<<

FI:GCP/RLA/140/JPN

TECHNICAL DOCUMENT No. 8

FAO/GOVERNMENT COOPERATIVE PROGRAMME

SCIENTIFIC BASIS FOR ECOSYSTEM-BASED MANAGEMENT IN THE LESSER ANTILLES INCLUDING INTERACTIONS WITH MARINE MAMMALS AND OTHER TOP PREDATORS

THE APPLICATION OF STABLE IN MARINE ECOSYSTEMS

FOOD AND AGRICULTURE ORGANIZATION OF THE UNITED NATIONS Barbados, 2008

FI:GCP/RLA/140/JPN

TECHNICAL DOCUMENT No. 8

FAO/GOVERNMENT COOPERATIVE PROGRAMME

SCIENTIFIC BASIS FOR ECOSYSTEM-BASED MANAGEMENT IN THE LESSER ANTILLES INCLUDING INTERACTIONS WITH MARINE MAMMALS AND OTHER TOP PREDATORS

THE APPLICATION OF STABLE ISOTOPE ANALYSIS IN MARINE ECOSYSTEMS

Report prepared for the

Lesser Antilles Pelagic Ecosystem Project

(GCP/RLA/140/JPN)

by

M. Aaron MacNeil

School of Marine Sciences and Technology, University of Newcastle, Newcastle upon Tyne, NE1 7RU, UK

FOOD AND AGRICULTURE ORGANIZATION OF THE UNITED NATIONS Barbados, 2008 The designations employed and the presentation of material in this information product do not imply the expression of any opinion whatsoever on the part of the Food and Agriculture Organization of the United Nations (FAO) concerning the legal or development status of any country, territory, city or area or of its authorities,or concerning the delimitation of its frontiers or boundaries. The mention of specific companies or products of manufacturers, whether or not these have been patented, does not imply that these have been endorsed or recommended by FAO in preference to others of a similar nature that are not mentioned.

The views expressed in this information product are those of the author(s) and do not necessarily reflect the views of FAO.

All rights reserved. Reproduction and dissemination of material in this information product for educational or other non-commercial purposes are authorized without any prior written permission from the copyright holders provided the source is fully acknowledged. Reproduction of material in this information product for resale or other commercial purposes is prohibited without written permission of the copyright holders. Applications for such permission should be addressed to: Chief Electronic Publishing Policy and Support Branch Communication Division FAO Viale delle Terme di Caracalla, 00153 Rome, Italy or by e-mail to: [email protected]

© FAO 2008

ABSTRACT

Scientific Basis for Ecosystem-Based Management in the Lesser Antilles Including Interactions with Marine Mammals and Other Top Predators: The application of stable isotope analysis in marine ecosystems, by M. Aaron MacNeil. FAO, Barbados, 2008. xii + 97 pp. , 10 Tables and 31 Figures. FI:GCP/RLA/140/JPN. Technical Document No. 8 Stable isotope analysis has become common in , particularly in the aquatic sciences, to describe energy flow and food-web structure. Biochemical properties of stable affect their distribution in biological materials and, through processes of synthesis and metabolism, record an ’s dietary history in its tissues. and have been the primary isotopes used in studies, recording source and trophic information respectively. But the application of stable isotopes in ecological studies is new, and many of their biochemical properties have yet to be analysed. Problems such as missed sources, uncertainties in fractionation, and differences among and tissues must be recognized. This paper reviews the basics of stable isotope chemistry and the roles of nitrogen and carbon isotopes in ecosystem studies. It also provides examples from a decade of stable isotope research while presenting the most current stable isotope models, and provides a bibliography of stable isotope literature. Sampling protocols for marine systems are included in an appendix. While stable isotopes can be powerful tracers in aquatic systems, substantial care is needed in their application and interpretation. Keywords: Ecosystem Approach to Fisheries; Diet Analysis; Stable Isotope; Lesser Antilles; Pelagic Ecosystem

iii

iv TABLE OF CONTENTS

LIST OF FIGURES...... VII LIST OF TABLES...... XI LIST OF TABLES...... XI GLOSSARY ...... XII GLOSSARY ...... XII 1 BACKGROUND ...... 1 PART 1 – REVIEW OF STABLE ISOTOPES IN MARINE ECOLOGY ...... 3 1 STABLE ISOTOPE ANALYSIS IN ECOLOGY ...... 3 2 STABLE ISOTOPE CHEMISTRY ...... 5 2.1 Nitrogen ...... 7 2.1.1 Nitrogen as a Biomarker ...... 7 2.1.2 Physiological Properties...... 8 2.1.3 Spatial Considerations...... 10 2.2 Carbon ...... 11 2.2.1 Carbon as a Biomarker ...... 11 2.2.2 Physiological Properties...... 12 2.2.3 Spatial Considerations...... 15 3 ECOLOGICAL APPLICATIONS ...... 17 3.1 Tracing Dietary Sources...... 17 3.1.1 Salmon Nutrients in Alaskan Streams ...... 17 3.1.2 Tracer Addition to Detect a Benthic-Pelagic Couple ...... 20 3.1.3 Feeding Habitats for Juvenile Reef Fish in Spanish Water Bay..... 21 3.2 Elucidating Food-Web Structure...... 24 3.2.1 Food Webs of the Rocky Mediterranean...... 24 3.2.2 Trophic Relationships in the Arctic ...... 26 3.2.3 Tracking Long-Term Changes in ...... 29 3.3 Applied Ecological Questions...... 31 3.3.1 Size Based Trophic Structure...... 31 3.3.2 Tracking Invasion Effects on Food Web Structure...... 34 3.3.3 Ecosystem Implications of Increases in Jellyfish ...... 37 3.4 Species-Specific Questions ...... 39 3.4.1 Tuna and Dolphin Associations in the Northeast Atlantic ...... 39 3.5 Source Models ...... 42

v 3.5.1 Two-Source Models...... 42 3.5.2 Multi-Source Models...... 43 3.6 Potential Ecosystem Prediction Models...... 51 4 SAMPLING CONSIDERATIONS FOR STABLE ISOTOPES IN MARINE ECOSYSTEMS ...... 52 4.1 Sample Quality ...... 52 4.1.1 Guidelines...... 54 4.2 Sample Quantity...... 54 4.2.1 Guidelines...... 54 4.3 Laboratory Preparation ...... 54 4.3.1 Guidelines...... 56 PART 2 - ANALYSIS OF STABLE ISOTOPES SAMPLES FROM THE LESSER ANTILLES PELAGIC ECOSYSTEM PROJECT ...... 57 1 OVERVIEW...... 57 2 SPATIAL STRUCTURE...... 57 3 SIZE-BASED TROPHIC STRUCTURE ...... 59 4 COMMUNITY METRICS ...... 61 5 ONTOGENETIC SHIFTS...... 64 6 FUNCTIONAL GROUP RESULTS ...... 64 7 REFERENCES...... 86 APPENDIX 1 FIELD SAMPLING PROTOCOL ...... 94 A 1.2 Sampling ...... 94 A 1.3 Sampling Invertebrates ...... 95 A 1.4 Sampling Fishes...... 96 A 1.5 Sampling Mammals ...... 97

vi LIST OF FIGURES

Figure 1 Ecological publications 1994-2003 that utilized stable isotope data. Publications were compiled from Web of Science (http://isiknowledge.com/; 9 December 2004) using search strings “stable isotope”, “trophic”, and “food web”...... 3 Figure 2 Lipid-corrected ∂13C (o; R2=0.77) and ∂15N (•; R2=0.85) signatures for Daphnia ...... 9 Figure 3 Distribution of 15N in the marine environment. Range shown as thin horizontal line, thick horizontal bar ± 1 SD. Mean, thin vertical line. Adapted from Owens (1987) and references are cited therein...... 10 Figure 4 Mean ∂15N ± SD versus reef site. Reefs with statistically indistinguishable ∂15N as identified by Bonferroni’s pairwise comparison test are grouped by horizontal bars at the top of the graph. Adapted from Heikoop et al. (2000). C.R. is Costa Rica (Caribbean Sea). ... 11 Figure 5. Stable isotope composition of tissues from experimental gerbils versus time. Dots represent the mean ∂13C value of a tissue during a given sample time. F-values given for each tissue are from a test for lack of fit of predicted values to observed values. An F-value followed by ns indicates that the values predicted by the equation do not depart significantly from the observed values. After Tieszen et al. (1983)...... 13 Figure 6 Relationship between organismal and dietary values for (•) and microbes (o). Best fit line (least-squares) for animal data only is Y = 0.94X - 1.68 (n=83). Confidence limits (95 percent) for the slope = ±0.046. Dashed lines are Y = X ± 2SD. Adapted from Fry and Sherr (1989)...... 14 Figure 7 Effect of lipid content on stable isotope ratios of tilapia carcasses. Adapted from Focken and Becker (1998) ...... 15 Figure 8 Observed ∂13C values of POC vs. latitude in the Atlantic Ocean. Adapted from Hofmann et al. (2000)...... 16 Figure 9 Percentage frequency distributions of stable carbon isotope ratios for (A) benthic and planktonic algae in marine environments, and (B) consumers collected from non-, non-estuarine coastal regions around the world. Adapted from France (1995) ...... 16 Figure 10 (A) Mean (±1 SE) foliar ∂15N in riparian Sitka spruce at spawning and reference sites in Snoqualmie River, Washington. (B) Annual basal area growth per unit area of riparian Sitka spruce at the same locations. Adapted from Helfield and Naiman (2001)...... 19 Figure 11 Distribution of tracer 15N in the food web of the oligohaline reach of the Parker River in Massachusetts. The shaded portion of the graph indicates the tracer continuous addition period. data are plotted on all three panels for comparison. (A) The pelagic food web; (B) benthic animals that depend on pelagic diatoms; (C) benthic

vii diatoms which have much lower 15N enrichment than phytoplankton and oligochaetes which accumulate label very slowly. After Peterson (1999) ...... 21 Figure 12 Mean (±SE) ∂15N and ∂13C values of potential fish food collected from (filled circles) and seagrass beds (open circles), and of 23 fish species collected from seagrass beds (filled squares). Adapted from (Nagelkerken and van der Velde, 2004a) ...... 22 Figure 13 Mean (±SE) ∂15N and ∂13C values of the 4 fish species and their most important food items caught during the daytime in mangroves (filled symbols) and seagrass beds (open symbols), Grey symbols indicate the selection of the seagrass fishes of similar size. Adapted from Nagelkerken and van der Velde, (2004b)...... 23 Figure 14 ∂15N and ∂13C for plants (hexagons), benthic invertebrates (rectangles), pelagic invertebrates (diamonds), fishes (circles), and Phalacrocorax aristotelis (star) from the Bay of Calvi, Corsica. Abbreviations are listed in Table 4. After Pinnegar and Polunin, (2000).....26 Figure 15 Relationship of ∂13C and ∂15N values of groups of marine food- web organisms from the Barrow Strait-Lancaster Sound study area. After Hobson and Welch (1992) ...... 28 Figure 16 Dual isotope plot trajectories of ten year running averages of raw haddock isotopic data from George’s Bank. Adapted from Wainright et al. (1993) ...... 30 Figure 17 Scatter plot of muscle ∂13C (lipid-normalized) and ∂15N stable isotopes ratios for five species of adult Pacific salmon collected in 1997. After Satterfield and Finney (2002)...... 31 Figure 18 Relationships between the ∂15N of white muscle tissue (mean ± CL) and maximum weight of (A) Celtic Sea and (B) North Sea fishes. Adapted from Jennings et al. (2001)...... 33 Figure 19 Relationship between the ∂15N of white muscle tissue and size class for the northern North Sea fish community. After Jennings et al. (2001) ...... 33 Figure 20 Stable isotope food web diagrams for Lake Tahoe representing 5 distinct time periods: (A) 1872-1894 — pre-exotic conditions with Lahontan cutthroat trout (LCT) as the native pelagic top predator, (B) 1904-1919 — similar to A, (C) 1927- 1942 — Lahontan cutthroat trout extirpated during this period, (D) 1959-1966 — lake trout are the top predator and are supported by a mix of benthic and pelagic carbon sources, (E) 1998-2000 — Mysis have established in the , and the trophic position of lake trout is suppressed. Cascade Lake is at bottom left. Species codes are in Table 7. Adapted from Vander Zanden et al. (2003) ...... 36 Figure 21 (metric tons) of medusae collected in the total National Marine Fisheries Service (NMFS) sampling area during 1975 and from

viii 1979 on the Eastern Bering Sea shelf. Also shown are the totals for the SE middle shelf and NW middle shelf only. After Brodeur et al. (2002)...... 37 Figure 22 Map of the study area and location of collections in 1996, 1997 and 1999 for stable isotope analysis. After Brodeur et al. (2002)...... 38 Figure 23 Plot of ∂13C versus ∂15N for 1996 NMFS and 1999 ’Oshoro Maru’ summer collections from the SEBS. Points are mean values for each measurement and error bars represent 1 SD of the mean. Adapted from Brodeur et al. (2002) ...... 39 Figure 24 The ∂13C and ∂15N stable isotope values in the muscles of tunas, striped, and common dolphins. After Das et al. (2000) ...... 40 Figure 25 The ∂13C and ∂15N stable isotope values in the livers of tunas, striped, and common dolphins. After Das et al. (2000) ...... 41 Figure 26 Mean ∂15N differences between elasmobranch cartilage and liver tissues relative to muscle tissue. Baseline values (0) set to equal muscle ∂15N means. Error bars representing one standard deviation and 95 percent confidence intervals (•) were calculated using non-parametric bootstrap estimates; (o) are muscle bootstrap 95 percent confidence intervals. Shortfin mako, blue shark, and common thresher tissues were sampled from near Cape Cod, MA, July 2002. After MacNeil et al. (2005b)...... 42 Figure 27 Plot of dual isotopic compositions of food sources A, B, C, and consumer D. A’, B’ and C’ represent the food source isotopic composition after adjustment for trophic fractionation. Plot values from Szepanski et al. (1999). After Ben-David and Schell (2001)...... 44 Figure 28 Scatter plot matrix showing isotopically feasible contributions of seven food sources in the diet of spring coastal mink in SE Alaska. Each panel shows a scatter plot of the feasible contributions of two food sources on a 0-1 scale. After Phillips and Gregg (2003)...... 46 Figure 29 Histograms of the distribution of feasible contributions of the seven autotrophs for (A) Acanthopagrus australis, (B) Silago ciliata, and (C) Silago maculata, after correcting fish values for ∂15N trophic level fractionation. Values in boxes are 1-percentile to 99-percentile ranges for the distributions. Adapted from Melville and Connolly (2003)...... 47 Figure 30 All physically possible combinations of a three-source mixture lie within the boundary of the large triangle, bounded by (1,0,0) (0,1,0) and (0,0,1). The corner point solutions delineate the solution area (shaded triangle) and are used to compute the centre of mass estimate. After Lubetkin and Simenstad (2004) ...... 48 Figure 31 (a) A food web for Sapelo Island based on ∂13C, ∂15N, and ∂34S data. (b) A food web for Sapelo Island based on ∂13C and ∂15N data. Arrows go from the source (prey) to the consumer (predator). For clarity, estimated trophic interactions representing less than 7 percent of a consumer’s diet are not shown. The lightest lines represent interactions accounting for 7 to 15 percent of an organism’s diet. Medium weight

ix lines represent food items contributing between 15 and 25 percent to an organism’s diet. Heavy lines indicate that more than 25 percent of an organism’s diet is estimated to come from a particular source/prey item. BC, from Bighole Creek; LC from Lab Creek. After Lubetkin and Simenstad (2004)...... 51

x LIST OF TABLES

Table 1 Advantages and disadvantages of gut contents and stable isotope analyses for elucidating the trophic relationships of consumers. Modified from Polunin and Pinnegar (2002)...... 4 Table 2 Average terrestrial abundances of the stable isotopes of major elements...... 5 Table 3 Isotope ratios of periphyton, insects, and fishes from Sashin Creek, Alaska. Adapted from Kline et al. (1990)...... 18 Table 4 ∂15N and ∂13C composition of fishes, invertebrates, and plants (mean ±SE) and size-ranges of fishes (total length) in the Bay of Calvi, Corsica. Adapted from Pinnegar and Polunin (2000)...... 24 Table 5 Stable-isotopic compositions of selected components of the Barrow Strait-Lancaster Sound marine food web. Values are given as X ± SD in ‰. Trophic level is based on isotopic model using a ∂15N trophic enrichment values of + 2.4 ‰ for marine birds and +3.8 ‰ for all other organisms. After Hobson and Welch (1992) ...... 26 Table 6 Proportion of body mass sampled by size class. After Jennings et al. (2001)...... 32 Table 7 Native and Introduced Fishes for Lake Tahoe and Cascade Lake. Adapted from Vander Zanden et al. (2003)...... 37 Table 8 Specific guidelines for multi-element mixing models. Adapted from Phillips and Koch (2002) ...... 45 Table 9 Estimated trophic levels for Sapelo Island consumers using SOURCE and STEP with ∂13C, ∂15N, and ∂34S or ∂13C and ∂15N. BC, from Bighole Creek; LC from Lab Creek. After Lubetkin and Simenstad (2004) 50 Table 10 Summary of stable isotope results for functional groups in the LAPE project. TPn is the trophic position indicated by nitrogen isotopes, TPc is the expected trophic position based on size and community regression. ΔTP is the difference between TPn and TPc and ΔTD is change in community trophic diversity due to the group...... 83

xi GLOSSARY

Biomarker - A chemical tracer that records some biological process in tissue. Depleted - A relative term used to identify a sample with a lower ∂-value relative to a sample with a higher ∂-value. Discrimination - A pattern of observed difference in the ratio of heavy to light isotopes at some point during a reaction. Enriched - A relative term used to identify a sample with a higher ∂-value relative to a sample with a lower ∂-value. Equilibrium Fractionation - Fractionation due to differences in bond energies between isotope species. Bonds between heavy isotopes are slightly stronger than bonds between light isotopes. Fractionation - a process that alters the relative proportions of heavy to light isotopes between products and reactants in a reaction. Isotopically heavy - A relative term for comparing two isotope forms of an element that refers to the species with more neutrons. Isotopically light - A relative term for comparing two isotope forms of an element that refers to the species with fewer neutrons. Kinetic Fractionation - Fractionation due to differences in reaction rates between isotope species. Heavy isotopes have slower reaction rates than light isotopes. Trophic Level - An integer value reflecting the height, in number of feeding linkages from basal energy resources, of a given consumer within a food chain. Trophic Position - A continuous and weighted measure reflecting the average number of all feeding linkages from basal energy resources to consumers within a food web.

xii 1 BACKGROUND

The implementation of the Ecosystem Approach to Fisheries (EAF) entails important changes in the way fisheries management is conceived and practiced. The FAO technical guidelines for the ecosystem approach to fisheries (2003) define EAF as follows: “An ecosystem approach to fisheries strives to balance diverse societal objectives, by taking into account the knowledge and uncertainties about biotic, abiotic and human components of ecosystems and their interactions and applying an integrated approach to fisheries within ecologically meaningful boundaries”. Although the main principles that characterize EAF are not new, but already embedded in a number of international agreements and conference documents, there is limited practical experience in implementing them. The project GCP/RLA/140/JPN (Scientific Basis for Ecosystem-based Management in the Lesser Antilles Including Interactions with Marine Mammals and Other Top Predators) addresses one of the challenges related to the implementation of the ecosystem approach to fisheries, i.e. the development of management strategies that take into account biological interactions among species, including cetaceans and other top predators and any species that may be of no direct importance to fisheries but yet, may play an important role in maintaining ecosystem structure and functioning. The medium-term objective of the project is to enable fishery institutions in the region, by 2007, to carry out improved assessments and monitoring of the status of the pelagic resources and fisheries and the ecosystem of which they form a part, for continuous adaptation and improvement of optimum management strategies. Immediate objectives include: 1. obtaining improved estimates of the abundance of key components of the Lesser Antilles pelagic ecosystem, including cetaceans and other top predators; 2. the formulation of a food web model of the ecosystem as a means of investigating ecosystem interactions and impacts; 3. the development of an ecosystem management plan for the pelagic waters of the EEZs of the participating countries, which will include management strategies for key species of fishery interest in the sub- region, as well as for other affected and dependent species, and 4. the development of research and management capacity for ecosystem- based management of their pelagic waters at a national and sub- regional level. Project activities in support of Objective 1 have included cetacean sighting surveys, both regional and offshore as well as national, nearshore surveys. There was a pelagic acoustic/trawling survey to obtain estimates of abundance of

1 forage species and environmental information. Work towards Objective 2 included collection, compilation and analysis of data to estimate model parameters regarding diets, physiology, fisheries and . These were incorporated into a mass-balance model of the pelagic food web using the Ecopath with Ecosim software. To address Objective 3 the LAPE project first completed a series of stakeholder consultations in each of the participating countries to identify fisheries management issues with a particular view towards ecological issues and prioritizing the identified issues. In most countries this process continued by developing performance reports, including specific indicators, for at least one of the high-priority issues. There remains work to be done in each case to complete this process for the pelagic fisheries, and other sectors have not been started. The development of national and sub-regional capacity in this regard (Objective 4) primarily included training sessions associated with specific activities i.e. 'on-the-job' training. There was also training for smaller groups involved in specific tasks e.g. GIS modellers or diet analysts. Amongst the immediate objectives listed above, is the formulation of a food web model of the ecosystem as a means for investigating ecosystem interactions and impacts. The structure of such a model is defined by the trophic, i.e. predator- prey relationships that exist between species or species groups. Thus the trophic structure is defined by the diets of the many species of fish, mammals, birds and invertebrates that make up the pelagic system. The LAPE project is using existing information published in the scientific literature as well as traditional diet studies, (i.e. stomach contents analysis) to accomplish this. However, recent studies of a similar nature have augmented the traditional diet information by the use of Stable Isotope Analysis. These include the UNDP/GEF funded project entitled “Food web study of the western and central Pacific Ocean tuna ecosystem” (see weblink http://www.spc.org.nc/OceanFish/Html/TEB /EcoSystem/foodweb.htm) and a closely related project of the Pelagic Fisheries Research Program of the University of Hawaii “Trophic Structure and Tuna Movements in the Cold Tongue-Warm Pool Pelagic Ecosystem of the Equatorial Pacific” (see weblink http://www.soest.hawaii.edu/PFRP /ocean/allain.html). In addition the GLOBEC project Impacts on Oceanic TOp Predators (CLIOTOP, weblink http://www.pml.ac.uk/globec/structure/regional/cliotop/ cliotop.htm) has incorporated Stable Isotope Analysis in the activities of Working Group 3 (Trophic pathways in open ocean ecosystems). The purpose of the first part of this report is to provide a review of the basics, methods and applications of stable isotope analysis to the understanding of the trophic structure of marine pelagic ecosystems. The second part of this report includes the analysis and results of the stable isotope sampling conducted by the LAPE project in conjunction with the conventional diet studies by the project.

2 PART 1 – REVIEW OF STABLE ISOTOPES IN MARINE ECOLOGY

1 STABLE ISOTOPE ANALYSIS IN ECOLOGY

The use of stable isotopes in ecology increased dramatically during the 1990s (Figure 1) and stable isotopes have become important tools for observing trophic relationships between consumers and their diets. While stable isotope analysis had previously been used primarily in the earth sciences, two factors — their power to discern recorded trophic information from animal tissue and declines in the costs associated with sample analysis — have made them widely applicable for ecologists.

Figure 1 Ecological publications 1994-2003 that utilized stable isotope data. Publications were compiled from Web of Science (http://isiknowledge.com/; 9 December 2004) using search strings “stable isotope”, “trophic”, and “food web”. The use of stable isotope information has increased particularly quickly in the aquatic sciences where isotopes have been effective in four broad categories of ecological questions, namely (i) quantifying sources of energy into food webs, (ii) elucidating the structure of food webs, (iii) investigating the mechanisms operating within food webs, and (iv) tracing migration of animals through the environment. Previously, data to address the first three of these categories was gathered principally from stomach contents analysis. Stomach contents techniques have been extensively developed for fishes and have provided a wealth of ecological information on species interactions in aquatic systems. But stomach contents

3 analysis has been criticized for providing only a relative ‘snapshot’ of diet composition (Pinnegar and Polunin, 2000), for not capturing true interaction dynamics (Deb, 1997), and for neglecting particular dietary items that quickly break down (Polunin et al., 2001). Stable isotopes can address many of these problems (Table 1) and, particularly when used in conjunction with stomach contents data, have proven to be robust across many aquatic environments (Post, 2002).

Table 1 Advantages and disadvantages of gut contents and stable isotope analyses for elucidating the trophic relationships of consumers. Modified from Polunin and Pinnegar (2002) Information Gut Contents Stable Isotopes Resolution of principal trophic Can be good where individual Can be good if pathways well pathways in food webs sources are not identifiable distinguished by ∂13C of the (e.g. indigestible hard parts) basal materials, poor if more than two pathways

Connectance (proportion of Good but only for individual Poor because only broad linkages that are realized) sources that are identifiable categories distinguishable as a rule

Measure of nutritional roles of Poor because diet, not actual Can be good because isotopes different dietary items absorption, quantified are in materials that have been assimilated

Measure of short-term Potentially good because data Poor because tissue turnover differences in diet of large are only short term is slow * predators

Measure of spatial differences Will be good where major Will be good where shifts in diet of large predators items are identifiable occur in items with distinct ∂13C and/or in trophic level

Measure of trophic level Often inaccurate because diet Can be accurate if basal incompletely described materials identified, and change in ∂15N per trophic level validated *This can be mitigated somewhat by sampling high-turnover tissues such as blood or liver

Because they are relatively easy to sample and inexpensive to analyse, carbon (12C and 13C) and nitrogen (14N and 15N) isotopes are most commonly used in food-web studies. In principle, because the ratio of heavy to light isotopes tends to be conserved in carbon and enriched in nitrogen from diets to consumers, these isotope pairs can track basal energy sources and estimate trophic position for individual species in a given ecosystem. In practice there are confounding factors in interpreting carbon and nitrogen isotope results and some published work has failed to account for potential problems. Yet, studies that have carefully sampled stomach contents and stable isotopes from aquatic communities have found broad agreement between these methods (Post, 2002; Vander Zanden et al., 1997). In addition, stable isotopes of other elements — such as oxygen and

4 sulphur — have proven useful in tracing migration and diet sources and have been employed in a variety of aquatic environments. This review covers five general topics: (i) the basics of stable isotope chemistry, (ii) interpretation of stable isotope results, (iii) how stable isotopes have been used previously in food-web ecology, (iv) modelling of food-webs with isotopes and (v) field sampling methods. An emphasis has been made on the marine environment throughout the text, although some important work — particularly regarding validation — has been conducted in freshwater systems.

2 STABLE ISOTOPE CHEMISTRY

Stable isotopes are naturally occurring forms of elements that differ in the number of neutrons present in their nuclei and do not show radioactive decay. For any given element, isotope forms with additional neutrons are referred to as isotopically heavy and forms with fewer neutrons are deemed isotopically light. One form is usually more abundant than the others (Table 2), and it is the ubiquity of a single isotope species that allows the other, rarer forms to be used as biochemical tracers in ecology (Peterson and Fry, 1987).

Table 2 Average terrestrial abundances of the stable isotopes of major elements Element Isotope Abundance (%) Hydrogen 1H 99.985 2H 0.015 Carbon 12C 98.890 13C 1.110 Nitrogen 14N 99.634 15N 0.376 Oxygen 16O 99.759 17O 0.037 18O 0.204 Sulphur 32S 95.000 33S 0.760 34S 4.220 36S 0.014

The relative proportions of heavy to light isotopes present in a sample (as measured by ) are conventionally expressed using ∂X‰ notation (McKinney et al., 1950): ( RR •− 1000) ∂X = sample standard Rstandard where R is the ratio of heavy to light isotope of element X for either the sample or a reference standard containing known proportions of the two isotope forms. This notation is necessary because the actual differences measured by a mass spectrometer are quite small — often to the 5th decimal point — and ∂ values

5 make these fractions simple to tabulate and compare among analyses (Schoeller, 1999). Critical to the use of stable isotopes in ecology is the concept of fractionation — a process that alters the relative abundance of the isotope form between the product and reactant of a reaction. Fractionation occurs because, while isotope forms have similar chemical properties, they are not identical, and the difference in mass between them leads to unique bond energies and reaction kinetics (Peterson and Fry, 1987). Differences in bond energies lead to equilibrium fractionation, where bonds including the heavier isotope form are stronger than bonds without and are therefore statistically less likely to be broken. Kinetic fractionation occurs because the heavier isotope form has slower reaction rates which, depending on the element and reaction process, leads to discrimination between isotope forms during a reaction. An important distinction must be made between discrimination, which refers to any pattern of differences between products and reactants, and fractionation, which is the process causing a change in values between products and reactants. The maximum discrimination between two isotope forms can be expressed as:

D ≈ ∂substrate −∂ product for any given biological reaction (Cerling and Harris, 1999). The degree of fractionation can be expressed as the value of D at some time t; most fractionation values presented in ecology are considered to be steady-state values of D caused solely by fractionation. Reactions that proceed to completion do not show fractionation because all of the atoms present in the product must, by definition, be present in the reactant and the discrimination that occurred during the reaction process has equalized; i.e. the slower species eventually catches up. For fractionation to occur, the products and reactants must be separated at some stage during the reaction process. The degree of fractionation in any chemical reaction is dependent upon conditions such as temperature and (potentially) concentration. The terms depleted and enriched are often used to distinguish between samples containing, respectively, fewer or greater concentrations of the heavier isotope species. There are two primary fractionation steps considered in animals. The first occurs where, as an animal consumes a given food, some proportion of the diet is absorbed into the body (reacted product) and the remaining proportion (unreacted product) remains in the digestive tract in the form of faeces. This absorption step enriches faeces relative to the absorbed diet (Polunin and Pinnegar, 2002). A second important fractionation step occurs during metabolic processes within the body where catabolism, transamination, and deamination can lead to isotopically depleted metabolites such as carbon dioxide, ammonia, and urea, and — as a result — isotopically enriched tissues (Polunin and Pinnegar, 2002). These two processes are most frequently cited as the causes of stable isotope differences between diet and consumer (Polunin and Pinnegar, 2002; Gannes et al., 1998; DeNiro and Epstein, 1981) and, although enrichment

6 due to catabolic pathways is neither necessary nor sufficient to lead to enrichment in tissues, remain viable explanations for metabolic fractionation (Ponsard and Averbuch, 1999). Researchers should be cautious however — much regarding the biochemistry of stable isotopes is not yet known (Lorrain, 2002; Ponsard and Averbuch, 1999, M.A. MacNeil, unpublished). While multiple elements have stable isotope forms, carbon and nitrogen have primarily been used in ecological studies. Sulphur, is generally conserved among trophic steps and has proven useful in tracer studies, particularly in ; oxygen ratios have been shown to record ambient water temperatures in otoliths. The chemical properties of nitrogen and carbon are, however, important background material for those conducting stable isotope analysis.

2.1 NITROGEN

Nitrogen occurs naturally in two stable forms, 14N and 15N, with 14N being by far the more common species (Table 2). The 15N concentration of air is constant at 0.366 ‰ and is therefore used as the internal standard for mass spectrometry analysis of nitrogen (Ehleringer and Rundel, 1989). The range of ∂15N values found in nature is between approximately -50 to 50‰, although most values are between -20 to 20‰ (Owens, 1987).

2.1.1 Nitrogen as a Biomarker

Stable nitrogen isotopes are used as biomarkers in ecosystem ecology where, when they are consistently enriched at each trophic step in a food chain, they become powerful indicators of trophic position. The use of ∂15N as a biomarker stems from the work of DeNiro and Epstein (1981), who found that 15N concentrations of consumers tend to become enriched over the 15N concentrations of their diet. This result was built upon by the seminal work of Minagawa and Wada (1984) who showed an average 3.4 ± 1.1‰ enrichment (fractionation factor) from diet to consumer across four aquatic ecosystems. Minagawa and Wada (1984) provided the general framework for the employment of stable nitrogen isotopes in food-web research, where enrichment in ∂15N within a given system is assumed to occur in each consumer over its diet in a consistent and predictable way. Once the ∂15N value of a known trophic level organism is determined, the trophic position of each member of the ecosystem can be estimated. The trophic position calculation from stable nitrogen isotopes is given as

(∂ 15 N − ∂ 15 N ) TP λ += consumer baseline i Δ n

15 15 15 15 where ∂ Nconsumer is the ∂ N of the species of interest, ∂ Nbaseline is the ∂ N of a known trophic-level species, λ is the trophic level of the baseline species, and Δn is the fractionation at each feeding link in the system. The average 3.4‰ enrichment per trophic level found by Minagawa and Wada (1984) has been used in ecosystem studies from environments as different as marine food-webs in the

7 Canadian Arctic (Hobson and Welch, 1992) to soil food-webs in German forests (Schneider et al., 2004). The immediate use of 3.4‰ in a new system should be viewed as tentative however, as the range of fractionation factors observed by Minagawa and Wada (1984; 1.3 to 5.3‰) suggested considerable differences in nitrogen enrichment factors for particular species. Surprisingly few studies have tested to see if analysed species have actually had 3.4‰ enrichment in ∂15N over their diets and more research is needed in this area — particularly for marine ecosystems. Several studies of well quantified lake systems have demonstrated that a 3 to 3.4‰ enrichment in ∂15N is, in general, widely applicable in freshwater systems (Post, 2002; Vander Zanden and Rasmussen, 2001; Vander Zanden et al., 1997). Post (2002) cautions however, that isotope-derived estimates of trophic position are highly sensitive to the fractionation factor used.

2.1.2 Physiological Properties

A few empirical studies have explored how nitrogen fractionation factors change with species, physiology, and environmental conditions. Debate about causes of ∂15N enrichment have argued primarily about the roles of metabolism and growth in determining ∂15N values. Ponsard and Averbuch (1999) modelled possible routes for isotope turnover and concluded that young animals that eat more than their adult weight as they grow will have similar ∂15N values to adults fed on the same diet. They demonstrated that their model applied to growing rats and mice, concluding that tissue turnover was a significant process leading to ∂15N enrichment. Gaye-Siessegger et al. (2004) compared ∂15N values among carp (Cyprinus carpio) maintained at different feeding rates and found that underfed fish became enriched in 15N without showing appreciable protein loss. They concluded that enrichment was due to protein recycling within muscle tissue. Vander Zanden et al. (2003) found consistent differences in ∂15N fractionation of consumers depending on their form of nitrogen excretion (i.e. urea, uric acid, ammonia, guanine, and amino acids) and cited this as evidence of a metabolic pathway for fractionation. Hesslein et al. (1993) first proposed that, in rapidly growing fish, the isotopic concentration of tissues after a diet switch will be dominated by the addition of new material from the diet, thereby overwhelming metabolic processes. Evidence of growth-mediated ∂15N enrichment has been observed only in the smallest and most rapidly growing fishes. Fish larvae have little tissue to turnover, and it is not surprising that most of the change in their ∂15N values are due to growth. In diet switching experiments on larval red drum (Sciaenops ocellatus), Herzka and Holt (2000) modelled both metabolism and growth processes, demonstrating that growth alone was sufficient to produce ∂15N enrichment. Similarly, Bosley et al. (2002) and MacAvoy et al. (2001) modelled diet switches in larval winter flounder (Pseudopleuronectes americanus) and blue catfish (Ictalurus furcatus) respectively, finding that 15N enrichment was due solely to growth. A single field experiment on juvenile gobies (Rhinogobius spp.) determined from mass-balance modelling that growth could account for more than 80 percent of ∂15N change in muscle after a naturally occurring diet switch (Marcogliese, 2001). It may be that growth

8 processes dominate the effects of metabolic processes in many fishes because they are poikilothermic and have indeterminate growth (Marcogliese, 2001). An important line of evidence has developed on the role of amino acids in determining ∂15N values. Amino acid distributions are tissue specific, and each performs a unique role in physiological processes. Gaebler et al. (1966) found that ∂15N enrichment in rat (Rattus norvegicus) liver was significantly greater in non- essential than essential amino acids, implying that 15N uptake is amino acid specific. Bulk stable nitrogen isotope ratios in from the tropical and subtropical Atlantic Ocean are largely dependent on the proportions of individual amino acids, where some, such as glutamic acid, are enriched between diet and consumer while others, such as phenylalanine, remain constant (McClelland and Montoya, 2002). It has been suggested that amino acids may drive inter-tissue differences in ∂15N values within species. Pinnegar and Polunin (1999) have found ultimate inter-tissue differences in rainbow trout (Oncorhynchus mykiss) fed a consistent diet from birth and they considered amino acid differences among tissues as a possible mechanism. Nitrogen isotope ratios have also been shown to depend upon sampling conditions. Power et al. (2003) found in water fleas (Daphnia magna) and amphipods (Hyalella spp.) consuming identical diets that their body-averaged ∂15N values will decline dramatically at temperatures outside their normal range (Figure 2). The potential causes for the decline included decreased nitrogen excretion with increased growth and reduced nitrogen assimilation efficiencies with increased metabolism. In addition, factors such as diet quality (Adams and Sterner, 2000) and feeding rate (Gaye-Siessegger et al., 2004) can affect observed ∂15N ratios in consumers.

Figure 2 Lipid-corrected ∂13C (o; R2=0.77) and ∂15N (•; R2=0.85) signatures for Daphnia

9 While ultimate inter-tissue variation in ∂15N can be attributed to different tissue amino acid compositions, tissues with high metabolic rates (e.g. liver, blood) more rapidly reflect changes in dietary ∂15N than less metabolically active tissues (e.g. muscle, ) and as a result, will reflect a shorter, more recent period of dietary feeding (Tieszen et al., 1983; MacNeil et al., 2005a). These metabolically- driven differences among tissues have been exploited to provide temporal data about diet and will be explored below.

2.1.3 Spatial Considerations

Because ∂15N values in food webs tend to be amplified from primary producers through to top consumers, spatial variability in primary producer ∂15N has been shown to be an important factor in determining consumer ∂15N. Post (2002) 15 advocated the use of a baseline species of known trophic level (λ and ∂ Nbaseline in Equation (3) above) to standardize local species to the ecosystem in question. The ∂15N of suspended particulate matter has been shown to vary considerably in the marine environment (Figure 3), with the bulk of ∂15N values lying between 0 and 20 ‰ (Owens, 1987).

Figure 3 Distribution of 15N in the marine environment. Range shown as thin horizontal line, thick horizontal bar ± 1 SD. Mean, thin vertical line. Adapted from Owens (1987) and references are cited therein. Schmidt (2003) found that copepod ∂15N values in the Southern Ocean varied by 4 to 6‰ from 49ºS, 20ºE to 60ºS, 10ºE, attributing the differences to changes in the ∂15N values of local particulate organic matter (POM). Similarly, Heikoop et al. (2000) have shown that the ∂15N of coral tissue changes dramatically at sites across the equator (Figure 4). Heikoop et al. (2000) concluded that differences

10 among reef nutrient sources were the most likely causes of the spatial pattern observed. For these reasons it is important to establish a ∂15N baseline for a new target ecosystem and to understand the degree of ∂15N variability within it.

Figure 4 Mean ∂15N ± SD versus reef site. Reefs with statistically indistinguishable ∂15N as identified by Bonferroni’s pairwise comparison test are grouped by horizontal bars at the top of the graph. Adapted from Heikoop et al. (2000). C.R. is Costa Rica (Caribbean Sea).

2.2 CARBON

Stable carbon isotopes exist in two forms, 12C and 13C (Table 2) and have been extensively studied by oceanographers to help understand the role of POM in the ocean (Eadie and Jeffery, 1973). PeeDee Belamite (PDB) — a Cretaceous marine fossil, Belemnitella americana, from the PeeDee fossil formation in South Carolina — is used as an isotopic standard for carbon due to its uniform ratio of 12C to 13C. PDB is the baseline for mass spectrometry analysis and, because it has a higher ∂13C signature than almost all other sources, ∂13C values are normally reported as negative values. The global distribution of carbon isotopes ranges from -90 ‰ for light methanes to 20 ‰ for heavy carbonates. In the ocean, ∂13C values range between -30 to 0 ‰.

2.2.1 Carbon as a Biomarker

In ecosystem studies, carbon isotopes have been used as a source tracer through food webs where ∂13C has been found to be largely conserved (±1‰) from diet to consumer. Once a ∂13C value has been established for the base of a food-web, consumers utilizing the resource can be traced through a given ecosystem. This means that the ∂13C signatures of consumers are ultimately determined by the ∂13C values of primary production in bacteria and plants. Plants have two distinct processes to fix carbon and these differences often lead to distinct ∂13C signatures.

11 In aquatic plants, three factors appear to be most important, (1) the isotopic composition of the dissolved inorganic carbon (DIC) in the water, (2) the isotopic discrimination of the carboxylating enzyme in the plant, and (3) diffusion limitations. The first factor, DIC, is known to vary widely in the ocean, where nearshore, estuarine waters have ∂13C values between -10 to -5 ‰ and open ocean ∂13C tends to be between 0 to 4 ‰ (Hofmann et al., 2000). The second factor, the kind of carboxylating enzyme present, can be of two forms, as rubisco (C3 plants) or as phosphoenolpyruvate (PEP) carboxylase (C4 plants). While rubisco discriminates against 13C by ~ -29 ‰, PEP-carboxylase discriminates against 13C by ~ -6 ‰ (Fry and Sherr, 1989). The actions of these enzymes lead to isotopic differences between C3 and C4 plants. The third factor, diffusion, is the most commonly- cited source of ∂13C variation in the marine environment. In benthic algae, the boundary layer surrounding a plant can be as wide as 1mm, thereby limiting the rate of fixation during and elevating ∂13C values (France, 1995). In turbulent waters however, this boundary layer can be as small as 10µm, thereby increasing the rate of carbon fixation and driving the ∂13C values down to the limits of the carboxylating enzyme present (France, 1995). In most cases these three factors interact to determine the observed differences in marine ∂13C.

2.2.2 Physiological Properties

A few laboratory studies have investigated how ∂13C values change over time in consumer tissues relative to a consistent diet. An early study by Tieszen et al. (1983) showed that metabolically distinct gerbil (Meriones unguienlatus) tissues (liver, fat, muscle, hair, and brain) had different rates of carbon-isotope uptake following experimental switching to a lower ∂13C diet (Figure 5). The study was the first to suggest that consideration of tissue type was important for interpreting ∂13C results and that turnover-times may confound isotope interpretation when the consumer’s diet varies through time. Hesslein et al. (1993) conducted a diet-switch experiment in juvenile broad whitefish (Coregonus nasus) to estimate the rate of ∂13C, ∂15N, and ∂34S turnover in fish muscle and found a 0.0018 d−1 rate of ∂13C turnover corresponding to half-turnover times in muscle of more than one year. These long turnover rates have been frequently cited in field experiments on fishes to suggest that muscle ∂13C values represent a year-long dietary average (e.g. Pinnegar and Polunin, 2000).

12

Figure 5. Stable isotope composition of tissues from experimental gerbils versus time. Dots represent the mean ∂13C value of a tissue during a given sample time. F-values given for each tissue are from a test for lack of fit of predicted values to observed values. An F-value followed by ns indicates that the values predicted by the equation do not depart significantly from the observed values. After Tieszen et al. (1983) The consistency of a ±1 ‰ fractionation factor between consumers and their diets has been observed by many ecosystems. Fry and Sherr (1989) surveyed field and laboratory studies with well-quantified carbon-isotope values to find that consumer ∂13C values were within ±0.7 ‰ of their diets (Figure 6). Vander Zanden and Rasmussen (2001) conducted a similar survey of aquatic experiments and found consumer ∂13C signatures within ±0.8 ‰ of their diets. Rau et al. (1983) observed that ∂13C was linearly correlated with the trophic level of pelagic fauna in the equatorial Pacific and concluded that accurate detection of ∂13C increases could trace consumer trophic position. With the relative consistency among results, most studies consider the enrichment in ∂13C to be negligible relative to differences among ∂15N values, however, investigators must be aware that a ±1 ‰ enrichment may affect their study species at each trophic step.

13

Figure 6 Relationship between organismal and dietary values for animals (•) and microbes (o). Best fit line (least-squares) for animal data only is Y = 0.94X - 1.68 (n=83). Confidence limits (95 percent) for the slope = ±0.046. Dashed lines are Y = X ± 2SD. Adapted from Fry and Sherr (1989). Laboratory analyses of ∂13C variability among species have centred primarily on the role of fats in determining carbon-isotope ratios. Focken and Becker (1998) conducted a feeding experiment on tilapia (Oreochromis niloticus) and carp to show that fish ∂13C was affected by the fat content of its diet (Figure 7); they recommended lipid-extraction on all fish muscle samples to reduce the biasing effects. Pinnegar and Polunin (1999) found similar results in experiments on rainbow trout, where the range of ∂13C values among muscle and liver tissues could be substantially explained by their lipid content. The discrimination in lipid ∂13C towards more negative values is thought to be due to the equilibrium effects of lipid synthesis (see above).

14

Figure 7 Effect of lipid content on stable isotope ratios of tilapia carcasses. Adapted from Focken and Becker (1998)

2.2.3 Spatial Considerations

∂13C of POM varies considerably by latitude in the marine environment (Figure 8) — by 7‰ between the equator and the southern ocean and by 2‰ between the equator and the Arctic ocean (Goericke and Fry, 1994). These variations have been attributed to differences in phytoplankton growth rates driven by CO2 concentrations and, ultimately, temperature (Hofmann et al., 2000; Rounick and Winterbourn, 1986). France (1995) showed that global differences between benthic and planktonic algal ∂13C could be traced into higher-level consumers (Figure 9). It is these benthic-planktonic differences in ∂13C that have provided much of the resolution in marine food webs, where carbon sources can be traced from each environment into the surrounding community. It is therefore essential to evaluate the spatial distribution of primary producer-∂13C in any study ecosystem.

15

Figure 8 Observed ∂13C values of POC vs. latitude in the Atlantic Ocean. Adapted from Hofmann et al. (2000)

Figure 9 Percentage frequency distributions of stable carbon isotope ratios for (A) benthic and planktonic algae in marine environments, and (B) consumers collected from non-seagrass, non- estuarine coastal regions around the world. Adapted from France (1995)

16 3 ECOLOGICAL APPLICATIONS

The frequent use of stable isotopes to study food webs has led to a wide range of ecological applications. Each of these sections below, provide outlines of particular studies that are representative of a given application of stable-isotope analysis. The list is by no means exhaustive — many additional examples can be found relating to similar issues and species. These publications were selected for their novelty of application and frequency of citation in the stable isotope literature.

3.1 TRACING DIETARY SOURCES

Multiple stable isotopes have been used to trace nutrient inputs over a wide range of aquatic systems. The strength of stable isotopes lies in their ability to identify the source of organic matter fuelling food webs at scales up to oceanic basins (Peterson, 1999). Multiple diet sources can be followed through consumers, from one ecosystem to another, and many new dietary pathways have been discovered using stable isotope tracers. The weaknesses of stable- isotope tracers are that, in general, n − 1 tracers are needed to resolve n dietary sources (see modelling section), and that additional spatial or temporal sampling may be required to aid in interpretation of the isotopic distributions (Peterson, 1999).

3.1.1 Salmon Nutrients in Alaskan Streams

Stable isotope tracers can been used to trace energetic pathways between disparate environments. In one of the earliest applications in this area, stable isotopes were used to quantify the contribution of marine-derived nitrogen (MDN) to freshwater systems. The work began with Kline et al. (1990), who wished to estimate how important the MDN transported upstream by spawning pink salmon (Oncorhynchus gorbuscha) was to the food webs of Alaskan streams. The importance of marine-derived nutrients into aquatic systems in Alaska had been recognized in the 1930s, but quantifying a direct link between marine and freshwater nitrogen had not been previously possible (Naiman et al., 2002; Kline et al., 1990). Kline et al. (1990) chose Sashin Creek, Alaska for their field site, where runs of approximately 30 000 pink salmon fill the creek in August and September each year. They established two sampling stations each above and below a 30 m-high waterfall located 1200m upstream from the creek entrance at the Gulf of Alaska. The waterfall effectively blocked pink salmon from the upper portions of the river and created natural control (above waterfall) and treatment (below waterfall) reaches. Kline et al. (1990) opportunistically sampled periphyton, insects, and fishes for ∂15N and ∂13C at the four stations from July 1985 to April 1986 to determine whether MDN could be detected in animals resident within the stream. Muscle tissue samples were taken from the largest species, while the algae and small insects were analysed whole. The results showed a significant enrichment of ∂15N in organisms from the treatment reaches over those sampled

17 from control areas (Table 3). With MDN being the only high-∂15N source in the area, Kline et al. (1990) concluded that migrating pink salmon were primary nutrient sources for food webs in Alaskan salmon streams.

Table 3 Isotope ratios of periphyton, insects, and fishes from Sashin Creek, Alaska. Adapted from Kline et al. (1990). Station ∂15N SD n ∂13C SD n Control Section Periphyton – Nov/85 4 1.5 0.1 2 -34.3 0.8 2 D 0.5 0.1 3 -24.0 0.1 2 3 0.5 - 1 -23.2 0.1 5

Stonefly Nymphs – Apr/86 3 5.1 0.4 4 -30.0 0.1 4

Rainbow Trout – Jul/85 4 7.4 - 7 -33.0 - 7 3 7.5 - 10 -28.1 - 10 3 10.0 - 11 -27.6 - 11 Spawning Section Periphyton – Nov/85 2 1.4 0.2 3 -16.2 0.1 7 1 5.4 - 1 -23.7 0.1 2 C 6.2 0.4 3 -22.6 0.4 7 B 6.2 - 1 - - - Stonefly Nymphs – Apr/86 1 12.8 0.2 2 -25.8 1.8 9

Rainbow Trout – Jul/85 2 11.3 - 9 -23.3 - 9 1 13.4 - 5 -23.3 - 5

Subsequent research found similar results in Kvichak River, Alaska (Kline et al., 1993) and Snoqualmie River, Washington (Bilby et al., 1996). Most recently, ∂15N analysis in Southeast Alaska by Helfield and Naiman (2001) showed in fact, that trees and shrubs adjacent to salmon streams derived 22 to 24 percent of their nitrogen from dead salmon and that trees growing at sites without salmon nitrogen input suffered reduced growth as a result (Figure 10).

18

Figure 10 (A) Mean (±1 SE) foliar ∂15N in riparian Sitka spruce at spawning and reference sites in Snoqualmie River, Washington. (B) Annual basal area growth per unit area of riparian Sitka spruce at the same locations. Adapted from Helfield and Naiman (2001) All of these studies relied on mixing models of nitrogen sources from terrestrial and marine sources. The model used by Helfield and Naiman (2001) took the form SAM − TEM %MDN = •100 MEM − TEM where percent MDN is the percentage of MDN in a given sample, SAM is the observed ∂15N of the sample, TEM is the terrestrial end member (a ∂15N value representing 0 percent MDN), and MEM is the marine end member (a ∂15N value representing 100 percent MDN). By standardizing terrestrial nitrogen values to equal 0, the model considered any enrichment of ∂15N to be from the only remaining source — MDN. The Alaskan salmon example shows how stable isotope methods can directly quantify consumer diet fractions from isotopically distinct resource pools through a given food web.

19 3.1.2 Tracer Addition to Detect a Benthic-Pelagic Couple

Tracing energy sources through food webs by adding highly enriched stable isotopes has been widely successful in freshwater systems (Peterson et al., 2001; Hall and Meyer, 1998), this technique has however, rarely been used in marine systems where isotope tracers would quickly disperse in the large water column. Despite these limitations, estuaries are (largely due to their physical constraints) good potential sites for tracer addition in partially marine conditions. Peterson (1999) added highly ∂15N-enriched inorganic nitrogen to Parker River-Plum Island Sound, Massachusetts, to show how stable isotopes could be used to elucidate the structure of a planktonic-trophic pathway (from phytoplankton to copepods to alewife, Alosa pseudoharengus) known to be present in the estuary. ∂15N-enriched tracers are most commonly in the form of ammonium chloride 15 15 ( NH4C) dripped into target systems; Peterson (1999) added ∂ N continuously for 27d using a peristaltic pump flowing into the oligohaline layer of the estuary. Peterson (1999) sampled phytoplankton, alewife, copepods, amphipods, mud , diatoms, and oligochaetes from around the estuary at 2 to 5d intervals during and after the addition of the tracer (Figure 11). As expected, the ∂15N signal taken up by the phytoplankton was quickly (2 to 7d) detected in copepods and alewife, showing clearly the pathway among those species. Surprisingly however, the ∂15N tracer also revealed a previously unknown trophic connection between the dominant benthic macrofauna in the estuary (amphipods and mud crabs) and the planktonic food web (Figure 11). Peterson (1999) concluded that sinking, pelagic diatoms were a vector for energy from the plankton to the and were serving as a primary food source for the benthic macrofauna. Peterson (1999) also showed that oligochaete worms and the mats of benthic diatoms upon which they fed constituted an entirely separate food web from the plankton-alewife and plankton-mud webs found in the estuary. The 15N tracer addition made these observations simple and straightforward to observe with a minimum of sampling effort. The application of tracer additions to marine systems may be limited however, to constrained areas where tracer dilution is not a significant factor.

20

Figure 11 Distribution of tracer 15N in the food web of the oligohaline reach of the Parker River estuary in Massachusetts. The shaded portion of the graph indicates the tracer continuous addition period. Phytoplankton data are plotted on all three panels for comparison. (A) The pelagic food web; (B) benthic animals that depend on pelagic diatoms; (C) benthic diatoms which have much lower 15N enrichment than phytoplankton and oligochaetes which accumulate label very slowly. After Peterson (1999)

3.1.3 Feeding Habitats for Juvenile Reef Fish in Spanish Water Bay

Stable isotopes have been used to differentiate between competing energy sources in many marine ecosystems. Differences can be detected between benthic and pelagic environments (Lawson and Hobson, 2000) as well as inshore and offshore locations (France, 1995). Seagrass beds and areas have both been shown to be important habitats for juvenile reef fishes (Duffy et al., 2001; Odum and Heald, 1975), but understanding the relative importance of these habitats to specific species in specific areas can help fisheries managers, for instance, in focusing marine protected areas (MPAs) on the most critical sites. In two studies, (Nagelkerken and van der Velde, 2004a,b), stable isotopes were used to show the importance of seagrass and mangrove diet sources in Spanish Water Bay, Curaçao. In the first study, Nagelkerken and van der Velde (2004a) collected 383 fish (23 species) from local seagrass beds and sampled them for both gut contents and stable isotopes. Stable isotope samples were also taken from potential prey species (crabs, echinoderms, amphipods, and tanaidaceans) both in mangroves and on seagrass beds. As usual for such analyses, muscle samples were taken from animals large enough to do so and the smallest species were pooled and analysed whole. Nagelkerken and van der Velde (2004a) then

21 examined ∂15N vs. ∂13C results for both predators and prey and found that the majority of seagrass fishes (18 spp.) relied almost exclusively on seagrass-derived organic matter for their nutrition (Figure 12). Two species, doctorfish (Acanthurus chirurgus) and smallmouth grunt (Haemulon chrysargyreum), fed primarily in the mangrove habitat, and two other species, yellow goatfish (Mulloidichthys martinicus) and yellowtail snapper (Ocyurus chrysurus), derived nutrients from both habitats. Characteristically for such data, the largest predator in the region, barracuda (Sphyraena barracuda), was enriched in ∂15N above the prey fishes and had a ∂13C signature biased toward the majority of the available prey species (Figure 12).

Figure 12 Mean (±SE) ∂15N and ∂13C values of potential fish food collected from mangroves (filled circles) and seagrass beds (open circles), and of 23 fish species collected from seagrass beds (filled squares). Adapted from (Nagelkerken and van der Velde, 2004a) The Nagelkerken and van der Velde (2004a) results demonstrated the importance of the seagrass beds in supplying nutrients to the majority of fishes sampled from above the beds during the daytime. In an MPA context, this would suggest that seagrass beds were more important than mangrove areas for juvenile reef fishes in Curaçao. The links were direct and the trophic pathway from the seagrass beds to the fishes found above them was clear. But further isotope analysis found a more complex pattern for fishes in Spanish Water Bay. Nagelkerken and van der Velde (2004b) sampled four of the most abundant juvenile fish species — blue- striped grunt (H. sciurus), french grunt (H. Flavolineatum), schoolmaster (Lutjanus apodus), and yellowtail snapper — over both seagrass beds and in mangrove

22 areas. Fish prey (decapods and tanaidacids) were also collected from both habitats and compared for ∂15N and ∂13C with the fish results. Once again, muscle samples were taken where possible and the rest of the samples were analysed whole. Surprisingly, the juvenile fishes in Spanish Water Bay had distinct seagrass and mangrove sub-populations (Figure 13), where fishes found in the mangroves did not generally forage on the seagrass beds and fishes on the seagrass beds did not appear to forage in mangrove areas (Nagelkerken and van der Velde, 2004b). The presence of these distinct feeding groups had not been detected from previous diet data in the bay, and the results gave a more accurate picture of juvenile feeding than from the initial, localized study. The stable isotope analyses in Spanish Water Bay were essential for teasing apart the foraging behaviour of fishes in the system. Stomach contents sampling were not location-specific, and with both the fishes and their prey present in both habitats determining nutrient sources for the juvenile fishes would have been difficult. The result that juvenile fishes of the same species were exploiting two distinct trophic pathways suggested that they had strong feeding-site fidelity within the bay. These two studies also demonstrated that stable isotope results presented in spatial isolation can produce an incomplete picture of diet sources and that sampling must have good spatial coverage to be reliable for analysis of any target community.

Figure 13 Mean (±SE) ∂15N and ∂13C values of the 4 fish species and their most important food items caught during the daytime in mangroves (filled symbols) and seagrass beds (open symbols), Grey symbols indicate the selection of the seagrass fishes of similar size. Adapted from Nagelkerken and van der Velde, (2004b)

23 3.2 ELUCIDATING FOOD-WEB STRUCTURE

An important aspect of stable isotopes has been their ability to resolve complex or previously unknown feeding interactions within a given food web. This attribute has been applied extensively in the marine environment and has led to unexpected results in some instances.

3.2.1 Food Webs of the Rocky Mediterranean

Stable isotope and dietary data have been used together to determine food web structure of marine fish communities, and each technique used has revealed features that would not otherwise have been detected from the other method alone (Polunin and Pinnegar, 2002). In the Bay of Calvi, Corsica, Pinnegar and Polunin (2000) conducted a stable-isotope study to compare the quality of trophic information available from stable isotope analysis with published dietary data and to estimate the relative importance of plankton and detrital energy sources for resident fishes. Pinnegar and Polunin (2000) sampled tissues from 125 plants and animals in the bay between March to May and August to September 1997 for stable isotope analysis (Table 4). The Bay of Calvi was chosen as being representative of the Mediterranean rocky-littoral zone, and dietary information was available for local fish species. Muscle samples were taken from each fish and invertebrate, and samples too small to remove a single-tissue sample were pooled and ground whole for analysis.

Table 4 ∂15N and ∂13C composition of fishes, invertebrates, and plants (mean ±SE) and size- ranges of fishes (total length) in the Bay of Calvi, Corsica. Adapted from Pinnegar and Polunin (2000) Species Abbreviation ∂15N ∂13C n Birds Phalacrocorax aristotelis PA 9.18 -18.81 1 Fish Apogon imberbis AI 9.77±0.17 -16.38±0.10 3 Oblada melanura OM 9.26±0.12 -18.68±0.26 3 Chromis chromis (adult) CC 6.90±0.15 -19.13±0.23 3 C. chromis (juvenile) CCj 4.63±0.02 -18.79±0.04 3 Spicara maena SM 7.02±0.32 -19.59±0.19 3 Boops boops BB 6.68±0.36 -18.69±0.09 3 Atherina presbyter AP 7.66±0.12 -16.54±0.18 3 Symphodud ocellatus OS 8.92±0.23 -18.56±0.28 3 S. tinca ST 8.40±0.40 -16.27±0.22 3 Diplodus annularis DA 8.39±0.27 -18.16±0.26 3 Mullus surmuletus MS 7.02±0.17 -17.06±0.24 3 Coris julis CJ 8.60±0.35 -16.27±0.31 3 Dentex dentex (juvenile) DD 8.85 -16.67 1 Muraena helena MH 9.48±0.34 -16.56±0.40 3 Conger conger CO 9.49±0.19 -17.25±0.14 2 Scorpaena porcus SP 7.93±0.44 -16.98±0.29 3 Serranus scriba SE 8.99±0.20 -16.99±0.28 3 Sarpa salpa SS 6.97±0.15 -17.74±0.39 3 Plankton

24 Species Abbreviation ∂15N ∂13C n Zooplankton(>200µm) P1 3.51±0.29 -22.34±0.34 6 Anchylomera blossevilleia P2 4.56±0.05 -22.08±0.10 3 Siriella spp.a P3 4.56±0.05 -22.08±0.10 3 Pelagia noctiluca P4 5.02 -21.13 1 Chelophyes appendiculata P5 3.94±0.12 -21.26±0.05 3 Benthic Invertebrates Paracentrotus lividus B1 1.69±0.12 -18.41±0.83 3 Sphaerechinus granularis B2 2.26±0.13 -20.15±0.25 3 Haliotis lamellosa B3 2.36±0.35 -17.77±0.64 3 Cerithium vulgatum B4 2.95±0.18 -18.45±0.23 3 Bittium reticulatuma B5 3.51±0.18 -17.00±0.24 3 Eunice harassi B6 3.58±0.07 -17.20±0.12 3 Benthic amphipodsa(mixed) B7 3.17±0.03 -20.52±0.04 3 Leptomysis livingula B8 4.94±0.02 -17.77±0.03 3 Pagurus chevreuxiia B9 4.45±0.11 -18.81±0.17 3 Lysmata seticaudata B10 6.54±0.19 -19.74±0.46 3 poressa B11 3.43±0.12 -18.76±0.22 3 Plants Cystosiera balearica A1 2.26±0.10 -20.35±0.18 3 Cladophora prolifera A2 0.95±0.06 -17.71±0.11 3 Halopteris scoparia A3 2.21±0.21 -20.38±0.08 3 Dictyota dichotoma A4 1.92±0.33 -17.97±0.19 3 Padina pavonia A5 2.91±0.19 -8.59±0.18 3 Sphaerecoccus coronopifolius A6 2.19±0.05 -33.74±0.04 3 oceanica A7 2.92±0.04 -14.04±0.23 3 aPooled samples of small animals

From plots of the stable-isotope results (Figure 14), Pinnegar and Polunin (2000) found that the gut and isotope data were each able to resolve trophic roles invisible to the other. For instance, stomach-contents data resolved two benthic- carnivorous fish groups (piscivore/decapod and small invertebrate feeders) that stable isotope results grouped together (Pinnegar and Polunin, 2000). Conversely, stable isotopes distinguished between wrasse (C. julis and S. tinca) and sparid (D. sargus and D. annularis) diets (Figure 14) while stomach contents had them grouped together. The stable-isotope data also suggested that the nocturnal- feeding sand-smelt (A. presbyter), which had previously been assumed to be planktivorous, fed primarily on meroplankton (benthic musids, amphipods, and copepods) known to migrate up from depth each night. Pinnegar and Polunin (2000) concluded that the strengths of stable-isotope and stomach-contents data complemented each other to provide a more accurate trophic structure of Mediterranean rocky-littoral zone fishes. Gut-contents provided the best resolution of diet and the trophic roles of benthic-carnivorous fishes, while stable-isotopes were useful for determining trophic similarities among species and for revealing un-described trophic linkages (Pinnegar and Polunin, 2000). The study’s conclusions emphasized the importance of comparing stomach- content and stable isotope results to achieve a complete dietary picture of the local fish community.

25

Figure 14 ∂15N and ∂13C for plants (hexagons), benthic invertebrates (rectangles), pelagic invertebrates (diamonds), fishes (circles), and Phalacrocorax aristotelis (star) from the Bay of Calvi, Corsica. Abbreviations are listed in Table 4. After Pinnegar and Polunin, (2000)

3.2.2 Trophic Relationships in the Arctic

Stable isotopes have been used to discern feeding relationships in extreme environments where conventional stomach-contents sampling has been logistically difficult. This property was exploited by Hobson and Welch (1992) to study the food-web structure of Lancaster Sound, Nunavut in the Canadian Arctic. Lancaster Sound became ecologically important in the 1990s as global warming models predicted the eventual opening of the adjacent Northwest Passage for commercial shipping and oil development. Previous studies had discerned the importance of Arctic cod (Boreogadus saida) in supporting upper trophic level birds and mammals in the area, but almost nothing was known about how energy production at lower trophic levels fed into the system. To investigate the food-web structure of the (then pristine) sound, Hobson and Welch (1992) took tissue samples from 322 animals (43 spp.) in July-August from 1988 to 1990 (Table 5).

Table 5 Stable-isotopic compositions of selected components of the Barrow Strait-Lancaster Sound marine food web. Values are given as X ± SD in ‰. Trophic level is based on isotopic model using a ∂15N trophic enrichment values of + 2.4 ‰ for marine birds and +3.8 ‰ for all other organisms. After Hobson and Welch (1992) Sample n 13C±SD 15N±SD Trophic level Primary producers Particulate organic matter 5 -21.6±0.3 5.4±0.8 1 Ice algae 2 -20.7±0.9 7.5±0.1 1 Kelp Laminaria solidungula 5 -20.1±0.4 7.1±1.3 1 L. longicruris 5 -20.0±0.6 7.6±0.9 1 Agarum cribosum 5 -20.1±0.3 7.7±0.3 1 Alaria sp. 5 -19.4±0.5 9.4±0.5 1

26 Sample n 13C±SD 15N±SD Trophic level Invertebrates Barnicles Balanus sp. 3 -20.8±0.4 10.4±0.3 2.3 Bivalves Mya truncata 7 -19.0±0.4 9.5±0.7 2.1 Serripes groenlandica 7 -19.0±0.4 9.5±0.7 1.9 Hiatella arctica 10 -18.9±0.8 9.8±0.5 2.2 Macoma calcarea 4 -17.5±0.5 10.8±0.5 2.4 Musculus discors 4 -20.5±0.5 7.9±0.4 1.7 Molluscs Buccinum sp. 5 — 12.6±0.7 2.9 Echinoderms Crossaster papposus 3 — 16.1±1.0 3.8 Leptasterias sp. 3 — 11.5±0.6 2.6 Unident. cucumber 3 -19.7±1.2 9.5±0.5 2.1 Strongylocentrotus sp. 3 -16.5±0.3 9.9±0.3 2.2 Copepods Calanus hyperboreus 6 -20.4±0.4 9.2±0.5 2 Mysids Mysis oculata 2 — 10.3±0.3 2.3 Amphipods Rozinante fragilis 5 -19.8±0.5 11.5±0.3 2.6 Gammarus wilkitzkii 5 -19.1±0.4 11.5±0.3 2.6 Parathemisto libellula 6 -20.3±0.4 11.7±0.7 2.7 Onsismus glacialis 4 -18.2±1.1 11.4±0.5 2.6 Stegocephalus inflatus 5 -15.0±0.7 15.1±0.6 3.6 Anenomes and ctenophores Mertensia ovum 5 -19.8±0.5 13.9±0.5 3.2 Anemone urticina 3 -18.9±0.5 14.0±0.4 3.3 Decapods Lebbeus polaris 5 -16.9±0.3 14.5±0.3 3.4 Fish (Range in fork length (mm) given in brackets) Boreogadus saida (48-247) 26 -18.9±1.0 15.2±0.7 3.6 B. saida (14-16) 1 -19.8 11.1 2.5 Gymnellus virides (101-156) 4 -15.7±0.2 16.2±0.5 3.8 Liparis sp. (46-125) 4 -17.4±0.5 15.0±0.4 2.5 Icelus bicornis (39-81) 4 -17.1±0.4 14.6±1.5 3.4 Myoxocephalus scorpiodes (92-145) 1 -18.1 15.2 3.4 Marine Birds Cepphus Grille 25 -17.3±1.2 15.4±0.7 3.9 Uria Lomvia 25 -18.4±0.6 15.8±0.7 4.1 Fulmarus Glacialis 25 -19.2±0.9 15.3±0.9 3.9 Rissa Tridactyla 25 -18.3±0.7 15.4±0.9 4 Larus Hyperboreus 25 -17.6±0.5 17.0±0.9 4.4 Marine Mammals Odobenus rosmarus 6 -17.8±0.3 12.5±0.6 2.9 Phoca hispida 9 -17.3±0.7 17.3±1.1 4.1 Erignathus barbatus 4 -16.6±0.5 16.8±0.2 4 Monodon monoceros 4 -18.0±0.4 15.8±0.7 3.7 Delphinapterus leucas 6 -18.1±0.5 16.6±0.6 3.9 Ursus maritimus 3 -18.0±0.6 21.1±0.6 5.1

27 Muscle tissue was taken from each animal, and the smallest zooplankton were mixed into groups of ten and analysed whole. Hobson and Welch (1992) plotted ∂13C values for each functional group and showed that little enrichment above an initial upward shift from primary producers to filter feeders had occurred (Figure 15).

Figure 15 Relationship of ∂13C and ∂15N values of groups of marine food-web organisms from the Barrow Strait-Lancaster Sound study area. After Hobson and Welch (1992) This indicated that a single energetic pathway existed in the Lancaster Sound food web. Plots of nitrogen-isotope results found that ∂15N was consistently enriched by 3.8‰ between known diet couples in the system — ringed seal (Phocia hispida) → polar bear (Ursus maritimus) and POM → copepod (Calanus hyperboreus) — and this was assumed to be an integrated average level of enrichment over the entire food web (Hobson and Welch, 1992). From this simplifying assumption, Hobson and Welch (1992) estimated the trophic position of each species (Table 5). The study found general agreement between the isotope results and previous stomach-contents data that had also indicated five trophic levels for the system. The isotope data did however show that local narwhal (Monodon monoceros) and beluga (Delphinapterus leucas) — thought to feed exclusively on Arctic cod — had in fact consumed substantial quantities of lower trophic level prey. The Hobson and Welch (1992) study was among the first to apply stable isotope analysis to a marine food web and to estimate trophic position. As Hobson and Welch emphasized, fractionation estimates used to calculate trophic position need to be validated for each new community. Their study illustrated one form of

28 (pseudo-) validation, where well-documented diets of polar bears and copepods were compared with ∂15N results to lend support to their use of a 3.8‰ fractionation factor. While calculating the trophic positions of marine animals is tenuous using stable isotopes alone, the methods used by Hobson and Welch (1992) can be used to provide substantial support for trophic position estimates in some marine systems.

3.2.3 Tracking Long-Term Changes in Trophic Level

Stable isotopes have considerable potential for monitoring long-term changes in trophic structure of, for instance, commercially important fisheries. Substantial tissue archives of marine species are rare, but a few studies have exploited archived fish scales to reconstruct past trophic information. George’s Bank, off the coast of New England, USA is a highly productive continental-plateau that extends northward from Cape Cod into the Gulf of Maine. The bank has supported extensive demersal fisheries for over 150 years (Overholtz, 2002), in particular flatfish and haddock (Melanogrammus aegelfinus). As part of a long-term ageing programme, the National Oceanic and Atmospheric Administration (NOAA) Fisheries branch has collected fish scales and catch data from commercially important species. Wainright et al. (1993) recognized the potential of the NOAA scale-archive and selected seven demersal species to trace trophic changes on George’s Bank at intervals of between 2 to 58 years. The study focused particular attention on the long and detailed time series for haddock for which there were 45 years of scale data available, from 1929 to 1987. Wainright et al. (1993) identified two potential sources of change in long-term isotope values namely, variation in phytoplankton isotope ratios at the base of the food web and consumer-level changes in diet. They used canonical correlation analysis to estimate the effects of six potential variables (haddock stock size, recruitment age two, mean weight at age two, temperature, the North Atlantic Oscillation (NAO), and the Greenland Regional Pressure Anomaly (GRPA) on the stable-isotope ratios of archived fish scales. Time-series of stable-isotope values and the variables used in the canonical analysis showed considerable short-term variation (< 8 yrs) among longer-term trends; the trend scatter was improved when plotted as a dual-isotope plot of ten- year running averages (Figure 16). Wainright et al. (1993) found that the environmental factors (the NAO and the GRPA in particular) had strong effects on isotope ratios and suggested that changes in phytoplankton values were the vector for the observed trends. They also concluded that real differences had occurred in haddock trophic level over the time series and that the net decline in ∂15N suggested a collapse of the trophic structure to a simpler food web. Subsequent research on other George’s Bank species have supported these conclusions (Fogarty and Murawski, 1998).

29

Figure 16 Dual isotope plot trajectories of ten year running averages of raw haddock isotopic data from George’s Bank. Adapted from Wainright et al. (1993) Similar scale-sampling methods have been used on Pacific salmon in Alaska, where Satterfield and Finney (2002) applied stable isotope analysis to categorize five Onchorhynchus species into functional feeding groups and to investigate long-term trends in sockeye (Oncorhynchus nerka) trophic status. The Alaskan Department of Fish and Game (ADFG) had been archiving sockeye salmon scales from Red Lake, Alaska for 33 years and Satterfield and Finney (2002) used scale- based trends in ∂15N and ∂13C as a basis for trophic inference. In addition to the archived data, the authors sampled 164 fish from lakes in Southern Alaska from 1990 to 1999 and collected muscle and scale samples for isotope analysis. The stable isotope results from Satterfield and Finney (2002) identified three functional feeding groups for Onchorhynchus spp. in Southern Alaska (Figure 17). Chinook (king salmon; O. tshawytcha)— known from diet studies to be the highest trophic-level species — had the most enriched ∂15N and ∂13C values. The unusual enrichment of ∂13C in larger salmon was found to be consistent among all the Onchorhynchus species that the authors attributed to feeding at different locations in the Northern Pacific. Satterfield and Finney (2002) concluded that the delineation of three trophic groups for Onchorhynchus fishes in Alaska was consistent with previous stomach-contents data but that additional isotope data would greatly enhance the robustness of their results. The sockeye time-series from the lakes showed minor (0.5‰) variability in ∂13C and a significant (3‰) decrease in ∂15N values from 1966 to 1999. Satterfield and Finney (2002) suggested this decline could be evidence of nearly a trophic level decline in the average trophic position of sockeye salmon over the study period, but concluded that the change may have been due to climatic forces altering phytoplankton values at the base of the food web.

30

Figure 17 Scatter plot of muscle ∂13C (lipid-normalized) and ∂15N stable isotopes ratios for five species of adult Pacific salmon collected in 1997. After Satterfield and Finney (2002) Both of the Wainright et al. (1993) and Satterfield and Finney (2002) studies demonstrated the potential utility of stable-isotopes to track long-term changes in trophic position over time. Similar data has been used in the Mediterranean to suggest that reported ‘fishing down’ of trophic levels there (Pauly et al., 1998) was, in fact, due to the aggregation of aquaculture and fisheries landings rather than excessive fishing pressure (Pinnegar et al., 2003). However, as both the George’s Bank and Alaskan salmon examples stated, caution should be exercised in studying long-term trends independent of stable isotope values for the base of the target community’s food web.

3.3 APPLIED ECOLOGICAL QUESTIONS

Stable isotope analysis has been used as an empirical tool for ecological questions other than trophic structure and diet sources. Primary theoretical ideas can be explored using stable isotope methods and marine researchers are increasingly applying them to these ends.

3.3.1 Size Based Trophic Structure

Ecologists have long recognized that larger animals consume larger prey (Elton, 1927), but little quantitative evidence has emerged of the intuitively simple relationship between body size and trophic level. Jennings et al. (2001) recognized that, if ∂15N is taken to be a relative index of trophic position, stable isotopes could be used to empirically quantify the relationship between trophic level and body size of fishes in marine communities. Jennings et al. (2001) sampled 48 fish species from 136 research-trawls in the Celtic and North Seas during February to March and August to September 2000. At minimum, three fish of each species

31 were sampled for ~2g of white dorsal muscle. The study standardized the life- history stage of sampled fishes by including only individuals that were between 60 to 80 percent of their maximum-recorded length among the sample set. Under these conditions, Jennings et al. (2001) divided the catch into 15 size classes, the smallest of which required whole-body sampling to obtain tissue of sufficient size for stable-isotope analysis (Table 6). From these data, the study examined both species-based and sized-based relationships with trophic position.

Table 6 Proportion of body mass sampled by size class. After Jennings et al. (2001). % individual body mass Number of fish sampled in Size class (g) sampled Composition of sample size class Whole fish excluding >1-2 20% head; tail and guts 20-25 (all in catch if <20) Whole fish excluding >2-4 20% head; tail and guts 20-25 (all in catch if <20) >4-8 20% Whole muscle and skin 20-25 (all in catch if <20) >8-16 20% Whole muscle and skin 20-25 (all in catch if <20) >16-32 10% Whole muscle and skin 20-25 (all in catch if <20) >32-64 5% Whole muscle 20-25 (all in catch if <20) >64-128 5% Whole muscle 20-25 (all in catch if <20) >128-256 2% Whole muscle 20-25 (all in catch if <20) >256-512 1% Whole muscle 20-25 (all in catch if <20) >512-1024 0.50% Whole muscle All in catch >1024-2048 0.20% Whole muscle All in catch >2048-4096 0.10% Whole muscle All in catch >4096-8192 0.05% Whole muscle All in catch >8192-16384 0.05% Whole muscle All in catch >16384-32768 0.05% Whole muscle All in catch

15 Cross-species plots of ∂ N versus log10(mass body) showed that there was little relationship between trophic position and the sizes of individual species (Figure 18). Considered alone, this result suggested that small species feed across all trophic levels in temperate fish communities and that no relationship existed between trophic position and body size. While this species-specific result seemed inconsistent with established ecological theory (Cohen et al., 1993), a second, species independent plot of the data found that, by size class, the results were dramatically different (Figure 19). The community-based plot of ∂15N versus log10(mass body) showed that, at the community level, there is powerful size-based structuring of marine communities (Jennings et al., 2001). The differences between Figures 18 and 19 demonstrated that small species feeding at a high trophic level were relatively unimportant in the North Sea community and that, in general, body size can be used to infer trophic position for temperate marine fishes.

32

Figure 18 Relationships between the ∂15N of white muscle tissue (mean ± CL) and maximum weight of (A) Celtic Sea and (B) North Sea fishes. Adapted from Jennings et al. (2001)

Figure 19 Relationship between the ∂15N of white muscle tissue and size class for the northern North Sea fish community. After Jennings et al. (2001) Subsequent research built upon the results to address other applied questions. Jennings et al. (2002a) examined stable isotope, size-spectra, and community metrics to investigate long-term trends in North Sea fishes. The study examined stable-isotope data from fishes caught in the North Sea from January to September 2001 and compared the results to North Sea size data from the International Bottom Trawl Survey. The results showed significant declines in the average trophic level of demersal fishes between 1982 and 2000, although the result required careful analysis of the size structure of the North Sea fish community over the study interval (Jennings et al., 2002a). As had been previously demonstrated (Jennings et al., 2001, Figure 18), fish species have size- independent trophic levels, and Jennings et al. (2002a) found that increases in herring biomass during the late 1980’s exerted considerable influence on the average community trophic level.

33 In another study, Jennings et al. (2002b) used size-based production data and stable-isotope analysis to estimate trophic transfer efficiencies between predators and prey in the North Sea food web. Sampling using corers, dredges, and nets, they collected infaunal invertebrates, epifaunal invertebrates, bottom-dwelling fishes, squid, and pelagic fishes. Their results estimated transfer efficiencies of between 3.7 to 12.4 percent, and a mean predator-prey body-mass ratio of 109:1 (Jennings et al., 2002b). These values broadly agreed with previous ecosystem- based modelling and supported the use of stable-isotopes in estimating trophic transfer efficiencies. Additional information on size-based applications of stable isotopes can be found in Jennings and Warr (2003a,b). The example conveys an important general property of stable isotope data: consistent trophic enrichment in ∂15N appears to be a community level, not a species level, phenomenon.

3.3.2 Tracking Invasion Effects on Food Web Structure

There has been great ecological interest in describing the effects of species invasions on native aquatic communities (Ricciardi and MacIsaac, 2000). Stable isotopes have been used to quantify these effects, where archived museum specimens from before and after invasions have recorded much about historical food-web structure. Vander Zanden et al. (2003) applied isotope methods to two lakes from the Truckee River watershed in Nevada to investigate the effects of species introductions. Lake Tahoe is a large (500 km2), deep lake in the Sierra Nevada mountains of California that has been extensively studied due to the historical presence of the (now threatened) Lahontan cutthroat trout (Oncorhynchus clarki henshawi, LCT). LCT was displaced from Lake Tahoe in the 1930’s by an introduction of lake trout (Salvelinus namaycush) in 1888 to increase recreational fishing. The Lake Tahoe food web was affected again in the 1960’s with the introduction of Mysis relicta as a forage base for lake trout and, with considerable interest in re-introducing LCT into the Truckee River drainage, Vander Zanden et al. (2003) attempted to quantify the historical food-web structure to aid in restoration efforts. Tissue samples from fish and invertebrates had been archived from the lake (at intermittent intervals) since 1872. A control lake, free of both lake trout and Mysis was selected from upstream of Lake Tahoe for comparison to tissue samples from museum collections. Cascade Lake is a small (0.86 km2) deep lake that had been protected from lake trout and Mysis introductions by public access restrictions imposed by private landowners. Although small numbers of rainbow trout and brown trout (Salmo trutta) were stocked in Cascade Lake in the early 1900’s, the species composition of the lake was close to that of Lake Tahoe at the start of the 20th century. For stable isotope comparisons, Vander Zanden et al. (2003) sampled fish, zooplankton, and invertebrates from Tahoe and Cascade lakes between 1999 and 2001. One g of dorsal muscle was sampled from each fish for analysis, while zooplankton and benthic invertebrates were separated by species and analysed whole. Historical food web samples (~0.5 g of dorsal muscle) were obtained from three field museum collections (California Academy of Sciences, the Smithsonian Institute, and the University of Michigan); much research has been done on estimating

34 how stable isotope values change after formalin fixation and storage (Edwards et al., 2002; Sarakinos et al., 2002; Arrington and Winemiller, 2002; Kaehler and Pakhomov, 2001) and the analyses have found a predictable rate of change in stable isotope signatures over time. Regression curves have subsequently been developed for calibrating archived tissues. Vander Zanden et al. (2003) chose a different approach however, and chose to standardize each year of archived samples to a relative trophic position (TP) value

(∂15N −∂15N ) TP = consumer baseline + 2 i 3.4

15 15 where 3.4 is the perceived per trophic level enrichment in ∂ N, and ∂ Nbaseline was established from linear regression curves for each lake, which were

∂15N =−0.094 •∂13C + 0.898 for Lake Tahoe, and

∂15N =−0.123•∂13C + 0.860

15 for Cascade Lake. These equations were used to calculate each lake’s ∂ Nbaseline using the baseline consumer’s ∂13C value. To compare food webs among lakes and time-periods, a two end-member mixing model estimated the percentage of benthic-carbon (PB) in the diet of each species. The model took the form

13 13 (∂ C fish −∂ Cpelagic ) PB = 13 13 •100. (∂ Cbenthic −∂ Cpelagic )

13 13 For Lake Tahoe, the ∂ Cbenthic was chosen as the mean ∂ C signature of 13 13 amphipods, crayfish, mayflies, snails, and clams; the ∂ Cpelagic was the mean ∂ C signature of zooplankton and Mysis. For Cascade Lake, snails and amphipods 13 were used for ∂ Cbenthic; zooplankton and some profundal benthic invertebrates 13 were used for ∂ Cpelagic. Plotting TP versus PB for each sampling year and lake provided a clear map of the changes in food-web structure that had occurred in Lake Tahoe during the 20th century (Figure 20). In the initial three time periods, fishes were relying on predominantly (80 percent) benthic production. The LCT results showed that these fishes were relative pelagic-specialists in the lake, and their elevated TP values shows that they were also pisciviorous Vander Zanden et al. (2003). By the fourth time period (1959-1966), the remaining LCT were gone, and the lake was dominated by lake trout at the highest TP and drawing nutrients almost equally from benthic and pelagic sources. The fifth plot (1998-2000) shows that, with the introduction of Mysis, the average TP of lake trout dropped, and their diets shifted to increased reliance on pelagic nutrients. These changes followed the addition of brown trout as the top predator in Lake Tahoe, and a general shift to a Mysis-based food web. The structure of Cascade Lake was shown to be similar to the historic structure of Lake Tahoe (Figure 20), but with the addition of brown trout as a top predator. From these results, Vander Zanden et al. (2003) argued that, with the large recreational fishery in Lake Tahoe and the evidently

35 drastic changes in its food-web structure, restoration efforts for LCT should be directed to Cascade and surrounding lakes rather than in Lake Tahoe. The study used archived tissue samples to illustrate the structural effects of species introductions on a large aquatic environment.

Figure 20 Stable isotope food web diagrams for Lake Tahoe representing 5 distinct time periods: (A) 1872-1894 — pre-exotic conditions with Lahontan cutthroat trout (LCT) as the native pelagic top predator, (B) 1904-1919 — similar to A, (C) 1927- 1942 — Lahontan cutthroat trout extirpated during this period, (D) 1959-1966 — lake trout are the top predator and are supported by a mix of benthic and pelagic carbon sources, (E) 1998-2000 — Mysis have established in the pelagic zone, and the trophic position of lake trout is suppressed. Cascade Lake is at bottom left. Species codes are in Table 7. Adapted from Vander Zanden et al. (2003)

36 Table 7 Native and Introduced Fishes for Lake Tahoe and Cascade Lake. Adapted from Vander Zanden et al. (2003). Lake Cascade Species Latin Name Code Tahoe Lake Native Fishes Tahoe sucker Catostomus tahoensis suc X X Lahontan speckled dace Rhinichthys osculus robustus dac X X Lahontan redside Richardsonius egreginus rec X X Lahontan cutthroat trout Oncorhynchus clarki henshawi lct X X Mountain whitefish Prosopium williamsoni whi X X Tui chub — benthic Gila bicolor obesa tui-b X Tui chub — pelagic Gila bicolor pectinifer tui-p X Paiute sculpin Cottus beldingii scu Established Non-native salmonids Rainbow trout Oncorhynchus mykiss rbt X X Brown trout Salmo trutta bt X X Kokanee salmon Oncorhynchus nerka kok X Lake trout Salvelinus namaycush lt X

3.3.3 Ecosystem Implications of Increases in Jellyfish

Many researchers have been working to assess the ecosystem impacts of drastic changes in marine fish populations, but little attention has been paid to how invertebrate predators affect the surrounding community. Although not commercially important, these animals can account for a considerable proportion of mortality in larval fishes (Arai, 1988). The Bering Sea experienced dramatic increases in walleye pollock (Theragra chalcogramma) in the 1970s and in large gelatinous medusae during the 1980s (Figure 21).

Figure 21 Biomass (metric tons) of medusae collected in the total National Marine Fisheries Service (NMFS) sampling area during 1975 and from 1979 on the Eastern Bering Sea shelf. Also shown are the totals for the SE middle shelf and NW middle shelf only. After Brodeur et al. (2002)

37 While considerable attention was focused on pollock population dynamics, the ecosystem effects of the jellyfish increases remained unexplored (Brodeur et al., 2002). Brodeur et al. (2002) sought to estimate the impact the jellyfish Chrysaora melanaster had on pollock and zooplankton populations in the eastern Bering Sea. While biomass estimates were available for all species from American and Japanese research surveys in the area, linking these estimates together required detailed dietary information from the jellyfish and potential prey. Brodeur et al. (2002) collected larval (age 0) pollock, macrozooplankton, and C. melanaster over 3 seasons from 15 sampling stations along the southeastern Bering Shelf (Figure 22) using a Methot frame-trawl. Dorsal muscle tissue from the larval fish, umbrellar tissue from the jellyfish, and whole macrozooplankton were sampled for stable isotope analysis. Diet data was also taken by haphazardly sampling medusae guts collected in 1997 and 1999. The ∂13C results were broadly consistent among sampling locations and across years, indicating that age-1+ pollock, C. melanaster, age-0 pollock, small hydromedusae, and euphausiids comprised a single trophic pathway in the Eastern Bering Sea (Figure 23). Brodeur et al. (2002) considered this result in conjunction with the moderately elevated (0.02 to 1.4 ‰) ∂15N values in C. melanaster over age-0 pollock to conclude that the jellyfish consumed a considerable proportion of larval pollock in the Bering Sea.

Figure 22 Map of the study area and location of collections in 1996, 1997 and 1999 for stable isotope analysis. After Brodeur et al. (2002)

38

Figure 23 Plot of ∂13C versus ∂15N for 1996 NMFS and 1999 ’Oshoro Maru’ summer collections from the SEBS. Points are mean values for each measurement and error bars represent 1 SD of the mean. Adapted from Brodeur et al. (2002) Combining their diet and biomass results, Brodeur et al. (2002) estimated that C. melanaster could directly consume 2.8 percent of the available age-0 pollock biomass per d and could indirectly affect pollock stocks by consumption of up to 31 percent of total zooplankton biomass annually. Two additional factors, a gross undersampling of medusae by NMFS trawls and the isotopic overlap between C. melanaster and both pollock groups, suggested that the net effect of the recent jellyfish increases on pollock was perhaps even greater than indicated by those estimates. The study showed that stable-isotopes can provide a critical trophic- link between consumers within a given food web. This connection is particularly difficult to achieve when sampling gelatinous species for which little diet data is generally available.

3.4 SPECIES-SPECIFIC QUESTIONS

Stable isotope studies can be tailored to species-specific questions in instances where stomach-contents sampling is logistically difficult. This is particularly the case for large-pelagic fishes and whales which are threatened or endangered. Often these analyses have made use of fishery catches (either accidental or directed), or strandings to acquire samples and have included other biochemical tracers such as metals and fatty-acids.

3.4.1 Tuna and Dolphin Associations in the Northeast Atlantic

Dietary information on dolphins has traditionally been conducted with stomach contents sampling of dead animals or from direct observation of feeding events at sea (Perrin et al., 1973; Au and Pitman, 1986) but it is often difficult to gather sufficient sample sizes from these methods. In the Northeast Atlantic, striped (Stenella coeruleoalba) and common (Delphinus delphis) dolphins have often been

39 caught in association with tunas during purse-seining operations. There has been considerable interest in reducing dolphin bycatch in tuna fisheries since the early 1990s, and US restrictions on non ‘dolphin-safe’ tuna have greatly reduced incidents to below 5,000 per year (Buck, 1997). Despite these efforts, significant levels of dolphin bycatch persist in some areas. To explore the ecological roles of these predators, Das et al. (2000) conducted a stable-isotope study to determine how much overlap was present in dolphin and tuna diets in the Bay of Biscay, Spain. Das et al. (2000) sampled 23 striped and ten common dolphins, and 20 albacore tuna (Thunnus alalunga) from commercial seining operations between May and September 1993. Each animal was sampled for muscle and liver tissue for analysis. The tunas were all juveniles, between four to five years old (75 to 81 cm), and the dolphins ranged between 0.1 to 20 years of age (no lengths provided). All samples were lipid-extracted prior to analysis. The plotted muscle ∂15N and ∂13C values revealed distinct feeding niches between albacore tuna and the dolphins in the Bay of Biscay (Figure 24). While the lack of overlap between the muscle-isotopes appeared to track long-term differences in feeding niche, the liver-isotope plot largely overlapped (Figure 25). Das et al. (2000) concluded that the overlapping liver-isotope results, in conjunction with metal data, indicated “two different [albacore] food habits were co-existing sympatrically and temporally” (pp. 107). However, a simpler and equally plausible explanation not put forward by Das et al. (2000) is that, due to the differences in metabolic turnover between tissues, the liver values represented the most recent period of consumer feeding in the Bay of Biscay while the muscle tissue reflected an average of yearly differences in diet. This conclusion would seem sound given that the dolphins and tuna were caught feeding together during the summer.

Figure 24 The ∂13C and ∂15N stable isotope values in the muscles of tunas, striped, and common dolphins. After Das et al. (2000)

40

Figure 25 The ∂13C and ∂15N stable isotope values in the livers of tunas, striped, and common dolphins. After Das et al. (2000) This kind of seasonality has been observed in the Northwest Atlantic where a known annual diet switch has been observed in shortfin mako (Isurus oxyrinchus) muscle and liver isotope signatures (MacNeil et al., 2005b). Shortfin mako diet has been studied since the early 1980’s by NOAA scientists in Narragansett, RI and much is currently known about their feeding habits. The sharks exist in two distinct on- and off-shelf populations year-round. The inshore population feeds almost exclusively on cephalopods during the winter and switches to an entirely bluefish (Pomatomus saltatrix) diet in the late spring when the prey move northward on their annual migration. MacNeil et al. (2005b) opportunistically sampled liver and muscle from shortfin mako, blue shark (Prionace glauca), and common thresher (Alopias vulpinus) from recreational fishing derbies in 2002. Comparison between shortfin mako liver and muscle ∂15N clearly showed the diet switch between the ∂15N-enriched bluefish and ∂15N-depleted squid (Figure 26). Both the Das et al. (2000) and MacNeil et al. (2005b) studies showed that temporal information can be compared between tissues and that stable-isotope analysis is particularly appropriate for large-pelagic consumers. Both studies suggest that two tissues of different metabolisms should be sampled where possible.

41

Figure 26 Mean ∂15N differences between elasmobranch cartilage and liver tissues relative to muscle tissue. Baseline values (0) set to equal muscle ∂15N means. Error bars representing one standard deviation and 95 percent confidence intervals (•) were calculated using non- parametric bootstrap estimates; (o) are muscle bootstrap 95 percent confidence intervals. Shortfin mako, blue shark, and common thresher tissues were sampled from near Cape Cod, MA, July 2002. After MacNeil et al. (2005b)

3.5 SOURCE MODELS

While stable isotopes are frequently cited for their potential to directly quantify the contribution of potential diet sources, there are surprisingly few models available to accomplish this in practice. Those that do exist can be broadly divided into two-source and multi-source models. As implied by the name, two- source models are applicable to systems where clear differences are present between two well-defined diet sources, as was the case for the salmon derived nitrogen example above. Multi-source mixing models can be used to differentiate among n diet sources provided n−1 stable isotopes are available as tracers. Most of the multi-source mixing models are relatively new and their use has thus far been limited.

3.5.1 Two-Source Models

Two-source mixing models are relatively simple expressions that calculate the percent ∂X from source Si versus source Sj in a given organism. The calculation requires only that Si and Sj differ by more than their individual variances. The most common model form for the percent contribution of one source (here Si) is

42 ∂X sample −∂X j %Si = •100 ∂X i −∂X j where ∂Xsample is the stable isotope signature of element X in a sample, ∂Xj is the stable isotope signature of Sj , and Xi is the stable isotope signature of Si. This model form is simple to calculate and has been applied in many systems and is appropriate when comparing between potential diet sources.

3.5.2 Multi-Source Models

Multi-source mixing models currently exist under two model forms, those that rely on Euclidian distance methods (e.g. Kline et al., 1993; Ben-David and Schell, 2001) and those that rely on linear mixing models (e.g. Phillips, 2001; Lubetkin and Simenstad, 2004). Criticisms of Euclidian distance models have been made Phillips (2001), stemming largely from a requirement that all input diets be used (i.e. some proportion of each diet source will be included in the results regardless of whether the diet is actually eaten).

3.5.2.1 Euclidian Distance Models Euclidian distance models rely on the assumption that the shorter the distance (in terms of ∂X’s) between a consumer and a given food source, say Si, the greater the proportion of that food source in the consumer’s diet. These methods appear to be useful only in three-source situations, after which additional end members cannot be uniquely resolved. The models have several structural forms that differ in how they represent distances between end-members, but a general form (e.g. Ben-David et al., 1997) estimates the percent contribution of a given diet (% Sdiet) as 1/DS ′ %S = •100 diet 1/DA ′ +1/DB ′ +1/DC ′ where S ′ is the fractionation-corrected ∂X of the food source of interest (A, B, or C), D is the ∂X value of the sample, and DS ′ is the corresponding distance from D to S ′ (Figure 27). The S ′ values are calculated to represent the expected value of D if the consumer was to feed on a pure diet of food S. This correction factor is necessary to remove the biasing effects of differential fractionation among food sources. Ben-David and Schell (2001) evaluated the performance of a Euclidian distance model in estimating the relative contributions of three known experimental diets in mink (Mustela vison). Their study found that the Euclidian model failed to accurately describe the actual proportions of each diet in mink blood, and suggested that incomplete mixing of diets during digestion. From these results, Ben-David and Schell (2001) cautioned that Euclidian models should be used to obtain only a general index of animal diets rather than to calculate exact proportions of food sources. They also concluded that Euclidian models are more appropriate than linear models for studies where accurate fractionation values for each diet are not known. However, Phillips (2001) has criticized the use of

43 Euclidian models in estimating more than one diet source, where relative contributions of each end member are not calculable and multiple proportion- solutions are possible.

Figure 27 Plot of dual isotopic compositions of food sources A, B, C, and consumer D. A’, B’ and C’ represent the food source isotopic composition after adjustment for trophic fractionation. Plot values from Szepanski et al. (1999). After Ben-David and Schell (2001)

3.5.2.2 The Phillips Linear-Mixing Models Linear-mixing models use systems of mass-balance equations to solve for unknown parameters — in this case the fractional contributions of food sources to a given consumer’s diet. The simplest linear-mixing model assumes (as do Euclidian models) that the partitioning of food sources into sampled tissues is the same for each stable isotope (e.g. C or N). This assumption is a good one provided the fractionation (∆) values between consumer and diet are well known (this is not always the case). The application of linear-mixing models have been strongly advocated by D.L. Phillips (e.g. Phillips, 2001), and the models presented here have all been developed from work associated with his research. The series of simple linear-mixing model equations take the form

∂Xi,D = f A∂X i,A + f B∂X i,B + fC∂X i,C

∂X j,D = f A∂X j,A + f B∂X j,B + fC∂X jC

1= fA + f B + fC where ∂X is the of element X in food source A, B, C, or consumer D, and f is the fraction of food source A, B, C in the diet of consumer D. In this three food-source model, two elements, Xi and Xj, are required and these are most often ∂15N and ∂13C. Given ∂15N and ∂13C values for A, B, C, and D, the f’s for each diet can be calculated as

44 15 15 13 13 13 13 15 15 (∂ NC −∂ NB )(∂ CD −∂ CB ) − (∂ CC −∂ CB )(∂ ND −∂ NB ) fA = 15 15 13 13 13 13 15 15 (∂ NC −∂ NB )(∂ CA −∂ CB ) − (∂ CC −∂ CB )(∂ NA −∂ NB ) 13 13 13 13 (∂ CD −∂ CC ) − (∂ CA −∂ CC ) f A fB = 13 13 (∂ CB −∂ CC )

fC =1− f A − fB

Ben-David and Schell (2001) assessed the performance of these linear-mixing models in evaluating the relative contributions of three known experimental diets in mink. Similar to a Euclidian model tested, the linear-mixing model also failed to accurately describe the actual proportions of each diet in mink blood. Again, Ben-David and Schell (2001) inferred that incomplete mixing of diets during digestion was to blame, and that this assumption is probably widely violated in nature. The study also found that the linear-mixing models were unable to cope with out-of-bounds values (i.e. ∂XD values outside the solution space) that resulted from the intrinsic variability of mink isotope ratios. Ben-David and Schell (2001) were adamant however, that both Euclidian and simple linear-mixing models require accurate parameterization of model inputs and that this information is often not known to researchers. Phillips and Koch (2002) addressed the assumption that equal proportions of elements are absorbed in a given consumer by developing concentration- weighted mixing-models that account for such differences. The solution set added a third fraction, fB,diet, to the simple linear model set to represent the fraction, by biomass, of each diet absorbed by consumer D. This added an entirely new set of linear equations that could be combined to solve for the source fractions of assimilated carbon (in, for example, diet A, fA,C) and nitrogen (fA,N). These new equation sets could then be solved by matrix algebra to obtain possible solution sets. Phillips and Koch (2002) concluded that this model could accurately account for differences among source contributions for specific elements and provided specific guidelines for its use (Table 8). Details of the modeling scheme are presented in Phillips and Koch (2002) and a spreadsheet for performing the model calculations is available at http://www.epa.gov/wed /pages/models.htm.

Table 8 Specific guidelines for multi-element mixing models. Adapted from Phillips and Koch (2002) Specific Guidelines 1 For clarity, always report the isotopic signatures and fractionation values used for each source and the mixture 2 Always measure and report elemental concentrations for each source 3 If the concentrations differ substantially among the sources, use a concentration-weighted mixing model to account for these differences 4 For unique solutions of the contributions from n sources, n−1 isotope systems (elements) must be used. The concentration-weighted mixing model can be generalized to any number of isotope systems

45 Phillips and Gregg (2003) took the linear-mixing model concept further by developing a feasible contribution matrix plot that provided isotopically-feasible solutions to fractions of food sources in consumer tissues. In essence, their plots provided fractional contributions of >n+1 food sources determined from n isotopes relative to a given fractional contribution from one of the potential sources (Phillips and Gregg, 2003). The plots created incrementally by calculating each possible solution to the simple linear mixing models by varying each source by 1 or 2 percent contribution. Each of these solution sets was then compared with the observed signatures and, if viable, stored in a data set for plotting. The scatter plot matrix of viable solutions could then be analysed for ranges of possible diet contributions for each food source. Phillips and Gregg (2003) created the program IsoSource to perform the calculations and it is available at http://www.epa.gov /wed/pages/models.htm. The appeal of the IsoSource procedure is its ability to estimate diet contribution ranges for potential food sources when there are >n+1 sources with n tracers. For omnivorous species this model scheme is particularly useful. For example, in Figure 28, seven potential food sources for Alaskan mink are presented. The potential range of fractional contribution for each food source is indicated by the

Figure 28 Scatter plot matrix showing isotopically feasible contributions of seven food sources in the diet of spring coastal mink in SE Alaska. Each panel shows a scatter plot of the feasible contributions of two food sources on a 0-1 scale. After Phillips and Gregg (2003)

46 lines on the diagonal panels. By fixing the fractional contribution of one food source — in this example, fish (0.68) — the fractional contributions of the other food sources can be estimated. This example shows at fish=0.68, the fractional contributions of the other food sources are 0.19 (mussel) and 0.13 (crab). The remaining shrimp, rodent, amphipod, and duck values are all 0. These secondary contributions can be single values (when at a corner-point of the plot) or within a particular range. The IsoSource model has been used in only three studies, twice on bears in Alaska (Felicetti et al., 2003; Ben-David et al., 2004) and once on mangrove fishes in Australia (Melville and Connolly, 2003). The Melville and Connolly (2003) study used isotope data to determine which of seven autotrophic sources (SG seagrass, EPI seagrass epiphytes, MAN mangroves, MPB microphyto-benthos, POM particulate organic matter, SMG saltmarsh grass, and SMU saltmarsh succulents) contribute to diets of three fish species (bream, Acanthopagrus australis; sand whiting, Silago ciliata; and winter whiting, S. maculata) in unvegetated . The resulting ∂15N and ∂13C values for autotrophs were input into IsoSource and feasible contribution distributions were plotted for each fish (Figure 29). These whole-estuary results were then compared with spatial analysis of ∂13C for inference about the fish diets.

Figure 29 Histograms of the distribution of feasible contributions of the seven autotrophs for (A) Acanthopagrus australis, (B) Silago ciliata, and (C) Silago maculata, after correcting fish values for ∂15N trophic level fractionation. Values in boxes are 1-percentile to 99-percentile ranges for the distributions. Adapted from Melville and Connolly (2003) The Melville and Connolly (2003) whole-estuary analysis found, from its narrow contribution distribution, that MAN-derived energy contributed little to the diets of any of the fishes and, due to their broad contribution distributions, that SG and EPI were most likely significant contributors to all fish diets. The only difference

47 among fishes was an increased contribution of SMG in sand whiting diets over the other fishes. The spatial analysis results provided more detailed information however, and Melville and Connolly (2003) found that mangrove habitats contributed significantly, but in a low percentage of total contribution, to the diets of bream and sand whiting. While informative, Melville and Connolly (2003) concluded that the whole estuary (IsoSource) analysis required detailed spatial analysis to be made in order to resolve the food source contributions among fishes.

3.5.2.3 Lubetkin-Simenstad Linear-Mixing Model Lubetkin and Simenstad (2004) developed two linear-mixing models, SOURCE and STEP, to quantify food web structure and sources with stable isotopes. The models were constructed with 3 goals in mind, (i) to identify the dominant sources of organic matter supporting the food web, (ii) to determine the sequence of trophic steps in the food web, and (iii) to evaluate specific predator-prey linkages within the food web. SOURCE and STEP require the same assumptions inherent in the simple linear-mixing models of Phillips (2001), including equal fractionation rates for each distinct tracer and that all tracers in the system be identified.

Figure 30 All physically possible combinations of a three-source mixture lie within the boundary of the large triangle, bounded by (1,0,0) (0,1,0) and (0,0,1). The corner point solutions delineate the solution area (shaded triangle) and are used to compute the centre of mass estimate. After Lubetkin and Simenstad (2004) SOURCE is a mass-balance model that estimates the relative proportions of each autotrophic food source in the diet of a given consumer — much like the models developed by Phillips and Gregg (2003). The model uses a system of linear equations that provide potential solution sets for each food source; it also explicitly incorporates fractionation values for the consumer’s trophic level. SOURCE takes a model form closely related to Equation (13), which — for three tracers and four sources – is

48 1= s1 + s2 + s3 + ...+ sn C = s t + s t + s t + ...+ s t + α(L) t 1 1 2 2 3 3 n n Cu = s1u1 + s2u2 + s3u3 + ...+ sn un + β(L)

Cv = s1v1 + s2v2 + s3v3 + ...+ snvn + γ(L) where each source i, contributes some fraction si of a given isotope tracer, t, u, or v to the consumer’s overall tracer level Ci. α, β, and γ represent the fractionation of each tracer into the consumer from any given source. Solution sets are created by solving the linear equations while setting each of the sources to 0. This procedure provides corner-point solutions for the three remaining sources that can then be used to outline a possible solution space (Figure 30). In compiling a solution set for each possible food source to consumer combination some solutions will not be physically possible (i.e. negative values) and will be discarded. From the final solution set, weighted averages can be calculated from the valid corner-point solutions to find a centre of mass estimate that incorporates any trends in the complied solutions. STEP differs from SOURCE as it is used to model metabolic fractions between a consumer and each food source to determine the direct links between consumers and their diets. Again, STEP takes a model form similar to those above, namely

1= f1 + f2 + f 3 + ...+ f n C −α = f t + f t + f t + ...+ f t t 1 1 2 2 3 3 m m Cu − β = f1u1 + f 2u2 + f 3u3 + ...+ f m um

Cv − γ = f1v1 + f 2v2 + f 3v3 + ...+ f mvm where the fm’s represent the fractions of each food source, m, in the consumer’s diet. The pool of m’s can include everything in the target food web and the model permits cannibalism (consuming one’s own tracer signatures). STEP is solved simultaneously for all possible food web combinations, although this candidate set of connections can (and should) be thinned for biological realism, e.g. carnivores need not consume plants. STEP is applied to each consumer in the system according to individual prey availability and, once all direct links have been estimated, the model estimates the consumer’s trophic level (TL) as a weighted average of prey trophic levels (TLi) plus 1, such that

TL = f1TL1 + f2TL2 + f 3TL3 + ...+ fmTLm +1 Lubetkin and Simenstad (2004) tested their models on a well-studied estuarine ecosystem in Sapelo Island, Georgia that allowed them to compare SOURCE and STEP models between two and three isotope datasets. Sapelo Island, on the southeast coast of the United States, is surrounded by extensive and highly- productive marshlands and research there has focused on nutrient flows among marshland species and marine fishes. Extensive sampling records have been collected, and an early stable isotope study by Peterson and Howarth (1987) provided a convenient data set for comparison and analysis.

49 Table 9 Estimated trophic levels for Sapelo Island consumers using SOURCE and STEP with ∂13C, ∂15N, and ∂34S or ∂13C and ∂15N. BC, from Bighole Creek; LC from Lab Creek. After Lubetkin and Simenstad (2004) Model SOURCE STEP Isotopes used C,N,S C,N C,N,S C,N Suspension feeders Geukensia demissa 0.73 0.71 1.36 1.16 Crassotrea virginiana (BC) 0.47 0.99 1.07 1.32 Crassotrea virginiana (LC) 0.83 1.14 1.23 1.44 Suspension feeders Mugil cephalus (small) 0.83 0.71 — 1.16 Palaeomonetus pugio 1 1.21 1.38 1.39 Penaeus setiferus 1.28 1.47 1.55 1.52 Mugil cephalus (large) 1.89 1.89 1.82 1.69 Deposit feeders Uca pugnax 0.86 0.39 1 0.98 Ilyanassa obsoleta 0.7 0.79 1.29 1.16 Littorina irrorata 0.61 0.4 — 0.96 Omnivores and predator Callinectes sapidus 1.57 1.77 1.67 1.73 Fundulus heteroclitus 1.81 2.12 1.87 2.01 Ocypode quadrata 2.45 2.37 2.23 2.36 Bairdiella crysura (predator) 1.35 1.61 1.63 1.6

The Sapelo Island data had six primary producer sources catalogued, saltmarsh cordgrass (Spartina alterniflora), black needlerush (Juncus romerianus), loblolly pine (Pinus taeda), live oak (quercus virginiana), creekbank algae, and phytoplankton. Each source had ∂13C, ∂15N, and ∂34S data available (except creekbank algae for which Lubetkin and Simenstad (2004) estimated a ∂34S value of 15 ‰) and these were compared to isotope ratios for 14 consumer species (Table 9). The SOURCE and STEP results were largely in agreement, although STEP tended to quantify a number of mathematically possible but biologically unlikely trophic linkages. The inclusion of ∂34S data resolved some of these problems however, providing more precise linkage estimates and removing weak or spurious links (Lubetkin and Simenstad, 2004, Figure 31). Lubetkin and Simenstad (2004) concluded that SOURCE and STEP gave realistic food web structures provided all significant nodes in the system are sampled. They also suggested that both packages could be used with pooled stable isotope and, e.g., fatty-acid or amino-acid tracers for still-more accurate food webs. The integration of multiple biochemical tracers is an open area for research and tools such as SOURCE and STEP should be considered initial tools to help bring these elements together. The S-plus implementation of SOURCE and STEP are available at http://staff.washington.edu/lubetkin/SOURCE_And_STEP/index.html.

50

Figure 31 (a) A food web for Sapelo Island based on ∂13C, ∂15N, and ∂34S data. (b) A food web for Sapelo Island based on ∂13C and ∂15N data. Arrows go from the source (prey) to the consumer (predator). For clarity, estimated trophic interactions representing less than 7 percent of a consumer’s diet are not shown. The lightest lines represent interactions accounting for 7 to 15 percent of an organism’s diet. Medium weight lines represent food items contributing between 15 and 25 percent to an organism’s diet. Heavy lines indicate that more than 25 percent of an organism’s diet is estimated to come from a particular source/prey item. BC, from Bighole Creek; LC from Lab Creek. After Lubetkin and Simenstad (2004)

3.6 POTENTIAL ECOSYSTEM PREDICTION MODELS

The use of stable isotopes in ecology and ecosystem-based modelling are both under development in many systems. Integration of the two concepts, whereby stable isotope data is used to inform ecosystem-based assessment models, has yet to be developed. Direct integration of stable isotope data is possible however, through Ecopath (with Ecosim) software that attempts to project the effects of fisheries-management decisions on a target ecosystem. It has been widely applied to food web and ecosystem studies in the marine sciences (Christensen, 1995) and currently has more than 1200 users world-wide (Ecopath web statistics, 17 February 2005; http://www.ecopath.org). Ecopath is a straight-forward, mass-

51 balance procedure that provides estimates of biomass flows between ecosystem elements and can be used with Ecosim to predict ecosystem impacts of biomass changes on a given food web. A potential exists for stable isotope data of diet fractions to be included in Ecopath models that has not been exploited to date. The Ecopath approach models the trophic interactions among functional groups in an ecosystem as 0 = production by (i) – predation on (i) – non predation losses of (i) – export of (i) which can be expressed as

0 = Pi − M2i − Pi (1 − EE) − EX i

where Pi is the proportion of i; M2i is the predation mortality of i; EEi is the ecotrophic efficiency of i (the fraction of the production i that is consumed within the system or exported or harvested); 1 − EE is the unknown mortality and EXi is the export of i. Equation 26 can also be expressed as

0 = BiPBi −Σj B j (QB j DC ji ) − PBiBi (1 − EE) − EX i where PBi is the production:biomass ratio, QBi is the consumption:biomass ratio, and DCji is the fraction of prey i in the average diet of predator j. Equation 27 is then grouped for each species n into n linear equations which can then be solved using matrix algebra. While many of the parameters can be estimated from within Ecopath, the DCji and EXi data cannot. In many systems, assigning DCji values is problematic but it is precisely the DCji information that can be estimated using stable isotopes and linear-mixing models like SOURCE and STEP. All of the components are available to integrate stable isotope trophic-linkage information into Ecopath and, due to their relative lack of complexity; this combination should be particularly effective for application to pelagic ecosystems.

4 SAMPLING CONSIDERATIONS FOR STABLE ISOTOPES IN MARINE ECOSYSTEMS

4.1 SAMPLE QUALITY

Maintaining sample quality for stable isotope analysis is considerably simpler than for many chemical tracers (e.g. heavy metals, organochlorines, fatty-acids). Because stable isotopes do not break down (they are elements) and mass spectrometry measurements are of relative, rather than absolute, quantities, tissue samples need only to be buffered against foreign agents (e.g. bacteria) that could alter the isotope ratios given enough time. Most studies choose to ice and subsequently freeze samples prior to laboratory analysis, and this method is certainly the most practical and inexpensive tissue preservation method available. Large survey ships and commercial fishing boats are ideal sites for sample collection, as they are often equipped with blast freezers and have

52 adequate space for sample storage. Freezing is not, however, necessary if tissues can be alternately fixed. Some studies (see Brodeur et al., 2002, for instance) chose to simply dry their samples on-site in a drying oven (set to 60ºC) and others have even used an electric hair dryer (Pinnegar and Polunin, 1999). Consideration must be made regarding proper labelling of samples prior to freezing. Samples can be easily lost due to label failure and many have underestimated the potential for such problems to occur. Even the most advanced labelling systems will break down in freezer conditions when seawater has affected them. Time is often short aboard fishing vessels and, given that multispecies data is the norm, one can easily be overwhelmed by the volume of samples to process from a given catch. Sample bags should be prepared in advance and labels should be double or triple redundant. If multiple tissues are sampled from each animal, it is advisable to enclose each tissue or stomach- content sample in an individually-labelled WhirlPac™-style bag and group these in a larger bag labelled for each animal. If any one label fails within the bag, the others can serve as reference and the sample will be saved. Ideally each individually labelled bag should be marked on the outside and should contain a duplicate label written in pencil on waterproof paper within the bag itself. Redundancy is the key — seawater and freezing are hard on plastics and ink. Consistency in the choice of tissue and sampling location may be important. While no studies have yet ascertained the degree of isotopic variance among sites within a given tissue, considerable differences have been found in ∂15N and ∂13C values among tissues, including differences between red and white muscle (Pinnegar and Polunin, 1999). Pinnegar and Polunin (1999) reported ultimate ∂15N and ∂13C differences among rainbow trout tissues in fish at steady-state with their diet. Their results suggest that tissues can vary in ∂15N by more than 1 ‰ and in ∂13C by up to 2.5 ‰. Biochemical differences may exist among tissue regions and this uncertainty can be addressed using consistent sampling locations. As outlined in sections above, metabolically-distinct tissues capture different periods of dietary feeding, with high-metabolism tissues, such as liver and blood, turning over faster than low-metabolism tissues, such as muscle or bone. These metabolic differences can be exploited to achieve greater temporal coverage by sampling multiple tissues than from muscle tissue alone. Sampling additional tissues aboard a ship will generally take little time and few resources to complete and should therefore be considered in study designs. Sampling large-pelagic fishes for multiple tissues may take considerably more effort, but muscle biopsy and blood sampling require similar logistics and can therefore be sampled together.

53 4.1.1 Guidelines

• Samples should be taken from one low and one high metabolism tissue in animals of sufficient size; small animals and plankton are conventionally sampled whole • Sampling locations should be standardized on each tissue type and among species groups (e.g. fishes, cephalopods) • Samples should be iced, frozen, or dried within approximately one hour of sampling

4.2 SAMPLE QUANTITY

Sample quantity for stable isotopes have frequently been on the order of three to five samples per species, tissue, and location. The choice of these sample sizes has been driven largely by the natural variability of the samples (typically ±0.4 ‰) and of mass spectrometry measurements (±0.2 ‰). However, given that sample analysis is usually the cost-limiting factor of any stable isotope study, up to ten samples for each species — and, for individuals >5 g, each tissue — should be collected to ensure good confidence interval estimates and in case further analysis is desired. Archived tissues are invaluable, particularly if carefully collected and catalogued. A few studies (e.g. Vander Zanden et al., 2003; Wainright et al., 1993) have extracted trophic time-series from stable isotopes in archived tissues and a tissue collection for a given system may have considerable utility for future research. Sampling guidelines are summarized for various classes of organisms is Appendix 1.

4.2.1 Guidelines

• For surveys landing multiple species, sample all available individuals per species ≤10. • Sample all available individuals of mammals, sharks, large tunas, and billfishes. • Plankton sampling should pool ≤20 individuals by species at each location for spatial analysis.

4.3 LABORATORY PREPARATION

While mass spectrometry is generally conducted by independent laboratories, sample preparation is usually conducted by the research group. Preparation is straightforward, requiring initial drying, homogenization, chemical treatment, desiccation, and weighing. Considerable time is required to complete these steps however, when thousands of samples need to be processed in isolation. An experienced technician can complete the total process in, cumulatively, about 15 min per sample. Firstly, frozen samples must be dried completely prior to processing. ~5 g samples should be used as this will provide a sufficient supply of stable samples for both the current and any subsequent isotope analysis.

54 Drying may be most efficiently done in a freeze dryer in 12 to 24 hrs. Care must be taken with liver and blood samples as they can easily clog a freeze-drying unit and these tissues should therefore be buffered by a filter according to the specifications of the machine. If freeze-drying is not available, samples can be left in a drying oven at 60 ºC until completely dry. Bone or cartilage samples must be taken from the vertebral centrum to the outside edge in a consistent thickness. Because vertebrae are laid down in rings, each may have a distinct and sampling from the centre outward will ensure each ring contributes to the overall sample. Once samples are in a dried and stable state, each sample must be pulverized to homogenize the tissue. Homogenization is essential to ensure a truly average tissue sample with no contamination bias. This can most efficiently be accomplished in a shaker mill or high-energy ball mill, where ball bearings are shaken along with the sample within a steel canister for two to five minutes, leaving behind a fine powder. Alternately, dried samples may be pulverized by hand using a mortar and pestle. Cross-contamination can easily occur during this process and care must be taken to thoroughly clean instruments between samples. Ball-milling canisters should be cleaned with soap and dried in a fume hood. When processing dry samples of the same species, mortar and pestle can be carefully wiped clean with a Kimwipe™. Between species or between oily samples, mortar and pestle must be cleaned with soap and water and dried in a fume hood. Chemical treatments are sometimes applied to stable isotope samples to help correct for biochemical differences among tissues. Lipid-extraction is by far the most commonly applied procedure, the theory being that lipid synthesis preferentially includes 12C relative to 13C and therefore lipid-rich tissues will be biased toward lighter ∂13C values (DeNiro and Epistein, 1977). Complicating this issue however, are results that have found that lipid extraction can differentially shift muscle ∂15N values in freshwater fishes (Sotiropoulos et al., 2004); this suggests that species-specific biases in lipidized tissues may add, rather than reduce, bias among fish tissues. Pinnegar and Polunin (1999) found that systematic lipid-extraction led to higher variance among samples and a subsequent loss of statistical power. Most studies on fishes currently lipid-extract to compare among tissues. To deal with the uncertainties surrounding lipid extraction, a subsample of treated or un-treated (depending on which is the dominant technique) tissue should be analysed to account for the effects of the preparation used. Lipid extraction is conventionally done by twice-agitating pulverized samples in a 2:1 methanol:chloroform solution (enough to submerge the sample) leaving the preparation to sit overnight, and removing the resulting supernatant in a centrifuge (toluene is sometimes used in this process). After the second methanol:chloroform treatment samples can then be left to air-dry in a fume hood. Acidification to remove carbonates has also been conducted in some situations, although it has been applied less frequently than lipid-extraction. The motivation

55 for acidification appears to be an artefact of stable-isotope analyses conducted before continuous-flow mass spectrometry (CFIRMS). Prior to this technology, isotope ratios were calculated individually, and the carbonates were removed from sediments, shells, and fish scales for ∂13C analysis to eliminate traces of non- dietary sources. This practice was carried on in subsequent multiple-isotope studies (e.g. Hobson and Clark, 1992) but acidification to remove extraneous carbonates from CF-IRMS-analysed samples has been shown to have unintended effects on ∂15N values. Bunn et al. (1995) analysed shrimp (Metapenaeus bennettae) samples before and after acidification and found that the process enriched their ∂15N values by 1‰; Pinnegar and Polunin (1999) found similar results in rainbow trout. In addition, both Bunn et al. (1995) and Pinnegar and Polunin (1999) concluded that acidification did not, in fact, alter the ∂13C values of their fish tissue samples. Schmidt (2003) chose not to acidify plankton samples on similar grounds. While some exploration of acidification may be useful in a newly- sampled food-web, it appears not to be an important step in the preparation of most marine-derived tissues. Following chemical treatments, samples can generally be left to air-dry in a fume hood or be heated in a drying oven at 60 ºC. Once dry, the powdered samples are weighed out on a micro-balance to the specifications of the laboratory conducting the analysis. Often samples are weighed to between 1 to 2 µg, enclosed in tin capsules, and placed in special index trays to keep track of each sample. The index trays are then sent to the analytical laboratory for analysis.

4.3.1 Guidelines

• 5 g of previously-frozen tissue should be freeze-dried for processing. • Dried tissue must be homogenized either in a ball-milling unit or with a mortar and pestle. • Samples (other than bone or cartilage) should have lipids extracted x2 with a 2:1 solution of methanol:chloroform and dried in a fume hood after supernatant removal. • Acidification is not generally recommended. • Samples must be weighed to the specifications of the mass- spectrometry lab thus an analytical balance, tin capsules, and trays are required.

56 PART 2 - ANALYSIS OF STABLE ISOTOPES SAMPLES FROM THE LESSER ANTILLES PELAGIC ECOSYSTEM PROJECT

1 OVERVIEW

The complete dataset consisted of 516 records, of 96 species comprising 464 individuals (or pooled zooplankton) collected between 8 June and 24 November 2006. Data were collected opportunistically in two ways: 1) prey species were collected during research trawls (during the LAPE ecosystem survey, FAO 2007) conducted in April/May 2006, at stations determined by other aspects of the LAPE project; 2) predator species were collected during day-fishing trips conducted from individual islands in June (hereafter predator survey). An important limitation of the stable isotope work was that the data were confounded spatially and temporally to some extent. In particular, predator species collected in the south (Trinidad) were highly enriched in δ15N and were not collected from another location at any time, making it difficult to assess the relative trophic status of these species (see Spatial Structure below). Similarly, species collected in predator and prey surveys did not overlap in time, making it problematic to assess any degree of temporal trend within species. The greatest problem with the dataset however, was low replication within species, making it difficult to assess the presence of ontogenetic shifts at size for a given species—a primary study objective. While this was undoubtedly due to the opportunistic nature of sampling, targeted sampling effort for key predator species (e.g. flyingfish; yellowfin tuna) would be desirable in any future sampling programme. In particular, a systematic effort to sample target species from throughout the archipelago would greatly help to ascertain the spatial extent of δ15N enrichment near the South American continent. Despite the low sample replication and spatial issues, at the community level 15 strong trends were observed, particularly in terms of the δ N-log2(weight) relationship (see Size Structure below). From the data there were clearly three isotopically-distinct communities present, providing novel trophic information that was unlikely to have been otherwise observed.

2 SPATIAL STRUCTURE

Of the 11 countries near or from which the prey and predator surveys were made, only samples from the predator survey in Trinidad showed a clear spatial effect on δ15N (Figure 2.1). Three species, Caranx hippos, Scomberomorus brasiliensis, and Scomberomorus cavalla were present in the Trinidad samples, and data from all three were enriched in δ15N by an average of 5‰, relative to the rest of the

57 pelagic community. Unfortunately these species were not observed elsewhere, somewhat confounding interpretation of their enrichment. However, oceanographic data from near Trinidad has demonstrated that exceptionally-high primary production occurs near Trinidad relative to the other Lesser Antilles islands (Forget, 2007), making terrestrial input as the most likely source of enrichment in Trinidad predator species.

Given this explanation, the elevated δ15N observed in both the muscle and liver tissue samples provide good evidence to suggest that all three fish species are resident in the Trinidad area (or other enriched locations to the south) and are not feeding elsewhere in the LAPE region. For both Scomberomorus spp. this is not surprising as they are known to be resident in the area and are not caught elsewhere in the LAPE system. Caranx hippos however, has a wide distribution and, should samples from elsewhere in the region be δ15N depleted relative to those from Trinidad, could confirm the residence hypotheses proposed by these data.

Figure 2. 1 δ15N vs. δ13C biplot for prey and predator surveys. Countries are Barbados (red), Saint Kitts (light green), Trinidad (pink), Barbuda (dark blue), Dominica (purple), Guadeloupe (light blue), Saint Vincent (green), Grenada (orange), Tobago (yellow), Martinique (purple), and Saint Lucia (aquamarine).

58 Contrary to the δ15N-enrichment observed, little spatial pattern was observed in the δ13C results. δ13C is frequently cited as a source tracer in aquatic systems (Melville and Connoly, 2003), however no differences in δ13C range were observed in prey or predator species from throughout the region. Predators had a slightly wider (+3.03‰) range of δ13C values, however most of this was from liver samples collected from predator species that are routinely depleted in δ13C due to their high lipid content (Post et al., 2007).

3 SIZE-BASED TROPHIC STRUCTURE

Because of the low species-level resolution present, the greatest utility of the dataset was for discerning size-based trophic structure within the LAPE region. Body size distributions have had a long history for describing energy allocation in ecosystems (Kerr, 1974), where larger individuals are generally expected to feed at higher trophic levels (Dickie et al., 1987). This intuitive property of trophic structure has been supported by research in the North Sea which demonstrated that, although δ15N data are uninformative about cross-species trophic relationships, body size is an excellent predictor of trophic level across a well- defined community (Jennings et al., 2001). Fishes collected for stable isotope analysis were divided into two environmental communities based on their environmental category on FISHBASE (fishbase.org; accessed 27 August 2007). Fishes described as reef-associated, demersal, oceanadromous, or pelagic were classified as pelagics and fishes described as mesopelagic or bathypelagic were classified as mesopelagics. Given the assumption that the LAPE region defines an ecologically-meaningful community and following the methods of Jennings et al. (2001), fishes within the sample data were analyzed in two ways: 1) as raw data, where each individual is a data point in their respective community, and 2) as binned data, where individuals were binned into log2-size classes by weight (kg) for the pelagic and mesopelagic communities. With the exception of fishes from Trinidad (see Spatial Results), these categories were supported by linear mixed-effects model results where AIC values strongly (>7, Table 2.1) favoured random intercepts and slopes between the pelagic and mesopelagic groups. Therefore, these two groups and their associated within- group regressions were used to assess the community-wide relationship between 15 δ N and log2(weight) for both species and functional groups. Because of their demonstrated enrichment, Trinidad data were analyzed separately.

59 Table 2.1 Model selection summaries for mixed-effects models of the relationship between raw 15 δ N and log2(weight). Subscripts denote observation i within environmental community j (pelagic or mesopelagic). Model Name log(L) K AIC ΔAIC 15 δ Nij=β0+ β1log2(weight)ij Fixed -693.35 5 1394.7 55.0 15 δ Nij=β0,j+ β1log2(weight)ij Random intercept -669.46 4 1346.9 7.2 15 δ Nij=β0j+ β1jlog2(weight)ij Random intercept and slope -664.88 3 1339.7 0.0

Jennings et al. (2001) analyzed their community data using binned data, stating 15 that cross-species δ N-log2(weight) relationships were relatively weak but, when averaged across size classes, the community-level relationship is exceptionally strong. However, model summaries for each group in the LAPE data (Table 2. 2) showed little difference between the raw and binned data, making this argument moot. It may be that the extensive surveys conducted by Jennings et al. (2001) in the North Sea represented many more rare or unusual species than observed here, making the raw, species-level data difficult to interpret. This did not appear to affect these data, and the raw data is presented throughout the report.

Table 2. 2 Summary statistics (95% CI) for linear regressions of raw and binned LAPE stable isotope data from three environmental regions. Parameter Raw Binned a) Trinidad Intercept 10.71 (9.13, 12.29) 10.50 (8.76, 12.24) Slope 0.22 (0.07, 0.37) 0.25 (0.06, 0.45) b) Pelagic Intercept 5.81 (5.46, 6.16) 5.41 (4.65, 6.18) Slope 0.24 (0.19, 0.29) 0.31 (0.24, 0.38) c) Mesopelagic Intercept 6.47 (6.07, 6.87) 6.88 (5.66, 8.10) Slope 0.44 (0.34, 0.55) 0.42 (0.17, 0.66)

The strongest conclusion that can be drawn from this study relates to the differences in both slope and intercept between the a priori-defined pelagic and mesopelagic communities. The nature of the δ15N enriched mesopelagic community relative to the pelagic indicates strongly that few of the larger pelagic species are relying on mesopelagic prey as an important food source in the LAPE 15 system. Many mesopelagic species having log2(weight)<5 have similar δ N signatures to pelagic predators with log2(weight)>15 (Figure 2. 2). Given the clear size-based relationships observed in both communities, it is highly unlikely that mesopelagics represent a major pelagic food source. The only alternative explanation is that pelagic species are feeding on a third, unsampled community (e.g. reef-associated species) that is greatly depleted in δ15N relative to the pelagic community, thereby masking any mesopelagic signal. Clearly, further sampling of other potential prey communities would be required to resolve this issue. The mesopelagic and pelagic communities also differed in their regression slopes 15 for the δ N-log2(weight) relationship, suggesting that the mesopelagic community had tighter trophic structuring by size than the pelagic community.

60 This difference has a caveat however, in that few large (>1kg) mesopelagic species were collected, making it difficult to assess how large mesopelagics would affect this relationship. It is however, clear from Figure 2. 2 that the observed mesopelagic community is enriched in δ15N relative to the pelagic. The Trinidad samples of three (FISBASE-defined) pelagic species were highly enriched relative to the pelagic community elsewhere, however they had very similar slope estimates to the pelagic samples (Table 2. 2). Although there are few Trinidad data available, these results support the contention that the δ15N enrichment of Trinidad samples was due to baseline enrichment in primary production rather than any gross differences in trophic structure in the area. As such, the differences between the Trinidad and pelagic community intercepts (5.1‰) is the best estimate of baseline enrichment in the Trinidad area.

Figure 2. 2 Linear regression plots for raw data from Trinidad (pink), mesopelagic (green), and pelagic (blue) fishes in the LAPE stable isotope survey.

4 COMMUNITY METRICS

Both pelagic and mesopelagic communities were mapped using the isotope community metrics suggested by Layman et al. (2007). It is important to note however, that these communities of interest were likely to have been affected by differences in sampling effort between the predator and prey surveys, where mesopelagics were sampled almost entirely during the prey survey and the large (log2(weight) >8) predators were sampled almost entirely during the predator

61 survey. The sampling undertaken may therefore have introduced some unknown level of bias into the community metric results, particularly in limiting the trophic-isotope space represented in the mesopelagic community. Trinidad samples were excluded due to the observed enrichment of species from that area. The general methodology consisted of calculating six metrics:

1. δ15N range (NR), defined as the distance between the most enriched and depleted δ15N values and suggesting more trophic levels in the system;

2. δ13C range (CR), defined as the distance between the most enriched and depleted δ13C values and suggesting niche diversification at the bottom of the food web; 3. total area (TA), defined as the convex hull area of all species in isotope bi-plot space and suggesting the total extent of trophic diversity within the system; 4. mean distance to centroid (CD) defined as the average Euclidian distance of each species to the δ15N- δ13C convex-hull centroid, suggesting the overall degree of trophic diversity in the system; 5. mean nearest neighbour distance (NND) defined as the mean Euclidian distance to each species' neighbour, a measure of species niche packing within the system; and 6. standard deviation of the nearest neighbour distance (SDNND), suggesting the evenness of trophic niches within the system. Relative to the pelagic community, the mesopelagic community showed relatively dense, even species packing (lower TA, NND and SDNND values), indicating that trophic niches within the sampled mesopelagic food web are tightly-structured, with more specialized feeding behaviour than in the pelagic zone (Figure 2. 3). The extreme points in TA were defined primarily by the multi- species small mesopelagics functional group, although in terms of NR much of this diversity may have been due to few samples being observed from a limited size range. Given the small range of sizes classes in the mesopelagic samples, the range of trophic diversity appeared to be relatively high, with NR values approaching those of the much larger pelagic community. This observation was supported by the size-based regression slopes, where the mesopelagic estimate was nearly double that of the pelagic community (Table 2. 2).

62

Figure 2. 3 Stable-isotope bi-plots of pelagic species collected from the LAPE regional surveys. Symbols indicate functional group membership. Summary statistics provided below points include: δ15N range (NR), δ13C range (CR), total convex hull area (TA), mean distance to centroid (CD), nearest neighbour distance (NND), and standard deviation of nearest neighbour distance (SDNND).

63 The pelagic community was more trophically-diverse than the mesopelagic community, with nearly double the TA and somewhat greater CR and NR ranges (Figure 2. 3). This was to be expected however, as the pelagic species observed were much larger in size, capturing the top portion of the food web that was unobserved in the mesopelagic samples. This may also reflect however, a wider feeding scope for the pelagic species that are thought to have wider spatial ranges than mesopelagic species (Helfman et al., 1997). The extreme points of TA were driven by unique functional groups, with (LAPE defined) small mesopelagic species representing the most depleted values and coastal species having some of the most enriched values. Again, this may likely represent some effect of sampling effort as well as the larger size classes observed in the predator surveys. Of particular note were strong differences in the carbon signatures of coastal pelagic predators, suggesting some degree of feeding specialization within the group.

5 ONTOGENETIC SHIFTS

Few species had sufficient samples across sizes to assess ontogenetic shifts, however one species, Thunnus atlanticus, demonstrated obvious differences among size classes - following predictions from the size-based analysis within each community. Similar to the functional group results (see below), observed δ15N-based trophic position and fitted regression-based trophic position estimates were made for the two size classes observed (Table 2. 3).

Table 2. 3 Ontogenetic shift estimates for blackfin tuna; mean [range] esitmates are calculated for individuals within each size class. Estimates include δ15N-estimated trophic position (TPn), community regression estimated TP (TPc), and the net difference between TPn and TPc (ΔTP).

Species TPn TPc ΔTP Thunnus atlanticus ~0.035 kg 3.05 [2.91, 3.05] 3.00 [3.00, 3.00] -0.05 Thunnus atlanticus ~2.32 kg 3.64 [3.38, 3.90] 3.16 [3.10, 3.22] -0.48

6 FUNCTIONAL GROUP RESULTS

The major objective of this analysis is to guide Ecopath with Ecosim (EwE) modelling of functional groups within the LAPE ecosystem. Therefore, given the spatial and size-based results above, species were allocated according to the functional groups as defined by LAPE technical reports 1SW and 2SW and presented in the context of for the a priori (FISHBASE-defined) community (pelagic or mesopelagic) to which they had been assigned (Trinidad samples were plotted independently). The comparison between previously assigned functional groups and the isotope-based community regressions provides a basis for assessing both species- and functional group-level trophic position.

64 15 For each species collected, the raw δ N vs. log2(weight) relationship was plotted with a line for the community specific regression. Along with summaries for each functional group, trophic position (TP) estimates were calculated for each species by assuming the regression intercept for each community equalled TP 3. Relative to this value, two estimates were calculated: TPn, the observed δ15N-based estimate for TP given the community intercept value and an assumed trophic step enrichment of 3.4‰; and TPc, the fitted weight-based estimate for δ15N using the regression equation for the specific community and an assumed trophic step enrichment of 3.4‰. TPc is generally more conservative as it is defined by the entire community and would be the same among species of the same weight. Also calculated for each species was the percent effect of its absence from the community on the mean distance to centroid (CD) value of Layman et al., (2007); i.e. the effect on community CD had the species not been observed. Positive values suggest (in isotopic terms) that the species is functionally redundant, while negative values suggest the species is functionally unique. Although there are few data for each species observed, the strength of the community-level relationships provides a reasonable basis for exploring differences among species. The following figures give the results of stable isotope analysis for species collected during the LAPE Project. Each figure includes the results for all species within a function group as indicated in the figure caption. The tabular results with each figure include the estimates of δ15N-estimated trophic position (TPn), community regression estimated TP (TPc), net difference between TPn and TPc (ΔTP), and the estimated effect on community trophic diversity (ΔTD).

65

Species TPn TPc ΔTP ΔTD Istiophorus albicans 4.21 [NA] 4.03 [NA] 0.177 0.013 Makaira nigricans 4.40 [4.17, 4.68] 4.21 [4.18, 4.26] 0.188 0.018 Billfish 4.35 [4.17, 4.68] 4.17 [4.03, 4.26] 0.185 -0.135 Figure 2. 4 Functional group billfish. Species are Istiophorus albicans (up triangle) and Makaira nigricans (diamond); origin countries are Barbados (black) and Tobago (red). Regression lines for pelagic community.

66

Species TPn TPc ΔTP ΔTD Thunnus atlanticus 3.40 [2.77, 4.22] 3.52 [3.30, 3.83] -0.118 -0.019

Figure 2. 5 Functional group blackfin tuna (Thunnus atlanticus) origin countries are Barbados (black), Grenada (red), Saint Kitts (green), and Saint Vincent (blue). Regression lines for pelagic community.

67

Species TPn TPc ΔTP ΔTD Caranx crysos 3.63 [3.58, 3.69] 3.19 [3.16, 3.21] 0.439 -0.008 Caranx lugbris 3.98 [NA] 3.73 [NA] 0.253 0.003 Carynx hippos 5.07 [4.51, 5.46] 3.71 [3.39, 3.88] 1.36 0.025 Eligatis bipinnulata 3.30 [3.26, 3.33] 3.68 [3.67, 3.68] -0.383 -0.003 Lobotes surinamensis 3.58 [3.44, 3.77] 3.74 [3.66, 3.80] -0.161 -0.006 Ocyurus chrysurus 3.96 [NA] 3.70 [NA] 0.256 0.000 Sphyraena barracuda 4.19 [NA] 3.86 [NA] 0.324 0.022 Lutjanus sp. 3.43 [3.33, 3.53] 3.07 [3.07, 3.07] 0.361 0.001 Coastal pelagic predators 4.21 [3.26, 5.46] 3.62 [3.07, 3.88] 0.597 -0.138 Figure 2. 6 Functional group coastal pelagic predators. Species are Caranx crysos (up triangle), Caranx lugubris (filled diamond), Carynx hippos (large filled circle), Eligatis bipinnulata (small filled circle), Lobotes surinamensis (open circle), Ocyurus chrysurus (open diamond), Sphyraena barracuda (open triangle) and Lutjanus sp. (open square); origin countries are Saint Vincent (red), Barbados (green), Saint Lucia (red), Grenada (orange), Trinidad (yellow), and Guadeloupe (blue). Regression lines for pelagic community(lower) and Trinidad (upper).

68

Species TPn TPc ΔTP ΔTD Coryphaena hippurus 4.33 [4.13, 4.46] 3.94 [3.93, 3.95] 0.392 0.015 Figure 2. 7 Functional group dolphinfish (Coryphaena hippurus) origin country is Barbados. Regression lines for pelagic community.

69

Species TPn TPc ΔTP ΔTD Cypselurus melanurus 3.84 [3.70, 4.04] 3.60 [3.59, 3.60] 0.246 -0.002 Hirundichthys affinis 3.73 [3.47, 3.98] 3.49 [3.47, 3.50] 0.234 0.013 Flyingfish 3.77 [3.47, 4.04] 3.53 [3.47, 3.60] 0.239 -0.182 Figure 2. 8 Functional group flyingfish. Species are Cypselurus melanurus (up triangle) and Hirundichthys affinis (diamond); origin country is Barbados. Regression lines for pelagic community.

70

Species TPn TPc ΔTP ΔTD Euphausiacea 3.51 [3.38, 3.78] 3.05 [3.00, 3.16] 0.452 -0.021 Sergestidae 3.31 [2.96, 3.76] 3.07 [3.00, 3.11] 0.248 -0.011 Krill 3.42 [2.96, 3.78] 3.06 [3.00, 3.16] 0.359 -0.210 Figure 2. 9 Functional group krill. Genera are Euphausiacea (up triangle) and Sergestidae (diamond); origin countries are Barbados (red), Martinique (orange), Saint Vincent (yellow), and Saint Lucia (green). Regression lines for the pelagic community.

71

Species TPn TPc ΔTP ΔTD Brama sp. 3.66 [2.65, 4.22] 3.24 [3.00, 3.62] 0.424 -0.008 Dactylopterus volitans 3.71 [2.80, 4.63] 3.31 [3.07, 3.56] 0.400 -0.004 Nemichthyidae 3.36 [1.89, 4.46] 3.20 [3.07, 3.32] 0.161 -0.006 Psenes arafurensis 3.55 [NA] 3.63 [NA] -0.085 -0.002 Psnes pellucidus 4.00 [NA] 3.81 [NA] 0.188 0.019 Large mesopelagics 3.56 [1.89, 4.63] 3.27 [3.00, 3.81] 0.287 -0.137 Figure 2. 10 Functional group large mesopelagics. Species are Brama sp. (up triangle), Dactlyopterus volitans (diamond), Nemichthyidae (large circle), Psnes arafurensis (small circle) and Psenes pellucidus (open circle); origin countries are Dominica (red), Barbuda (orange), Saint Lucia (yellow), Saint Vincent (green), and Martinique (teal). Regression lines for the mesopelagic community.

72

Species TPn TPc ΔTP ΔTD Scomberomorus regalis 3.79 [3.78, 3.81] 3.71 [3.71, 3.71] 0.084 -0.002 Scomberomorus brasiliensis * 5.18 [4.76, 5.51] 3.71 [3.59, 3.76] 1.47 NA** Scomberomorus cavalla 5.08 [4.93, 5.40] 3.74 [3.65, 3.86] 1.34 NA** Mackerels * 5.02 [3.78, 5.51] 3.72 [3.59, 3.86] 1.30 -0.171 Figure 2. 11 Functional group mackerels. Species are Scomberomorus regalis (up triangle), Scomberomorus brasiliensis (diamond), and Scomberomorus cavalla (circle); origin countries are Barbados (red) and Trinidad (orange). Regression lines for pelagic (lower) and Trinidad (upper) communities.

73

Figure 2. 12 Functional group other offshore predators. Species are Auxis thazard thazard (up triangle) and Canthidermis maculatus (circle); origin countries are Barbuda (red), Saint Kitts (orange), and Barbados (yellow). Regression lines for the pelagic community.

74

Species TPn TPc ΔTP ΔTD Acanthostracion polygonius 3.56 [NA] 3.62 [NA] -0.058 -0.013 Acanthurus bihianus 2.58 [NA] 3.00 [NA] -0.424 0.021 Acanthurus sp. 2.76 [2.71, 2.82] 3.00 [3.00, 3.00] -0.237 0.011 Cantherines pullus 2.74 [2.74, 2.74] 3.09 [3.07, 3.11] -0.351 0.011 Heteropriacanthus cruentatus 2.79 [2.77, 2.81] 3.03 [3.00, 3.07] -0.242 0.005 Lutjanus vivanus 3.19 [2.94, 3.64] 3.00 [3.00, 3.00] 0.190 -0.007 Polymyxia nobilis 4.06 [NA] 3.38 [NA] 0.675 0.121 Reef associated 3.09 [2.58, 4.06] 3.08 [3.00, 3.62] 0.003 -0.190

Figure 2. 13 Functional group reef associated. Species are Acanthostracion polygonius (up triangle), Acanthurus bahianus (diamond), Acanthurus sp. (large circle), Heteropriacanthus cruentatus (open circle), and Lutjanus vivanus (open square), Polymixia nobilis (open diamond). Origin countries are Grenada (red), Barbuda (orange), Saint Kitts (yellow), Saint Lucia (green), and Guadeloupe (teal). Regression lines for the pelagic community.

75

Species TPn TPc ΔTP ΔTD Caridea 3.05 [3.02, 3.09] 3.11 [3.11, 3.11] -0.056 -0.008 Oplophoridae 3.41 [2.84, 3.94] 3.08 [3.00, 3.16] 0.328 -0.024 Penaeidae 2.69 [2.64, 2.73] 3.07 [3.07, 3.07] -0.385 0.015 Shrimp 3.14 [2.64, 3.94] 3.09 [3.00, 3.16] 0.054 -0.189

Figure 2. 14 Functional group shrimp. Families are Caridea (up triangle), Oplophoridae (diamond), and Penaeidae (circle); origin countries are Barbuda (black), Grenada (red), and Martinique (green). Regression lines for the pelagic community.

76

Species TPn TPc ΔTP ΔTD Hemiramphus balao 3.19 [3.12, 3.25] 3.39 [3.37, 3.39] -0.201 -0.009 Selar crumenophthalmus 2.56 [2.56, 2.56] 3.20 [3.20, 3.20] -0.634 0.014 Small coastal pelagics 3.06 [2.56, 3.25] 3.35 [3.20, 3.39] -0.288 -0.162 Figure 2. 15 Functional group small coastal pelagics. Species are Hemiramphus balao (up triangle) and Selar crumenophthalmus (diamond); origin countries Barbados (red) and Barbuda (orange). Regression lines for the pelagic community.

77

Figure 2. 16 Functional group small mesopelagics. Species are too numerous to plot individually; origin countries are Martinique (black), Saint Vincent (red), Barbuda (green), Dominica (blue), Grenada (teal), Saint Lucia (pink), Guadeloupe (yellow), and Saint Kitts (black). Regression lines for the mesopelagic community. Species TPn TPc ΔTP ΔTD Acropomatidae 3.68 [3.47, 3.78] 3.16 [3.14, 3.18] 0.519 -0.014 Argyropelecus aculeatus 3.62 [2.39, 4.30] 3.20 [3.00, 3.31] 0.418 -0.013 Argyropelecus affinis 3.31 [NA] 3.00 [NA] 0.306 -0.005 Aristostomias sp. 3.99 [3.59, 4.30] 3.34 [3.32, 3.4] 0.651 0.003 Astronesthes sp. 3.90 [NA] 3.18 [NA] 0.716 -0.003 Bathophilus sp. 4.27 [4.00, 4.48] 3.32 [3.30, 3.33] 0.957 0.017 Bathylagidae 3.43 [3.26, 3.74] 3.23 [3.21, 3.26] 0.201 -0.012 Bethymyrinae 2.32 [2.09, 2.51] 3.21 [3.21, 3.21] -0.893 0.039 Benthodesmus nasutus 3.71 [NA] 3.11 [NA] 0.601 -0.011 Benthosema sp. 3.88 [NA] 3.16 [NA] 0.717 0.010 Bonapartia sp. 3.55 [3.49, 3.60] 3.11 [3.11, 3.11] 0.438 -0.013 Bregmaceros sp. 3.65 [NA] 3.00 [NA] 0.647 -0.017 Chaetodon ocellatus 3.97 [NA] 3.00 [NA] 0.967 -0.013 Chauliodontidae 3.40 [3.04, 4.18] 3.23 [3.18, 3.28] 0.172 -0.004 Chauliodus sloani 3.60 [3.44, 3.98] 3.35 [3.31, 3.39] 0.250 -0.008

78 Species TPn TPc ΔTP ΔTD Chiasmodontidae 3.29 [2.92, 3.81] 3.20 [3.14, 3.25] 0.087 -0.003 Cubiceps gracilis 3.46 [2.89, 3.95] 3.32 [3.18, 3.36] 0.145 -0.021 Cubiceps sp. 2.87 [2.74, 3.01] 3.17 [3.07, 3.20] -0.301 0.002 Decapterus tabl 3.28 [2.78, 3.54] 3.07 [3.00, 3.29] 0.206 -0.011 Diretmoides sp. 3.72 [3.62, 3.87] 3.29 [3.27, 3.32] 0.434 -0.013 Eustomias sp. 3.74 [NA] 3.11 [NA] 0.633 -0.009 Gonostomatidae 3.79 [3.27, 4.09] 3.24 [3.18, 3.29] 0.558 -0.007 Lepidocybium 3.24 [2.84, 3.48] 3.35 [3.18, 3.47] -0.112 -0.015 flavobrunneum Leptostomias sp. 3.46 [3.36, 3.51] 3.28 [3.28, 3.30] 0.176 -0.008 Lestidiops sp. 3.23 [NA] 3.18 [NA] 0.049 0.000 Malacosteidae 4.27 [3.86, 4.75] 3.47 [3.42, 3.54] 0.803 0.018 Manducus maderensis 3.84 [NA] 3.30 [NA] 0.533 -0.007 Myctophidae 4.10 [3.53, 4.50] 3.13 [3.00, 3.24] 0.971 0.007 Neonesthes sp. 4.25 [3.58, 4.61] 3.34 [3.23, 3.40] 0.912 0.015 Nesiarchus nasutus 2.69 [2.49, 2.87] 3.26 [3.22, 3.40] -0.573 0.031 Nomeus gronovii 3.07 [2.81, 3.25] 3.03 [3.00, 3.14] 0.033 0.012 Notosudidae 3.51 [NA] 3.07 [NA] 0.439 -0.011 Odontostomops sp. 3.75 [3.56, 4.09] 3.17 [3.00, 3.27] 0.584 -0.007 Pollichthys sp. 2.98 [2.81, 3.15] 3.25 [3.20, 3.30] -0.267 0.014 Polydactylus sp. 3.69 [3.63, 3.79] 3.09 [3.00, 3.14] 0.593 -0.010 Promethichthys 2.83 [2.64, 2.79] 3.23 [3.20, 3.30] -0.398 0.008 prometheus Scombrolabrax heterolepis 3.48 [2.98, 4.20] 3.27 [3.16, 3.37] 0.207 -0.013 Scopelarchus sp. 3.54 [NA] 3.11 [NA] 0.424 -0.016 Scopelosaurus sp. 3.29 [NA] 3.16 [NA] 0.129 -0.004 Serrivomeridae 3.82 [3.52, 4.06] 3.27 [3.21, 3.39] 0.545 -0.006 Stenotomus caprinus 3.35 [NA] 3.00 [NA] 0.347 0.045 Sternoptyx sp. 3.40 [3.27, 3.48] 3.10 [3.07, 3.11] 0.303 -0.008 Small mesopelagics 3.55 [2.09, 4.75] 3.22 [3.00, 3.54] 0.331 -0.035

79

Species TPn TPc ΔTP ΔTD Enoploteuthis sp. 3.27 [3.18, 3.36] 3.00 [3.00, 3.00] 0.269 -0.003 Histioteuthis dolfleini 3.59 [3.11, 4.06] 3.38 [3.20, 3.56] 0.211 -0.001 Hyaloteuthis pelagica 3.20 [3.06, 3.34] 3.28 [3.22, 3.33] -0.074 0.001 Loligo plei 3.86 [3.75, 4.01] 3.19 [3.11, 3.29] 0.677 -0.006 Lycoteuthis sp. 3.04 [2.67, 3.41] 3.08 [3.00, 3.23] -0.040 0.014 Omnastrephes bartramii 3.93 [3.70, 4.40] 3.46 [3.36, 3.63] 0.470 -0.000 Onycoteuthis sp. 2.77 [NA] 3.16 [NA] -0.392 0.028 Ornithoteuthis antillarum 3.19 [3.12, 3.25] 3.39 [3.37, 3.39] -0.201 -0.009 Spriula spirula 2.56 [2.56, 2.56] 3.20 [3.20, 3.20] -0.634 0.014 Squid 3.17 [2.56, 3.54] 3.19 [3.00, 3.39] -0.023 -0.163 Figure 2. 17 Functional group squid. Species are Enoploteuthis sp. (closed up triangle), Histioteuthis dolfleini (closed diamond), Hyaloteuthis pelagica (large circle), Loligo plei (small closed circle), Lycoteuthis sp. (open circle), Ornithoteuthis sp. (open square), Ornithoteuthis antillarum (open diamond), and Spriula spirula (open up triangle); origin countries are Martinique (black), Barbuda (red), Saint Lucia (green), Grenada (blue), and Saint Vincent (teal). Regression lines for the mesopelagic community.

80

Species TPn TPc ΔTP ΔTD Xiphias gladius 4.35 [NA] 4.10 [NA] 0.252 0.022 Figure 2. 18 Functional group swordfish.(Xiphias gladius) origin country is Saint Kitts. Regression lines for the pelagic community.

81

Species TPn TPc ΔTP ΔTD Acanthocybium solandri 3.91 [3.40, 4.26] 3.91 [3.84, 3.97] -0.003 -0.004 Figure 2. 19 Functional group Wahoo (Acanthocybium solandri) origin countries are Barbados (red) and Guadelouope (green). Regression lines for the pelagic community.

82 Table 10 Summary of stable isotope results for functional groups in the LAPE project. TPn is the trophic position indicated by nitrogen isotopes, TPc is the expected trophic position based on size and community regression. ΔTP is the difference between TPn and TPc and ΔTD is change in community trophic diversity due to the group. Functional Group / Species TPn TPc ΔTP ΔTD Swordfish 4.35 [NA] 4.10 [NA] 0.252 0.022 Xiphias gladius Billfish 4.35 [4.17, 4.68] 4.17 [4.03, 4.26] 0.185 -0.135 Istiophorus albicans 4.21 [NA] 4.03 [NA] 0.177 0.013 Makaira nigricans 4.40 [4.17, 4.68] 4.21 [4.18, 4.26] 0.188 0.018 Blackfin Tuna 3.40 [2.77, 4.22] 3.52 [3.30, 3.83] -0.118 -0.019 Thunnus atlanticus Other Offshore Predators Canthidermis maculatus 3.03 [2.84, 3.24] 3.66 [3.63, 3.69] -0.628 -0.001 Auxis thazard thazard 2.67 [2.53, 2.85] 3.25 [3.14, 3.33] -0.575 0.009 Mackerels 5.02 [3.78, 5.51] 3.72 [3.59, 3.86] 1.30 -0.171 Scomberomorus regalis 3.79 [3.78, 3.81] 3.71 [3.71, 3.71] 0.084 -0.002 Scomberomorus brasiliensis * 5.18 [4.76, 5.51] 3.71 [3.59, 3.76] 1.47 NA** Scomberomorus cavalla 5.08 [4.93, 5.40] 3.74 [3.65, 3.86] 1.34 NA** Wahoo 3.91 [3.40, 4.26] 3.91 [3.84, 3.97] -0.003 0.004 Acanthocybium solandri Dolphinfish 4.33 [4.13, 4.46] 3.94 [3.93, 3.95] 0.392 0.015 Coryphaena hippurus Flyingfish 3.77 [3.47, 4.04] 3.53 [3.47, 3.60] 0.239 -0.182 Hirundichthys affinis 3.73 [3.47, 3.98] 3.49 [3.47, 3.50] 0.234 0.013 Cypselurus melanurus 3.84 [3.70, 4.04] 3.60 [3.59, 3.60] 0.246 -0.002 Coastal pelagic predators 4.21 [3.26, 5.46] 3.62 [3.07, 3.88] 0.597 -0.138 Ocyurus chrysurus 3.96 [NA] 3.70 [NA] 0.256 0.000 Caranx crysos 3.63 [3.58, 3.69] 3.19 [3.16, 3.21] 0.439 -0.012 Caranx lugbris 3.98 [NA] 3.73 [NA] 0.253 0.002 Caranx hippos 5.07 [4.51, 5.46] 3.71 [3.39, 3.88] 1.360 NA Sphyraena barracuda 4.19 [NA] 3.86 [NA] 0.324 0.022 Lutjanus sp. 3.43 [3.33, 3.53] 3.07 [3.07, 3.07] 0.361 0.001 Elegatis bipinnulata 3.30 [3.26, 3.33] 3.68 [3.67, 3.68] -0.383 -0.008 Lobotes surinamensis 3.58 [3.44, 3.77] 3.74 [3.66, 3.80] -0.161 -0.010 Small coastal pelagics 3.06 [2.56, 3.25] 3.35 [3.20, 3.39] -0.288 -0.162 Decapterus tabl 3.28 [2.78, 3.54] 3.07 [3.00, 3.29] 0.206 -0.011 Hemiramphus balao 3.19 [3.12, 3.25] 3.39 [3.37, 3.39] -0.201 -0.009 Selar crumenophthalmus 2.56 [2.56, 2.56] 3.20 [3.20, 3.20] -0.634 0.014 Reef Associated 3.09 [2.58, 4.06] 3.08 [3.00, 3.62] 0.003 -0.190 Acanthostracion polygonius 3.56 [NA] 3.62 [NA] -0.058 -0.013 Polymixia nobilis 4.06 [NA] 3.38 [NA] 0.675 0.008 Acanthurus bahianus 2.58 [NA] 3.00 [NA] -0.424 0.022 Acanthurus sp. 2.76 [2.71, 2.82] 3.00 [3.00, 3.00] 0.519 0.011 Cantherines pullus 2.74 [2.74, 2.74] 3.09 [3.07, 3.11] -0.351 0.011 Heteropriacanthus cruentatus 2.79 [2.77, 2.81] 3.03 [3.00, 3.07] -0.242 0.005 Lutjanus vivanus 3.58 [3.44, 3.77] 3.74 [3.66, 3.80] -0.161 -0.010

83 Functional Group / Species TPn TPc ΔTP ΔTD Small mesopelagics 3.55 [2.09, 4.75] 3.22 [3.00, 3.54] 0.331 -0.035 Acropomatidae 3.68 [3.47, 3.78] 3.16 [3.14, 3.18] 0.519 -0.014 Argyropelecus aculeatus 3.62 [2.39, 4.30] 3.20 [3.00, 3.31] 0.418 -0.013 Argyropelecus affinis 3.31 [NA] 3.00 [NA] 0.306 -0.005 Aristostomias sp. 3.99 [3.59, 4.30] 3.34 [3.32, 3.4] 0.651 0.003 Astronesthes sp. 3.90 [NA] 3.18 [NA] 0.716 -0.003 Bathophilus sp. 4.27 [4.00, 4.48] 3.32 [3.30, 3.33] 0.957 0.017 Bathylagidae 3.43 [3.26, 3.74] 3.23 [3.21, 3.26] 0.201 -0.012 Bethymyrinae 2.32 [2.09, 2.51] 3.21 [3.21, 3.21] -0.893 0.039 Benthodesmus nasutus 3.71 [NA] 3.11 [NA] 0.601 -0.011 Benthosema sp. 3.88 [NA] 3.16 [NA] 0.717 0.010 Bonapartia sp. 3.55 [3.49, 3.60] 3.11 [3.11, 3.11] 0.438 -0.013 Bregmaceros sp. 3.65 [NA] 3.00 [NA] 0.647 -0.017 Chaetodon ocellatus 3.97 [NA] 3.00 [NA] 0.967 -0.013 Chauliodontidae 3.40 [3.04, 4.18] 3.23 [3.18, 3.28] 0.172 -0.004 Chauliodus sloani 3.60 [3.44, 3.98] 3.35 [3.31, 3.39] 0.250 -0.008 Chiasmodontidae 3.29 [2.92, 3.81] 3.20 [3.14, 3.25] 0.087 -0.003 Cubiceps gracilis 3.46 [2.89, 3.95] 3.32 [3.18, 3.36] 0.145 -0.021 Cubiceps sp. 2.87 [2.74, 3.01] 3.17 [3.07, 3.20] -0.301 0.002 Dactylopterus volitans 3.71 [2.80, 4.63] 3.31 [3.07, 3.56] 0.400 -0.012 Decapterus tabl 3.28 [2.78, 3.54] 3.07 [3.00, 3.29] 0.206 -0.011 Diretmoides sp. 3.72 [3.62, 3.87] 3.29 [3.27, 3.32] 0.434 -0.013 Eustomias sp. 3.74 [NA] 3.11 [NA] 0.633 -0.009 Gonostomatidae 3.79 [3.27, 4.09] 3.24 [3.18, 3.29] 0.558 -0.007 Lepidocybium flavobrunneum 3.24 [2.84, 3.48] 3.35 [3.18, 3.47] -0.112 -0.015 Leptostomias sp. 3.46 [3.36, 3.51] 3.28 [3.28, 3.30] 0.176 -0.008 Lestidiops sp. 3.23 [NA] 3.18 [NA] 0.049 0.000 Malacosteidae 4.27 [3.86, 4.75] 3.47 [3.42, 3.54] 0.803 0.018 Manducus maderensis 3.84 [NA] 3.30 [NA] 0.533 -0.007 Myctophidae 4.10 [3.53, 4.50] 3.13 [3.00, 3.24] 0.971 0.007 Nemichthyidae 3.36 [1.89, 4.46] 3.20 [3.07, 3.32] 0.161 -0.008 Neonesthes sp. 4.25 [3.58, 4.61] 3.34 [3.23, 3.40] 0.912 0.015 Nesiarchus nasutus 2.69 [2.49, 2.87] 3.26 [3.22, 3.40] -0.573 0.031 Nomeus gronovii 3.07 [2.81, 3.25] 3.03 [3.00, 3.14] 0.033 0.012 Notosudidae 3.51 [NA] 3.07 [NA] 0.439 -0.011 Odontostomops sp. 3.75 [3.56, 4.09] 3.17 [3.00, 3.27] 0.584 -0.007 Pollichthys sp. 2.98 [2.81, 3.15] 3.25 [3.20, 3.30] -0.267 0.014 Polydactylus sp. 3.69 [3.63, 3.79] 3.09 [3.00, 3.14] 0.593 -0.010 Promethichthys prometheus 2.83 [2.64, 2.79] 3.23 [3.20, 3.30] -0.398 0.008 Scombrolabrax heterolepis 3.48 [2.98, 4.20] 3.27 [3.16, 3.37] 0.207 -0.013 Scopelarchus sp. 3.54 [NA] 3.11 [NA] 0.424 -0.016 Scopelosaurus sp. 3.29 [NA] 3.16 [NA] 0.129 -0.004 Serrivomeridae 3.82 [3.52, 4.06] 3.27 [3.21, 3.39] 0.545 -0.006 Stenotomus caprinus 3.35 [NA] 3.00 [NA] 0.347 0.045 Sternoptyx sp. 3.40 [3.27, 3.48] 3.10 [3.07, 3.11] 0.303 -0.008 Large Mesopelagics 3.56 [1.89, 4.63] 3.27 [3.00, 3.81] 0.287 -0.137 Dactylopterus volitans 3.71 [2.80, 4.63] 3.31 [3.07, 3.56] 0.400 -0.012 Nemichthyidae 3.36 [1.89, 4.46] 3.20 [3.07, 3.32] 0.161 -0.008 Psenes arafurensis 3.55 [NA] 3.63 [NA] -0.085 -0.002 Psenes pellucidus 4.00 [NA] 3.81 [NA] 0.188 0.019 Brama sp. 3.66 [2.65, 4.22] 3.24 [3.00, 3.62] 0.424 -0.013

84 Functional Group / Species TPn TPc ΔTP ΔTD Squid 3.17 [2.56, 3.54] 3.19 [3.00, 3.39] -0.023 -0.163 Enoploteuthis sp. 3.27 [3.18, 3.36] 3.00 [3.00, 3.00] 0.269 -0.003 Histioteuthis dolfleini 3.59 [3.11, 4.06] 3.38 [3.20, 3.56] 0.211 -0.001 Hyaloteuthis pelagica 3.20 [3.06, 3.34] 3.28 [3.22, 3.33] -0.074 0.001 Loligo plei 3.86 [3.75, 4.01] 3.19 [3.11, 3.29] 0.677 -0.006 Lycoteuthis sp. 3.04 [2.67, 3.41] 3.08 [3.00, 3.23] -0.040 0.014 Onycoteuthis sp. 2.77 [NA] 3.16 [NA] -0.392 0.028 Ornithoteuthis antillarum 3.19 [3.12, 3.25] 3.39 [3.37, 3.39] -0.201 -0.009 Spriula spirula 2.56 [2.56, 2.56] 3.20 [3.20, 3.20] -0.634 0.014 Ommastrephes bartramii 3.93 [3.70, 4.40] 3.46 [3.36, 3.63] 0.470 0.000 Krill 3.42 [2.96, 3.78] 3.06 [3.00, 3.16] 0.359 -0.210 Euphausiacea 3.51 [3.38, 3.78] 3.05 [3.00, 3.16] 0.452 -0.021 Sergestidae 3.31 [2.96, 3.76] 3.07 [3.00, 3.11] 0.248 -0.011 Shrimp 3.14 [2.64, 3.94] 3.09 [3.00, 3.16] 0.054 -0.189 Caridea 3.05 [3.02, 3.09] 3.11 [3.11, 3.11] -0.056 -0.008 Oplophoridae 3.41 [2.84, 3.94] 3.08 [3.00, 3.16] 0.328 -0.024 Penaeidae 2.69 [2.64, 2.73] 3.07 [3.07, 3.07] -0.385 0.015

85 7 REFERENCES

Adams, T. S. and Sterner, R. W. (2000). The effect of dietary nitrogen content on trophic level 15N enrichment. and Oceanography, 45(3): 601– 607. Arai, M. (1988). Interactions of fish and pelagic coelenterates. Canadian Journal of Zoology, 66: 1913–1927. Arrington, D. A. and Winemiller, K. O. (2002). Preservation effects on stable isotope analysis of fish muscle. Transactions of the American Fisheries Society, 131: 337–342. Au, D. and Pitman, R. (1986). interactions with dolphins and tuna in the eastern tropical pacific. The Condor, 88: 304–317. Ben-David, M., Hanley, T., Kline, D., and Schell, D. (1997). Seasonal changes in diets of coastal and riverine mink: the role of spawning pacific salmon. Canadian Journal of Zoology, 75: 803-811. Ben-David, M. and Schell, D. (2001). Mixing models in analyses of diet using multiple stable isotopes: a response. Oecologia, 127: 180–184. Ben-David, M., Titus, K., and LaVern, R. (2004). Consumption of salmon by Alaskan brown bears: a trade-off between nutritional requirements and the risk of infanticide? Oecologia, 138: 465–744. Bilby, R., Fransen, B., and Bisson, P. (1996). Incorporation of nitrogen and carbon from spawning coho salmon into the trophic system of small streams: evidence from stable isotopes. Canadian Journal of Fisheries and Aquatic Sciences, 53: 164–173. Bosley, K. L., Whitting, D. A., Chambers, R. C., and Wainright, S. C. (2002). Estimating turnover rates of carbon and nitrogen in recently metamorphosed winter flounder pseudopleuronectes americanus with stable isotopes. Marine Ecology-Progress Series, 236: 233–240. Brodeur, R., Sugisaki, H., and Hunt Jr., G. (2002). Increases in jellyfish biomass in the Bering Sea: implications for the ecosystem. Marine Ecology-Progress Series, 233: 89–103. Buck, E. (1997). Dolphin protection and tuna seining. Report 97-588 ENR, Congressional Research Service, Washington. Bunn, S., Loneragan, N., and Kempster, M. (1995). Effects of acid washing on stable isotope ratios of C and N in penaeid shrimp and seagrass: implications for food-web studies using multiple stable isotopes. Limnology and Oceanography, 40(3): 622–625. Cerling, T. and Harris, J. (1999). Carbon isotope fractionation between diet and bioapatite in ungulate mammals and implications for ecological and paleontological studies. Oecologia, 120: 347–363. Christensen, V. (1995). Ecosystem maturity - towards quantification. Ecological Modelling, 77: 3–32.

86 Cohen, J., Pimm, S. L., Yodzis, P., and Saldana, J. (1993). Body sizes of animal predators and animal prey in food webs. Journal of Animal Ecology, 62: 67–78. Das, K., Lepoint, G., Loizeau, V., Debacker, V., Dauby, P., and Bouquegeneau, J. (2000). Tuna and dolphin associations in the northeast Atlantic: evidence of different ecological niches from stable isotope and heavy metal measurements. Marine Pollution Bulletin, 40(2): 102–109. Deb, D. (1997). Trophic uncertainty vs parsimony in food web research. Oikos, 78: 191–194. DeNiro, M. and Epistein, S. (1977). Mechanism of carbon isotope fractionation associated with lipid synthesis. Science, 197: 261–263. DeNiro, M. J. and Epstein, S. (1981). Influence of diet on the distribution of nitrogen isotopes in animals. Geochimica et Cosmochimica Acta, 45: 341– 351. Dickie L.M., Kerr, S.R., and Boudreau, P.R. (1987) Size-dependent processes underlying regularities in ecosystem structure. Ecological Monographs, 57: 233-250. Duffy, J., MacDonald, K., Rhode, J., and Parker, J. (2001). Grazer diversity, functional redundancy, and in seagrass beds: an experimental test. Ecology, 82(9): 2417–2434. Eadie, B. J. and Jeffery, L. M. (1973). ∂13C analyses of oceanic particulate organic matter. Marine Chemistry, 1: 199–209. Edwards, M. S., Turner, T. F., and Sharp, Z. D. (2002). Short- and long-term effects of fixation and preservation on stable isotope values (∂13C, ∂15N, ∂34S) of fluid-preserved museum specimens. Copeia, 2002(4): 1106–1112. Ehleringer, J. R. and Rundel, P. W. (1989). Chapter 1. Stable isotopes: History, units, and instrumentation. In Rundel, P. W. and Nagy, K. A., editors, Stable isotopes in ecological research, 1-15 pg. Springer-Verlag, Berlin. Elton, C. S. (1927). Animal Ecology. Sidgwick and Jackson, London. Felicetti, L., Schwartz, C., Rye, R., Haroldson, M.A., Gunther, K., Phillips, D., and Robbins, C. (2003). Use of sulfur and nitrogen stable isotopes to determine the importance of whitebark pine nuts to Yellowstone grizzly bears. Canadian Journal of Zoology, 81: 763-770. Focken, U. and Becker, K. (1998). Metabolic fractionation of stable carbon isotopes: implications of different proximate compositions for studies of the aquatic food webs using ∂13C data. Oecologia, 115: 337–343. Fogarty, M. and Murawski, S. (1998). Large-scale disturbance and the structure of marine systems: fishery impacts on Georges Bank. Ecological Applications, 8(1): S6–S22. Forget, M.-H. (2007) Phytoplankton community and primary production in Caribbean waters: The biological oceanography component of the LAPE project. LAPE Technical Document 5.

87 France, R. L. (1995). Carbon-13 enrichment in benthic compared to planktonic algae: food web implications. Marine Ecology-Progress Series, 124: 307– 312. Fry, B. and Sherr, E. B. (1989). Chapter 12 ∂13C measurements as indicators of carbon flow in marine and freshwater ecosystems. In Rundel, P. W. and Nay, K. A., editors, Stable isotopes in ecological research, 197–229 pg. Springer-Verlag, Berlin. Gaebler, O. H., Vitti, T. G., and Vukmirovich, R. (1966). Isotope effects in metabolism of 14N and 15n from unlabeled dietary proteins. Canadian Journal of Biochemistry, 44: 1249–1257. Gannes, L., Martinez del Rio, C., and Koch, P. (1998). Natural abundance variations in stable isotopes and their potential uses in animal physiological ecology. Comparative Biochemistry and Physiology A, 119(3): 725–737. Gaye-Siessegger, J., Focken, U., Muetzel, S., Abel, H., and Becker, K. (2004). Feeding level and individual metabolic rate affect ∂13C and ∂15N values in carp: implications for food web studies. Oecologia, 138: 175–183. Goericke, R. and Fry, B. (1994). Variations of marine plankton ∂13C with latitude, temperature, and dissolved CO2 in the world ocean. Global Biogeochemical Cycles, 8(1): 85–90. Hall, R. and Meyer, J. (1998). The trophic significance of bacteria in a detritus- based stream food web. Ecology, 79(6): 1995–2012. Heikoop, J., Dunn, J., Risk, M., Tomascik, T., Schwarcz, H., Sandeman, I., and Sammarco, P. (2000). ∂15N and ∂13C of coral tissue show significant inter- reef variation. Coral Reefs, 19: 189–193. Helfield, J. and Naiman, R. (2001). Effects of salmon-derived nitrogen on riparian forest growth and implications for stream productivity. Ecology, 82(9): 2403–2409. Helfman, G.S., Collette, B.B. and Facey, D.E. (1997) The Diversity of Fishes. Blackwell Publishing, Malden, USA. Herzka, S. and Holt, G. (2000). Changes in isotopic composition of red drum (Sciaenops ocellatus) larvae in response to dietary shifts: potential applications to settlement studies. Canadian Journal of Fisheries and Aquatic Sciences, 57: 137–147. Hesslein, R. H., Hallard, K. A., and Ramlal, P. (1993). Replacement of sulfur, carbon, and nitrogen in tissue of growing broad whitefish (Coregonus nasus) in response to a change in diet traced by ∂34S, ∂13C, and ∂15N. Canadian Journal of Fisheries and Aquatic Sciences, 50: 2071–2076. Hobson, K. A. and Clark, R. G. (1992). Assessing avian diets using stable isotopes ii: factors influencing diet-tissue fractionation. The Condor, 94: 189–197. Hobson, K. A. and Welch, H. E. (1992). Determination of trophic relationships within a high arctic marine food web using ∂13C and ∂15N analysis. Marine Ecology-Progress Series, 84: 9–18.

88 Hofmann, M., Wolf-Gladrow, D., Takahashi, T., Sutherland, S., Six, K., and Maier-Reimer, E. (2000). Stable carbon isotope distribution of particulate organic matter in the ocean: a model study. Marine Chemistry, 72: 1311- 150. Jennings, S., Greenstreet, S., Hill, L., Piet, G., Pinnegar, J., and Warr, K. (2002a). Long-term trends in the trophic structure of the north sea fish community: evidence from stable-isotope analysis, size-spectra and community metrics. Marine , 141: 1085–1097. Jennings, S., Pinnegar, J., Polunin, N., and Boon, T. (2001). Weak cross-species relationships between body size and trophic level belie powerful size- based trophic structuring in fish communities. Journal of Animal Ecology, 70: 934–944. Jennings, S. and Warr, K. (2003a). Smaller predator-prey body size ratios in longer food chains. Proceedings of the Royal Society of London, B, 270: 1413–1417. Jennings, S. and Warr, K. J. (2003b). Environmental correlates of large-scale spatial variation in the ∂15N of marine animals. , 142: 1131– 1140. Jennings, S., Warr, K. J., and Mackinson, S. (2002b). Use of size-based production and stable isotope analyses to predict trophic transfer efficiencies and predator-prey body mass ratios in food webs. Marine Ecology-Progress Series, 240: 11–20. Kaehler, S. and Pakhomov, E. (2001). Effects of storage and preservation on the ∂13C and ∂15N signature of selected marine organisms. Marine Ecology- Progress Series, 219: 299–304. Kerr, S.R. (1974) Theory of size distribution in ecological communities. Journal of the Fisheries Research Board of Canada 31: 1850-1862. Kline, T.C. Goering, J., Mathisen, O., Poe, P., and Parker, P. (1993). Recycling of elements transported upstream by runs of pacific salmon: ii. ∂15N and ∂13C evidence in the Kvichakk River watershed, Bristol Bay, south-western Alaska. Canadian Journal of Fisheries and Aquatic Sciences, 50: 2350–2365. Kline, T., Goering, J., Mathisen, O., and Poe, P. (1990). Recycling of elements transported upstream by runs of pacific salmon: 1. ∂15N and ∂13C evidence in Sashin Creek, south-eastern Alaska. Canadian Journal of Fisheries and Aquatic Sciences, 47: 136–144. Lawson, J. and Hobson, K. (2000). Diet of harp seals (Pagophilus groenlandicus) in nearshore northeast Newfoundland: inferences from stable-carbon (13C) and nitrogen (15N) isotope analysis. Science, 16: 578–591. Lorrain, A., Paulet, Y.M., Chauvaud, L., Savoye, N., Donval, A., and Saout, C. (2002). Differential ∂13C and ∂15N signatures among scallop tissues: implications for ecology and physiology. Journal of Experimental Marine Biology and Ecology, 275: 47–61. Lubetkin, S. and Simenstad, C. (2004). Multi-source mixing models to quantify food web sources and pathways. Journal of , 41: 996–1008.

89 MacAvoy, S. E., Macko, S. A., and Garman, G. C. (2001). Isotopic turnover in aquatic predators: quantifying the exploitation of migratory prey. Canadian Journal of Fisheries and Aquatic Sciences, 58: 923–932. MacNeil, M., Drouillard, K., and Fisk, A. (2005a). Variable uptake and elimination of stable nitrogen isotopes. IN REVIEW. MacNeil, M., Skomal, G., and Fisk, A. (2005b). Tissue types reflect trophic inferences from stable isotopes in sharks. IN REVIEW. Marcogliese, D. J. (2001). Pursuing parasites up the food chain: Implications of food web structure and function on parasite communities in aquatic systems. Acta Parasitologica, 46(2): 82–93. McClelland, J. W. and Montoya, J. P. (2002). Trophic relationships and the nitrogen isotopic composition of amino acids in plankton. Ecology, 83(3): 2173–2180. McKinney, C., McRea, I., Epistein, S., Allen, H., and Urey, H. (1950). Improvements in mass spectrometers for measurement of small differences in isotope abundance ratios. Review of Scientific Instruments, 21: 724–730. Melville, A. and Connolly, R. (2003). Spatial analysis of stable isotope data to determine primary sources of nutrition for fish. Oecologia, 136: 499–507. Minagawa, M. and Wada, E. (1984). Stepwise enrichment of 15N along food chains: Further evidence and the relation between ∂15N and animal age. Geochemica et Cosmochimica Acta, 48: 1135–1140. Nagelkerken, I. and van der Velde, G. (2004a). Are Caribbean mangroves important feeding grounds for juvenile reef fish from adjacent seagrass beds? Marine Ecology-Progress Series, 274: 143-151. Nagelkerken, I. and van der Velde, G. (2004b). Relative importance of interlinked mangroves and seagrass beds as feeding habitats for juvenile reef fish on a Caribbean island. Marine Ecology-Progress Series, 274: 153–159. Naiman, R., Bilby, R., Schindler, D., and Helfield, J. (2002). Pacific salmon, nutrients, and the dynamics of freshwater and riparian ecosystems. Ecosystems, 57: 399–417. Odum, W. E. and Heald, E. J. (1975). The detritus-based food web of an estuarine mangrove community. Estuarine Research: 1(1): 256–286. Overholtz, W. (2002). The Gulf of Maine-Georges Bank Atlantic herring (Clupea harengus): spatial pattern analysis of the collapse and recovery of a large marine fish complex. Fisheries Research, 57: 237–254. Owens, N. J. P. (1987). Natural variations in 15N in the marine environment. Advances in Marine Biology, 24: 389–451. Pauly, D., Christensen, V., Dalsgaard, J., Froese, R., and Torres, F. (1998). Fishing down marine food webs. Science, 279: 860–863. Perrin, W., Warner, R., Fiscus, C., and Holts, D. (1973). Stomach contents of porpoise, Stenella sp., and yellowfin tuna, Thunnus albacares in mixed species aggregations. Fishery Bulletin, 71: 1077–1091.

90 Peterson, B. (1999). Stable isotopes as tracers of organic matter input and transfer in benthic food webs: a review. Acta Oecologica, 20(4): 479–487. Peterson, B. and Howarth, R. (1987). Sulfur, carbon, and nitrogen isotopes used to trace organic matter flow in the salt-marsh estuaries of Saplelo island, Georgia. Limnology and Oceanography, 32: 1195–1213. Peterson, B., Wolheim, W., Mulholland, P., Webster, J., Meyer, J., Tank, J., Marti, E., Bowden, W., Valett, H., Hershey, A., McDowell, W., Dodds, W., Hamilton, S., Gregory, S., and Morrall, D. (2001). Control of nitrogen export from watersheds by headwater streams. Science, 292: 86–90. Peterson, B. J. and Fry, B. (1987). Stable isotopes in ecosystem studies. Annual Reviews in Ecology and Systematics, 18: 293–320. Phillips, D. and Gregg, J. (2003). Source partitioning using stable isotopes: coping with too many sources. Oecologia, 136: 261–269. Phillips, D. and Koch, P. (2002). Incorporating concentration dependence in stable isotope mixing models. Oecologia, 130: 114–125. Phillips, D. L. (2001). Mixing models in analyses of diet using multiple stable isotopes: a critique. Oecologia, 127: 166–170. Pinnegar, J., Polunin, N., and Badalamenti, F. (2003). Long-term changes in the trophic level of western Mediterranean fishery and aquaculture landings. Canadian Journal of Fisheries and Aquatic Sciences, 60: 222–235. Pinnegar, J. K. and Polunin, N. V. C. (1999). Differential fractionation of ∂13C and ∂15N among fish tissues: implications for the study of trophic interactions. Functional Ecology, 13: 225–231. Pinnegar, J. K. and Polunin, N. V. C. (2000). Contributions of stable-isotope data to elucidating food webs of Mediterranean rocky littoral fishes. Oecologia, 122(3): 399–409. Polunin, N., Morales-Nin, B., Pawsey, W., Cartes, J., Pinnegar, J., and Moranta, J. (2001). Feeding relationships in the Mediterranean bathyal assemblages elucidated by stable nitrogen and carbon isotope data. Marine Ecology Progress Series, 220: 13–23. Polunin, N. and Pinnegar, J. (2002). Chapter 14 Trophic ecology and the structure of marine food webs. In Hart, P. and Reynolds, J., editors, Handbook of Fish and Fisheries, Volume 1 301–320 pg. Blackwell Science Ltd., Oxford. Ponsard, S. and Averbuch, P. (1999). Should growing and adult animals fed on the same diet show different ∂15N values? Rapid Communications in Mass Spectrometry, 13: 1305–1310. Post, D. L. (2002). Using stable isotopes to estimate trophic position: models, methods, and assumptions. Ecology, 83(3): 703–718. Post, D.M., Layman, C.A., Albrey Arrington, D., Takimoto, G. Quattrochi, J. and Montana, C.G. (2007) Getting to the fat of the matter: models, methods and assumptions for dealing with lipids in stable isotope analysis. Ocecologia 152: 179-189.

91 Power, M., Guiguer, K., and Barton, D. (2003). Effects of temperature on isotopic enrichment in Daphnia magna: implications for aquatic food-web studies. Rapid Communications in Mass Spectrometry, 17: 1619–1625. Rau, G., Mearns, A., Young, D., Olson, R., Schafer, H., and Kaplan, I. (1983). Animal 13C/12C correlates with trophic level in pelagic food webs. Ecology, 64(5): 1314–1318. Ricciardi, A. and MacIsaac, H. (2000). Recent mass invasion of the North American great lakes by Ponte-Caspian species. Trends in Ecology and Evolution, 15(2): 62–65. Rounick, J. and Winterbourn, M. (1986). Stable carbon isotopes and carbon flow in ecosystems. BioScience, 36(3): 171–177. Sarakinos, H., Johnson, M., and Vander Zanden, M. (2002). A synthesis of tissue- preserved effects on carbon and nitrogen stable isotope signatures. Canadian Journal of Zoology, 80: 381–387. Satterfield, F. and Finney, B. (2002). Stable isotope analysis of Pacific salmon: insight into trophic status and oceanographic conditions over the last 30 years. Progress in Oceanography, 53: 231–246. Schmidt, H. L. (2003). Fundamentals and systematics of the non-statistical distributions of isotopes in natural compounds. Naturwissenschaften, 90: 537–552. Schneider, K., Migge, S., Norton, R., Scheu, S., Langel, R., Reineking, A., and Maraun, M. (2004). Trophic niche differentiation in soil microarthropods (oribatida, acari): evidence from stable isotope ratios (15N/14N). Soil Biology and Biochemistry, 36: 1769–1774. Schoeller, D. A. (1999). Isotope fractionation: why aren’t we what we eat? Journal of Archaelogical Science, 26: 667–673. Sotiropoulos, M., Tonn, W., and Wassenaar, L. (2004). Effects of lipid extraction on stable carbon and nitrogen isotope analyses of fish tissues: potential consequences for food web studies. Ecology of , 13: 155– 160. Szepanski, M., Ben-David, M., and Van Ballenberghe, V. (1999). Assessment of anadromous salmon resources in the diet of the Alexander archipelago wolf using stable isotope analysis. Oecologia, 120: 327-335. Tieszen, L. L., Boutton, T. W., Tesdahl, K. G., and Slade, N. A. (1983). Fractionation and turnover of stable carbon isotopes in animal tissues: implications for ∂13C analysis of diet. Oecologia, 57: 32–37. Vander Zanden, M. J., Cabana, G., and Rasmussen, J. B. (1997). Comparing trophic position of freshwater fish calculated using stable nitrogen isotope ratios (∂15N) and literature dietary data. Canadian Journal of Fisheries and Aquatic Sciences, 54: 1142–1158. Vander Zanden, M. J., Chandra, S., Allen, B. C., Reuter, J. E., and Goldman, C. R. (2003). Historical food web structure and restoration of the native aquatic communities in the lake Tahoe (California-Nevada) basin. Ecosystems, 6: 274–288.

92 Vander Zanden, M. J. and Rasmussen, J. B. (2001). Variation in ∂15N and ∂13C trophic fractionation: implications for aquatic food web studies. Limnology and Oceanography, 46(8): 2061–2066. Wainright, S. C., Fogarty, M. J., Greenfield, R. C., and Fry, B. (1993). Long-term changes in the Georges Bank food web: trends in stable isotopic compositions of fish scales. Marine Biology, 115: 481–493.

93

APPENDIX 1 FIELD SAMPLING PROTOCOL

A 1.1 SUGGESTED EQUIPMENT • • Scalpels • • Syringes • • Plastic (PCR-style) capped test tubes • • Oyster Knife • • WhirlPac™-style bags • • Sample tags

A 1.2 SAMPLING PLANKTON Rigorous methods exist for plankton sampling and any study deploying plankton nets has more than adequate procedures in place for stable isotope analysis. Plankton sampling for isotopes does not require biomass estimates per se, although these may be required for subsequent food web analysis. The most important guideline is to separate each group to the lowest possible taxonomic level. After this, samples are generally pooled by sampling location for analysis.

Tissue Sample Size Guidelines POM Whole Pooled per Collect POM from glass fibre disc filters. Under a station microscope (120°—) remove obvious zooplankton and inorganic materials. Ice or freeze disc contents for desiccation in the laboratory. Macroalgae Whole 10 (all in catch if Collect and freeze identifiable fronds and algal detritus <10) into species groups. Unidentifiable materials should be collected and pooled together. Freeze in sample bags. Zooplankton Whole 10 (all in catch if Separate by species and pool into individual sample <10) containers. Hold live overnight in seawater to allow the organisms to void their stomach contents. Strain into vials and freeze.

94 A 1.3 SAMPLING INVERTEBRATES The wide-variety of invertebrates found in the ocean make standardizing sampling methods difficult as there may be exceptional species captured. But as muscle is the primary target tissue for invertebrates this should be the priority for sampling. Other tissues (gonads, digestive glands) can also be of importance, depending on the research question. It is recommended to gather one extra tissue during sampling to increase sampling efficacy.

Tissue Sample Size Guidelines Parasites Whole All in catch Collect all visible parasites from affected tissue. Sample affected tissue also to estimate fractionation between parasite and host. Freeze in sample bag. Bivalves Muscle 10 (all in catch Using an oyster knife separate shell halves. Using a scalpel if <10) carefully remove entire adductor muscle and clean with freshwater. Freeze in sample bag. Gonads 10 (all in catch After removal of muscle tissue carefully cut entire gonads if <10) from surrounding tissue with a scalpel. Freeze in sample bag. Digestive 10 (all in catch Rinse stomachs of dead animal with fresh seawater injected gland if <10) gently into the mouth. Once no new particulate material has been observed exiting the stomach locate digestive gland and carefully remove. Freeze in sample bag. Muscle 10 (all in catch Select standard location per species for maximum muscle mass if<10) (e.g. claws on crabs or tails of shrimp). It may be easier to sample these parts with exoskeleton attached for subsequent removal in the laboratory. Place sample in sample bag for freezing. Cephalopods Muscle 10 (all in catch Pull tentacles and internal organs from mantle. Remove pen. if<10) Cut ~5 g cross-section of muscle from centre of mantle and freeze in sample bag. Jellyfish (small) Umbrella 10 (all in catch Wearing gloves carefully remove tentacles from the bell using if<10) a scalpel. Rinse with fresh water and freeze in sample bag. Jellyfish (large)

Umbrella 10 (all in catch Wearing gloves carefully remove tentacles from the bell using if<10) a scalpel. Cut a 2-cm-wide slice of umbrellar tissue from the centre of the bell to outer edge and remove. Rinse with fresh water and freeze.

95 A 1.4 SAMPLING FISHES Fish tissue sampling can occur in a research survey or from commercial or recreational fisheries. Fish should be sampled fresh— i.e. if fit for consumption, the fish has not likely deteriorated and is suitable for stable isotope analysis. Gravid females should not be used as they are metabolically distinct from other fishes. If gravid fishes constitute a significant portion of the catch they should be sampled as a separate group.

Tissue Sample Size Guidelines Fishes 1 to 5 g Whole 10 (all in catch Due to their small size these fish must be sampled whole. if <10) Where possible remove the head; tail; and guts prior to sampling. This will leave a muscle-dominated sample. Fishes 6 to 30 g Muscle with skin 10 (all in Cut away a the entire side of body musculature with a catch if <10) scalpel as if filleting the fish. the skin cannot easily be removed from fishes of this size and is therefore retained on the sample. Fishes 31 to 500 g Muscle 10 (all in Excise 5 g of white muscle from just in front of the fish’s catch if<10) dorsal fin. Remove skin and freeze. Do not sample red muscle as this is isotopically-distinct from white muscle and will bias results. Liver 10 (all in Remove ~5 g (where possible) of liver tissue from the catch if<10) end of the liver lobe. Blood 10 (all in With syringe ready cut tail off just in front of caudal catch if<10) peduncle and insert syringe tip into exposed vertebral aorta vertebral aorta and draw to 10 cc. Transfer to PCR- tube and freeze. Fishes > 500 g Muscle All in catch Excise ~5 g of white muscle from just in front of the fish’s dorsal fin. Remove skin and freeze. Do not sample red muscle as this is isotopically-distinct from white muscle and will bias results. Liver All in catch Remove ~5 g of liver tissue from the end of the liver lobe. Freeze. Blood All in catch With syringe ready cut tail off just in front of caudal peduncle and insert syringe tip into exposed vertebral aorta and draw 2 x 10 cc. Transfer to PCR-tube and freeze. Fishes: large-pelagics (dead) Muscle All in catch Excise ~5 g of white muscle from just in front of the fish’s dorsal fin. Remove skin and freeze. Do not sample red muscle as this is isotopically-distinct from white muscle and will bias results. Liver All in catch Remove ~5 g of liver tissue from the end of a liver lobe. Freeze in sample bag. Blood All in catch With syringe ready cut tail off just in front of caudal peduncle and insert syringe tip into exposed vertebral aorta vertebral aorta and draw to 10 cc. Transfer to PCR- tube and freeze. Bone; Cartilage All in catch From just above 5th gill slit (sharks) or anterior operculum edge (tunas; billfish) remove vertebral aorta section of 2 to 3 pieces. Freeze cartilage in sample bag.

96 Tissue Sample Size Guidelines Fishes: large-pelagics (live) Muscle All in catch Carefully cut a 3 cm slit just in front of the fish’s dorsal fin. Open wound and remove a 2 to 3 g sliver of muscle from one side. Freeze. Apply veterinary glue to the wound edge. Suture together wound ensuring no space remains between muscle surfaces. Air/fresh water pockets will cause infection and disease. Do not sample red muscle as this is isotopically-distinct from white muscle and will bias results. Blood All in catch Live blood samples are collected from caudal venipuncture. Carefully insert syringe tip into exposed blood vessel and draw 2 x 10 cc. transfer to PCR-tube and freeze.

A 1.5 SAMPLING MAMMALS Mammal sampling is necessarily expensive and requires specialized equipment. Considerable use can be made of beached animals and fishery catches and tissue samples can readily be obtained during necropsy.

Tissue Sample Size Guidelines Mammals (live) Skin/blubber All possible Skin/blubber sampling on live whales requires a biopsy crossbow or pole. Biopsy crossbows must be custom made for the project. Tethered arrows are used with a hollow tip. This is shot into the body of the animal to obtain a muscle sample. Depth is controlled by an expandable disc on the arrow. Alternately a long biopsy pole may be used from a zodiac Sloughed skin All available Collect sloughed skin samples in dip-nets at surfacing site. The sloughed-skin samples are the most cost effective sample for live whales and have a linear relationship with muscle tissue however this must be validated with muscle samples (taken from dead animals) for each new species. Mammals (dead) Muscle All possible Cut through skin and blubber tissue from just in front of the whale’s dorsal fin. Excise ~10 g of muscle tissue and freeze in sample bag. Liver All possible Remove ~5 g of liver tissue from near the end of liver lobe. Freeze in sample bag. Bone All possible Remove single vertebrae from below dorsal fin for laboratory analysis.

97