantic salmon (Salmo Salar) concluded that the lipid composition of the salmon scales were dominated by phospholipids (81%), triacylglycerides (12%) free fatty acids (FAs) (7%) (Otto Grahl-Nielsen, 2012). Thus, it is possible that lipids existing in scales are reabsorbed during the spawning migration of sockeye salmon, explaining the wide ranges in C:N ratios observed in our time-series on sockeye salmon scales. Since lipids in primary and secondary consumers are depleted in δ13C by up to 9% relative to whole tissues of an organism (Veefkind, 2003), samples with high lipid content would be biased toward more depleted 13C values. This could be falsely interpreted in our estimates of dietary sources and therefore it is desirable to correct for the variability in 13C values caused by lipid content. Full details of the correction for lipids are given in the section 2.4.2 of the methodology. To further investigate the differences in 13C values between pre-spawn and post-spawn fish, in 2013 samples of both scales and muscle tissue of pre-spawn and post spawn sockeye salmon were collected for stable isotope analysis. Both muscle and scale tissue samples of pre-spawn salmon were collected from the Wuikinuxv food fishery as described above. For the collection 20  of scales and muscle tissues of post-spawn fish we joined the DFO during their 2013 Rivers Inlet sockeye salmon routine stock assessments based at Genesee camp. Sockeye salmon are enumerated in non-glacial spawning streams of Owikeno Lake through visual counts and gill-net drifts. Streams are surveyed approximately every ten day and are usually walked and boated to a regular site where the survey is terminated. These sites exist where obstacles prevent fish from continuing up the river or the crews cannot proceed.  During stock assessment surveys in Rivers Inlet, the DFO team not only enumerates sockeye salmon but also samples post-spawn sockeye salmon for scales, otoliths, muscle tissue and other measurements used to determine age, sex and size composition. We joined the DFO sampling crew during the creek walks of the Ashlum, Neechanz and Genesee creeks and we collected scales and muscle from 75 spawned-out sockeye that were either moribund or freshly dead. Wedges of muscle measuring approximately 1cm2 were excised with a scalpel from the dorsal musculature just posterior to the head. Scales were removed from the same area. Samples were kept in Eppendorf tubes and frozen at -20C prior to arrival to the processing lab at UBC. Both scales and muscle tissue were thawed at air temperature before preparation.   2.3 Sample preparation Whole scales were briefly soaked in deionized water. Any residue was removed from the surface of the scales using forceps. Scales were then dried at air temperature and stored in Eppendorf tubes prior to stable isotope analysis. 21  Muscle tissues were air dried at 50C for approximately 15 days before being ground to a fine powder with a mortar and pestle. Ground samples were stored in Eppendorf tubes before stable isotope analysis.  2.4  Measurement of stable isotopes Both muscle and scale samples were analyzed using a Europa Scientific 20-20 continuous flow isotope ratio mass spectrometer at the IsoEnvironmental lab in Rhodes University, South Africa.  All isotopic ratios are expressed in following standard notation: X (‰) = (Rsample / Rstandard – 1) x 1000 where X is 13C or 15N and Rsample is the 13C/12C or 15N/14N respectively. Rstandard for 13C is Pee Dee Belemnite; for 15N it is atmospheric N2 (air). The measurement precision was estimated at 0.13‰ for 15N and 0.10‰ for 13C. C:N ratios were determined from percentage element weight. 2.4.1 13C correction for Suess Effect High concentrations of anthropogenic CO2 (which is depleted in δ13C) from the burning of fossil fuels have been released into the atmosphere since the start of the industrial revolution in 1850. As a consequence, the 13C of dissolved inorganic carbon (DIC) in the world’s oceans is gradually depleting. This is commonly referred to as the “oceanic Suess effect” (Kroopnick, 1985). The degree of 13C change from the Suess effect differs spatially in the ocean due to processes such as ocean circulation and mixing. The high latitude zone of the Pacific Ocean, 22  where sockeye salmon complete their at sea growth, is subject to upwelling and deep water mixing in the winter. Where upwelling and winter deep mixing occur, the Suess effect is diminished (Schell, 2001), and indeed, small decreases in 13C have been observed in the North Pacific Ocean (-0.012‰ decade-1) (Tanaka, 2003). Nonetheless, all 13C values were adjusted for anthropogenic 13C effects according to the equation below (Misarti et al., 2009): Suess Effect Correction Factor = a*exp(b*0.027) where a is the maximum annual rate of 13C decrease in the North Pacific (in this case -0.012 derived from Tanaka (2003); b is the year represented by the death of the animal minus 1850 (the start of the Industrial revolution); and 0.027 describes the curve presented by Gruber et al. (1999) for change in the 13C of the worlds’ oceans from 1945 to 1997. 2.4.2  13C correction for C:N As previously discussed in the methodology, C:N ratios of our time-series of Rivers Inlet sockeye salmon scales ranged from values of 2 to  values of 7 in some pre-spawn samples, most likely due to variation in lipid content. Given the effect of C:N, and lipids, on 13C values, we implemented a C:N correction to standardise our data. Species and tissue-specific lipid normalization models based on bulk C:N ratios are known to be a reliable tool to predict lipid-corrected 13C values (Logan et al., 2008).   23   Figure 2.3. Relationship between C:N ratios and 13C of Rivers Inlet sockeye scales fitted using an exponential model for (a) all sampled years and, (b) 2013 pre-spawn and post-spawn samples.    (a)   (b) 24  There was a significant correlation between 13C and C:N ratios of scales for all sampled years (R2= 0.67, N=769) (Figure 2.3. (a)). However, the relationship between 13C and C:N ratios for 2013 only, containing both pre-spawn and post-scales (Figure 2.3 (b)) fitted the exponential model better (R2=0.87, N=159). Thus, the latter relationship was used to predict 13C values corresponding to C:N ratios. The predicted 13C values were subsequently used to generate the lipid-corrected 13C (13C ‘) equation below: (13C – 13Cpredicted) + 13CC:N=3 = 13C’ being 13C the original 13C values obtained in our stable isotope analysis; 13Cpredicted the 13C values predicted by the exponential equation (Figure 2.3 (b)) using the relationship between 13C and C:N ratios for 2013 only; 13CC:N=3 the predicted 13C value at a “normal” or standard C:N ratio (we used C:N=3 since it is the approximate average C:N ratio of our entire time series); and 13C’ the 13C values standardised to C:N = 3. In other words, the equation extracted the variation between original and predicted 13C value and applied this variation to values standardised to C:N = 3. In this way, variation in the 13C values between years resulting from variation in the condition (lipid content) of the fish was removed.  Since C:N ratios of muscle tissue for 2011, 2012 and 2013 also ranged from values of 2 to values of 7, a C:N correction was also implemented to standardize the 13C values.   25   Figure 2.4. Relationship between C:N ratios and 13C of Rivers Inlet sockeye muscle tissue fitted using an exponential model for 2013 pre-spawn and post-spawn samples.   An exponential model was the best fit of the relationship between C:N ratios and 13C of Rivers Inlet sockeye salmon muscle tissue observed in 2013 for both pre-spawn and post-spawn samples (R2=0.82, N=175) (Figure 2.4). The 13C’ values of muscle tissue were obtained using the same equation applied to standardize the 13C values of scales described above. The only difference was that 13Cpredicted values of 2011, 2012 and 2013 Rivers Inlet sockeye muscle tissue were generated using the muscle tissue-specific exponential relationship between C:N ratios and 13C observed in Figure 2.4.   26  2.4.3 Muscle and scale isotopic offset The tissue-based offset (13Cm-sc and 13C’m-sc and 15Nm-sc) was defined here as the difference between the muscle and the scale stable isotope signatures for 2011, 2012 and 2013, and it was determined by subtracting the scale isotope values from that of muscle, as: Xm-sc= Xmuscle - Xscale where X is 13C, 13C’ or 15N.  2.5 Statistical analyses  All statistical analyses were completed using R (R Core Development Team 2014). The normality and variance homogeneity of the data analyzed were assessed prior to statistical analysis using Shapiro-Wilks goodness of fit test and Levene’s test (package Rcmdr), respectively. Where variance homogeneity was not found, appropriate alternatives to standard statistical tests were used.  Level of significance was set at p<0.001 in all analyses of variance. Welch analysis of variance (ANOVA) was used to test differences in mean 13C and 15N between pre-spawn and post-spawn scales and muscle tissue in 2013 because of unequal variances (stats package).  Paired t-tests were used to compare muscle tissue and scale stable isotope signatures of carbon and nitrogen for 2011, 2012, 2013 pre-spawn and 2013 post-spawn Rivers Inlet sockeye salmon. In other words, paired t-tests were used to test whether 13Cm-sc, 13C’m-sc and 15Nm-sc were significantly different from zero. One-way ANOVAS were used to test differences in 13Cm-sc, 13C’m-sc and 15Nm-sc among 2011, 2012, 2013 pre-spawn and 2013 post-spawn Rivers Inlet sockeye salmon, with Tukey’s post-hoc test used to assess significance in differences between 27  groups. Simple linear regression was used to test for significant linear relationships between the carbon and nitrogen time-series, and between the stable isotope time series and the resulting scores of the principal component analysis (PCA). The significance of correlation was set at p<0.05.   2.6 Principal component analysis (PCA) Principal component analysis (PCA) is a multivariate statistical procedure that reduces a group of original variables (likely a large number of variables) to a smaller number of dimensions or Principal Components (PCs) that account for most of the variance in the original set of variables. The new generated dimensions or PCs are defined by two elements: eigenvectors and eigenvalues. Eigenvectors indicate the direction of the new generated dimension or PC and eigenvalues indicate how much of the variance in the data is explained by that PC. The eigenvector with the highest eigenvalue is the first Principal Component (PC1). From the PCA we can extract loadings and scores. Loadings are the contributions of each original variable to the PC. Large positive loadings indicate that the variable is positively correlated to the PC and large negative loadings indicate that the variable is negatively correlated to the PC. Loadings close to zero indicate that the variable does not contribute to the PC. Scores are the reconstructed data points in the new reduced dimension or PC.  PCA is a useful tool to explore common trends in the original set of variables. Hare and Mantua (2000) evaluated 100 environmental time series to determine the evidence of climatic regime shifts in the North Pacific Ocean between 1965 and 1997. Using PCA, they were 28  able to identify the 1977 and 1989 regime shifts (Hare and Francis, 1995; Hare and Mantua, 2000; Mantua et al., 1997). More recently, Litzow and Mueter (2014) updated the work of Hare and Mantua (2000) and were able to identity another regime shift in 2007. The effects of such large-scale shift in climate in the physical and biological properties of the Northeast Pacific have been well documented. Therefore, a PCA of environmental indices relevant to the oceanic domains of the Gulf of Alaska, where maturing sockeye salmon reside during their pelagic stage of their life cycle (Figure 1.1), presents a powerful tool to distinguish between the factors responsible for the variation observed in our stable isotope time-series.  The specific goals of our Principal Component Analysis were to: (1) identify and evaluate the common patterns of variability in 33 time-series (spanning four decades) of physical and biological variables in the North East Pacific Ocean and Gulf of Alaska, (2) compare our resulting scores with well documented climate data (e.g. Hare and Mantua, 2000; Litzow and Mueter, 2014), and (3) assess any potential links between the resulting scores and our stable isotope ratios for a more accurate interpretation of the stable isotope data. 2.6.1 Data and methods. The environmental time series selected for the PCA comprised 15 physical and 18 biological series (Table 2.1). Time series were chosen based on potential indicators of decadal scale climate/ecosystem change in the geographic range of BC sockeye salmon and / or a priori determined correlation with variation in the growth or production of salmon.  29  Table 2.1. Numeric and alphabetic abbreviations for the 33 environmental time-series used in this analysis. A brief description and source of information are provided for each time-series in Appendix A. The time-series are plotted geographically in Figure 2.5.        No Abbreviation Full name 1 PNA Pacific North American Index 2 EP East Pacific – North Pacific Index 3 SOI Southern Oscillation Index 4 NP North Pacific Index 5 SSTWIN Pine Island Sea Surface Temperature – winter average 6 SSTSUM Pine Island Sea Surface Temperature – summer average  7 PDOWIN Pacific Decadal Oscillation – winter average 8 PDOSUM Pacific Decadal Oscillation – summer average 9 SSSWIN Pine Island Sea Surface Salinity – winter average 10 SSSSUM Pine Island Sea Surface Salinity – summer average 11 NINO34WIN ENSO3.4 – winter average 12 NINO34SUM ENSO3.4 – summer average 13 NPGOWIN North Pacific Gyre Oscillation – winter average 14 NPGOSUM North Pacific Gyre Oscillation – summer average 15 PTI Ocean Station Papa trajectory index 16 SOUTHAKPINK Southeast Alaska pink salmon catch 17 SOUTHAKCHUM Southeast Alaska chum salmon catch 18 SOUTHAKSOCK Southeast Alaska sockeye salmon catch 19 CENTRALAKPINK Central Alaska pink salmon catch 20 CENTRALAKCHUM Central Alaska chum salmon catch 21 CENTRALAKSOCK Central Alaska sockeye salmon catch 22 RISOCKTBM Rivers Inlet sockeye salmon total biomass 23 FRASOCKTBM Fraser River sockeye salmon total biomass 24 TYEESOCK Skeena River Tyee test-fishery sockeye salmon index 25 BCAREA4PINK British Columbia statistical area 4 pink salmon total biomass 26 BCAREA4CHUM British Columbia statistical area 4 chum salmon total biomass 27 BCAREA4SOCK British Columbia statistical area 4 sockeye salmon total biomass 28 BCAREA7PINK British Columbia statistical area 7 pink salmon total biomass 29 BCAREA7CHUM British Columbia statistical area 7 chum salmon total biomass 30 BCAREA8PINK British Columbia statistical area 8 pink salmon total biomass 31 BCARE8CHUM British Columbia statistical area 8 chum salmon total biomass 32 BCAREA11SOCK British Columbia statistical area 11 sockeye salmon total biomass 33 COLUMBIASOCK Columbia River sockeye salmon escapement  30    Figure 2.5. Geographic distribution of our 33 environmental time-series. Their location on the map indicates where each variable was measured or has influence. See Table 2.1 for a definition of each abbreviation.  We attempted to select a broadly representative set of environmental indicators. However, several time series had missing data points either near the beginning or the end of the records. We did not use a time series in the subsequent calculations if there were more than 5 consecutive years of missing data for the corresponding variable. This reduced the number of variables used in the analysis as well as the span of the time-series used on each PCA. Our physical time series (total of 15) represent mostly winter averages of atmospheric and oceanic conditions while most of the biological time series (total of 18) represented annual values of salmonid stocks that are known to feed at a similar trophic level (Johnson and Schindler, 2008) 1 PNA 2 EP 3 SOI 4 NP 5 SSTWIN 6 SSTSUM 7 PDOWIN 8 PDOSUM 9 SSSWIN 10 SSSSUM 11 NINO34WIN 12 NINO34SUM 13 NPGOWIN 14 NPGOSUM 15 PTI 16 SOUTHAKPINK 17 SOUTHAKCHUM 18 SOUTHAKSOCK 19 CENTRALAKPINK 20 CENTRALAKCHUM 21 CENTRALAKSOCK 22 RISOCKTBM 23 FRASOCKTBM 24 TYEESOCK 25 BCAREA4PINK 26 BCAREA4CHUM 27 BCAREA4SOCK 28 BCAREA7PINK 29 BCAREA7CHUM 30 BCAREA8PINK 31 BCAREA8CHUM 32 BCAREA11SOCK 33 COLUMBIASOCK 31  and overlap in the Gulf of Alaska during their maturing years (French et al., 1996; Burgner, 1991).  We ran separate PCAs for the 15 physical time-series, the 18 biological time-series and the 33 physical and biological time-series combined. These independent PCAs were run for the periods 1950-2013, 1970-2012 and 1970-2012 respectively. Running separate PCAs permitted us to recognize patterns of variability in physical and biological variables independently from each other. This approach helped us to perform a more rigorous comparison between the modes of variability observed in each PCA and our carbon and nitrogen stable isotope time-series. Principal components analysis (PCA) was performed using the exploratory data analysis package FactoMineR for R (R Core Development Team 2014).  In order to deal with the missing data we used the MissMDA package for R (R Core Development Team 2014), which imputes the missing data of a mixed data set using the iterative PCA algorithm (method="EM") or the regularised iterative PCA algorithm (method="Regularized"). We used the regularized method. This method imputes missing values with initial values such as the mean of the variable. If the argument seed is set to a specific value, a random initialization is performed: the initial values are drawn from a gaussian distribution with mean and standard deviation calculated from the observed values. The output of the algorithm can be used as an input of the PCA function of the FactoMineR package in order to perform PCA on an incomplete dataset. Scores were normalized prior to plotting and statistical analyses by dividing each score by the square root of the eigenvalue for the relevant principal component. The standardized scores were subsequently correlated with the stable isotope time-series. 32  2.7 Additional data sets  Some data sets had too many data gaps to be included in the PCA but were crucial to interpret the variability observed in our carbon and nitrogen stable isotope time-series. 2.7.1 Baseline Isotope data. In order to interpret the feeding habits and migration patterns of consumers based on stable isotopes data, it is necessary to know the isotopic values of the basal food web where the consumer feeds (Post, 2002; Vander Zanden and Rasmussen). Maturing sockeye salmon in the Gulf of Alaska are known to consume a high variety of prey items, mainly squids (Berryteuthys anonychus) and fish but they are also known to consume euphausiids, hyperiid amphipods, copepods (Neocalanus spp.), pteropods, crustacean larvae and pelagic polychaetes (Brodeur, 1990). Data on potential prey items and suspended particulate organic matter (SPOM) used to provide basic isotope data for the Gulf of Alaska were obtained from existing studies: SPOM, copepods, pteropods, euphausiids, amphipods and chaetognaths isotope data collected between May 1991 and May 1993 along the Line P and at Ocean Station Papa (OSP) (Wu et al., 1996; 1998) and squid samples taken during spring and summer of 1992 and 1994 from the Gulf of Alaska (Continental shelf) (Hobson et al., 1997).     33  2.7.2 Nutrient data. Information on surface nitrate concentration (NO3) is needed for a more accurate interpretation of the variability observed in the nitrogen stable isotope time-series. A comprehensive NO3 time series of the North Pacific Ocean for the period 1985-2010 was facilitated by the Department of Fisheries and Oceans Canada (DFO). Because most data came from ships of opportunity (SOO) running between North America and Japan sampling the entire longitude of the North Pacific Ocean, we only selected NO3 data that was collected in the North East Pacific Ocean relevant to BC sockeye salmon distribution during their maturing years at sea. Since complete spatial coverage was not always possible due to weather conditions, we chose a range of longitudes that were sampled in all years. Thus, we calculated the yearly average of NO3 in surface water out of all macronutrient samples between 50-57N and 145-160W for the period 1985-2010.         34   Results 33.1 Comparison of 2013 pre-spawn and post-spawn samples The non-lipid-normalized 13C of muscle tissue of pre-spawn Rivers Inlet sockeye salmon (N=97) ranged from -25.17‰ to -20.78‰ with a mean of -22.81±0.88‰, while non-lipid-normalized 13C for muscle tissue of post-spawn fish (N=78) ranged from -22.45‰ to -20.26‰ with a mean of -21.25±0.47‰ (Figure 3.1 (a)). The non-lipid-normalized 13C isotope ratios for muscle tissue from pre-spawn Rivers Inlet sockeye salmon were significantly more depleted and more variable than those of post-spawn fish (Figure 3.1 (a), ANOVA, p<0.001). Analysis of the non-lipid-normalized 13C isotope ratios of scale tissues also revealed significant differences between pre-spawn and post-spawn fish (Figure 3.1 (b), ANOVA, p<0.001). Pre-spawn scales were significantly more depleted in 13C and more variable than those of post-spawn fish.  The non-lipid-normalized 13C of pre-spawn scales (N=89) ranged between -24.18‰ and -17.48 with a mean of -20.12±1.35‰, while non-lipid-normalized scale tissue 13C of post-spawn fish (N=70) ranged between -18.93‰ and -17.33‰ with a mean of -17.97±0.35‰ (Figure 3.1 (b)).       35   Figure 3.1. Boxplot of 2013 non-lipid-normalized 13C stable isotope ratios for Rivers Inlet sockeye salmon (a) muscle of pre-spawn and post-spawn fish and, (b) scales of pre-spawn and post-spawn salmon. Median values (solid horizontal line), interquartile range (box outline), and 95% confidence intervals (whiskers) are shown. (a) (b) 36  The C:N ratios for muscle tissue of pre-spawn fish (N=97) ranged from 3.33 to 6.37 with a mean value of 4.16±0.73, while C:N ratios for muscle tissue of post-spawn fish (N=78) ranged from 2.98 to 3.95 with a mean value of 3.28±0.21 (Figure 3.2 (a)). The C:N ratios for muscle tissue of pre-spawn Rivers Inlet sockeye salmon were significantly higher and more variable than those of post spawn fish (ANOVA, p<0.001). The C:N ratios in scale tissues behaved similarly, ranging from 6.07 to 2.80 with a mean value of 3.48±0.59 in pre-spawn fish (N=89), and ranging from 3.38 to 2.44 with a mean value of 2.69±0.19 in post-spawn fish (N=70) (Figure 3.2 (b)). C:N ratios for scale tissue of pre-spawn fish were also significantly higher and more variable than those of post-spawn fish (ANOVA, p<0.001).  37    Figure 3.2. Boxplot of 2013 C:N ratios for Rivers Inlet sockeye salmon (a) muscle of pre-spawn and post-spawn fish and, (b) scales of pre-spawn and post-spawn salmon. Median values (solid horizontal line), interquartile range (box outline), and 95% confidence intervals (whiskers) are shown. (b) (a) 38  We observed that the tissue of pre-spawn fish, which had significantly higher C:N ratios than those of post-spawn fish, also had significantly more depleted values of 13C. This relationship was reflected in the strong correlation between C:N ratios and 13C values in both scale (Figure 2.3 (b))  and muscle tissue (Figure 2.4) used to standardize our data, as described in section 2.4.2. After lipid-normalizing the 13C values (13C’) for both muscle and scale tissue of 2013 pre-spawn and post-spawn Rivers Inlet sockeye salmon, 13C’ values for muscle tissue of pre-spawn fish (N=97) ranged from -21.60‰ to -19.04‰ with a mean of -20.68±0.53‰, while 13C’ for muscle tissue of post-spawn fish (N=78) ranged from -21.34‰ to -18.99‰ with a mean of -20.67±0.44‰ (Figure 3.3 (a)). No significant differences in 13C’ isotope ratios were observed between lipid-normalized muscle tissue of pre-spawn and post-spawn Rivers Inlet sockeye salmon (Figure 3.3 (a), ANOVA, p=0.88). Analysis of 13C’ isotope ratios of scales also showed no significant differences between pre-spawn and post-spawn fish (Figure 3.3 (b), ANOVA, p=0.27). The 13C’ of pre-spawn scales (N=89) ranged between -20.00‰ and -17.16‰ with a mean of -18.88±0.63‰, while 13C’ for scale tissue of post-spawn fish (N=70) ranged between -19.77‰ and -17.04‰ with a mean of -18.98±0.51‰ (Figure 3.3 (b)).   39   Figure 3.3. Boxplot of 2013 lipid-normalized 13C (13C ‘) stable isotope ratios for Rivers Inlet sockeye salmon (a) muscle of pre-spawn and post-spawn fish and, (b) scales of pre-spawn and post-spawn salmon. Median values (solid horizontal line), interquartile range (box outline), and 95% confidence intervals (whiskers) are shown.   (a) (b) 40  The 15N of muscle tissue of pre-spawn RI sockeye salmon (N=97) ranged from 9.52‰ to 12.12‰ with a mean of 10.77±0.54‰ and 15N values for muscle tissue of post-spawn RI sockeye salmon (N=78) ranged from 9.42‰ to 11.87‰ with a mean of 10.51±0.52‰ (Figure 3.4 (a)). No significant difference in 15N was observed between muscle tissue of pre-spawn and post-spawn fish (ANOVA, p=0.0014). For scale tissues, 15N values of pre-spawn fish (N=89) ranged from 8.86‰ to 13.81‰ with a mean of 10.54±0.94‰ and 15N of post-spawn fish (N=70) ranged from 8.87‰ to 12.25‰ with a mean of 10.59±0.68‰ (Figure 3.4 (b)). No significant difference in 15N was observed between scales of pre-spawn and post-spawn fish (ANOVA, p=0.70).   41   Figure 3.4. Boxplot of 2013 15N stable isotope ratios for Rivers Inlet sockeye salmon (a) muscle of pre-spawn and post-spawn fish and, (b) scales of pre-spawn and post-spawn salmon. Median values (solid horizontal line), interquartile range (box outline), and 95% confidence intervals (whiskers) are shown.   (a) (b) 42  3.2 Isotopic relatioship between muscle and scale tissue Non-lipid normalized 13C, lipid-normalized 13C (13C’) and 15N obtained from muscle and scale tissue samples of 2011, 2012, 2013 pre-spawn and 2013 post-spawn Rivers Inlet sockeye salmon (Table 3.1) were used to calculate the differences in isotopic signatures between muscle and scales as described in section 2.4.3. Mean differences between the non-lipid-normalized 13C in muscle tissue and scales (13Cm-sc) were -4.47±0.55, -0.53±1.27, -2.65±1.48 and -3.28±0.36 for 2011, 2012, 2013 (pre-spawn) and 2013 (post-spawn) respectively (Figure 3.5). The paired t-tests indicated significant differences between non-lipid-normalized muscle and scale 13C values for all years (t-test, p<0.001) (Table 3.2), with scales being enriched in 13C values compared to those of muscle. One-way ANOVA results indicated significant differences in 13Cm-sc among years (ANOVA, p<0.001). Tukey’s post-hoc HSD indicated that the 13Cm-sc for all years were significantly different from one another (HSD, p<0.001).  43   Figure 3.5. Boxplot of differences in non-lipid-normalized 13C stable isotope ratios between muscle and scales (13Cm-sc) of 2011, 2012, 2013 pre-spawn and 2013 post-spawn Rivers Inlet sockeye salmon. Median values (solid horizontal line), interquartile range (box outline), and 95% confidence intervals (whiskers) are shown.  Mean differences between the lipid-normalized 13C (13C ‘) in muscle tissue and scales (13C’m-sc) were -1.93±0.37, -1.92±0.63, -1.81±0.58 and -1.67±0.76 for 2011, 2012, 2013 (pre) and 2013 (post) respectively (Figure 3.6). The paired t-tests indicated significant differences between muscle and scale 13C’ values for all years (t-test, p<0.001) (Table 3.2), with scales being enriched in 13C’ values compared to muscle. One-way ANOVA results showed no significant differences in 13C’m-sc among years (ANOVA, p=0.027). A consistent enrichment in 13C’ of scales compared to muscle tissue was observed among years. 44   Figure 3.6. Boxplot of differences in lipid-normalized 13C (13C ‘) stable isotope ratios between muscle and scales (13C’m-sc) of 2011, 2012, 2013 pre-spawn and 2013 post-spawn Rivers Inlet sockeye salmon. Median values (solid horizontal line), interquartile range (box outline), and 95% confidence intervals (whiskers) are shown.  Mean differences between the 15N in muscle tissue and scales (15Nm-sc) were 0.52±0.47, 0.45±0.35, 0.24±1.00 and -0.09±0.75 for 2011, 2012, 2013 (pre-spawn) and 2013 (post-spawn) respectively (Figure 3.7). Significant differences were observed between muscle and scale 15N in all years (t-test, p≤<0.01) with scales being depleted in 15N compared to muscle tissue. (Table 3.2). No significant differences were observed in the 15N signatures between muscle and scale samples of 2013 post spawn fish (t-test, p=0.17) (Table 3.2). One-way ANOVA results indicated significant differences in 15Nm-sc among years (ANOVA, p<0.001). Tukey’s post-hoc HSD indicated that only 2013 post-spawn 15Nm-sc differed significantly from 15Nm-sc of 2011 and 2012 (HSD, p<0.001).  45   Figure 3.7. Boxplot of differences in 15N stable isotope ratios between muscle and scales (15Nm-sc) of 2011, 2012, 2013 pre-spawn and 2013 post-spawn Rivers Inlet sockeye salmon. Median values (solid horizontal line), interquartile range (box outline), and 95% confidence intervals (whiskers) are shown.    46   Table 3.1. Summary statistics of 13C, 13C’ and 15N obtained from muscle and scale tissue samples of 2011, 2012, 2013 pre-spawn and 2013 post-spawn Rivers Inlet sockeye salmon.  13C  13C’  15N Tissue Year N Mean SD Min Max  Mean SD Min Max  Mean SD Min Max Muscle  2011 88 -23.08 0.59 -24.73 -21.83  -20.76 0.31 -21.31 -19.89  10.28 0.34 9.52 10.98  2012 82 -23.12 0.85 -25.01 -20.66  -20.74 0.49 -21.71 -19.22  10.84 0.58 9.71 12.78  2013 pre 97 -22.81 0.88 -25.17 -20.78  -20.68 0.53 -21.60 -19.04  10.77 0.54 9.52 12.12  2013 post 78 -21.25 0.47 -22.45 -20.26  -20.67 0.44 -21.34 -18.99  10.51 0.52 9.42 11.87 Scale  2011 84 -18.59 0.43 -20.17 -17.34  -18.82 0.30 -19.34 -17.97  9.77 0.51 8.59 11.07  2012 82 -22.59 1.31 -25.62 -19.39  -18.82 0.73 -20.04 -16.74  10.39 0.59 9.00 11.78  2013 pre 89 -20.12 1.35 -24.18 -17.48  -18.88 0.63 -20.00 -17.16  10.54 0.94 8.86 13.81  2013 post 70 -17.97 0.36 -18.93 -17.33  -18.98 0.51 -19.77 -17.04  10.59 0.68 8.87 12.25 N= sample size, SD= standard deviation.   Table 3.2. Paired t-test results comparing stable isotope signatures from scales and muscle tissue samples of 2011, 2012, 2013 pre-spawn and 2013 post-spawn Rivers Inlet sockeye salmon.  13C  13C’  15N Year N t-stat P  t-stat P  t-stat P 2011 84 -73.88 <0.001  -48.04 <0.001  10.06 <0.001 2012 82 -3.73 <0.001  -27.20 <0.001  11.67 <0.001 2013 pre 89 -16.86 <0.001  -29.29 <0.001  2.25 0.01 2013 post 70 -47.52 <0.001  -18.33 <0.001  -0.95 0.17 N=sample size, P=p-value.    47  3.3 Carbon and nitrogen stable isotope time-series from rivers inlet sockeye salmon scales  As described in the Materials and Methods section, variability observed in the C:N ratios of Rivers Inlet sockeye salmon scales was a source of significant 13C variability among years (Figure 3.8). 13C values were therefore standardized to take C:N ratios into account (Figure 3.9). Remarkably, the unusual high C:N ratios of Rivers Inlet sockeye salmon scales in 2012 (Figure 3.8) coincide with unusual depleted  13C values of Rivers Inlet sockeye salmon scales in 2012 (Figure 3.9). Yearly mean 13C values were also corrected for Suess effect (see Materials and Methods section).  Figure 3.8. Yearly C:N mean of Rivers Inlet sockeye salmon scales for the period from 1915-2013. Error bars represent standard deviation.  48    Figure 3.9. Yearly 13C (‰) mean of Rivers Inlet sockeye salmon scales for the period from 1915-2013.  The lipid-normalized carbon (13C’) and nitrogen stable isotope values of 768 Rivers Inlet sockeye salmon scales were within the range reported in previous studies (Table 3.3). During 1915-2013, 13C’ values of Rivers Inlet sockeye scales varied between -17.9‰ and -15.7‰ with an average of -17±0.77‰ (Figure 3.10 (a)). During the same period, 15N values ranged from 9.25‰ to 11.11‰ with an average of 10.11±0.61‰ (Figure 3.10 (b)).   49  Table 3.3. Observed range of 13C’ and 15N in the stable isotope analysis of three different populations of sockeye salmon.  The 13C’ values varied significantly throughout the time-series (Figure 3.10 (a)). We observed lower and more constant 13C’ values from 1915 to 1970, prior to the 1970s regime shift (-17.34±0.3‰). Between 1986 and 2010 13C’ values were typically higher and more variable (mean -16.2±0.7‰).  Notably, in 1990 and 2011-2013 13C’ values were similar to the pre 1970’s period. The 15N values (Figure 3.10 (b)) also varied significantly between 1915 and 2013. Values were relatively low and stable prior to 1960 (mean 15N was 9.58±0.27‰), and were higher and more variable from 1965 to 2013 (mean 10.46±0.51‰). Similar to 13C’, 15N returned to pre-regime shift levels in 1990.  After 1990, 15N progressively increased and remained constant until 2011 when values demonstrated a sudden decline before increasing again after 2011. Both 13C’ and 15N time series followed a similar pattern and were significantly correlated (R2=0.36, p<0.05) (Figure 3.11).    Study Location Species 13C’(‰) RANGE 15N(‰) RANGE This study (2014) Central North Pacific (Rivers Inlet, BC) Sockeye -18 to  -16 9.25 to 11.11 Johnson and Schindler (2008) Southwest Alaska (Bristol Bay) Sockeye  -18.5 to  -16  10.5 to 13  Satterfield and Finney (2002) South West Alaska (Red Lake, AK) Sockeye -18.5 to -16.5  8.2 to 11.7  50   Figure 3.10. Yearly (a) 13C’ (‰) and (b) 15N (‰) mean of Rivers Inlet sockeye salmon scales for the period from 1915-2013. Dashed vertical line indicates the 1977 regime shift. Error bars represent standard deviation. 13C data are lipid-normalized and corrected for Suess effect (see Materials and Methods section).    (a) (b) 51    Figure 3.11. Relationship between the 13C’ and 15N time-series of RI sockeye salmon scales for the period 1915-2013.     52  3.4 Principal component analysis (PCA)  3.4.1 PCA on the 33 environmental time-series (PCAenv) We aimed to isolate the most important patterns of common variability in the 33 environmental time-series for the period 1970-2012. Eigenvalue analyses (Table 3.4) showed that the two first principal components accounted for 68% of the total variation.  Table 3.4. Eigenvalues, percentage of variance and cumulative percentage of variance for the first five principal components of the principal component analysis performed to the 33 environmental variables.  eigenvalue percentage of variance cumulative percentage of variance PC1 26.05472 39.47684 39.47684 PC2 19.09089 28.92559 68.40243 PC3 3.433644 5.202491 73.60493 PC4 2.506362 3.797518 77.40244 PC5 2.067397 3.13242 80.53486    53   Figure 3.12. The first two principal component scores from a principal component analysis of the 33 environmental time-series. Blue colored areas correspond to negative phases while red colored areas correspond to positive phases. Vertical dashed lines indicate the 1977 and 1989 regime shifts. 54  The first principal component (PC1) was negative from 1970 until 1977 when it transitioned to a positive phase (Figure 3.12). This phase generally remained positive until 2007 with the exception of an abrupt shift to a short-lived negative phase which only lasted from 1999 to 2003 (Figure 3.12). PC1 remained negative from 2007 to 2012 (Figure 3.12). The time-series shows a minimum in 1971 and a maximum in 1992 (Figure 3.12).  Although none of our time-series were highly correlated with the principal components in this PCA based on the 33 environmental variables (PCAenv), we can still differentiate between lower (r<|0.08|) and higher (r>|0.1|) loadings (analogous to Pearson’s r) (Table 3.5). PC1 was mainly explained by physical variables. High loadings occurred for 9 of the physical time-series and 2 of the biological time-series. Another 5 variables had loadings of between |0.08| and |0.1|. Among the large-scale climate variables the largest loadings correspond to both summer averages of SST and PDO which were positively correlated to PC1. Other large-scale climate variables were also highly correlated to PC1. The ENSO 3.4 winter-average and PTI were positively correlated to PC1 while the NPGO, SOI and NP also had high loadings but were inversely correlated to PC1. SSSs had negligible loadings on PC1. Among the biological indices we observed that both Southeast Alaska and Central Alaska sockeye catch time-series had a relatively strong positive relationship with PC1. The second principal component (PC2) showed two distinct trends for the period 1970-2012. PC2 was negative from 1970 to 1994 when it transitioned to a positive phase until 2012 (Figure 3.12). PC2 was mainly explained by biological indices. Relatively high loadings (r>|0.1|) occurred in 9 of the environmental variables, all of them being biological indices (Table 3.5). It is 55  remarkable that all Alaskan biological indices had positive loadings while the remaining biological indices belonging to British Columbia and Washington State had negative loadings with the exception of Fraser River sockeye salmon total biomass and Columbia River sockeye salmon escapement time-series which had negligible loadings (Table 3.5).     56  Table 3.5. The 33 environmental time-series and their corresponding loadings of the first two principal components of the PCA. High loadings (r>|0.1|) are underlined.    PC1 PC2 Pacific North American Index 0.092752 0.058985 East Pacific – North Pacific Index 0.095567 -0.06925 Southern Oscillation Index -0.13264 0.037576 North Pacific Index -0.11232 -0.01288 Pine Island Sea Surface Temperature – winter average 0.142823 0.070402 Pine Island Sea Surface Temperature – summer average  0.176775 0.047903 Pacific Decadal Oscillation – winter average 0.127474 0.017106 Pacific Decadal Oscillation – summer average 0.154849 -0.03404 Pine Island Sea Surface Salinity – winter average -0.02442 -0.02148 Pine Island Sea Surface Salinity – summer average -0.00112 -0.04549 ENSO3.4 – winter average 0.124042 -0.05122 ENSO3.4 – summer average 0.051968 0.00362 North Pacific Gyre Oscillation – winter average -0.08504 0.093383 North Pacific Gyre Oscillation – summer average -0.11527 0.089584 Ocean Station Papa trajectory index 0.114664 0.011327 Southeast Alaska pink salmon catch 0.08071 0.10129 Southeast Alaska chum salmon catch 0.020306 0.170139 Southeast Alaska sockeye salmon catch 0.144135 0.028172 Central Alaska pink salmon catch 0.028159 0.173225 Central Alaska chum salmon catch -0.01087 0.147168 Central Alaska sockeye salmon catch 0.108895 0.104254 Rivers Inlet sockeye salmon total biomass -0.00692 -0.16575 Fraser River sockeye salmon total biomass 0.026804 0.09663 Skeena River Tyee test-fishery sockeye salmon index -0.0512 -0.02719 British Columbia statistical area 4 pink salmon total biomass 0.054354 -0.04745 British Columbia statistical area 4 chum salmon total biomass 0.02741 -0.07738 British Columbia statistical area 4 sockeye salmon total biomass 0.069304 -0.00071 British Columbia statistical area 7 pink salmon total biomass -0.03316 -0.15946 British Columbia statistical area 7 chum salmon total biomass -0.04804 -0.14003 British Columbia statistical area 8 pink salmon total biomass 0.059668 -0.08667 British Columbia statistical area 8 chum salmon total biomass 0.091367 -0.04634 British Columbia statistical area 11 sockeye salmon total biomass 0.068533 -0.13088 Columbia River sockeye salmon escapement  -0.06119 0.096372   57  3.4.2 PCA on the 15 physical time-series (PCAphys) Eigenvalue analysis (Table 3.4) of the PCA of the 15 physical time-series (PCAphys) for the period 1950-2013 indicated that only the first two principal components were meaningful. PC1 and PC2 explained 48.6% and 29.8% of the total variance respectively (Table 3.6).  Table 3.6. Eigenvalues, percentage of variance and cumulative percentage of variance for the first five principal components of the principal component analysis performed to the 15 physical time-series.  eigenvalue percentage of variance cumulative percentage of variance PC1 14.57107 48.57024 48.57024 PC2 8.95212 29.8404 78.41064 PC3 1.28645 4.288166 82.6988 PC4 1.113105 3.710349 86.40915 PC5 1.010513 3.368377 89.77753  58   Figure 3.13. The first two principal component scores from a principal component analysis of the 15 physical time-series (PCAphys). . Blue colored areas correspond to negative phases while red colored areas correspond to positive phases. Vertical dashed lines indicate the 1977 and 1989 regime shifts.   59   The first principal component (PC1) scores of PCAphys showed many similarities with the scores of PC1 of the full data set (Figure 3.13).  PC1 was generally negative from 1950 until 1977 and then strongly positive until 2007 with the exception of an abrupt shift to a short-lived negative phase in 1999 which only lasted four years (Figure 3.13). In 2007 PC1 became negative again until 2013 (Figure 3.13). The time series shows a minimum in 1950 and a maximum in 1983.  We observed higher loadings (r>|0.2|) for PCAphys than for PCAenv. Again, SSTs and PDO (winter average) were the variables that showed higher loadings for PC1 (r>2) (Table 3.7). PNA was also highly correlated to PC1 (r=2) and NP appeared to be negatively correlated to PC1 (r= -2.1). We didn’t observe any clear temporal trend in PC2. PC2 scores were more variable, alternating between positive and negative phases along the time-series (Figure 3.13). Both the winter and summer average of the NPGO had the highest loadings but were inversely correlated to PC2 (Table 3.7).  Table 3.7. The 15 physical time-series and their corresponding loadings of the first two principal components of the PCA. High loadings (r>|0.2|) are underlined.    PC1 PC2 Pacific North American Index 0.199785 -0.14871 East Pacific – North Pacific Index 0.101465 0.154076 Southern Oscillation Index -0.18106 -0.07368 North Pacific Index -0.2108 0.079596 Pine Island Sea Surface Temperature – winter average 0.20617 -0.08744 Pine Island Sea Surface Temperature – summer average 0.228484 0.031732 Pacific Decadal Oscillation – winter average 0.215604 -0.0544   60   PC1 PC2 Pacific Decadal Oscillation – summer average 0.184641 0.110626 Pine Island Sea Surface Salinity – winter average -0.08449 0.166356 Pine Island Sea Surface Salinity – summer average -0.05292 0.152531 ENSO3.4 – winter average 0.168607 0.158112 ENSO3.4 – summer average 0.065366 0.174324 North Pacific Gyre Oscillation – winter average 0.005629 -0.27077 North Pacific Gyre Oscillation – summer average -0.03503 -0.29287 Ocean Station Papa trajectory index 0.154236 -0.04801   3.4.3 PCA on the 18 biological time-series (PCAbiol) Eigenvalue analysis (Table 3.6) of the PCA of the 18 biological time-series (PCAbiol) for the period 1970-2012 indicated that only the first two principal components were meaningful. PC1 and PC2 explained 44.4% and 27.3% of the total variance respectively (Table 3.8).  Table 3.8. Eigenvalues, percentage of variance and cumulative percentage of variance for the first five principal components of the principal component analysis performed to the 18 biological time-series.         eigenvalue percentage of variance cumulative percentage of variance PC1 15.99907 44.44186 44.44186 PC2 9.847628 27.35452 71.79639 PC3 1.634104 4.539177 76.33556 PC4 1.522066 4.227962 80.56353 PC5 1.368377 3.801046 84.36457 Table 3.7. Continued 61   Figure 3.14. The first two principal component scores from a principal component analysis of the 18 biological time-series. . Blue colored areas correspond to negative phases while red colored areas correspond to positive phases. Vertical dashed lines indicate the 1977 and 1989 regime shifts. 62  The first principal component (PC1) scores of the 18 biological time-series showed a similar but inverse pattern to PC2env (Figure 3.14). PC1 showed a remarkable linear decline along the time-series, with high and positive scores during the early 1970’s that progressively decreased until 1993 (with a few exceptions) when they became negative and kept increasing in negative scores all the way until 2012 (Figure 3.14). The time-series showed a minimum in 2010 and a maximum in 1973 (Figure 3.14). All Alaskan biological indices had negative loadings while all the remaining variables corresponding to British Columbia and Washington State had positive loadings with the exception of Fraser River sockeye salmon total biomass which showed a relatively negative loading (r=-0.11) and both British Columbia Area 4 sockeye salmon total biomass and Columbia River sockeye salmon escapement which had negligible loadings (Table 3.9). The variables that best explained PC1 were Rivers Inlet sockeye salmon total biomass which was positively correlated to PC1 (r>0.18) and both Southeast Alaska chum salmon catch and Central Alaska pink salmon catch which were negatively correlated to PC1 (r<-0.18) (Table 3.9). PC2 scores showed a gradual transition from a negative phase starting in 1970 to a positive phase starting in 1983 which remained until 1998 when it turned negative again until 2012 (Figure 3.14). Southeast Alaska pink salmon catch, Southeast Alaska sockeye salmon catch, Central Alaska sockeye salmon catch and British Columbia Area 4 sockeye salmon total biomass had the highest loadings (r>0.19) and they were all positively correlated to PC2 (Table 3.9).   63  Table 3.9. The 18 biological time-series and their corresponding loadings of the first two principal components of the PCA. High loadings (r>|0.18|) are underlined.    PC1 PC2 Southeast Alaska pink salmon catch -0.12458 0.202135 Southeast Alaska chum salmon catch -0.18945 0.047454 Southeast Alaska sockeye salmon catch -0.05898 0.270285 Central Alaska pink salmon catch -0.19095 0.032351 Central Alaska chum salmon catch -0.141 -0.05067 Central Alaska sockeye salmon catch -0.14068 0.186304 Rivers Inlet sockeye salmon total biomass 0.187539 -0.00829 Fraser River sockeye salmon total biomass -0.11396 0.012971 Skeena River Tyee test-fishery sockeye salmon index 0.038275 -0.04333 British Columbia statistical area 4 pink salmon total biomass 0.052047 0.178445 British Columbia statistical area 4 chum salmon total biomass 0.110344 0.091418 British Columbia statistical area 4 sockeye salmon total biomass -0.00422 0.196343 British Columbia statistical area 7 pink salmon total biomass 0.175038 -0.02549 British Columbia statistical area 7 chum salmon total biomass 0.154712 -0.06284 British Columbia statistical area 8 pink salmon total biomass 0.117692 0.148012 British Columbia statistical area 8 chum salmon total biomass 0.05279 0.13779 British Columbia statistical area 11 sockeye salmon total biomass 0.142574 0.15807 Columbia River sockeye salmon escapement  -0.09203 -0.10391      64  3.5 Relationships between environmental conditions and stable isotope time-series  In order to assess any potential relationships between the changing environmental conditions in the Northeast Pacific Ocean during the period of our study and the carbon and nitrogen stable isotope time-series, we performed simple linear regression between our PCA results and the stable isotope time-series. Since (1) the majority of sockeye salmon returning to spawn at Rivers Inlet have spent two years at sea and (2) the stable isotope data obtained from whole scales reflect ocean conditions and diet biased towards the last season of growth (last years at sea) (Trueman and Moore, 2007), we expected to see an immediate or short-lag response between scale isotope ratios and environmental changes experienced in the open ocean.  Therefore, we compared our carbon and nitrogen stable isotope time-series with 0, 1 and 2 years lag PCA scores.             65  Table 3.10. Statistical results from simple linear regression of nitrogen and carbon stable isotope time-series with 0, 1 and 2 year lag PCA scores. Correlation coefficient (r), the probability (p-value) that the correlation was greater (or less) than 0, degrees of freedom (df), and number of data points compared (N) are given. Correlations significant at < 0.05 are shaded.   13C   15N Principal Component     r p-value df N  r p-value df N PC1env 0-lag PC1phys 0-lag PC1bio 0-lag -0.11 0.05 -0.24 0.76 0.87 0.5 8 11 8 10 13 10 0.57 0.64 -0.03 0.07 0.01 0.92 8 11 8 10 13 10 PC1env 1 year lag PC1phys 1 year lag PC1bio 1 year lag -.0.01 -0.02 0.04 0.9 0.93 0.93 7 11 7 9 13 9 0.46 0.4 0.44 0.2 0.17 0.22 7 11 7 9 13 9 PC1env 2 year lag PC1phys 2 year lag PC1bio 2year lag -0.07 0.04 -0.56 0.85 0.87 0.11 7 11 7 9 13 9 0.02 0.18 -0.02 0.95 0.56 0.97 7 11 7 9 13 9 PC2env 0-lag PC2phys 0-lag PC2bio 0-lag 0.26 -0.15 -0.31 0.47 0.63 0.38 8 11 8 10 13 10 -0.01 0.14 0.35 0.97 0.65 0.31 8 11 8 10 13 10 PC2env 1 year lag PC2phys 1 year lag PC2bio 1 year lag -0.02 -0.07 -0.1 0.95 0.83 0.8 7 11 7 9 13 9 -0.5 0.28 0.43 0.16 0.36 0.24 7 11 7 9 13 9 PC2env 2 year lag PC2phys 2 year lag PC2bio 2year lag 0.56 -0.15 -0.53 0.1 0.63 0.14 7 11 7 9 13 9 -0.01 -0.01 -0.09 0.97 0.99 0.82 7 11 7 9 13 9  There was a strong and significant positive relationship between 0-lag PC1phys and the 15N time-series (r=0.64, p<0.05) (Table 3.10). No significant relationships were observed between the remaining PCAs and the 15N and 13C time-series (Table 3.10). 66  3.6 Baseline isotopic data Available isotope data for potential sockeye salmon prey and suspended particulate organic matter (SPOM) collected at different places and times in the Gulf of Alaska allowed us to stablish a trophic hierarchy that can be used to assess any potential changes in the feeding habits that Rivers Inlet sockeye salmon might have experienced between the period of our study (Figure 3.15). Given that 15N NO3- decreases from east to west across the Gulf of Alaska (Schell, 1998), gonatid squid (Berryteuthis anonychus) sampled in the continental shelf may be expected to be enriched in 15N compared to the same squids sampled in the oceanic stations of the Gulf of Alaska (assuming similar trophic level) (Wu et al., 1996). Thus, the 15N signatures of gonatid squid collected in the continental shelf were not directly comparable to the 15N signatures of sockeye salmon prey collected in the oceanic stations of the Gulf of Alaska. We compared the 15N signatures of copepods sampled in the shelf (e.g., Hobson et al., 1997) versus copepods sampled in the oceanic stations (e.g., Wu et al., 1996). Copepods sampled in the shelf were 27% enriched in 15N compared to those sampled in the oceanic stations. Assuming similar trophic level, we applied this percentage of enrichment to the 15N values of squid sampled in the shelf to find the equivalent oceanic 15N signatures. Gonatid squid appeared to be situated at a higher trophic level than copepods (between copepods and amphipods). This is consistent with the trophic positioning of squid in the continental shelf observed by Hobson et al. (1997).  67   Figure 3.15. Mean (± SD) nitrogen stable isotope ratios from sockeye salmon prey and suspended particulate organic matter (SPOM) collected between May 1991 and May 1993 along the Line P and at Ocean Station Papa (OSP) (Wu et al., 1996; 1998); squid samples taken during spring and summer of 1992 and 1994 from the continental shelf of the Gulf of Alaska (δ15N shelf signatures were converted to δ15N oceanic signatures based on the ∆δ15N between oceanic and shelf copepod signatures -Hobson et al., 1997); and from sockeye salmon scales collected for the period 1915-2013 (this study). Sample sizes given in parentheses.      68  3.7 Nutrient data The average yearly surface nitrate concentration within 50-57N and 145-160W decreased in the late 1980s, reaching the lowest value in 1994. Nitrate concentration appears to recover in the 2000s, reaching the highest concentrations in 2001, 2007, and 2010 (Figure 3.16). No significant relationships were observed between the nitrogen stable isotope time-series and the 0, 1, and 2 year lag NO3 time-series (Table 3.11).   Figure 3.16. Black dashed line shows average yearly nitrate concentration (M) in surface waters in the Gulf of Alaska between 50-57N and 145-160W for the period 1985-2010. Red line shows yearly 15N (‰) mean of Rivers Inlet sockeye salmon scales for the period 1986-2010.     69  Table 3.11. Statistical results from simple linear regression of nitrogen stable isotope time-series for the period 1986-2010 with 0, 1 and 2 year lag NO3 time-series. Correlation coefficient (r), the probability (p-value) that the correlation was greater (or less) than 0, degrees of freedom (df), and number of data points compared (N) are given.    15N  NO3     r p-value df N  NO3 0-lag NO3 1 year lag NO3 2 year lag 0.48 0.25 -0.48 0.32 0.58 0.32 4 5 4 6 7 6    70   Discussion 44.1 Comparison of pre-spawn and post-spawn isotope data: physiological implications The enrichment in 13C observed in the muscle tissue of the post-spawn salmon relative to the one of the pre-spawn salmon (Figure 3.1 (a)) was likely caused by a reduction in the lipid content of muscle tissues during the upstream migration. Indeed, Doucett et al. (1999) observed a progressive enrichment in 13C values of the red muscle tissue as salmon migrated upstream. In fact, a range of at least 4-6‰ can be expected in the 13C values of anadromous salmonids during their freshwater spawning migration (Doucett et al., 1999). This isotopic enrichment in migrating salmon results from mobilization, reorganization and catabolism of stored lipids (Doucett et al., 1999). Since C:N ratios are  a good proxy for lipid content (Bodin et al., 2007; Post, 2002; Schmidt et al., 2003), the significantly lower C:N ratios observed in muscle tissue of post-spawn sockeye salmon relative to pre-spawn salmon (Figure 3.2 (a)) supports reduced lipid content as the cause of 13C enrichment. The similar 15N values observed in muscle tissue of both pre-spawn and post-spawn salmon (Figure 3.4 (a)) may be an indicator that there was not any protein turnover or recycling in muscle tissue of sockeye salmon during their freshwater spawning migration. Similarly, Doucett et al. (1999) did not observe any 15N enrichment in red muscle of Atlantic salmon (Salmon salar) during the upstream migration. Moreover, protein losses were observed only in overwintering Atlantic salmon but not during the spawning migration (Doucett et al., 1999), indicating that proteins were not metabolized in red muscle tissue during the upstream migration.  71  A similar behavior regarding the isotopic differences between muscle of pre-spawn and post-spawn fish was observed in scale tissues. Scales of post-spawn salmon were significantly enriched in 13C relative to scales of pre-spawn fish (Figure 3.1 (b)). C:N ratios in scales of post-spawn fish were also significantly lower when compared to scales of pre-spawn fish (Figure 3.2 (b)). These findings suggest that the lipid constituents of scale tissue may also be catabolized during the upstream migration of salmonids. Since scales are reabsorbed by all species of salmon during the spawning migration (Persson et al., 1999), it is possible that lipids found in scales were catabolized during the resorption process, leading to both enriched 13C values and lower C:N ratios in scales of post-spawn fish relative to scales of pre-spawn fish. The lack of 15N enrichment in scale tissue of post-spawn fish relative to pre-spawn fish (Figure 3.4 (b)) suggests that scale collagen was not catabolized during scale resorption in the upstream migration. These facts suggest that changes in the stable isotope ratios of both scale and muscle tissues of salmon can result from losses in lipid content as adults migrate upstream. Therefore, it is important to recognize the physiological status of the sampled fish and if necessary, account for lipid content in both muscle and scale tissue of anadromous salmonids or any fish species undergoing extensive migrations or any other excessive metabolic cost during a period of nutritional stress. Failure to take in account the importance of this “nutritional effect” could result in incorrect interpretations of the resulting stable isotope ratios in food web studies.   72  4.2 Isotopic relationship between muscle and scale tissue An understanding of the relationship between muscle- and scale-derived stable isotope signatures is important for the confident interpretation of scale isotopic results. For all years (2011, 2012 and 2013), the scale 13C values were significantly higher than those of muscle tissue (Table 3.1). However, this offset of carbon signatures between the two tissue types (13Cm-sc) was not constant across years (Figure 3.5), suggesting the influence of factors (e.g. lipid accumulation) that contribute to the 13C signature. Despite not being constant across years, 13Cm-sc was consistent with the findings of previous studies where salmon scales are enriched in 13C by 2.2-4.0‰ relative to muscle tissue (Johnson and Schindler, 2012; Satterfield and Finney, 2002), except in 2012 where 13Cm-sc was significantly lower than that observed in the literature (Figure 3.5). This coincides with both the unusually high C:N ratios  (Figure 3.8) and depleted 13C values (Figure 3.9) observed in 2012. Thus, we conclude that scale samples in 2012 contained lipid residues (e.g. mucus) that altered the 13C signature. The fact that after lipid normalization, the offset of carbon signatures between the two tissue types (13C’m-sc) is constant across all years and is similar to the 2.7‰ enrichment in 13C of scales relative to muscle tissue observed after lipid normalization in previous studies (Figure 3.6, Sinnatamby et al., 2008), indicates that the correction for C:N applied to the 13C values of muscle tissue and scales of Rivers inlet sockeye salmon successfully accounted for the effect of lipid content on the 13C signature of both tissues.  The 15N values from scales of 2011, 2012 and 2013 pre-spawn Rivers Inlet sockeye salmon were significantly depleted relative to that of muscle tissue while no significant differences in 73  the 15N values between muscle and scale tissue were observed in 2013 post-spawn Rivers Inlet sockeye salmon (Table 3.2). Both results are consistent with the ones observed in previous studies. Sinnatamby et al. (2008) observed depleted 15N values in scales of Atlantic salmon (Salmo salar) relative to those of muscle tissue while Satterfield and Finney (2002) observed that the 15N of Pacific salmon (Oncorhynchus spp.) did not differ significantly between the two tissues. Notably, the difference in 15N between the two tissues that we observed in 2011, 2012 and 2013 pre-spawn Rivers Inlet sockeye salmon, and that was also observed by Sinnatamby et al. (2008) in Atlantic Salmon,  was small and probably not biologically significant (Sinnatamby et al., 2008).  The similarities of our results to the ones observed in the literature indicate that after lipid-normalization, scales of Rivers Inlet sockeye salmon are reliable resources for retrospective stable isotope analyses. 4.3 Principal component analysis (PCA) With the first principal component analysis of the 33 physical and biological variables (PC1env, Figure 3.12) we were able to identify the following shifts in the environmental conditions of the North East Pacific Ocean:  an abrupt shift to a positive phase in the late 1970’s, an abrupt shift to a negative phase in 1999 that lasted for three consecutive years, and another abrupt shift to a negative phase in 2007. The first principal component of the independent climate-only PCA (PC1phys, Figure 3.13) supports the same environmental shifts observed in PC1env since the latter is mostly explained by physical indices (Table 3.5). These findings coincide with the ones of Hare and Mantua (2000) and Litzow and Mueter (2013) who successfully identified an abrupt climate 74  regime shift of the North Pacific Ocean to a warm or positive phase in 1977 and another climate regime shift to a negative phase in 2007/2008 with similar characteristics to the ones that prevailed before the 1977 shift. The 1999 shift to a negative phase which lasted until 2002, observed in both PC1env (Figure 3.12) and PC1phys (Figure 3.13) coincides with the rapid transition to the 1999-2002 La Niña following the 1999-1998 El Niño. The environmental conditions set during the 1999-2002 La Niña are considered to qualify as a potential regime shift to a negative and highly productive phase of the Northeast Pacific Ocean (Batten and Welch, 2004; Bograd et al., 2000; Peterson and Schwing, 2003). The 1989 negative score followed by two consecutive years of low positive scores in PC1phys (Figure 3.13) coincides with the less important climate regime shift to “moderate” conditions of the Northeast Pacific in 1989 identified by Hare and Mantua (2000). The gradual shift from a negative to a positive phase observed in PC2env (Figure 3.12) is mirrored in PC1biol (Figure 3.14) which is characterized by a gradual shift from a positive to a negative phase. The fact that PC2env was mostly explained by biological indices (Table 3.5) explains the similarities between PC2env and PC1biol. Both PCAs successfully captured the “inverse production regimes” between Alaskan and West Coast salmon populations demonstrated by Hare et al. (1999).  One being the mirror image of the other was explained by both PCAs being positively correlated to biological indices with loadings of opposite sign. In other words, PC2env was positively correlated to Alaskan salmon stocks (Table 3.5) while PC1biol was positively correlated to West Coast salmon stocks (Table 3.9).  The gradual shift observed in both PC2env (Figure 3.12) and PC1biol (Figure 3.14) coincides with the gradual response of biological indices to abrupt shifts in climate observed by Hare and Mantua (2000) and Litzow 75  and Mueter (2014). This remarkable response can be explained either by a gradual transition of the biological indices to an equilibrium following a perturbation (e.g., Frank et al., 2011) or by a response of the biota to incremental forcing mechanisms such as anthropogenic climate change, fisheries and habitat loss (Litzow and Mueter, 2014).  The fact that the environmental shifts observed in our PCA coincide with the climate regime shifts identified by Hare and Mantua (2000) and Litzow and Mueter (2014), indicate that our principal component analyses are a reliable synthesis of the large-scale changes in climate affecting the North East Pacific Ocean. Therefore, the obtained PCA results can be used to test the relationship between our stable isotope data and the environmental changes affecting the North East Pacific Ocean. 4.4 Stable isotope variability The high variability observed in our carbon and nitrogen stable isotope time-series of Rivers Inlet sockeye salmon scales (Figure 3.10) suggests that there were substantial changes in the environmental conditions experienced by Rivers Inlet sockeye salmon in the high seas during the period 1915-2013. Three processes that may account for the variability observed in our stable isotope time-series are: (1) temporal changes in diet/trophic position of Rivers Inlet adult sockeye, (2) changes in their foraging location and (3) changes in oceanographic conditions over a large region where adult Rivers Inlet sockeye salmon forage.    76  Diet/trophic variability Sockeye salmon in the Gulf of Alaska are known to prey on a wide variety of organisms, and are known to feed at many different trophic levels by switching their diets from zooplankton to micronekton such as squids and fishes (Kaeriyama et al., 2000). Variation on their food habits is closely associated with spatial and/or temporal variation in prey abundance and availability caused by changing ocean conditions or in response to inter- and intra-specific competition for prey resources ( Aydin et al., 2000, Ruggerone et al., 2003; Tadokoro et al., 1996). For example, Kaeriyama et al. (2004) conducted an evaluation of salmon diets from 1994 to 2000 along the different oceanic domains of the Gulf of Alaska. They found that gonatid squid (Berryteuthis anonychus) was the dominant prey of sockeye salmon feeding in the Gulf of Alaska (Aydin et al., 2000; Kaeriyama et al., 2000; Pearcy et al., 1988). However, sockeye salmon in the Gulf of Alaska would shift to a more diverse zooplankton diet in response to a decrease in gonatid squid abundance/availability (Kaeriyama et al., 2000; Kaeriyama et al., 2004). Such reduced abundance/availability of squid in the Gulf of Alaska has been attributed to changes in the northern boundary of squid associated with the strong 1997 El Niño and 1999 La Niña events, reducing the geographic overlap between sockeye salmon and squid (Kaeriyama et al., 2004). Given the fact that whole scales reflect the isotopic signature of the integrated diet of that particular fish biased towards the period when most growth was accumulated (last year at sea) (Trueman and Moore, 2007), and the increase in both 15N and 13C  by 3-4‰ (average of 3.4‰ in marine food webs) and 0-1‰ (average of 1‰ in marine food webs) respectively per trophic transfer (Minagawa and Wada, 1984; Vander Zanden and Rasmussen, 2001), a shift of Rivers Inlet sockeye feeding at higher trophic level diet based mainly on gonatid squid (Figure 3.15) to 77  a lower trophic level diet richer on mesozooplankton (copepods, euphausiids and pteropods) (Figure 3.15), would translate in a decrease in both the nitrogen and carbon isotope ratios at a ratio of approximately 3.4:1 (Wada et al., 1987; Rau et al., 1990). In other words, if the variation observed in the stable isotope time-series was to be caused by a simple change in trophic level (assuming that other factors driving stable isotope signatures remain constant), the increase/decrease in 15N would be 3.4 times bigger than that of 13C (slope in 15N versus 13C close to 3.4). Changes in the isotopic values of Rivers Inlet sockeye salmon may also result from prey items themselves shifting foraging trophic level. The omnivorous copepod Neocalanus plumchrus will shift feeding strategy depending on ambient phytoplankton availability (Kobari et al., 2003). Since nitrate is the only source of nitrogen resupplied to surface waters in amounts great enough to support biomass production (Whitney et al., 1998), limitation of its supply will also limit phytoplankton production (Whitney et al., 1998). Thus, it is possible that during periods of nitrate depletion in a large area of the Gulf of Alaska, Neocalanus shifted from a diet richer in phytoplankton to a diet richer in microzooplankton, causing a lengthening of the food chain, which in turn would result in an increase of both 15N and 13C transferred up the food chain up to salmon. This transfer may be direct during periods when salmon preys more heavily on mesozooplankton, or indirect during periods when salmon preys more heavily on gonatid squid, since the latter also preys on zooplankton (Kaeriyama et al., 2004). Moreover, Mackas et al. (2007) observed that smaller subtropical species of copepods were more abundant and with a more northerly distribution during warm periods (e.g., the 1991 El Niño event). This change in 78  zooplankton community from bigger boreal calanoid copepod species to smaller sub-tropical species may also result in a lengthening of the food web.  Other studies on the trophic ecology of sockeye salmon in the Gulf of Alaska using stable isotope analysis also conclude that sockeye salmon feed opportunistically and broadly across the prey items available in the different oceanic domains of the Gulf of Alaska and that changes in trophic level may have occurred during periods of high climate variability (e.g. Johnson and Schindler, 2012; Satterfield and Finney, 2002). The fact that our carbon and nitrogen stable isotope time-series were strongly correlated (R2=0.36, Figure 3.11), together with the previously observed high interannual variability on the feeding habits of sockeye salmon in the Gulf of Alaska, suggests that Rivers Inlet sockeye salmon may have experienced changes in trophic level in response to periods of high climate variability. Since shifts in diet appear to be closely related to shifts in climate (e.g. Kaeriyama et al., 2000; Kaeriyama et al., 2004), the strong correlation between 15N and PCAphys (Table 3.10) suggests that changes in trophic level may have been responsible, in part, for the 15N variation observed in our stable isotope time-series.  Nonetheless, only some variability in the carbon and nitrogen stable isotope time-series can be explained by simple changes in diet/trophic level, since the expected 3.4:1 ratio of increase/decrease in nitrogen and carbon stable isotopes at each trophic level only occurs in a few periods within the stable isotope time-series. The remaining periods are characterized by a decoupling of the carbon and nitrogen time-series or by increases/decreases of both nitrogen and carbon stable isotopes inconsistent with the 3.4:1 increase/decrease in nitrogen and carbon stable isotopes expected at each trophic level. Therefore, we conclude that mechanisms 79  other than changes in diet/trophic level are also contributing to the variability observed in our stable isotope time-series. Foraging habits/location Pacific salmon are a highly migratory species that spend multiple years foraging widely in the North Pacific Ocean and thus potentially encountering enormous spatial variability of prey types and isotopic composition (Kaeriyama et al., 2000). The ocean migration of West Coast stocks of sockeye salmon is generally believed to involve a counterclockwise movement south in winter and north in summer following the currents of large oceanic gyres in the Gulf of Alaska (Burgner, 1991). Although interannual variability in migration routes has not been investigated, Pacific salmon stocks are believed to be restricted further north in the North Pacific ocean during warmer years (Blackbourn, 1987). Nearshore habitats are generally more productive and tend to support prey with enriched 13C and 15N compared with offshore habitats (Schell et al., 1998). Although there is unlikely to be a change between nearshore offshore feeding during the pelagic stage of sockeye salmon, the fact that sockeye salmon stocks may be distributed further north during warmer years may confer a change in the diet composition. In the offshore waters of the Gulf of Alaska, there are two distinct latitudinal “zones” of feeding for adult salmon (Aydin et al., 2000; Kaeriyama et al., 2000; Kaeriyama et al., 2004; Pearcy et al., 1988). In the southern zone (50-53N), sockeye salmon are known to feed mainly on the micronektonic squid Berryteuthis anonychus, while between 53N and the northern continental shelf, squid are largely absent and lower trophic level zooplankton dominate sockeye salmon diets (Aydin et al., 2000; Kaeriyama et al., 2000; Kaeriyama et al., 80  2004; Pearcy et al., 1988). Therefore, a change to a lower trophic level diet due to sockeye salmon feeding further north during warmer years may translate to a decrease of both carbon and nitrogen stable isotope values. The co-variation observed between our carbon and nitrogen stable isotope time-series (Figure 3.11) supports this. Thus, changes in the isotopic composition of Rivers Inlet sockeye salmon could be caused by fish feeding for longer periods of time in the northern oceanic domains of the Gulf of Alaska during warm periods.   Environmental influences The variations observed in the isotope data of Rivers Inlet sockeye salmon scales could be also explained by changes in oceanographic conditions altering the stable isotope composition at the base of the food web over a large region of the Gulf of Alaska. Interannual variability in 15N may reflect changes in nitrogen assimilation at the base of the food web. An inverse relationship has been observed between surface NO3- and the 15N of particulate organic matter (POM) (Rau et al., 1998), reflecting an increase in the 15N of the trophic baseline which would be transferred up the food chain to salmon. During the past decades, the upper ocean in the Northeast Pacific has undergone a series of changes in physical and chemical properties linked to large-scale changes in climate such as El Niño/La Niña cycle (Whitney et al., 1998). For example, during the 1989 La Niña, winter waters at Ocean Station Papa (OSP) were relatively cool, saline and nitrate rich (Whitney et al., 1998). However, with the onset of the 1991 El Niño period, winter waters at OSP were more saline by 0.3 psu, warmer by over 2C and nitrate depleted by 30% (Whitney et al., 1998). Moreover, the nitrate time-series covering a large area of the Gulf of Alaska (Figure 3.16) showed high interannual variability for the period 1985-2010. 81  The changes in surface nitrate concentration observed in the Gulf of Alaska during the past decades may therefore have contributed to the variation in nitrogen stable isotope signature of Rivers Inlet sockeye salmon scales.  The 13C values in top trophic level marine organisms are set by the composition of phytoplankton in the food web.  The isotopic composition of the phytoplankton is affected by the isotopic composition of the dissolved inorganic carbon (DIC) assimilated by primary producers (Burkhardt et al., 1999; Freeman and Hayes, 1992). During photosynthetic fixation of CO2 into organic material, algal cells discriminate against the heavier stable carbon isotope 13C (Burkhardt et al., 1999; Freeman and Hayes, 1992). Thus, 13C values have a strong inverse relationship with DIC concentration (Freeman and Hayes, 1992; Rau, 1994). Increased solubility of CO2 at lower temperatures increases the DIC concentration and availability of 12C for primary producers, resulting in lower baseline 13C values.  Based on the mean +0.4‰/C 13C sensitivity to temperature observed by Rau et al. (1997), the large year to year variations in 13C observed in our stable isotope time-series could be also caused by inter-annual variations in sea surface temperature during periods of high climate variability (e.g., El Niño/La Niña cycle, climate regime shifts). Factors other than CO2 may also have a significant impact on the carbon isotopic composition in phytoplankton. Differences in algal growth rate as well as taxon-specific differences would have an effect on the carbon fractionation process (e.g., Rau et al., 1989). The offshore Gulf of Alaska is dominated by small-sized phytoplankton (mostly dinoflagellates) whose populations stay relatively constant throughout the year (without typical spring blooms), relatively high 82  macronutrient concentrations year-round, and iron limitation of phytoplankton growth (Harrison et al., 2004).  Thus, it is unlikely that differences in carbon fractionation caused by changes in phytoplankton species were responsible for any of the variability observed in the carbon stable isotope time-series. However, despite phytoplankton communities being relatively constant in the offshore Gulf of Alaska, large-scale changes in climate affecting the Northeast Pacific are the cause of large variability in environmental factors such as temperature, depth of the mixed layer and nutrient supply, which at the same time control primary production. Since 13C increases as primary productivity rates increase (Laws et al., 1995), changes in primary productivity associated with large-scale changes in climate may contribute to the 13C variability observed in our stable isotope time-series. This is supported by the decoupling of our carbon and nitrogen stable isotope time-series during some periods. Generally, low primary production is associated with low nutrient levels; therefore, an increase in 15N caused by low surface nitrate concentration will be accompanied by a decrease in 13C as a consequence of lower primary production. The fact that 13C and PCAphys were not correlated (Table 3.10) suggests that mechanisms other than temperature contributed to the 13C signature. Thus, large changes in primary production occurred in the offshore waters inhabited by Rivers Inlet sockeye salmon during the period of our study likely contributed to the variation observed in the carbon stable isotope time-series.    83  Summary Changes in the trophic position of Rivers Inlet sockeye salmon and/or changes in the length of the food web are expected to be responsible for some of the isotope variability observed in the 13C and 15N time-series of Rivers Inlet sockeye salmon scales (Figure 3.10). Such trophic variation is likely associated with large-scale changes in climate. During periods of higher primary production rates (e.g., negative phases of the PDO and La Niña events), a shortening of the food web may result from macrozooplankton (Neocalanus spp.) feeding more heavily on phytoplankton (Kobari et al., 2003). On the contrary, reduced primary production during warmer periods (e.g., positive phases of the PDO and El Niño events) may result in a lengthening of the food web due to macrozooplankton feeding more heavily on microzooplankton (Kobari et al., 2003). A higher proportion of small sub-tropical copepods versus bigger boreal copepods in the Gulf of Alaska associated with “warm” ocean conditions (Mackas et al., 2007) may also contribute to a lengthening of the food web. Similarly, Rivers Inlet sockeye salmon may have experienced changes in trophic level by shifting from a higher trophic level diet richer in gonatid squid to a lower trophic level diet richer in mesozooplankton. Such variation is related to the availability of prey items (Aydin et al., 2000; Burgner, 1991), which in turn is associated with large-scale changes in climate (e.g. Kaeriyama et al., 2000; Kaeriyama et al., 2004). Changes in the northern boundary of squid is associated with strong El Niño/La Niña events (Kaeriyama et al., 2004), while sockeye salmon in the Gulf of Alaska tend to show a more northerly distribution during warm winters (Blackbourn, 1987), effecting the geographic overlap between sockeye salmon and squid. Therefore, a shift to a lower trophic level diet richer in mesozooplankton is expected during years of reduced geographic overlap 84  between sockeye salmon and squid. Furthermore, fluctuations in temperature, surface nitrate concentration, and primary productivity rates are also expected to contribute to the stable isotope variability observed in the 13C and 15N time-series of Rivers Inlet sockeye salmon scales (Figure 3.10). Warm periods may be associated with increased SST (increased stratification), resulting in decreased surface nitrate concentrations, and decreased primary production rates. On the contrary, cold periods may be associated with lower SST (decreased stratification), resulting in increased surface nitrate concentrations, and increased primary production rates. The potential factors forcing high seas sockeye salmon isotope ratio response to shifting ocean conditions are summarised in Figure 4.1.    Figure 4.1. Summary of expected interactions between open ocean environmental conditions and Rivers Inlet sockeye salmon scale stable isotope signatures in response to climate variability. 85  4.5 Stable isotopes and climate shifts Following the rules established in Figure 4.1 we re-valuated the PCA scores, NO3 time-series (Figure 3.16), baseline data (Figure 3.15), and additional climatic information from the literature, in an attempt to identify the environmental conditions experienced by Rivers Inlet sockeye salmon during the pelagic stage of their life cycle in relation to large-scale climate events. Although we have limited information regarding ocean conditions in the Gulf of Alaska for the beginning of our stable isotope time-series, abrupt switches in the sign of the PDO index were documented around 1925 and 1947 when the PDO shifted from a negative to a positive phase and from a positive to a negative phase respectively (Mantua et al., 1997). Both of these switches were indicators of regime shifts between periods of relatively stable climate, and were associated with substantial changes in the community ecology of the Gulf of Alaska (Mantua and Hare, 2000). Since higher SSTs and lower surface nitrate concentration are associated with a warmer ocean, we would expect higher values of 13C and 15N during the positive phase relative to the negative phase, assuming that the trophic level of Rivers Inlet sockeye salmon remained constant. However, both 13C and 15N increased between 1946 and 1960 (Figure 3.10). A shift of PC1phys to positive scores in both 1958 and 1959 (Figure 3.13) would explain the higher values of 13C and 15N observed in 1960 relative to 1946 (Figure 3.10). Such an occurrence reflects how the long-term effects of the PDO in the environmental conditions of the offshore Gulf of Alaska may be intensified, or in this case, reversed by high frequency events of climate variability (e.g. El Niño/La Niña cycles). 86  The increase of both 13C and 15N observed between 1960 and 1965 (Figure 3.10) is not consistent with the lower SSTs and higher surface concentration expected during the “colder” environmental conditions (larger negative scores in PC1phys) experienced by Rivers Inlet sockeye salmon in the offshore Gulf of Alaska in 1965, relative to 1960 (Figure 3.13). Instead, the increase in both 13C and 15N observed in 1965 relative to 1960 was consistent with a change in trophic level, since the ratio of increase in both nitrogen and carbon stable isotopes was close to the approximately 3.4:1 ratio of increase in both isotopes that would be expected when feeding at a higher trophic level (Wada et al., 1987; Rau et al., 1990). Whether this change in trophic level was due to Rivers Inlet sockeye salmon shifting prey or by the prey itself shifting trophic level it is unclear. The decrease in both 13C and 15N observed in 1970 relative to 1965 is not consistent with only a dietary change since the decrease in carbon is larger than that of nitrogen. Even though PC1phys scores were positive in 1970, PC1phys scores were largely negative in 1969 (Figure 3.13), when Rivers Inlet sockeye salmon returning to spawn in 1970 would have also been feeding in the open ocean of the Gulf of Alaska. On the contrary, fish returning to spawn in 1965 would have experienced a warmer ocean since PC1phys scores were positive in 1964. The lower SSTs and higher surface nitrate concentration characteristic of a colder and more productive ocean could explain the decrease in both carbon and nitrogen stable isotopes observed in 1970 relative to 1965 (Figure 3.10). The following period was characterized by the well documented 1977 regime shift to a “warm” or positive, which was characterized by an extensive reorganization of the Gulf of Alaska marine 87  ecosystem (Hare and Mantua, 2000; Mantua et al., 1997). In addition, the strong 1983 El Niño event intensified the warm conditions set by the 1977 regime shift. This was reflected in the large positives scores observed in both PC1env (Figure 3.12) and PC1phys (Figure 3.13) after the 1977 regime shift. Therefore, both an increase in SST and a decrease in surface nitrate concentration are to be expected between 1970 and 1986. The large increase of 13C during this period (Figure 3.10 (a)) is consistent with Rivers Inlet sockeye salmon experiencing higher SSTs in 1986 relative to 1970. However, the small increase in 15N observed between 1970 and 1986 (Figure 3.10 (b)) is not consistent with the large decrease in surface nitrate concentration that we would expect during this period, especially after the 1983 El Niño event. This suggests that a shortening of the food web and/or sockeye salmon feeding at a lower trophic level may have reduced the increase in 15N expected from a reduction in surface nitrate concentration. A shortening of the food web is unlikely due to reduced primary production rates associated with a warmer ocean. Instead, sockeye salmon may have shifted from a higher trophic level diet based on squid to a lower trophic level diet richer in mesozooplankton. In fact, zooplankton populations in the offshore Gulf of Alaska experienced a large increase coinciding with the 1977 regime shift, which lasted until the late 1980’s (Brodeur and Ware, 1992; Sugimoto and Tadokoro, 1997; Sugimoto and Tadokoro, 1998). This increase in abundance of zooplankton, together with the reduced geographic overlap between sockeye salmon and squid observed during El Niño events (e.g., Kaeriyama et al., 2004), may have resulted in Rivers Inlet sockeye salmon shifting to a lower trophic level diet richer in zooplankton. The subsequent sharp decrease in both 13C and 15N coincided with the onset of the 1989 La Niña event, when winter waters at Ocean Station Papa (OSP) were relatively cool, saline and 88  nitrate rich (Whitney et al., 1998). This is reflected in the negative scores observed in both PC1env (Figure 3.12) and PC1phys (Figure 3.13) in 1989. Although both a decrease in SSTs and an increase in surface nitrate in 1990 relative to 1986 were consistent with the sharp decrease in both 13C and 15N (Figure 3.10), a shortening of the food web could have also contributed to the decrease in both stable isotopes between 1986 and 1990. The high levels of surface nitrate concentration associated with the 1989 La Niña increased primary production in a large are of the Gulf of Alaska (Whitney et al., 1998). A shift of Neocalanus to a more herbivorouse diet would be transferred up the food chain up to salmon, which could also explain part of the decrease in both stable isotopes between 1986 and 1990.  The increase of both 13C and 15N between 1990 and 1992 coincided with the onset of the 1991 El Niño period, which was characterized by warmer (by over 2C) and nitrate depleted (by 30%) surface waters at OSP relative to 1989 (Whitney et al., 1998). That would explain both the sharp increase of 13C and the less marked increase of 15N in 1992 relative to 1990 observed in our time-series (Figure 3.10). A shift of Rivers Inlet sockeye salmon to a higher trophic level diet (e.g., higher proportion of gonatid squid) and/or a lengthening of the food web associated with a warmer ocean may have also contributed to the increase in both stable isotopes.  The decoupling in the carbon and nitrogen stable isotope time-series between 1992 and 1995 may be explained by changes at the base of the food web contributing to the carbon and nitrogen stable isotope signatures in opposite ways. As a consequence of the 1991 El Niño event, there was a 50% reduction in NO3 available to phytoplankton at OSP in the spring of 1994, compared with 1991 (Figure 3.16) (Whitney et al., 1998). This may explain the increase in 89  15N observed in our time-series (Figure 3.10 (b)). At the same time, the decrease in primary production associated with the low levels of surface nitrate concentration (Whitney et al., 1998) likely caused the decrease of 13C observed between 1992 and 1995 (Figure 3.10 (a)). A lengthening of the food web associated with a warmer ocean may also have contributed to the increase in 15N observed during this period.  1999 was characterized by the onset of the 1999 La Niña event. This was reflected in the large negative scores observed in PC1env (Figure 3.12) and PC1phys (Figure 3.13) in 1999. The increase in surface nitrate concentration in the Gulf of Alaska associated with the 1999 La Niña (Whitney and Welch, 2002) was not reflected in the nitrogen stable isotope time-series. Instead, we observed an increase in 15N between 1995 and 2000 (Figure 3.10 (b)). Therefore, a lengthening of the food web and/or sockeye feeding at a higher trophic level diet may have caused the increased in 15N during this period. The expected large decrease in 13C associated with lower SSTs in 2000 relative to 1995 was likely masked by the increase in primary production resulting from the replenishment of nutrients associated with the 1999 La Niña (Whitney and Welch, 2002). An abrupt reversion to a PDO-negative state occurred in the winter 2007/08. This switch was accompanied with widespread ecological changes in the Northeast Pacific Ocean (Litzow and Mueter, 2014). However, during the strong 2009/2010 El Niño event, the “cold” environmental conditions set by the 2007 regime shift were reversed. This was reflected in the positive scores observed in PC1phys (Figure 3.13) in 2010. This explains why the carbon and nitrogen stable isotope signatures barely fluctuated between 2004 and 2010 (Figure 3.10), since in both years 90  Rivers Inlet sockeye salmon experienced similar environmental conditions in the offshore Gulf of Alaska (similar PC1phys scores, Figure 3.13). After the 2009/2010 El Niño, the strong 2011 La Niña event intensified the “cold” environmental conditions set by the 2007 regime shift to a negative phase of the PDO (negative PC1phys scores, Figure 3.13). The lower SSTs and higher surface nitrate concentrations that Rivers Inlet sockeye salmon encountered in the offshore Gulf of Alaska during the strong 2011 La Niña relative to the 2010 El Niño  were reflected in the large decrease in both the carbon and nitrogen stable isotope time-series between 2010 and 2011 (Figure 3.10).  A shortening of the food web associated to a colder and more productive ocean may have contributed to the decrease in both carbon and nitrogen stable isotopes. The following increase of 15N observed in 2012 and 2013 may have resulted from reduced surface nitrate concentrations associated with more moderate (less cold) ocean conditions after the 2011 La Niña event (Figure 3.13). Moreover, a lengthening of the food web and/or a change to a higher trophic diet associated with a warmer ocean may have also contributed to the increase in 15N observed during this period (Figure 3.10 (b)). The increase in SSTs expected after the 2011 La Niña, especially in 2013 when environmental conditions were less “cold” (Figure 3.13), was not reflected in the carbon stable isotope time-series (Figure 3.10 (a)). Thus, other factors such as reduced primary productivity rates associated with warmer conditions may be masking the expected increase of 13C in 2013.   91  4.6 Environmental conditions and sockeye salmon returns Rivers Inlet sockeye salmon experienced highly variable open ocean conditions between 1915 and 2013. Low frequency (decadal scale) shifts in the physical environment of the offshore Gulf of Alaska associated with North Pacific climate (e.g., PDO) were reflected in the carbon and nitrogen stable isotope signatures of returning adult scales. Rivers Inlet sockeye experienced a warmer, and nutrient depleted ocean during positive phases of the PDO while a colder and more nutrient rich ocean was characteristic of negative phases of the PDO.  Overlaid on top of the low frequency cycles was the influence of high frequency shifts in climate (e.g., El Niño/La Niña events). The occurrence of El Niño events during positive phases of the PDO intensified the warm and nutrient depleted ocean conditions characteristic of positive PDO phases (e.g. the 1983 El Niño), while El Niño events occurring during negative phases of the PDO reversed the colder and nutrient rich conditions characteristic of negative PDO phases to a warmer and nutrient depleted ocean (e.g. the 2010 El Niño). Similarly, La Niña events occurring during cold phases of the North Pacific regime intensified the already cold and nutrient rich conditions characteristic of negative PDO phases (e.g. the 2011 La Niña) while la Niña events occurring during positive PDO phases reversed the warm and nutrient depleted ocean conditions characteristic of negative PDO phases to a colder and nutrient rich ocean (e.g. the 1989 La Niña). Such variation in the nutrient supply resulted in reduced primary production rates during warm periods (positive PDO phases and El Niño events) and increased primary production rates during cold periods (negative PDO phases and La Niña events). The feeding ecology of Rivers Inlet sockeye salmon changed in response to these changes in the physical (temperature and nutrient supply) and biological (primary production) environment of the offshore Gulf of Alaska. 92  Whether this variation in the trophic ecology of Rivers Inlet sockeye salmon was driven by Rivers Inlet sockeye salmon shifting prey and/or by the prey itself feeding at a different trophic level is unclear. The isotope data is consistent with a lengthening/shortening of the food web coinciding with warm and less productive periods, and with cold and more productive periods respectively. However, the isotope data also supports Rivers Inlet sockeye salmon shifting diet depending upon prey availability (e.g., shift to a lower trophic level diet coinciding with higher abundance of zooplankton after the 1977 regime shift). It is possible that a combination of these two factors was responsible for the changes in the stable isotope ratios of Rivers Inlet sockeye salmon during the period of this study. It is worth emphasizing that after correcting the 13C time-series of Rivers Inlet sockeye salmon scales for C:N ratios, not only the carbon and nitrogen stable isotope time-series were significantly correlated (Figure 3.11), but also the 13C time-series reflected the variability in both temperature and primary production rates expected in cold and warm periods of North Pacific climate (e.g., PDO and ENSO). Our results differed from previous stable isotope studies of sockeye salmon scales where a correction for C:N ratios was not applied. Both Satterfield and Finney (2002), and Johnson and Schindler (2012) did not observe any correlation between the carbon and nitrogen stable isotope time-series. Moreover, Satterfield and Finney (2002) observed minor fluctuations in the 13C time-series during a period of high climate variability, while Johnson and Schindler (2012) observed a relationship between the isotopic data and climate shifts but the magnitude of variation was small, also inconsistent with the high variability of North Pacific climate characteristic of the period of their study. This reaffirms the importance of applying a correction for lipid content in scale tissue of salmonids, and how not 93  applying this correction factor could lead to the incorrect interpretation of the resulting isotope ratios in food web studies.  The stable isotope analysis conducted during this study showed that there is a clear relationship between large-scale changes in climate and the feeding ecology of Rivers Inlet sockeye salmon. It has also been established that there is a relationship between large-scale changes in climate and sockeye salmon population dynamics (Francis and Hare, 1994; Hare and Mantua, 2000; Hare et al., 1999; Mantua et al., 1997). Therefore, could it be possible that the changes in the feeding ecology experienced by Rivers Inlet sockeye salmon in the offshore Gulf of Alaska contributed to the population declines associated with climate forcing? Rivers Inlet sockeye salmon started experiencing low and unstable returns coinciding with the 1977 regime shift before collapsing in the early 1990’s (Figure 1.2). The feeding ecology of Rivers Inlet sockeye salmon also changed during these periods. After the 1977 regime shift and the 1983 El Niño event, Rivers Inlet sockeye salmon likely shifted to a lower trophic level diet richer in zooplankton. Moreover, a reduction in primary production associated with the 1991 El Niño event likely caused a lengthening of the food web in the early 1990’s. Both a low trophic level diet richer in zooplankton as well as a longer and less efficient food web, coupled with higher metabolic costs associated with warmer temperatures (Hinch et al., 1995a) could potentially affect sockeye salmon growth in the open ocean. Low caloric diets in the open ocean have previously been associated with reduced growth of sockeye salmon (Aydin et al., 2000; Kaeriyama et al., 2000). Both Aydin et al. (2000) and Kaeriyama et al. (2000) observed that in years when the caloric value of prey consumed was higher (e.g., diet richer in gonatid squid) salmon grew significantly larger than in years when salmon fed more heavily on zooplankton. 94  Reduced growth of salmon could potentially increase their vulnerability to predation and energy depletion during the migration back to natal rivers, negatively affecting salmon survival rates. For example, Crossin et al. (2004) observed that the river entry energy states (somatic energy levels) of Fraser river sockeye salmon were higher during regimes characterized by good growing conditions (e.g., cold and more productive periods), and low during a regime with poorer, transitional growing conditions (e.g., warmer and less productive periods). Lower somatic energy levels at the onset of the river migration, as a consequence of warmer SSTs and poorer feeding conditions in the high seas during warm periods, may result in: risk of insufficient energy reserves to complete river migration and spawning (Hastler et al., 2011; Crossin et al., 2008); increased risk of disease due to a compromised immune system (Miller et al., 2014); reduced capacity to endure the higher metabolic demands associated with warmer riverine SSTs (Eliason et al., 2011); and reduced egg production (Healey, 1987). All of these potential effects of reduced growth/somatic energy levels associated with warmer SSTs and poor feeding conditions in the high seas may result in increased rates of in-river or pre-spawning mortality. Thus, the changes observed in Rivers Inlet sockeye salmon feeding ecology, together with higher metabolic demands associated with positive PDO phases and El Niño events, could potentially have a negative effect on sockeye salmon survival rates. A more comprehensive study of the foraging ecology of sockeye salmon in the open ocean and its relationship with salmon survival rates is needed for a better understanding of the mechanisms driving sockeye salmon population dynamics.  95  4.7 Conclusions  This study made several important advancements in the application of scale isotope analysis to ecological studies. We demonstrated that:  1. Both pre-spawn and post-spawn scale tissue of salmonids can be used in retrospective stable isotope analyses. 2. A correction for lipid content is needed when using muscle and/or scale tissue of salmonids or other species experiencing exhausting migrations as the large fluctuations in lipid content in their tissues will bias the 13C values. 3. Salmon scales incorporate the stable isotopes of carbon and nitrogen in a consistent way relative to muscle, allowing between tissue comparisons. The stable isotope analysis performed on Rivers Inlet sockeye salmon for the period 1915-2013 allowed us to draw several general conclusions: 1. Rivers Inlet sockeye experienced a warmer and less productive ocean during positive PDO phases and El Niño events. In contrast, Rivers Inlet sockeye salmon experienced a colder and more productive ocean during negative PDO phases and La Niña events. 2. Rivers Inlet sockeye salmon feeding ecology changed in response to both low frequency (PDO shifts) and high frequency (El Niño/La Niña cycle) shifts of North Pacific climate.  3. The observed changes in the feeding ecology of Rivers Inlet sockeye salmon associated with large-scale changes in climate are expected to significantly affect the condition and size of maturing fish, which in turn may affect sockeye salmon populations through 96  changes in survival rates in the high seas and in the freshwater phase of the return migration.                          97  References  Aydin, K.Y., Myers, K.W. and Walker, R.V., 2000. Variation insummer distribution of the prey of Pacific salmon (Oncorhynchus spp.) in the offshore Gulf of Alaska in relation to oceanographic conditions, 1991–98. N. Pac. Anadr. Fish Comm. Bull. 2: 43–54. Batten, S. D., Welch, D. W., 2004. Changes in oceanic zooplankton populations in the North-east Pacific associated with the possible climatic regime shift of 1998/1999. Deep Sea Res. II, 51: 863–873. Beamish, R. J., Mahnken, C., Neville, C. M., 1997. Hatchery and wild production of Pacific salmon in relation to large-scale, natural shifts in the productivity of the marine environment. ICES J Mar Sci 54(6):1200–1215. Blackbourn, D. J., 1987. Sea surface temperature and pre-season prediction of return timing in Fraser River sockeye salmon (Oncorhynchus nerka). Can. Spec. Pub. Fish. Aquat. Sci 96: 296–306. Bodin, N., Le Loc’h, F., Hily, C., 2007. Effect of lipid removal on carbon and nitrogen stable isotope ratios in crustacean tissues. Journal of Experimental Marine Biology and Ecology 341: 168–175.  Bograd, S., Digiacomo, P., Durazo, R., Hayward, T., Hyrenbach, K., Lynn, R., Mantyla, A., Schwing, F., Sydeman, W., Baumgartner, T., Lavaniegos, B. & Moore, C., 2000. The state of the California Current, 1999-2000: Forward to a new regime? Reports of California Cooperative Oceanic Fisheries Investigations, vol. 41: 26-52. Brodeur, R. D., 1990. A synthesis of the food habits and feeding ecology of salmonids in marine waters of the North Pacific. (INPFC Doe.) FRI-UW-9016, FRI. UW, Seattle. 38 pp. Brodeur, R. D. &  Ware, D., 1992. Long-term variability in zooplankton biomass in the subarctic Pacific Ocean. Fish. Oceanogr., 1: 32–38. Brodeur, R. D. & Ware, D. M.,1995. Interdecadal variability in distribution and catch rates of epipelagic nekton in the Northeast Pacific ocean. In: Beamish, R. J. (ed.) Climate and Northern Fish Populations. Can. Spec. Publ. Fish. Aquat. Sci. 121: 329-356. 98  Burgner, R.L., 1991. Life History of Sockeye Salmon. In: C. Groot and L. Margolis eds. Pacific Salmon Life Histories. University of British Columbia Press, Vancouver, BC. 3-117. Burkhardt, S., Riebesell, U. & Zondervan, I., 1999. Effects of growth rate, CO2 concentration, and cell size on the stable carbon isotope fractionation in marine phytoplankton, Geochimica et Cosmochimica Acta, vol. 63, no. 22: 3729-3741. Cabana, G., Rasmussen, J.B., 1996. Comparison of aquatic food chains using nitrogen isotopes. Proc. Natl. Acad. Sci. 93: 10844–10847. Chhak, K. & Di Lorenzo, E., 2007. Decadal variations in the California Current upwelling cells. Geophysical Research Letters, vol. 34, no. 14. Cox, S. P., Hinch, S. G., 1997. Changes in size at maturity of Fraser River sockeye salmon (Oncorhynchus nerka) (1952-1993) and associations with temperature. Canadian Journal of Fisheries and Aquatic Sciences, vol. 54, no. 5: 1159-1165. Crossin, G., Hinch, S., Cooke, S., Cooperman, M., Patterson, D., Welch, D., Hanson, K., Olsson, I., English, K. & Farrell, A. 2009. Mechanisms Influencing the Timing and Success of Reproductive Migration in a Capital Breeding Semelparous Fish Species, the Sockeye Salmon. Physiological and Biochemical Zoology, vol. 82, no. 6: 635-652. Crossin G. T., Hinch, S.G., Farrell, A. P., Higgs, D. A., Healey,  M. C., 2004. Somatic energy of sockeye salmon at the onset of upriver migration: a comparison among ocean climate regimes. Fisheries Oceanography 29: 22-33 Doucett, R.R., Booth, R.K., Power, G., McKinley, R.S., 1999. Effects of the spawning migration on the nutritional status of anadromous Atlantic salmon (Salmo salar): insights from stable-isotope analysis. Can. J. Fish. Aquat. Sci. 56: 2172–2180.  Eliason, E., Clark, T., Hague, M., Hanson, L., Gallagher, Z., Jeffries, K., Gale, M., Patterson, D., Hinch, S. & Farrell, A. 2011. Differences in Thermal Tolerance Among Sockeye Salmon Populations. Science, vol. 332, no. 6025: 109-112. Foskett, D.R., 1958. The Rivers Inlet Sockeye Salmon.  Journal of the Fisheries Research Board of Canada, vol. 15, no. 5: 867-889. 99  Francis, R. & Hare, S., 1994, Decadal-scale regime shifts in the large marine ecosystems of the North-east Pacific: A case for historical science, Fisheries Oceanography, vol. 3, no. 4: 279-291. Frank, K.T., Petrie, B., Fisher, J.A.D., Leggett, W.C., 2011. Transient dynamics of an altered large marine ecosystem. Nature 477: 86–89.  Freeman, K.H., Hayes, J.M., 1992. Fractionation of carbon isotopes by phytoplankton and estimates of ancient CO2 levels. Global Biogeochem. Cycles 6: 185–198.  French, R., Bilton, H., Osako, M., and Hartt, A.,1976. Distribution and origin of sockeye salmon (Oncorhynchus nerka) in offshore waters of the North Pacific Ocean. Int. N. Pac. Fish. Comm. Bull. 34. Groot, C., Margolis, L., 1991. Pacific salmon life histories. UBC Press, Vancouver. Gruber, N., Keeling, C., Bacastow, R., Guenther, P., Lueker, T., Wahlen, M., Meijer, H., Mook, W. & Stocker, T., 1999. Spatiotemporal patterns of carbon-13 in the global surface oceans and the oceanic Suess effect. Global Biogeochemical Cycles, vol. 13, no. 2: 307-335. Hare, S.R., Francis, R.C., 1995. Climate change and salmon production in the Northeast Pacific Ocean. Can. Spec. Publ. Fish. Aquat. Sci. 357–372. Hare, S.R., N. J. Mantua and R.C. Francis. 1999. Inverse production regimes: Alaskan and West Coast Salmon. Fisheries, 24: 6–14. Hare, S.R., Mantua, N.J., 2000. Empirical evidence for North Pacific regime shifts in 1977 and 1989. Prog. Oceanogr. 47: 103–145. Harrison P.J, Whitney F.A., Tsuda A, Saito H., Tadokoro K., 2004. Nutrient and Plankton Dynamics in the NE and NW Gyres of the Subarctic Pacific Ocean. J Oceanogr.; 60(1): 93-117. Hasler, C., Donaldson, M., Sunder, R., Guimond, E., Patterson, D., Mossop, B., Hinch, S. & Cooke, S. 2011. Osmoregulatory, metabolic, and nutritional condition of summer-run male Chinook salmon in relation to their fate and migratory behavior in a regulated river. Endangered Species Research, vol. 14, no. 1: 79-89. Healey, M.C., 1987. The adaptive significance of age and size at maturity in female sockeye salmon (Oncorhynchus nerka). In Sockeye salmon (Oncorhynchus nerka) population biology 100  and future management. Edited by H.D, Smith, L. Margolis, and C.C. Wood. Can. Spec. Publ. Fish. Aquat. Sci. No. 96. pp. 110-117. Higgs, D. A., J. S. MacDonald, C. D. Levings, and B. S. Dosanjh., 1995. Nutrition and feeding habits in relation to life history stage. 159–316 in C. Groot, L. Margolis, and W. C. Clarke, eds. Physiological ecology of Pacific salmon. Univ. of British Columbia Press, Vancouver, BC. Hinch, S. G., Healey, M. C., Diewert, R. E., and Henderson, M. A. 1995a. Climate change and ocean energetics of Fraser River sockeye (Oncorhynchus nerka). In Climate change and northern fish populations. Edited by R.J. Beamish.Can. Spec. Publ. Fish. Aquat. Sci. No. 121: 439–445 Hinch, S. G., Healey, M. C., Diewert, R. E., Henderson, M. A., Thomson, K. A., Hourston, R., Juanes, F., 1995b. Potential effects of climate change on marine growth and survival of Fraser River sockeye salmon. Canadian Journal of Fisheries and Aquatic Sciences, vol. 52, no. 12: 2651-2659. Hobson, K., Sease, J., Merrick, R. & Piatt, J., 1997. Investigating trophic relationships of pinnipeds in Alaska and Washington using stable isotope ratios of nitrogen and carbon, Marine Mammal Science, vol. 13, no. 1: 114-132. Hodgson, S., Quinn, T. P., Hilborn, R., Francis, R. C. & Rogers, D. E., 2006. Marine and freshwater climatic factors affecting interannual variation in the timing of return migration to fresh water of sockeye salmon (Oncorhynchus nerka). Fisheries Oceanography, 15: 1-24. Holt, C.A., Peterman, R.M., 2004. Long-term trends in age-specific recruitment of sockeye salmon (Oncorhynchus nerka) in a changing environment. Can. J. Fish. Aquat. Sci. 61: 2455–2470.  Ingraham, W. J. Jr., Ebbesmeyer, C. C., & Hinrichsen, R. A., 1998. Imminent climate and circulation shift in Northeast Pacific Ocean could have major impact on marine resources. EOS, 79: 197–201. Ishida, Y., S.-o. Ito, Y. Ueno, and J. Sakai., 1998. Seasonal growth patterns of Pacific salmon (Oncorhynchus spp.) in offshore waters of the North Pacific Ocean. N. Pac. Anadr. Fish. Comm. Bull. No.1: 66-80. 101  Johnson, S.P, and D.E. Schindler. 2008. Trophic ecology of Pacific salmon (Oncorhynchus spp.) in the ocean: a synthesis of stable isotope research. Ecological Research 24: 855-863. Johnson, S.P, and Schindler, D.E., 2012. Four decades of foraging history: stock-specific variation in the carbon and nitrogen stable isotope signatures of Alaskan sockeye salmon. Marine Ecology Progress Series 460: 155–167.  Jonsson, N., Jonsson, B., Hansen, L.P., 1991. Energetic cost of spawning in male and female Atlantic salmon (Salmo salar L.). Journal of Fish Biology 39: 739–744.  Kacem, A., Meunier, F.J., Baglinière, J.L., 1998. A quantitative study of morphological and histological changes in the skeleton of Salmo salar during its anadromous migration. Journal of Fish Biology 53: 1096–1109.  Kaeriyama, M., M. Nakamura, M. Yamaguchi, H. Ueda, G. Anma, Takagi, K.Y. Aydin, R.V. Walker, and K.W.Myers. 2000. Feeding ecology of sockeye and pink salmon in the Gulf of Alaska. N. Pac. Anadr. Fish Comm.Bull. 2: 55–63. Kaeriyama, M., M. Nakamura, R.R. Edphalina, J.R., Bower, H. Yamaguchi, R.V. Walker, and K.W. Myers. 2004.Change in feeding ecology and trophic dynamics of Pacific salmon (Oncorhynchus spp.) in the central Gulf of Alaska in relation to climate events. Fish. Oceanogr. 13: 197–207. Kobari, T., Ikeda, T., Kanno, Y., Shiga, N., Takagi, S., Azumaya, T., 2003. Interannual variations in abundance and body size in Neocalanus copepods in the central North Pacific. Journal of Plankton research 25: 483–494. Krogius, F. V., 1960. The rate of growth and age groupings of salmon in the sea. Vopr. Ikhtiol. 16:67-88. (Translated by the Canadian Dept. of Fisheries and Oceans, Translation Series No. 413.) Kroopnick, P., 1985. The distribution of 13C of ΣCO2 in the world oceans. Deep Sea Research Part A.Oceanographic Research Papers, vol. 32, no. 1: 57-84. Lara, R.J., Alder, V., Franzosi, C.A. & Kattner, G., 2010. Characteristics of suspended particulate organic matter in the southwestern Atlantic: Influence of temperature, nutrient and phytoplankton features on the stable isotope signature. Journal of Marine Systems, vol. 79, no. 1: 199-209. 102  Laws, E. A., Popp, B. N., Bidigare, R. R., Kennicutt, M. C., & Macko, S. A., 1995. Dependence of phytoplankton carbon isotopic composition on growth rate and [CO2]aq: Theoretical considerations and experimental results. Geochimica et Cosmochimica et Acta, 59: 1131–1138. Laws, E., Bidigare, R. & Popp, B., 1997. Effect of growth rate and CO2 concentration on carbon isotopic fractionation by the marine diatom Phaeodactylum tricornutum. Limnology and Oceanography, vol. 42, no. 7: 1552-1560. Litzow, M. A., 2006. Climate regime shifts and community reorganization in the Gulf of Alaska: how do recent shifts compare with 1976/1977? ICES Journal of Marine Science 63: 1386-1396. Litzow, M.A., Mueter, F.J., 2014. Assessing the ecological importance of climate regime shifts: An approach from the North Pacific Ocean. Progress in Oceanography 120: 110–119.  Logan, J.M., Jardine, T.D., Miller, T.J., Bunn, S.E., Cunjak, R.A., Lutcavage, M.E., 2008. Lipid corrections in carbon and nitrogen stable isotope analyses: comparison of chemical extraction and modelling methods. Journal of Animal Ecology 77: 838–846.  Mackas, D.L, Goldblatt, R., Lewis, A.G., 1998. Interdecadal variation in developmental timing of Neocalanus plumchrus populations at Ocean Station P in the subarctic North Pacific. Can J Fish Aquat Sci 55: 1878-1893. Mackas, D.L., Batten, S., and Trudel, M., 2007. Effects on zooplankton of a warmer ocean: Recent evidence from the Northeast Pacific. Prog. Oceanogr. 75(2): 223–252. Mantua, N.J., Hare, S.R., Zhang, Y., Wallace, J.M., Francis R.C., 1997. A Pacific interdecadal climate oscillation with impacts on salmon production. Bull Am Meteor01 Soc 78: 1069-1079 Mantua, N.J. and S.R. Hare, Y. Zhang, J.M. Wallace, and R.C. Francis, 1997. A Pacific interdecadal climate oscillation with impacts on salmon production. Bulletin of the American Meteorological Society, 78: 1069-1079. Mantua, N.J., Hare, S.R., 2002. The Pacific decadal oscillation. J. Oceanogr. 58, 35–44. McKinnell, S.M., Wood, C.C., Rutherford, D.T., Hyatt, K.D., Welch, D.W., 2001. The Demise of Owikeno Lake Sockeye Salmon. North American Journal of Fisheries Management 21: 774–791.  103  Miller, K., A. Teffer, S. Tucker, S. Li, A. Schulze, M. Trudel, F. Juanes, A.Tabata, K. Kaukinen, N. Ginther, T. Ming, S. Cooke, M. Hipfner, D. Patterson, and S.Hinch. 2014. Infectious disease, shifting climates and opportunistic predators: cumulative factors potentially impacting declining wild salmon populations. Evolutionary Applications 7: 812-855. Minagawa, M., Wada, E., 1984. Stepwise enrichment of 15N along food chains: Further evidence and the relation between δ15N and animal age. Geochim. Cosmochim. Acta 48: 1135–1140. Misarti, N., Finney, B., Maschner, H., and Wooller, M.J., 2009. Changes in northeast Pacific marine ecosystems over the last 4500 years: Evidence from stable isotope analysis of bone collagen from archeological middens. The Holocene 19:1139–1151. Otto Grahl-Nielsen, K.A.G., 2012. Fatty acids in fish scales 157: 1567–1576. Pearcy, W. G., 1997. Salmon production in changing ocean domains. 331–352 in D. J. Stouder, P. A. Bisson and R. J. Naiman, eds. Pacific salmon and their ecosystems: status and future options. Chapman and Hall, New York. Pearcy, W.G., R.D. Brodeur, J.M. Shenker, W.W. Smoker and Y. Endo. 1988. Food habits of Pacific salmon and steelhead trout, midwater trawl catches and oceanographic conditions in the Gulf of Alaska, 1980-1985. Bull. Ocean Res. Inst. 26(II):29-78. Persson, P., Bjornsson, B. & Takagi, Y., 1999. Characterization of morphology and physiological actions of scale osteoclasts in the rainbow trout. Journal of fish biology, vol. 54, no. 3: 669-684. Peterman, R.M., Pyper, B.J., Lapointe, M.F., Adkison, M.D., Walters, C.J., 1998. Patterns of covariation in survival rates of British Columbian and Alaskan sockeye salmon (Oncorhynchus nerka) stocks. Can. J. Fish. Aquat. Sci. 55: 2503–2517.  Peterson, B.J., Fry, B., 1987. Stable Isotopes in Ecosystem Studies. Annu. Rev. Ecol. Syst. 18: 293–320.  Peterson, W.T. & Schwing, F.B., 2003. A new climate regime in northeast pacific ecosystems", Geophysical Research Letters, vol. 30, no. 17: 1896.  Post, D.M., 2002. Using stable isotopes to estimate trophic position: models, methods, and assumptions. Ecology 83: 703–718. 104  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: 179–189.  Pyper, B.J., Peterman, R.M., Lapointe, M.F., Walters, C.J., 1999. Patterns of covariation in length and age at maturity of British Columbia and Alaska sockeye salmon (Oncorhynchus nerka) stocks. Can. J. Fish. Aquat. Sci. 56: 1046–1057.  R Core Team (2014). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.  Rau, G.H., 1994. Variations in Sedimentary Organic δ13C as a Proxy for Past Changes in Ocean and Atmospheric CO2 Concentrations, in: Zahn, R., Pedersen, T.F., Kaminski, M.A., Labeyrie, L. (Eds.), Carbon Cycling in the Glacial Ocean: Constraints on the Ocean’s Role in Global Change, NATO ASI Series. Springer Berlin Heidelberg. 307–321. Rau, G. H., Low, C., Pennington, J. T., Buck, K. R., & Chavez, F. P., 1998. Suspended particulate nitrogen δ15N versus nitrate utilization: observations in Monterey Bay, CA. Deep-Sea Research II, 45: 1603–1616. Rau, G.H., Riebesell, U., and Wolf-Gladrow, D., 1997. CO2aqdependent photosynthetic 13C fractionation in the ocean: a model versus measurements. Global Biogeochem. Cycles, 11: 267–278. Rau G., Takahashi T., and des Marais D. J., 1989. Latitudinal variations in plankton d13C: Implications for CO2 and productivity in past oceans. Nature 341: 516–518. Rodionov, S.N., Bond, N.A. & Overland, J.E., 2007. The Aleutian Low, storm tracks, and winter climate variability in the Bering Sea. Deep-Sea Research Part II, vol. 54, no. 23: 2560-2577 Ruggerone, G.T., Nielsen, J.L., Bumgarner, J., 2007. Linkages between Alaskan sockeye salmon abundance, growth at sea, and climate, 1955–2002. Deep Sea Res. Part II Top. Stud. Oceanogr. 54: 2776–2793.  Ruggerone, G.T., Zimmermann, M., Myers, K.W., Nielsen, J.L., Rogers, D.E., 2003. Competition between Asian pink salmon (Oncorhynchus gorbuscha) and Alaskan sockeye salmon (O. nerka) in the North Pacific Ocean. Fish. Oceanogr. 12: 209–219. 105  Rutherford, D. T., and C.C. Wood. 1995. Stock status and 1996 forecast of Smith Inlet sockeye salmon. PSARC Working Paper S95-8. Rutherford, D.T., S. McKinell, C.C. Wood, K.D. Hyatt, and R.D. Goruk. 1995. Assessment of the Status of Rivers Inlet Sockeye Salmon. PSARC Working Paper S95-5. Satterfield, F.R., Finney, B.P., 2002. Stable isotope analysis of Pacific salmon: insight into trophic status and oceanographic conditions over the last 30 years. Prog. Oceanogr. 53: 231–246. Schell, D.M., 2001. Carbon isotope ratio variations in Bering Sea biota: the role of anthropogenic carbon dioxide, Limnol. Oceanogr., 46: 999–1000. Schell, D.M., Barnett, B.A., Vinette, K.A., 1998. Carbon and nitrogen isotope ratios in zooplankton of the Bering, Chukchi and Beaufort seas. Mar. Ecol. Prog. Ser. 162: 11–23 Schmidt, K., Atkinson, A., Stübing, D., McClelland, J.W., Montoya, J.P., Voss, M., 2003. Trophic relationships among Southern Ocean copepods and krill: some uses and limitations of a stable isotope approach. Limnology and Oceanography 48: 277–289. Sinnatamby, R.N., Dempson, J.B., Power, M., 2008. A comparison of muscle- and scale-derived d13C and d15N across three lifehistory stages of Atlantic salmon, Salmo salar. Rapid Commun Mass Spectrom 22: 2773–2778. Sugimoto, T., Tadokoro, K., 1997. Interannual – interdecadal variations in zooplankton biomass, chlorophyll concentration and physical environment in the subarctic Pacific and Bering Sea. Fisheries Oceanography 6: 74-93. Sugimoto, T., Tadokoro, K., 1998. Interdecadal variations of zooplankton biomass and physical environment in the North Pacific. Fish. Oceanogr. 7: 289-299. Tadokoro, K., Ishida, Y., Davis, N.D., Ueyanagi, S., Sugimoto, T., 1996. Change in chum salmon (Oncorhynchus keta) stomach contents associated with fluctuation of pink salmon (O. gorbuscha) abundance in the central subarctic Pacific and Bering Sea. Fish. Oceanogr. 5: 89–99.  Tanaka, T., 2003. Oceanic Suess effect of δ 13 C in subpolar region: The North Pacific. Geophysical Research Letters. VOL. 30, NO. 22: 2159 pp. 106  Trenberth, K. E., and J. W. Hurrell, 1995. Natural Climate Variability on Decade-to-Century Time Scales. D.G. Martinson, K. Bryan, M. Ghil, M.M. Hall, T.R. Karl, E.S. Sarachik, S. Sorooshian, and L.D. Talley, eds. National Academy Press, Washington, D.C., 472-482. Trueman, C.N., Moore, A., 2007. Use of the stable isotope composition of fish scales for monitoring aquatic ecosystems. Stable Isot. Indic. Ecol. Change. 145–161. Vander Zanden, M.J., Rasmussen, J.B., 2001. Variation in δ15N and δ13C trophic fractionation: Implications for aquatic food web studies. Limnol. Oceanogr. 46: 2061–2066. Veefkind, R. J., 2003. Carbon isotope ratios and composition of fatty acids: Tags and trophic markers in pelagic organisms. PhD. University of Victoria. 272 pp. Wada, E., M. Terazaki, Y. Kabaya and T. Nemoto., 1997. 15N and 13C abundances in the Antarctic Ocean with emphasis on the biogeochemical structure of the food web. Deep Sea Research, 34: 829-841 Ware, D. M., and G. A. McFarlane., 1989. Fisheries production domains in the Northeast Pacific Ocean. In: R. J. Beamish and G. A. McFarlane [eds.]. Effects of ocean variability on recruitment and an evaluation of parameters used in stock assessment models, Rept. 108: 359-379 Waser, N. A., Harrison, W. G., Head, E. J. H., Nielson, B., Lutz, V. A., and Calvert, S. E., 2000. Geographic variations in the nitrogen isotope composition of surface particulate nitrogen and new production across the North Atlantic Ocean. Deep Sea Research I: Oceanographic Research Papers, 47: 1207–1226. Welch, D.W., Parsons, T.R., 1993. δ13C-δ15N values as indicators of trophic position and competitive overlap for Pacific salmon (Oncorhynchus spp.). Fisheries Oceanography, vol. 2, no. 1: 11-23. Wendler, G., 2012. The First Decade of the New Century: A Cooling Trend for Most of Alaska. The Open Atmospheric Science Journal, vol. 6, no. 1: 111-116. Whitney, F. A., Welch, D. W., 2002. Impact of the 1997-8 El Niño and 1999 La Niña on nutrient supply in the Gulf of Alaska. Prog. Oceanogr. 54: 405-421 Whitney, F.A., Freeland, H.J., 1999. Variability in upper-ocean water properties in the NE Pacific Ocean. Deep Sea Res. Part II Top. Stud. Oceanogr. 46: 2351–2370.  107  Wu, J., Calvert, S.E. & Wong, C.S., 1997. Nitrogen isotope variations in the subarctic northeast Pacific: relationships to nitrate utilization and trophic structure. Deep-Sea Research Part I, vol. 44, no. 2: 287-314 Wu, J., Calvert, S.E., Wong, C.S. & Whitney, F.A., 1999. Carbon and nitrogen isotopic composition of sedimenting particulate material at Station Papa in the subarctic northeast Pacific. Deep-Sea Research Part II, vol. 46, no. 11:2793-2832.              108  Appendix: The 33 physical and biological time-series used in the Principal Component Analysis (PCA). A.1 Physical time-series A.1.1 Pacific North American Index The Pacific-North American Index (PNA) is one of the most prominent modes of low-frequency variability in the Northern Hemisphere extratropics, appearing in all months except June and July. The PNA index is obtained by projecting the PNA loading pattern to the daily anomaly 500 millibar height field over 0-90°N.  Our PNA time-series consisted of a yearly winter average (Dec-Feb) for the period 1950-2013.  Data source: ftp://ftp.cpc.ncep.noaa.gov/wd52dg/data/indices/pna_index.tim (Accessed April 10, 2014) A.1.2 East Pacific – North Pacific Index The East Pacific Index (EP) is derived from a rotated principal component analysis (RPCA) of normalized 500-hPa height anomalies from the period 1950-2000. The resulting time series is then re-normalized to coincide with the 1981-2010 base period monthly means. The index is updated monthly. Our EPI time-series consisted of a yearly winter average (Jan-Feb) for the period 1950-2013. Data source: ftp://ftp.cpc.ncep.noaa.gov/wd52dg/data/indices/epnp_index.tim (Accessed April 10, 2014)  109  A.1.3 Southern Oscillation Index The Southern Oscillation Index (SOI) is a standardized index based on the observed sea level pressure differences between Tahiti and Darwin, Australia. The SOI is one measure of the large-scale fluctuations in air pressure occurring between the western and eastern tropical Pacific (i.e., the state of the Southern Oscillation) during El Niño and La Niña episodes. Our time-series consisted in a yearly winter average (Dec-Feb) for the period 1951-2013. Data source: http://www.cpc.ncep.noaa.gov/data/indices/soi (Accessed April 10, 2014) A.1.4 North Pacific Index The North Pacific Index (NP) is the area-weighted sea level pressure over the region 30°N 65°N, 160°E-140°W. It is approximately a mirror image of the NPA. Our time-series consisted of a yearly winter average (Dec-Feb) for the period 1948-2011. Data source: http://www.esrl.noaa.gov/psd/data/correlation/np.data (Accessed April 10, 2014) A.1.5 Pine Island Sea Surface Temperature – winter average Daily sea surface temperature (SST) observations (C) are recorded at several locations on the coast of British Columbia by the British Columbia Shore Station Oceanographic Program (BCSOP). There are 13 participating stations. Most of the stations are at lighthouses staffed by Fisheries and Oceans Canada (DFO). Pine Island is located in the middle of the entrance to 110  Queen Charlotte Strait from Queen Charlotte Sound and it was chosen over other stations in the vicinity because it is a site of deep tidal mixing where SSTs are more likely to be reflective of ocean temperatures. Our time-series consisted in a yearly winter average (Dec-Feb) for the period 1937-2013. Data source: http://www.pac.dfo-mpo.gc.ca/science/oceans/data-donnees/lighthouses-phares/data/pinet.txt (Accessed April 10, 2014) A.1.6 Pine Island Sea Surface Temperature – summer average See description above. SST summer average (Jun-Aug) for the period 1937-2013. Data source: http://www.pac.dfo-mpo.gc.ca/science/oceans/data-donnees/lighthouses-phares/data/pinet.txt (Accessed April 10, 2014) A.1.7 Pacific Decadal Oscillation – winter average The Pacific (inter-) decadal oscillation (PDO) Index consists of updated standardized values derived as the leading principal component (PC) of monthly SST anomalies in the North Pacific Ocean, poleward of 20N. The monthly mean global average SST anomalies are removed to separate this pattern of variability from any "global warming" signal that may be present in the data. Our time series consisted of a yearly winter average (Dec-Feb) for the period 1900-2013. Data source: http://jisao.washington.edu/pdo/PDO.latest (Accessed April 10, 2014) 111  A.1.8 Pacific Decadal Oscillation – summer average See description above. Our PDO-summer average time-series consisted of a yearly summer average (Jun-Aug) for the period 1900-2013. Data source: http://jisao.washington.edu/pdo/PDO.latest (Accessed April 10, 2014) A.1.9 Pine island Sea Surface Salinity – winter average Pine Island daily sea surface salinity (SSS) measured in units of PSU (Practical Salinity Unit). Our time series consisted of a yearly winter average (Dec-Jan) for the period 1937-2013. Data source: http://www.pac.dfo-mpo.gc.ca/science/oceans/data-donnees/lighthouses-phares/index-eng.html (Accessed April 10, 2014) A.1.10 Pine island Sea Surface Salinity – summer average Pine Island SSS yearly summer average (Jun-Aug) for the period 1937-2013. Data source: http://www.pac.dfo-mpo.gc.ca/science/oceans/data-donnees/lighthouses-phares/index-eng.html (Accessed April 10, 2014) A.1.11 El Niño Southern Oscillation 3.4 (NINO3.4) – winter average The NINO3.4 index is one of several El Niño/Southern Oscillation (ENSO) indicators based on sea surface temperatures. NINO3.4 is the average sea surface temperature anomaly in the region bounded by 5°N to 5°S, from 170°W to 120°W. This region has large variability on El Niño 112  time scales, and is close to the region where changes in local sea-surface temperature are important for shifting the large region of rainfall typically located in the far western Pacific.  An El Niño or La Niña event is identified if the 5-month running-average of the NINO3.4 index exceeds +0.4°C for El Niño or -0.4°C for La Niña for at least 6 consecutive months. Our time series consisted of a yearly winter average (Dec-Feb) for the period 1870-2014. Data source: http://www.esrl.noaa.gov/psd/gcos_wgsp/Timeseries/Data/nino34.long.data (Accessed January 20, 2015) A.1.12 El Nino Southern Oscillation 3.4 (NINO3.4) – summer average See description above. Our ENSO 3.4 – summer average time-series consisted of a yearly summer average (Jun-Aug) for the period 1870-2014.  Data source:  http://www.esrl.noaa.gov/psd/gcos_wgsp/Timeseries/Data/nino34.long.data (Accessed January 20, 2015) A.1.13 North Pacific Gyre Oscillation – winter average The North Pacific Gyre Oscillation (NPGO) is a climate pattern that emerges as the second empirical orthogonal function of sea surface height variability in the Northeast Pacific. The NPGO is significantly correlated with previously unexplained fluctuations of salinity, nutrients and chlorophyll-a measured in long-term observations in the California Current (CalCOFI) and Gulf of Alaska. Fluctuations in the NPGO are driven by regional and basin-scale variations in wind-driven upwelling and horizontal advection, the fundamental processes controlling salinity 113  and nutrient concentrations. Nutrient fluctuations drive concomitant changes in phytoplankton concentrations, and may force similar variability in higher trophic levels. The NPGO thus provides a strong indicator of fluctuations in the mechanisms driving planktonic ecosystem dynamics. Our time series consisted of a yearly winter average (Dec-Feb) for the period 1950-2014. Data source: http://climexp.knmi.nl/data/inpgo.dat (Accessed January 20, 2015) A.1.14 North Pacific Gyre Oscillation – summer average See description above. Our NPGO – summer average time-series consisted of a yearly summer average (Jun-Aug) for the period 1950-2014. Data source: http://climexp.knmi.nl/data/inpgo.dat (Accessed January 20, 2015) A.1.15 Papa Trajectory Index The Papa Trajectory Index (PTI) is based on modeled surface currents for the Gulf of Alaska. OSCURS is an Ocean Surface CURrent Simulator that computes Langrangian drift in the oceanic surface layer from daily atmospheric surface pressures. Using OSCURS, Ingraham et al. (1998) developed the Papa Trajectory Index (PTI) as a measure of the strength of the winter Alaska Gyre circulation. To compute the PTI, a passive drifter is released at Ocean Weather Station Papa (50°N, 145°W) on each December first and its movement in the surface currents simulated 114  for 90 days. The latitude of the drift end point is the PTI for that year. The trajectories fan out northeastwardly toward the North American continent and show a predominately bimodal pattern of separations to the north and south. To reveal decadal fluctuations in the oceanic current structure relative to the long-term mean latitude, the trajectories were smoothed in time with a 5-year running mean boxcar filter. Our time series consisted of a 5-year running mean for the period 1904-2012. Data source: W. James Ingraham (retired, personal communication)   A.2 Biological time-series A.2.1 Southeast Alaska pink salmon catch The Southeast Alaska pink salmon (Oncorhynchus gorbuscha) catch time-series included all commercial catch (number of fish) of pink salmon in the Southeast Alaska region for the period 1960-2014. The Southeast Alaska/Yakutat Region (Region I) consists of Alaska waters between Cape Suckling on the north and Dixon Entrance on the south.  Data source: Division of Commercial Fisheries, Alaska Department of Fish and Game (ADF&G). For more details contact Steve Heinl at: steve.heinl@alaska.gov A.2.2 Southeast Alaska chum salmon catch The Southeast Alaska chum salmon (Oncorhynchus keta) catch time-series included all commercial catch (number of fish) of chum salmon in the Southeast Alaska region for the 115  period 1960-2014. The Southeast Alaska/Yakutat Region (Region I) consists of Alaska waters between Cape Suckling on the north and Dixon Entrance on the south.  Data source: Division of Commercial Fisheries, Alaska Department of Fish and Game (ADF&G). For more details contact Steve Heinl at: steve.heinl@alaska.gov A.2.3 Southeast Alaska sockeye salmon catch The Southeast Alaska sockeye salmon (Oncorhynchus nerka) catch time-series included all commercial catch (number of fish) of sockeye salmon in the Southeast Alaska region for the period 1960-2014. The Southeast Alaska/Yakutat Region (Region I) consists of Alaska waters between Cape Suckling on the north and Dixon Entrance on the south.  Data source: Division of Commercial Fisheries, Alaska Department of Fish and Game (ADF&G). For more details contact Steve Heinl at: steve.heinl@alaska.gov A.2.4 Central Alaska pink salmon catch The Central Alaska pink salmon catch time-series included all commercial catch (number of fish) of pink salmon in the Central Alaska region for the period 1969-2014. Central Region Alaska commercial fisheries are composed of four distinct management areas that include Bristol Bay, Prince William Sound and Copper River, Upper Cook Inlet, and Lower Cook Inlet. Data source: Division of Commercial Fisheries, Alaska Department of Fish and Game (ADF&G). For more details contact Jennifer Shriver at: jennifer.shriver@alaska.gov  116  A.2.5 Central Alaska chum salmon catch The Central Alaska chum salmon catch time-series included all commercial catch (number of fish) of chum salmon in the Central Alaska region for the period 1969-2014. Central Region Alaska commercial fisheries are composed of four distinct management areas that include Bristol Bay, Prince William Sound and Copper River, Upper Cook Inlet, and Lower Cook Inlet. Data source: Division of Commercial Fisheries, Alaska Department of Fish and Game (ADF&G). For more details contact Jennifer Shriver at: jennifer.shriver@alaska.gov A.2.6 Central Alaska sockeye salmon catch The Central Alaska sockeye salmon catch time-series included all commercial catch (number of fish) of sockeye salmon in the Central Alaska region for the period 1969-2014. Central Region Alaska commercial fisheries are composed of four distinct management areas that include Bristol Bay, Prince William Sound and Copper River, Upper Cook Inlet, and Lower Cook Inlet. Data source: Division of Commercial Fisheries, Alaska Department of Fish and Game (ADF&G). For more details contact Jennifer Shriver at: jennifer.shriver@alaska.gov A.2.7 Rivers Inlet sockeye salmon total biomass Rivers Inlet (statistical area 9) sockeye salmon total biomass (commercial catch and escapement) data is collected yearly by the Department of Fisheries and Oceans Canada (DFO). Our time-series consisted of the yearly total Rivers Inlet sockeye biomass (number of fish) for the period 1948-2013. Data for the period 1948-1997 was summarized by D. Rutherford, C. Wood, and S. McKinnell in the “Rivers Inlet sockeye salmon: Status update” 1998 report. Data 117  for the period 1997-2013 was summarized by DFO in the “2012 post season review and 2013 planning framework for salmon of central coast (statistical areas 7-9)” 2013 report. 1948-1997 data source: http://www.dfo-mpo.gc.ca/CSAS/Csas/DocREC/1998/98_091_e.pdf (Accessed February 7, 2014) 1997-2013 data source: http://www.skeenafisheries.ca/wp-content/uploads/2013/04/2012-Salmon-Post-Season-Review-Booklet-Central-Coast-Areas-7.pdf (Accessed February 7, 2014) A.2.8 Fraser River sockeye salmon total biomass Our time series consisted of the yearly total Fraser River sockeye total biomass (commercial catch and escapement) for the period 1970- 2010. Data source: Fraser stock assessment, DFO (Department of Fisheries and Oceans Canada). For more details contact Tracy Cone at: Tracy.cone@dfo-mpo.gc.ca A.2.9 Skeena River Tyee test-fishery sockeye salmon index This program was developed to provide daily estimates of sockeye escapements through the commercial fishery. The data obtained from this operation, combined with estimates of the commercial catch in Area 4, provided a complete picture of the sockeye runs as they develop each year.  The index value is calculated as the daily catch per hour averaged over 1 to 4 sets per day. Daily escapement estimates are calculated for sockeye salmon using a multiplier based on the past 3 118  year's average. Our time-series consisted of a yearly average of the daily Skeena River Tyee test-fishery sockeye salmon index for the period 1956-2012. Data available at: http://www-ops2.pac.dfo-mpo.gc.ca/xnet/content/salmon/testfish/skeenatyee/LargeSockeye.html (Accessed February 7, 2014) A.2.10 British Columbia statistical area 4 pink salmon total biomass Our time series consisted of the yearly total biomass (commercial catch and escapement) of pink salmon in BC statistical area 4 for the period 1954- 2010. Data source: http://skeenasalmonprogram.ca/library/lib_263/ (Accessed January 20, 2015) A.2.11 British Columbia statistical area 4 chum salmon total biomass Our time series consisted of the yearly total biomass (commercial catch and escapement) of chum salmon in BC statistical area 4 for the period 1954- 2010. Data source: http://skeenasalmonprogram.ca/library/lib_263/ (Accessed January 20, 2015)    119  A.2.12 British Columbia statistical area 4 sockeye salmon total biomass Our time series consisted of the yearly total biomass (commercial catch and escapement) of sockeye salmon in BC statistical area 4 for the period 1960- 2008. Data source: http://skeenasalmonprogram.ca/library/lib_263/ (Accessed January 20, 2015) A.2.13 British Columbia statistical area 7 pink salmon total biomass Our time series consisted of the yearly total biomass (commercial catch and escapement) of pink salmon in BC statistical area 7 for the period 1960- 2012. Data source: http://www.skeenafisheries.ca/wp-content/uploads/2013/04/2012-Salmon-Post-Season-Review-Booklet-Central-Coast-Areas-7.pdf (Accessed January 20, 2015) A.2.14 British Columbia statistical area 7 chum salmon total biomass Our time series consisted of the yearly total biomass (commercial catch and escapement) of chum salmon in BC statistical area 7 for the period 1960- 2012. Data source: http://www.skeenafisheries.ca/wp-content/uploads/2013/04/2012-Salmon-Post-Season-Review-Booklet-Central-Coast-Areas-7.pdf (Accessed January 20, 2015)  120  A.2.15 British Columbia statistical area 8 pink salmon total biomass Our time series consisted of the yearly total biomass (commercial catch and escapement) of pink salmon in BC statistical area 8 for the period 1960- 2012. Data source: http://www.skeenafisheries.ca/wp-content/uploads/2013/04/2012-Salmon-Post-Season-Review-Booklet-Central-Coast-Areas-7.pdf (Accessed January 20, 2015) A.2.16 British Columbia statistical area 8 chum salmon total biomass Our time series consisted of the yearly total biomass (commercial catch and escapement) of chum salmon in BC statistical area 8 for the period 1960- 2012. Data source: http://www.skeenafisheries.ca/wp-content/uploads/2013/04/2012-Salmon-Post-Season-Review-Booklet-Central-Coast-Areas-7.pdf (Accessed January 20, 2015) A.2.17 British Columbia statistical area 11 sockeye salmon total biomass Our time series consisted of the yearly total biomass (commercial catch and escapement) of sockeye salmon in BC statistical area 11 for the period 1960- 2012. Data source: http://www.skeenafisheries.ca/wp-content/uploads/2013/04/2012-Salmon-Post-Season-Review-Booklet-Central-Coast-Areas-7.pdf (Accessed January 20, 2015)  121  A.2.18 Columbia River sockeye salmon escapement Our time series consisted of the yearly total escapement (number of fish) of sockeye salmon in the mouth of the Columbia River (Bonneville Dam) for the period 1938-2014. Data source: http://www.cbr.washington.edu/dart/wrapper?type=php&fname=adultannual_1421106204_263.php (Accessed January 20, 2015)  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0167721/manifest

Comment

Related Items