UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Effects of global fisheries on the biomass of marine ecosystems : a trophic-level-based approach Tremblay-Boyer, Laura 2010

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

Item Metadata

Download

Media
24-ubc_2010_fall_tremblayboyer_laura.pdf [ 5.12MB ]
Metadata
JSON: 24-1.0071043.json
JSON-LD: 24-1.0071043-ld.json
RDF/XML (Pretty): 24-1.0071043-rdf.xml
RDF/JSON: 24-1.0071043-rdf.json
Turtle: 24-1.0071043-turtle.txt
N-Triples: 24-1.0071043-rdf-ntriples.txt
Original Record: 24-1.0071043-source.json
Full Text
24-1.0071043-fulltext.txt
Citation
24-1.0071043.ris

Full Text

Effects of global fisheries on the biomass of marine ecosystems: a trophic-level-based approach  by Laura Tremblay-Boyer BSc Biology, McGill University, 2007  a thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in  the faculty of graduate studies (Zoology)  The University Of British Columbia (Vancouver) June 2010 c Laura Tremblay-Boyer, 2010  Abstract Marine fisheries have been occurring for centuries but the last 50 years have seen a drastic increase in their reach and intensity. Fisheries now impact marine ecosystems worldwide and it is necessary to understand their impact, both historical and current, at this scale. A global perspective provides an efficient inter-disciplinary communication tool and allows to summarize and validate our current understanding of ecosystem functioning. The objectives of this thesis are to generate global estimates of biomass for marine ecosystems and evaluate the effects that fisheries have had on ocean biomass since the 1950s. A simple but versatile ecosystem model was used to represent ecosystems as a function of energy fluxes through trophic levels. Using primary production data, sea surface temperature, fisheries catch and trophic level of species, the model was applied on a half-degree grid covering all oceans. Estimates of biomass by trophic levels were derived for marine ecosystems in an unexploited state, as well as for all decades since the 1950s. Trends in the decline of marine biomass from the unexploited state were analyzed for all oceans, with a special emphasis on predator species since they are highly vulnerable to fishing. This thesis is the first application of a trophic modelling approach to a worldwide estimation of the effects of fishing. It provides an independent confirmation of previous reports by other researchers that were based on proxies of biomass or on meta-analyses of local datasets. The results presented highlight three main trends about the global effects of fishing: (1) predators are more affected than organisms at lower trophic levels; (2) declines in ecosystem biomass are stronger along coastlines than in the High Seas; (3) the extent of fishing and its impacts have expanded from north temperate to equatorial and southern waters in the last 50 years. More specifically, this work shows that many oceans historically exploited by humans have seen a drastic decline in their predator biomass, with about half of the coastal areas of the North Atlantic and North Pacific showing a decline in predator biomass of more than 90%. The spatial and temporal trends presented in this work provide a global synthesis of the effects of fishing on the biomass of marine ecosystems and point to a potential state of the world’s oceans if industrial fisheries maintain their current trajectory.  ii  Table of Contents Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Table of Contents  ii  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  iii  List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  v  List of Figures  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  Acknowledgements  vi  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii  1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  1  1.1  Problem-statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  1  1.2  Literature review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  2  1.3  Research objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  3  2 Effects of industrial fisheries on the biomass of the world’s marine ecosystems  4  2.1  Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  4  2.2  Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  7  2.2.1  Model description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  7  2.2.2  Application of the model at a global scale . . . . . . . . . . . . . . . . . . . . 11  2.2.3  Parameters and data inputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13  2.2.4  Scenarios of ecosystem response to fishing . . . . . . . . . . . . . . . . . . . . 15  2.3  2.4  Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 2.3.1  Estimates of global biomass . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21  2.3.2  Temporal and spatial trends in ecosystem maximum catch . . . . . . . . . . . 22  2.3.3  General trends in the decline of global marine biomass . . . . . . . . . . . . . 22  2.3.4  Spatial trends of decline in predator biomass . . . . . . . . . . . . . . . . . . 23  2.3.5  Predator decline by regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25  2.3.6  Scenarios of ecosystem response to fishing . . . . . . . . . . . . . . . . . . . . 25  Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 2.4.1  Global marine biomass estimates . . . . . . . . . . . . . . . . . . . . . . . . . 30  iii  2.4.2  Trends in the effects of fishing on ecosystem biomass . . . . . . . . . . . . . . 32  2.4.3  Scenarios of ecosystem response to fishing . . . . . . . . . . . . . . . . . . . . 34  2.4.4  Issues and possible solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36  2.4.5  Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40  3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  42  3.1  Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42  3.2  Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43  3.3  Future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44  3.4  Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45  References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  47  Appendices  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  53  A Supporting figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  53  A.1 Inputs to Ecotroph from oceanographical data  . . . . . . . . . . . . . . . . . . . . . 54  A.1.1 Global maps of primary production and sea surface temperature . . . . . . . 54 A.1.2 Validation of the use of primary production from year 1998 . . . . . . . . . . 55 A.2 Trends in reconstructed global catch . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 A.3 Map of unexploited predator biomass . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 B Sensitivity analyses  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  59  B.1 Transfer efficiency parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 B.2 Top-down parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 B.3 Smooth parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 C Estimating transfer efficiency from time-series of catch data  iv  . . . . . . . . . .  67  List of Tables 2.1  Summary of the Ecotroph model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12  2.2  Description of the scenarios of ecosystem response to fishing . . . . . . . . . . . . . . 20  2.3  Comparison of published predictions of global ecosystem biomass . . . . . . . . . . . 21  B.1 Changes in total and predator biomass as a function of transfer efficiency . . . . . . 60 B.2 Changes in total and predator biomass as a function of the top-down parameter . . . 64 B.3 Changes in total and predator biomass as a function of the smooth parameter . . . . 65 B.4 Range of individual trophic levels under the default value of smooth . . . . . . . . . 65  v  List of Figures 2.1  Flow chart of the application of the Ecotroph model to the world’s oceans . . . . . . 18  2.2  Illustration of the filters applied to test for a signal of overexploitation . . . . . . . . 19  2.3  Map of the decade of maximum ecosystem catch recorded for the period 1950–2006.  2.4  Trends in global marine biomass since the 1950s  2.5  Maps of the decline in predator biomass from the 1950s to the 2000s . . . . . . . . . 26  2.6  Proportion of predator biomass left in the coastal areas of the world’s oceans . . . . 28  2.7  Global marine biomass under three scenarios of ecosystem response to fishing . . . . 29  22  . . . . . . . . . . . . . . . . . . . . 24  A.1 Primary production and sea surface temperature for the world . . . . . . . . . . . . 54 A.2 Primary production in 1998 compared to other years in 1950-2004 . . . . . . . . . . 56 A.3 Summary in the world’s catch trends 1950-2000 . . . . . . . . . . . . . . . . . . . . . 57 A.4 Density of unexploited predator biomass for the world . . . . . . . . . . . . . . . . . 58 B.1 Spatial trends in the response of individual cells to transfer-efficiency . . . . . . . . . 61 B.2 Spatial trends in the response of individual cells to the top-down parameter . . . . . 63 B.3 Spatial trends in the response of individual cells to the smooth parameter . . . . . . 66 C.1 Diagram of the distribution of production by trophic level in an ecosystem . . . . . . 68 C.2 Example of the estimation of transfer efficiency from catch and trophic level data . . 71  vi  Acknowledgements I would first like to thank my supervisor Daniel Pauly for providing valuable advice and sharing his global perspective on the multiple facets of interactions between humans and the marine realm. Many thanks also to Didier Gascuel, who shared his intuition about Ecotroph’s behavior and welcomed me for an extended stay in his laboratory in Rennes, France. My committee members, Steve Martell and Jon Shurin, provided insightful comments on both the on-going work and the written thesis. Villy Christensen, Maria Lourdes Palomares and Reg Watson gave me access to the global datasets without which my work would have been impossible. Etienne Rivot at Agrocampus Ouest listened to me patiently during a research down and taught me a valuable lesson about scientific communication. Wilf Swartz and Rachael Louton both served as sounding boards for ideas I was developing during this thesis. The administrative staff at the Zoology department, Allison Barnes and Alice Liou, were always helpful and patiently dealt with my apparent inability to give any form back on time. Thanks also to Janice Doyle, Grace Ong, Marina Campbell and Ann Tautz from the Fisheries Centre for their assistance with all types of administrative matters. Working at the Fisheries Centre is great and would not be the same without friends to procrastinate with: thanks to Robert Ahrens, Megan Bailey, Lucas Brotz, Brooke Campbell, Andr´es Cisneros, Sarika Cullis-Suzuki, Meaghan Darcy Bryan, Carie Hoover, Rachael Louton, Rhona Govender, Michelle Paleczny, Shannon Obradovich, Wilf Swartz and Nathan Taylor. Thanks also to Caroline Kostecki from Agrocampus Ouest for always being around for gossips, tea and chocolate. Milles mercis to my roommates from Rennes, Julie Stein, Estelle Rouquet, Arzhela Hemery and S´ebastien Jubeau, for making my stay in France so much more than just about the research. My Vancouver roommates, David Kohler and Patrick Thompson, get my most sincere and heartfelt thanks for patiently coping with the ups and downs of working on this thesis. Last but not least, I want to thank my family and especially my parents, Gilles Boyer and Denise Tremblay, for unconditional support and encouragements over the years. Financial support for this Master’s came from NSERC as an Alexander Bell Canadian Graduate Scholarship, the UBC Zoology department in the form of top-up awards and TAships, the Pew Foundation through a grant to the Sea Around Us Project, the French Embassy in Canada as a Bourse de collaboration de recherche France-Canada and the Pˆole Halieutique of Agrocampus Ouest. vii  Chapter 1  Introduction 1.1  Problem-statement  Modern fisheries have come a long way since humans first started hunting fish. From the modest bone harpoon of 90,000 years ago (Yellen et al. 1995) evolved longlines with thousands of hooks, trawls that reach deeper than 2,000 meters and factory boats that stay at sea for weeks. We are slowly starting to uncover evidence that humans in historical times could negatively affect marine populations despite rudimentary technologies (Jackson 2001; Roberts 2007). There is no doubt, therefore, that industrial fisheries today can have a profound and lasting impact on marine ecosystems worldwide. The potential consequences are serious: 2.9 billions people rely on marine products for food (FAO 2009). Moreover, despite widespread belief to the contrary, marine populations are not immune from ecological or actual extinction (Dulvy et al. 2003). This is of concern since marine biodiversity is the driver behind a number of ecosystem services (Worm et al. 2006). Understanding the current state of marine resources and the ecosystems that support them is an essential step towards the development of sustainable fisheries. Modern fisheries need to be regulated to be sustainable (Clark 1990; Gordon 1954). Efficient management relies on understanding how fishing affects ecosystems at a local scale because management initiatives should be tailored to the system at hand (Walters and Martell 2004). In addition, fisheries impact marine systems worldwide, hence a global perspective is required to extract general trends in the response of ecosystems to fishing (Pauly 2007). Global studies are a powerful 1  inter-disciplinary communication tool, a feature that is valuable in the context of modern fisheries where the realized application of prescribed management is often influenced by social, cultural, political and economic factors (Berkes et al. 2006; Hilborn 2007). Lastly, global fisheries are akin to a giant predator-removal experiment and synthesizing their global impacts allows to verify and identify gaps in our understanding of large-scale ecosystem functioning.  1.2  Literature review  Sampling marine ecosystems at a global scale would be inefficient, if at all possible. However, the observation of recurring, similar phenomena in a diverse array of systems suggests that both direct and indirect effects of fisheries can be generalized. For example, alternate states, defined as drastically different configurations of the same ecosystem, have occurred in a number of locations, most famously in coral reefs and kelp forests (Hughes 1994; Steneck et al. 2002), but also in the North Sea (Beaugrand 2004) and in the Benguela current (Cury and Shannon 2004). Drastic increases in jellyfish populations are commonplace (Richardson et al. 2009) and affect areas such as the Black Sea (Daskalov et al. 2007), the Caspian Sea (Stone 2005) and the Sea of Japan (Kawahara et al. 2006). Trophic cascades, wherein the removal of predators affect populations further than one trophic level below, have been observed, for example, in the Baltic Sea (Casini et al. 2008), the North Atlantic (Frank et al. 2005) and the open oceans (Baum and Worm 2009). Synthesis of local observations can contribute to our understanding of the general effects of fisheries but does not substitute for global studies. However, global studies are hampered by the intrinsic challenges (and potential pitfalls) of obtaining and summarizing data at very large scales. So far, two main approaches have been used to infer the effects of fisheries at a global scale. The first one is the meta-analysis of local data. Myers and Worm (2003) used time-series of catch per unit effort from different locales to show that marine predators worldwide were in decline (but see Walters 2003). The second approach is the analysis of a global dataset in which data can be used as proxy for the state of some ecosystem property. For example, Pauly and Christensen (1995) used primary production and global fisheries catch data from to estimate the proportion of worldwide primary production required by fisheries. Similarly, Pauly et al. (1998) analyzed the mean trophic  2  level of global catch data and found that it was declining. It was assumed that the change in global catch composition reflected a change in community composition and thus that the abundance of high trophic level species was declining (but see Essington et al. 2006). Ecosystem modelling is an approach widely used in local studies focused on the impact of fishing but it has yet to have a global application. This is not surprising since traditional ecosystem models are complex and require large amounts of data. However, models are tools built to represent a set of properties in light of specific objectives. The main goal of global studies is to extract general spatial and temporal trends. Simple ecosystem models that perform well in data-poor cases and use straight-forward assumptions are useful in this regard. Jennings et al. (2008) presented what, to my knowledge, is the only study that applied a marine ecosystem model globally. Their simple community model is based on sizes and assumptions of predator-prey size ratios and estimates theoretical ecosystem biomass for marine animals. However, the results of Jennings et al. (2008) do not account for the effects of fishing. An alternative strategy to model ecosystems at large scales is to focus on trophic levels instead of species. Doing so greatly simplifies the representation of the ecosystem, but still captures important properties such as biomass and processes such as production. In addition, the trophic-level-based approach is appropriate when assessing the impacts of modern fisheries since we know that they preferentially target high trophic level species, which are intrinsically more vulnerable to fishing (Cheung et al. 2005), and play a structuring role in the ecosystem (Heithaus et al. 2008).  1.3  Research objectives  The objectives of this Master’s thesis were to estimate the effects of fishing on marine ecosystem biomass at a global scale. The model developed generates a clear and simple representation of ecosystems and incorporates global datasets accounting for important features of marine ecosystems in the context of fishing (primary production, sea surface temperature, fisheries catch, trophic level of species caught). Spatial and temporal trends in the changes of ecosystem biomass due to fishing for the period 1950-2006 were assessed and summarized for all oceans with a special emphasis on predators.  3  Chapter 2  Effects of industrial fisheries on the biomass of the world’s marine ecosystems 2.1  Introduction  Understanding marine systems can be challenging because it is hard to observe their components directly. Models have been valuable tools to aid our comprehension of marine processes, from individual behavior such as foraging (Walters et al. 1997) to ocean-wide interactions between oceanic currents, nutrients and primary producers (Gregg and Casey 2004). Models are increasingly being used to address applied questions, particularly in the area of fisheries management and conservation (Plag´anyi 2007), and the recent recognition that the impacts of fishing extend beyond the targeted species has promoted new research into the use of ecosystem models. Scientists now agree on the importance of ecosystem-based management (see e.g., McLeod et al. 2005) and legal commitments by the Parties of the Convention of Biological Diversity further emphasize the need to understand the effects of fishing at the ecosystem scale. Ecosystem modelling requires strategic decisions about the scope of the model, the types of interactions to be represented and the unit of focus (e.g. species, functional group, trophic level,  4  etc.; see Plag´anyi 2007). Some approaches focus on key species groups (e.g., Multi-species Virtual Population Analysis; Vinther 2001); others include most or all biotic components of the system (e.g. Ecopath with Ecosim EwE; Walters et al. 1997), and some even link with an oceanographic model of the surrounding environment (e.g., Atlantis; Fulton et al. 2004). The challenge is to find the optimal combination of model complexity and necessary data requirements to produce results that are robust and relevant in terms of the research or policy objectives (Fulton et al. 2003). For example, EwE is a complex mass-balanced and dynamic model that can be useful for policy exploration through the simulation of broad ecological responses to various management options (Walters and Martell 2004). Multi-species models, such as the one used by Commission for the Conservation of Antarctic Marine Living Resources, are simpler but can capture fisheries responses for selected species and are detailed enough to be used to derive fishing quotas (Constable et al. 2000). The predictions made by ecosystem models are often complex and hard to interpret. In this regard, indicators can be useful tools to extract the relevant information from the various outputs of these models (Rice 2000; Rochet and Trenkel 2003). Alternatively, ecosystem models can be built in a simplified way that focuses on the ecosystem property of interest. Again, the challenge is to find the degree of complexity that both minimizes data requirement and yields results that are realistic and relevant. For instance, Ecotroph (Gascuel and Pauly 2009) is an ecosystem model that focuses on trophic levels instead of species, allowing the removal of much of the complexity that results from the accounting of intricate species-specific interactions. The model was shown to reliably reproduce trends in biomass by trophic levels in a shelf ecosystem off Guinea (Gascuel 2005) and can serve as an alternative way of detecting changes in ecosystem properties or structure. The global fisheries crisis poses a unique challenge in terms of ecosystem modelling. From an ecological perspective, the similarities in the types of ecological responses to fisheries observed worldwide, such as trophic cascades and regime shifts, highlight the existence of shared properties of ecosystem functioning (Crowder et al. 2008; Jiao 2009). On the other hand, detailed analysis of such responses emphasizes the complex interplay between fishing and local biotic and environmental conditions. For example, the drastic decline in coral cover observed in the 1990s in the  5  Caribbean was caused by a combination of overfishing of herbivorous fish and an epizootic that led to an extremely strong decline in the sea urchin population (Hughes 1994). Large fluctuations in anchovy and sardine biomass in the Pacific Ocean were primarily driven by changing regional climate regimes, with fishing acting only as a secondary factor (Chavez et al. 2003). Understanding the effects of fishing at a local scale is essential for efficient management, but a global overview allows the identification of important spatial and temporal patterns in the response of ecosystems to fishing (Pauly 2007). Additionally, a global perspective is a powerful tool to communicate with scientists and other people outside of fisheries science, an aspect that is valuable given the importance of social and economic factors acting both at local and global scales to affect the application and outcome of management (Berkes et al. 2006; Hilborn 2007). Current basin-wide or global understanding of the impacts of fisheries comes from two main sources. The first is the direct analysis of locally available catch and catch composition data. This approach has shown, for example, that the combination of increasing effort and declining global catch points to a global decline in marine ecosystem biomass (Watson and Pauly 2001). The second method is based on the meta-analysis of data originating from a representative set of systems. For instance, Frank et al. (2007) compared trophic cascades resulting from fishing in the North Atlantic and showed that temperature and species richness affected how strong these cascades were. While both ecologists and fisheries scientists agree on the value of ecosystem models, there has yet to be a global application of one that accounts for fishing, mostly due to the data intensive nature of existing models and their ecological complexity. However, these models were developed with local management in mind, whereas global studies are conducted for the purpose of extracting general trends. I suggest here that the first step towards modelling marine ecosystems at a global scale can be taken with a model that is vastly simplified in its structure, but uses sound ecological assumptions and is robust enough to yield realistic, large-scale trends. In this report I present the application of an ecosystem model to global oceans. I used Ecotroph (Gascuel and Pauly 2009) as a modelling framework to generate global estimates of marine biomass by trophic levels, both for their unexploited state and for all decades starting in 1950. Ecotroph represents ecosystems through three fundamental properties defined over trophic levels: biomass,  6  production and the kinetics, or rate of biomass turn-over. Production is assumed to flow continuously from primary producers to herbivores and predators, with losses occuring between trophic levels, because of natural factors such as consumption rates, metabolism and reproduction, and anthropogenic factors such as fisheries catch. Key input parameters include primary production, temperature, trophic transfer efficiency and fisheries catch. Additionally, Ecotroph can generate estimates of unexploited biomass by trophic levels from primary production estimates and empirical predictions of the turnover rate of biomass (see Gascuel et al. 2008). The estimation of unexploited biomass is a particularly interesting feature given that only anecdotal baseline data are available on the unfished state of marine ecosystems. I first present global estimates of change in marine biomass by trophic levels between fished and unexploited states for the period 1950 to 2006. These estimates are compared under different representations of ecosystem response to fishing. I discuss spatial trends, specifically in terms of declines in biomass for specific areas and how they compare between predators and lower trophic levels. This research assesses the global trends in the removal of ecosystem biomass due to fishing and can serve as a valuable step towards highlighting the gaps in our understanding of ecosystem functioning at large scales.  2.2  Methods  Estimates of unexploited and fished biomass for marine ecosystems were generated by applying the Ecotroph model to each cell of a 0.5 ◦ grid covering the world’s oceans. The main data inputs were marine primary production, sea surface temperature and fisheries catch data by trophic level and the results were produced for the period 1950-2006 at the temporal resolution of decades.  2.2.1  Model description  A detailed description of the Ecotroph model was published in Gascuel et al. (2009). Ecotroph uses the trophic level (TL or τ ) as its fundamental unit to model ecosystems, unlike most other ecosystem models which focus on individual species or group of species. The trophic level shows how far removed from primary production an individual feeds at and is a useful representation of the ecological function of a species in an ecosystem. Primary producers are defined as trophic 7  level 1 and herbivores as trophic level 2. The trophic level of consumers is one plus the average of the trophic level of their prey, which means that its value is ≥ 2 with very few species — if any — having a trophic level above 5. A unit of biomass enters an ecosystem at TL=1 through photosynthesis by primary producers or the recycling of detritus and is transferred to higher trophic levels through predation or ontogeny. Since at any point in time thousands of such transfers occur, the average of all trophic level transfers from primary producers to higher trophic levels can be described as a continuous process. Based on this principle, Ecotroph models ecosystems by defining three fundamental properties over trophic levels: biomass, production and kinetics. The biomass is the amount of organic matter (in tonnes) present at any moment at a given TL. The production is the quantity of biomass that passes through a TL in a year (tonnes year−1 ). The kinetics is the speed at which biomass moves through a TL, in TL year−1 or ∆τ /∆t. It is equivalent to the production/biomass value (P/B) in Ecopath models (Gascuel et al. 2009), and can be thought of as the turnover rate of biomass by TL. The distribution of biomass or production over the trophic levels of an ecosystem is called the trophic spectrum. In theory the trophic spectrum of a given ecosystem property is defined continuously over trophic levels. Here, I used a discrete approximation to the continuous form of the model in order to apply it to real ecosystems. I divided the trophic spectrum into discrete intervals of width ∆τ = 0.1. The value of biomass or production in each interval is simply the area under the curve bounded by the interval. Ecotroph’s most important equation states that the biomass B at any trophic level can be calculated as the ratio of production P to kinetics K for that trophic level (Gascuel et al. 2009), that is  Bτ =  Pτ Kτ  (2.1)  where Kτ = ∆τ /∆t and t is a unit of time. The Ecotroph ecosystem model can thus be used to produce estimates of ecosystem biomass if the production and kinetics by trophic level are known. The process used to calculate production 8  and kinetics by trophic level is described below, with the required data, parameters and equations summarized in Table 2.1. 1. Calculating production by trophic level: Ecotroph assumes that production flows continuously from primary producers to herbivores and predators and declines predictably between TL from natural energy losses due to digestion, respiration, excretion, partial consumption, etc. This rate of decline is called the transfer efficiency (TE). It is calculated as the proportion of production remaining after a transfer of one trophic level, that is, TE = P(τ +1)/P(τ ), which can also be defined as the exponential rate of decline µτ = -log(TE). If the ecosystem is exploited, production declines at a faster rate between the trophic levels targeted by fisheries. The fishing mortality Fτ and its corresponding exponential rate of decline (the fishing loss rate ϕτ ) are  Fτ = Yτ /Bτ  (2.2a)  ϕτ = Yτ /Pτ  (2.2b)  where Yτ is the fisheries catch at τ . With primary production P(τ = 1), µτ and ϕτ , estimates of production for TL ≥ 1 can be calculated from  Pτ +∆τ = Pτ · e−(µτ +ϕτ )∆τ  (2.3)  where µ and ϕ are exponential rates of decline respectively quantifying losses due to natural causes and fisheries. It can be difficult to obtain estimates of fishing loss rate for each trophic level. If catch data by trophic level are available, an alternative approach can be used based on the single-species Statistical Catch Analysis (Pope 1972). If we assume that the catch of a trophic level interval  9  is removed from the production at exactly half of the TL interval, then the production at τ is  Pτ +∆τ = Pτ · e−µτ ∆τ − Yτ · e−µτ  ∆τ 2  (2.4)  where Pτ −∆τ is the production of the TL just before τ , Yτ is the capture removed at exactly half of the trophic level interval and e−µτ ∆τ accounts for natural energy losses in ∆τ . This approximation is almost exact for values of µτ ∆τ smaller than 0.3. The production trophic spectrum can thus be solved starting with primary production at P(τ =1), removing catch as needed at each half trophic level interval and accounting for natural losses as set by the parameter µ. This approach was used in the current application of this model because of the availability of catch data at a global scale. Once Pτ is derived from (2.4), the fishing loss rate ϕτ can be calculated for each τ as:  ϕτ =  1 · log ∆τ  Pτ Pτ + ∆τ  − µτ  (2.5)  2. Calculating kinetics by trophic level: The kinetics Kτ are obtained from an empirical model that predicts Kτ as a function of sea surface temperature H( ◦ C) and trophic level. This model was parameterized by Gascuel et al. (2008) based on the P/B values (equivalent to the kinetics in Ecotroph) of 1718 groups from 55 published Ecopath models, and it explains 54% of the observed variance in kinetics. This model is assumed to correspond to the unexploited state of the system and is described in the following equation:  Kτ,unexpl = 20.19 · τ −3.26 · e0.041H  (2.6)  where τ is the trophic level and H is the sea surface temperature in ◦ C. When the ecosystem is exploited, the kinetics of the targeted trophic level increase because production is being removed from that TL at a faster rate. When catch or fishing mortality is known, the unexploited kinetics are adjusted to account for fishing mortality as 10  Kτ = Kτ,unexpl + Fτ  (2.7)  where Kτ,unexpl refers to the kinetics in the unexploited state (from equation 2.6) and Fτ is the fishing mortality at τ . Note that Kτ must be solved iteratively because the biomass at τ is required to calculate Fτ (see equation 2.2a). However, equation 2.7 can be re-arranged as Kτ = Kτ,unexpl + ϕτ Kτ , since Fτ = ϕτ Kτ (see equations 2.2a and 2.2b), and this version of the equation has the analytical solution Kτ = Kτ,unexpl /(1 − ϕτ ) for ϕτ < 1. Note that this is an update from the original formulation of the model presented in Gascuel et al. (2009). In summary, to estimate biomass by trophic level when the ecosystem is not under exploitation, production by trophic level is calculated from (2.3) with the primary production, the transfer efficiency and a fishing loss rate ϕ set at zero (initializing with P1 =PP). The unexploited kinetics for each trophic level can be calculated from (2.6) based on the sea surface temperature. The production and kinetics by trophic level are then used in equation 2.1 to derive the biomass in the unexploited state. If the ecosystem is being fished, production can be estimated from primary production, transfer efficiency and catch data from equation 2.4 (with P1 =PP). A first estimate of biomass is calculated with (2.1) by using the unexploited kinetics. This estimate is used to compute Fτ from equation 2.2a, which is then used to get an updated value for the kinetics according to (2.7). The last two steps are repeated until the kinetics converge to a constant value. Alternatively, it is also possible to use the analytical solution described above to solve for Kτ .  2.2.2  Application of the model at a global scale  The Ecotroph model was applied separately to all cells of a 0.5 ◦ grid covering the world for which either two of the following conditions were met: the water coverage was either greater than 50% or the Sea Around Us Project had at least one fisheries catch record for that cell. A total of 179,612 cells met at least one of these criteria. Each cell was assumed to represent an ‘ecosystem’ and there is no dispersal of production between cells. Cells that have a PP of zero have no ecosystem production under Ecotroph and were not included in the calculations. Because the distance between 11  Table 2.1: Summary of the application of the Ecotroph model to generate estimates of biomass by trophic levels for the unexploited state and all decades between 1950 and 2006. Description  In-text reference  Model initialized with: • Primary production, P1 =PP • Sea surface temperature, SST  Section 2.2.3 Section 2.2.3  Model conditioned with: • Catch data by trophic level  Section 2.2.3  Model parameter: • Transfer efficiency µ  Section 2.2.3  For τ in [2, 5], with ∆τ =0.1: I) 1. 2. 3.  Calculation of unexploited biomass: Pτ +∆τ,unexpl = Pτ,unexpl · e−µτ ∆τ , with P1 = PP Kτ,unexpl = 20.19 · τ −3.26 · e0.041H , where H=SST Bτ,unexpl = Pτ,unexpl /Kτ,unexpl  II) Calculation of fished biomass: ∆τ 4. Pτ +∆τ = Pτ · e−µτ ∆τ − Yτ · e−µτ 2 5.  1 τ ϕτ = ∆τ · log PτP+∆τ K Kτ = τ,unexpl 1−ϕτ  Eq. 2.3 Eq. 2.6 Eq. 2.1  Eq. 2.4 Eq. 2.5  − µτ  6. 7. Bτ = Pτ /Kτ 8. Repeat steps 4 to 7 for all decades between 1950 and 2006.  Eq. 2.7 Eq. 2.1  degrees of longitude decreases as we move away from the equator, the area of the cells varies over 3 orders of magnitude from about 3000 km2 in the tropics to 3 km2 close to the poles. Consequently, all of my results were standardized by area. Biomass, production and kinetics under unexploited and fished states were calculated for trophic levels between 2 and 5 at intervals of ∆τ =0.1. All units are in tonnes km−2 year−1 . For each cell, the model was initialized with P1 = PP (tonnes km−2 year−1 ). The model was run for each decade between 1950 and 2006 using catch data by trophic level averaged by decade. This allowed to produce biomass estimates that were both robust 12  to extremes in the annual catch values and representative of the general evolution of industrial fishing since the 1950s. Note that only one possible set of predictions of unexploited biomass for the period 1950-2006 was generated because I used one PP value per cell for the 1950-2006 period and the only factors that affect unexploited biomass are PP, sea surface temperature and TE (see equations 2.3 and 2.6).  2.2.3  Parameters and data inputs  Transfer efficiency A value of 10% was used for the transfer efficiency parameter TE. This value was selected as the mean of estimates from a wide range of ecosystems (Pauly and Christensen 1995). TE is converted to its log-scale equivalent through µτ = −log(T E). I assumed that µτ is constant for all trophic levels, which largely hold empirically (Christensen and Pauly 1993). A sensitivity analysis was conducted to determine how the value of TE affected the biomass derived for each cell in terms of its absolute value and compared to that of other cells (see Appendix B). Primary Production PP Primary production data for each cell were obtained from the Sea Around Us Project databases (www.seaaroundus.org), which were themselves derived from SeaWiFS chlorophyll data (seawifs. gsfc.nasa.gov/SEAWIFS.html) and photosynthetically active radiation (Bouvet et al. 2002) using a model described by Platt and Sathyendranath (1988) (see Appendix A). When missing, values where interpolated from neighboring cells (Lai 2004). Because I was primarily interested in the interaction between spatial trends in primary production and temporal trends in catch, I used one PP value (year = 1998) for all decades. To make sure that the chosen year 1998 did not represent extreme values for this time period, I used a temporal PP model developed by Sarmiento et al. (unpublished data) for which the absolute values of annual biomass are not as accurate as those obtained from the SeaWifs data but that captures well the temporal trends in PP between years. I calculated how each annual PP/cell prediction differed from the mean PP over 1950-2006 for that cell and confirmed that the PP values in 1998 were representative of the general spatial trends in  13  PP observed in the last 50 years (the proportion of cells in 1998 within one standard deviation of the mean cell PP over 1950-2004 was above the average of other years, see Appendix A). Sea Surface Temperature SST Annual SST was obtained from the NOAA World Ocean Atlas 2001 (www.nodc.noaa.gov/OC5/) and averaged for each cell over the last 50 years. See Appendix A for a global map of the SST values used in this study. Fisheries catch Species-specific catch data for 1950-2006 came from the Sea Around Us Project global catch database at a 30 minute resolution. Each record contains information about location, wet weight of catch, year and species caught. The database was built by integrating available catch data sets including data from the FAO, ICES and other organizations (Watson et al. 2004) and catch data reconstructed by the Sea Around Us Project for 12 countries (Watson and Pauly 2001; Zeller and Pauly 2007). Herein, the catch data was assigned to each cell by accounting for fishing access agreements between countries, species and effort distribution (when available) and a set of assumptions about catch trends within the species distribution. The Ecotroph model requires a catch trophic spectrum (i.e., catch data for each trophic level) so the Sea Around Us Project catch data had to be processed accordingly before being incorporated in the model. Species-specific catch observations for a given region were transformed into an approximation of a continuous catch trophic spectrum by assuming that the variation in TL of the individuals of a species is positively related to the species’ mean TL. In other words, as mean TL increases for a species, there is greater variation in the TL of the individuals in the population due, for example, to diet shifts in ontogeny (as an illustration, compare zooplankton and sharks). I modelled the distribution of trophic levels of individuals within a species as log-normal with a standard deviation S defined as Sτ = smooth · log(τ − 0.05). The constant smooth represents the strength of the smoothing as TL increases and τ is the mean trophic level (centered at ∆τ /2). I arbitrarily set smooth at 0.075 and a sensitivity analysis, available in Appendix B, showed that results are relatively robust to this parameter. 14  An approximation to a continuous distribution of catch over TL for a region was obtained by standardizing the log-normal distribution of the individuals of a species over TL. The contribution of each TL interval to the total area under the curve of the distribution was multiplied by the catch of the species. The ecosystem catch for each TL interval was then calculated by summing over all of the species-specific catch records that occurred in the interval, which resulteds in a catch trophic spectrum. Mean decade catch trophic spectra were built by first constructing annual catch trophic spectra for each cell for all years between 1950-2006 based on the fisheries catch data for that cell. The mean catch value of each TL interval was then taken over each of the six decades in the 1950-2006 period: the 1950s (1950-1959), the 1960s (1960-1969), the 1970s (1970-1979), the 1980s (1980-1989), the 1990s (1990-1999), the 2000s (2000-2006). The decadal catch trophic spectrum (in tonnes km−2 ) is thus the average of all of the year-specific catch trophic spectra for the cell of interest and it was assumed to be representative of the type of catch patterns applied to the cell for each decade. Trophic levels Trophic levels data for each species or group of species recorded as caught were obtained from the Sea Around Us Project (Deng Palomares, pers. comm). Individual species TL came from FishBase (www.fishbase.org), preferably from diet composition studies since they are more accurate. Trophic levels for groups of species were the average of available TL for species belonging to the group of interest.  2.2.4  Scenarios of ecosystem response to fishing  Global fisheries catches have been declining since the late 1980s if we account for catch overreporting by China (FAO 2009; Watson and Pauly 2001). In general, a decline in catch is caused either by lower fishing mortality or by a reduction in marine biomass. In its current formulation, Ecotroph assumes the former, that is, declining catch is caused by declining fishing mortality (see equations 2.2a and 2.2b). However, we know that in most instances the observed decline of catch in ecosystems is due to overexploitation (Mullon et al. 2005), although in certain systems management has been effective in reducing fishing mortality (Worm et al. 2009). A function accounting for fishing 15  mortality could be included in Ecotroph but no global data set of fishing mortality is currently available. In order to account for the possibility that declines in catches could be due to overexploitation and not declining mortality, I ran the Ecotroph model as described above and further processed its outputs according to a set of three scenarios. These scenarios were created to capture different interpretations of the observed worldwide decline in fisheries catch and can be qualitatively described as optimistic, intermediate and pessimistic explanations (see Figure 2.1 and Table 2): 1. Scenario 1 (optimistic): catches are related to declining fishing mortality as assumed by Ecotroph. The model’s estimates of biomass were therefore appropriate and were not adjusted; 2. Scenario 2 (intermediate): Catch declines because of overexploitation but biomass can recover after being driven down, with animals feeding at lower trophic levels recovering faster; 3. Scenario 3 (pessimistic): Declines in catch are caused by overexploitation as in Scenario 2 but biomass cannot recover once it has been driven down. Under scenarios 2 and 3, the biomass predictions generated by Ecotroph were processed posthoc for cells and decades where an overexploitation signal is detected. For the purpose of our analysis, I defined ecosystem overexploitation as a decline in catch between decades observed over a TL interval of 0.5 or more. In addition, this definition was only applied to cells and decades that satisfied the two following criteria (see Figure 2.2): 1. Exploitation level: Cells must be fully-accessed that is, for at least one decade in the 19502006 period, the observed catch must reduce ecosystem biomass by at least 10% of the cell’s biomass at Maximum Sustainable Yield (BM SY ). This allowed to filter out cells that are only marginally exploited by fisheries and where overexploitation is unlikely to have occurred even if a decline in catch is observed. Maximum sustainable yield was assumed to occur at 50% of the unexploited biomass of the ecosystem (a conservative estimate). 2. Relative catch: The catch previously recorded in the cell must be at least 20% of the maximum recorded catch in the cell for 1950-2006 or the decline must occur past the decade 16  where the maximum catch was recorded. This prevents small fluctuations in catch level (especially in early decades) from generating an overexploitation signal. All cell and decade combinations were tested for overexploitation according to the set of rules defined above. For both scenarios 2 and 3, I assumed that if ecosystem overexploitation occurs, fishing mortality F is the same as that of the previous decade. Since Ydec−1 = F · Bdec−1 and Ydec = F · Bdec , then: Bdec = Bdec−1 ·  Ydec Ydec−1  (2.8)  Therefore, whenever a signal of overexploitation was recorded for a cell, the Ecotroph predictions of biomass for that TL interval/decade combination were further processed to satisfy the definitions of scenarios 2 and 3 (see Table 2): • Scenario 2 (intermediate): In this scenario, catch declines because of overexploitation but biomass is allowed to recover after being driven down, especially at lower trophic levels. If an overexploitation signal was recorded, the biomass for the current decade of the corresponding TL interval was re-calculated according to (2.8). If the catch increased in subsequent decades, I assumed that ecosystem overexploitation had stopped for that TL interval. The biomass for the TL interval was allowed to recover from the previous decade’s biomass to the biomass predicted by Ecotroph (i.e. from Scenario 1) depending on trophic level. Recovery R to equilibrium biomass was a declining linear relationship of TL, with full recovery at TL=2 and no recovery at TL=5, as defined in the relationship R(τ ) = −0.33τ + 1.67 (where 0 ≤ R ≤ 1). • Scenario 3 (pessimistic): Declines in catch are caused by overexploitation as in Scenario 2, but biomass is not allowed to recover (i.e., declines are ratchet-like). In this scenario, when a signal of overexploitation was recorded, biomass was re-calculated with (2.8). If the signal stopped (i.e., catch increases after having previously declined), biomass was not allowed to recover to the equilibrium value predicted by Ecotroph and was kept constant unless a further decline in catch occurred. 17  Grid of 0.5° by 0.5° cells  Parameters: i. Transfer efficiency ii. Top-down effect  Data: Primary production Sea surface temperature Fisheries catch by species TL of species caught  Smooth to obtain catch trophic spectrum  ECOTROPH Default predictions of biomass by TL by decade  Is a positive signal of overexploitation recorded for any decade? Recalculate biomass according to scenarios 2 and 3 S C E N A R I O S  YES  NO  If no overexploitation recorded, biomass under all scenarios = default biomass = scenario 1  1 2 For a given scenario, map of fishing impact (biomass left %):  3 100% For each scenario, calculate fishing impact as: Biomass fished / Biomass unexploited Where biomass is aggregated over TL intervals: 1. Predators 3.5 <=TL<=5 2. Total 2 <= TL <= 5 0%  Figure 2.1: Flow chart representing the application of the Ecotroph model at a global scale using the waters around Ireland as an example (especially the Celtic Sea). For each cell data were extracted from the global datasets and Ecotroph was run to predict biomass by decade. Unexploited biomass was calculated by assuming there was no catch. By default, Ecotroph’s biomass predictions correspond to Scenario 1 (square). Each cell was then tested for an overexploitation signal. If a signal was detected, the biomass of the affected TL was recalculated according to Scenario 2 (triangle) and 3 (pentagon). If not, the biomass by TL for that cell was left at the default value predicted by Ecotroph for all scenarios. The % decline in biomass was calculated as the ratio of fished to unexploited biomass for TL intervals representing either the total ecosystem biomass (excluding phytoplankton) or the predators (TL>3.5). The resulting value corresponds to the percentage of unexploited biomass left in the system and lies in the (0, 100)% range, with a color key going from yellow (light grey) to dark blue (dark grey), respectively. 18  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igure 2.2: An illustration of the three sequential filters applied to each cell to test for a signal of overexploitation. For each TL interval of 0.5, starting at TL=2, I tested for the following three conditions: (1) Ecotroph must predict a decline of at least 10% of BM SY , (2) the catch recorded for this TL interval in the current decade has to be smaller than that of the previous decade, (3) the catch recorded so far must be at least 20% of the maximum catch recorded in the cell for the TL interval. If all three conditions were met, the TL interval for that cell in that decade was considered overexploited. If catch increased in subsequent decades, the biomass of the TL interval was assumed to be recovering.  19  Table 2.2: Description of the three scenarios of ecosystem response to fishing applied to Ecotroph’s predictions of biomass by decade (Bdec, default ). The labels are used in the text to qualitatively describe the scenarios. Ecotroph’s predictions of biomass for the current decade Bdec were adjusted according to the equations described below. In Scenario 2, a recovery factor Rτ is included to scale recovery as a function of TL.  Calculation of biomass Label Scenario 1 20 Scenario 2  Scenario 3  optimistic or default  intermediate  pessimistic  Explanation Catch declines because of a reduction in fishing mortality; biomass is predicted under levels of catch at equilibrium Catch declines because of overexploitation; biomass is allowed to recover as a function of TL if overexploitation stops Catch declines because of overexploitation; biomass is not allowed to recover if overexploitation stops  OE in past decade?  Overexploitation (OE) in current decade? No  Yes  No  Bdec = Bdec, default  Bdec = Bdec, default  Yes  Bdec = Bdec, default  Bdec = Bdec, default  No  Bdec = Bdec, default  Ydec Ydec−1  Yes  Bdec = Bdec−1 + Rτ · (Bdec, default − Bdec−1 )  Bdec = Bdec−1 · Bdec = Bdec−1 ·  Ydec Ydec−1  No  Bdec = Bdec, default  Bdec = Bdec−1 ·  Ydec Ydec−1  Yes  Bdec = Bdec−1  Bdec = Bdec−1 ·  Ydec Ydec−1  2.3  Results  I present here an analysis of the outputs of the Ecotroph model applied to the world’s oceans on a 0.5 ◦ grid. Unless otherwise mentioned, the results presented below are based on an analysis of the biomass predictions under the intermediate scenario (#2) which assumes that worldwide decline in catches are due to overexploitation but that ecosystem biomass can recover if overexploitation stops. Note that cells belonging to an Economic Exclusive Zone (EEZ) are defined as being ‘coastals’ and that the terms ‘coastal’ and ‘EEZ’ are used interchangeably.  2.3.1  Estimates of global biomass  Table 2.3 compares the predictions of marine global biomass made by the Ecotroph model to currently available published estimates. To my knowledge, no other study has estimated global marine biomass under exploitation so this comparison focuses on the unexploited state of the ecosystem. For the total biomass (marine animals, or TL ≥ 2), Ecotroph’s estimates of unexploited biomass are considerably greater than those of Jennings et al. (2008). The estimate for predators (teleosts, or TL ≥ 3.5) is also higher than that of Jennings et al. (2008), but is smaller than another estimate made by Wilson et al. (2009). See Appendix B for a global map of unexploited predator biomass. Table 2.3: Comparison of published predictions of global ecosystem biomass, including assumption about the exploitation status of ecosystems and the ecosystem subset for which biomass is estimated.  This study  Ecosystem status  Ecosystem subset estimated  unexploited  TL ≥ 2 TL ≥ 3.5 TL ≥ 2 TL ≥ 3.5 marine animals teleosts teleosts  in the 2000s Jennings et al. (2008) Wilson et al. (2009)  unexploited unexploited unexploited  ∗ defined as having weight >10−5 g  21  ∗  (x  Biomass tonnes)  109  11.82 1.56 10.98 1.08 2.62 0.90 2.05  Figure 2.3: Map of the decade of maximum ecosystem catch recorded for the period 1950–2006.  2.3.2  Temporal and spatial trends in ecosystem maximum catch  Figure 2.3 shows a map of the decade of highest observed catch for each cell. Many industrial countries had the highest ecosystem catch in the 1950-1960s. A number of tropical regions, including islands and the coast of West Africa, experienced their highest catch in 1970-1980s. Some locations in the High Seas had their highest catch in the 1990s, but for most areas, catch continues to increase with 2000 being the decade of maximum recorded catch.  2.3.3  General trends in the decline of global marine biomass  Figure 2.4, left, shows the ratio (in %) of fished to unexploited biomass for decades between 1950 and 2006, based on the intermediate scenario (#2). Predator biomass, defined as the biomass in trophic levels 3.5 and higher, declines faster than total biomass (TL ≥ 2). Predator decline was further assigned to High Seas and EEZ cells (defining a 200 miles coastal zone). This showed that predators within EEZs declined faster (Fig. 2.4, center). Lastly, predator biomass in EEZs were disaggregated into individual oceans and Figure 2.4, right, shows the rates of decline by oceans and  22  hemisphere (North or South). There is a latitudinal trend in biomass decline going from North to South, with the North Atlantic and the North Pacific oceans showing the strongest decline overall.  2.3.4  Spatial trends of decline in predator biomass  Ecotroph’s predictions of biomass by trophic levels for decades between 1950 and 2006 can be compared to the predicted unexploited biomass. The % decline in predator biomass is defined as the ratio of fished to unexploited biomass for TL ≥ 3.5. Figure 2.5 shows the spatial evolution of predator decline under the intermediate scenario (#2) from the 1950 to the 2000s1 . Zones of high decline (mapped in green-yellow) were already present in northern oceans in the 1950s, continued to spread to equatorial and tropical waters in the following decades and extended to the High Seas from the 1980s onwards.  1  A map of predator biomass under the unexploited state is available in Appendix A.  23  100  Total High Seas  Biomass (%)  80  Antarctic  Predators 60  Indian South Atlantic South Pacific  EEZ 40  24  North Atlantic North Pacific  20 0  1950  1960  1970  1980  1990  2000  1950  1960  1970  1980  1990  2000  1950  1960  1970  1980  1990  2000  Year  Figure 2.4: Global trends in biomass left compared to the unexploited state for (left, 1) total (TL 2-5) vs. predators (TL ≥ 3.5); (center, 2) predators in High Seas and coasts (EEZs); (right, 3) predators in coasts by oceans, from South to North. These trends are calculated by summing up both the fished and unexploited biomass of the relevant group (total or predators) over the defined areas and taking the ratio Bf ished /Bunexploited .  2.3.5  Predator decline by regions  Many individual cells have recorded important declines of predators, but these values are not accurately represented when declines are aggregated over large regions. Figure 2.6, bottom, shows the proportion of area for each ocean that contains fully-accessed cells2 . I calculate the proportion of the coastal area of each ocean where the predator decline is equal or greater to 75% under the intermediate scenario (#2) (recovery based on TL) and the pessimistic scenario (#3) (no recovery), as well as the proportion where the decline is equal or greater than 90% (Figure 2.6, top).The South and North Pacific have the highest proportion of their area fully-accessed by fisheries. Also, both the North Atlantic and Pacific register important declines for predators, with respectively 60.2% and 56.8% of the area with declines of more than 75%, and 48.4% and 48.7% of the area with declines of more than 90%.  2.3.6  Scenarios of ecosystem response to fishing  Ecotroph’s predictions of biomass were processed under three scenarios meant to represent different interpretations of the global decline in fisheries catch. The optimistic scenario (#1) assumes declines in catch are caused by a reduction in fishing mortality, while the intermediate scenario (#2) assumes declines in catch are related to overexploitation, but that ecosystem biomass is allowed to recover if overexploitation stops. The pessimistic scenario (#3) also assumes that declines in catch are related to overexploitation, but ecosystem biomass is not allowed to recover. These resulted in three sets of biomass estimates by trophic level. Figure 2.7 compares the predicted declines in global total and predator biomass in 2000, within and outside of EEZs under the different scenarios. The optimistic scenario consistently predicts the highest proportions of ecosystem biomass left under fishing, including almost no decline for total biomass in the High Seas. The intermediate and pessimistic scenarios predict the lowest proportions of biomass remaining, with less than 40% of the predator biomass left in coastal areas in both cases. The biomass predictions made by the intermediate scenario are slightly higher than that made by the pessimistic scenario, with the differences between these scenarios greatest for the predator biomass in the coasts. 2  As defined in the methods: a decline in default fished biomass of at least 10% of biomass at MSY, where MSY is assumed to occur at 50% of the cell’s total unexploited biomass  25  Figure 2.5: Maps of the proportion of predator biomass (TL ≥ 3.5) remaining after each decade of fishing from the 1950s to 2000s under the intermediate scenario of ecosystem response to fishing (Scenario 2)... 26  Figure 2.5: (continued from last page) ... FAO areas are outlined with a black line. Note that a map of unexploited biomass for predators is presented as supporting material for this figure in Appendix A. Cells in dark purple have an unexploited ecosystem production of zero. 27  1.0  Predator biomass remaining: < 25%, Scenario 3 < 25%, Scenario 2 < 10%, Scenario 3 < 10%, Scenario 2  0.8  0.6  Proportion of coastal area with <10%, 25% predator biomass left 0.4  0.2  0.0 North Atlantic  North Pacific  South Pacific  Antarctic  South Atlantic  Indian  % exploited by industrial fisheries: 1  ● fully−accessed ● under−accessed  1  Figure 2.6: Above: Proportion of the coastal area of each ocean in which predator decline is higher than 75% (orange) and 90% (red) under the intermediate and pessimistic scenarios. Below: Pie charts showing the fraction of each ocean (EEZ and High Seas) classified as fully-accessed (orange) (calculated over both coastal and High Seas areas).  28  ● Scenario 1 (optimistic) ● Scenario 2 (intermediate) ● Scenario 3 (pessimistic)  Proportion of biomass left  1.0  0.8  0.6  0.4  0.2  0.0 HS  EEZ  Total  HS  EEZ  Predators  Figure 2.7: Comparison of the predicted proportion of biomass left in the 2000s compared to unexploited level for the three different scenarios applied to Ecotroph’s outputs. The proportions of biomass remaining are calculated for total (TL ≥ 2) and predator biomass (TL ≥ 3.5), and disaggregated into High Seas (HS) and coastal (EEZ) regions.  29  2.4  Discussion  This study assesses the impacts of fishing on ecosystem biomass at a global scale. I estimated the biomass of marine ecosystems worldwide by using a model that gives a simple but useful representation of ecosystems. My estimates of biomass are based on basic principles of energy transfer between trophic levels and account for primary production, sea surface temperature, fisheries catch and trophic level of species caught. The focus on trophic levels is especially appropriate in a fisheries context: high trophic levels species have historically been under heavier exploitation because of their high market demand and vulnerability to fishing gear, as well as are intrinsically more vulnerable to fishing (Dulvy et al. 2003). Moreover, the recent shift towards lower trophic level species in the catch (Pauly et al. 1998) potentially reflects a widespread change in ecosystem structure.  2.4.1  Global marine biomass estimates  Ecotroph generates estimates of unexploited marine biomass by trophic level based on primary productivity and a rate of energy loss between trophic levels (transfer efficiency, TE). This approach resulted in an estimated unexploited biomass of 11.82 x 109 tonnes for trophic levels ≥ 2 (i.e. including zooplankton), and 1.56 x 109 tonnes for predators (TL ≥ 3.5). This estimate is sensitive to the value of TE, with estimates of biomass declining rapidly with the value of this parameter (see Appendix B). These results can be compared to that of two other studies who have likewise attempted to estimate worldwide unexploited biomass for either the whole or a subset of marine ecosystems. Jennings et al. (2008) used a size-based method and estimated the global biomass of ‘marine animals’ (weight >10−5 g) to be 2.62 x 109 tonnes. Assuming that ‘marine animals’ are equivalent to biomass at TL ≥ 2, this value is less than a fourth of my estimate. Jennings et al. (2008)’s model was very sensitive to the transfer efficiency as well as an additional parameter called the predator-prey mass ratio. Focusing on higher trophic level organisms, Jennings et al. (2008)’s estimate for teleosts biomass is 0.90 x 109 tonnes. Another study by Wilson et al. (2009) estimated the global biomass of teleosts to be 2.05 x 109 tonnes based on Ecopath models. Because Ecotroph does not make the distinction between species, it is not possible to isolate the teleost component from the biomass trophic spectrum. However, I can use my estimate of biomass for TL ≥ 3.5 to 30  make a rough comparison and the value of 1.56 x 109 tonnes I obtained for TL ≥ 3.5 is between those estimated in the two other studies. It is probable that my estimates for teleosts would be in the same order of magnitude if I excluded non-teleosts (marine mammals, rays and skates, etc.) and included teleosts of TL ≥ 3.5 to my biomass estimate. This calculation would require the knowledge of the proportion of the biomass by trophic level that is made of teleost fishes, which is not available to my knowledge. I note that estimates generated through Ecotroph and Ecopath are consistently greater than that of Jennings et al. (2008). Whether or not the match between Ecotroph and Ecopath with Ecosim estimates is considered to increase confidence in their output, or is simply a result of their basic similarity (Gascuel et al. 2009) cannot be decided at present. Estimates of fished biomass were obtained by accounting for catch by TL and decreased rates of residence time of production by TL under fishing (i.e., the kinetics or P/B). With the default, optimistic output of the Ecotroph model (scenario #1), I estimated a fished biomass for the global oceans in the 2000s of 10.98 x 109 tonnes for TL≥ 2 and 1.08 x 109 tonnes for TL ≥ 3.5, representing declines from unexploited biomass of 7.1% and 30.8% respectively. The predicted declines seem relatively low, especially for the total biomass. However, one must keep in mind that about a third of ecosystem biomass (excluding phytoplankton) has TL <2.3 and is mostly made of species not targeted by fishing, such as zooplankton. Moreover, my results most probably under-estimate the decline of marine biomass. I only accounted for the impacts of direct catch. ‘Illegal, Unreported and Unregulated’ catches (IUUs) were not included in the analysis, despite making up to 40% of reported catch in some regions (Agnew et al. 2009). Discards, which are conservatively estimated at 6.8 million tonnes year−1 (Kelleher 2005), were also not included. The estimates of decline presented above are based on the optimistic scenario which assumes that declines in fisheries catches are due to a reduction in fishing mortality and that biomass recovers fully as soon as fishing stops. Even in the pessimistic scenario (#3) though, fishing mortality when catch declines is assumed to be constant, even though it has often been shown to increase in such instances (Mullon et al. 2005)(see below for a more detailed comparison between scenarios). Lastly, I assumed that all the biomass at a given TL was vulnerable to fishing, whereas this is often not the case. If I had calculated the declines over the proportion of biomass for each TL that is either accessible and/or targeted by fishing, the  31  predicted declines would have been stronger.  2.4.2  Trends in the effects of fishing on ecosystem biomass  I analyzed the effects of fishing on global ecosystem biomass in terms of three different factors: space, time and trophic level. Fishing impact, quantified here as a reduction in biomass by trophic level, is much stronger for predators and within coasts/EEZs (Figure 2.4). Fisheries impact was already strong in the 1950s in the North Atlantic and in the Southern East China Sea, with a number of cells showing predator decline of 75% or more (Figure 2.5). Areas of high impact continued to increase throughout the 1960s and 1970s and showed a gradual spread towards Southern waters (Western Africa, South America, East Asia) (Figures 2.4-3 and 2.5). The High Seas near China and Japan were affected as early as the 1950s, with pockets of depletion appearing worldwide in the 1980s (Figure 2.5). The 2000s show an accentuation of existing trends, especially in terms of impact on the High Seas (Figure 2.5). Except for the Antarctic, biomass declined between decades at about the same rate for southern oceans, but both the North Atlantic and North Pacific showed a fast decline of predator biomass in coastal areas up to the 1970s, with a leveling at about 20% of the unexploited biomass in the last two decades (Figure 2.4-3). This is the first study that models the decline in biomass due to fishing at a global scale, so it is not possible to compare it quantitatively to the work of other authors. My results, however, confirm the trends documented in the literature for the last two decades. The initial high impact in both the North Atlantic and North Pacific is expected from the history of industrial fisheries (Lotze 2007; Roberts 2007) and has been reported by other authors at least for the North Atlantic (Christensen et al. 2003). Observing a consistently strong impact in the South China Sea is also expected given China’s intense exploitation of its marine resources, coupled with average primary productivity (Watson and Pauly 2001). The Antarctic south of Africa (FAO zone #58) shows an important depletion in the 1980s, followed by a recovery in later decades. This pattern is expected given the low primary productivity of this region and the peak in the krill catch that occured at the beginning of the 1980s. The ensuing depletion in higher trophic levels appears reasonable given the importance of krill to the Antarctic’s foodweb (Croxall et al. 1999). The gradual transition in impact from Northern to Southern waters is a result of increased 32  exploitation of tropical waters by distant fleets of Northern (i.e., developed) countries starting in the 1960s (Alder and Sumaila 2004). My model predicts a relatively low fishing impact in the High Seas compared to the coasts, with almost no cells having predator biomass decline beyond 40% of the unexploited biomass. This is expected from the extremely large areas (High Seas, defined as being outside of territorial waters, make up 64.6% of the global oceans in my model) and increased costs and technological challenges of fishing away from the coasts. However, it is probable that the trophic-level approach (and aggregation over large areas) failed to capture the drastic decline of many large predators species (Ward and Myers 2005). I discuss this issue in further details below. Lastly, examination of Halpern et al.’s (2008) map of human impact on the oceans shows that my study captures many of the same qualitative spatial trends, especially in terms of high impact in the North Atlantic and South China Sea, but also in terms of the High Seas being the least affected. My results consistently show that the effects of fishing are greater on predators than on lower trophic levels (Figure 2.4-1 and Figure 2.7). This is not surprising given the distribution of energy in ecosystems, with about 13% of the biomass of animals residing at TL greater than 3.5 vs. making up approximately 40% of the global catch since the 1950s (see Appendix A). In addition, the biomass of predators in Ecotroph is negatively impacted by the loss of food source when their prey is being fished. It would be expected that the progressive shift towards lower TL species in the catch has exacerbated this phenomenon. In the past 10 years, a number of studies have reported strong decline in the global abundance of marine predators, primarily from two types of data: proxies of biomass and time-series of abundance (Myers and Worm 2003; Pauly et al. 1998). My modelling approach adds a third, independent confirmation to this observation. Close examination of Figure 2.4-3 shows that my estimates of average predator decline are somehow conservative. For example, the North Atlantic, where I predict the worst decline, shows a decline of 80.3%, whereas other studies (e.g. Christensen et al. 2003) have documented declines of more than 90%. My results can be explained partly by the fact that I looked at entire trophic levels and not specific species so that the declines I calculated are averaged over entire sections of the foodweb. Also, I presented declines aggregated over large regions (for e.g., the smallest region, the North Atlantic outside of EEZs, has an area of 23.7 million km2 ), which failed to capture the heterogeneity of the  33  decline between individual cells. To compensate for this, Figure 2.6 looked at the proportion of the area of exploited coastal cells that show a decline in predator biomass of more than 75% and more than 90%. The North Atlantic and the North Pacific, which have historically been the most exploited, have respectively 60.2% and 56.8% of their area with a predator decline greater than 75% and 48.4% and 48.7% of their area with a decline higher than 90%. For the North Atlantic, the proportion where decline is higher than 75% climbs to 69.5% under my pessimistic scenario. These values compare to other studies that have estimated declines of more than 90% in predator biomass (Pauly et al. 1998; Myers and Worm 2003, but see Walters 2003). Local field studies have also shown similar results. Friedlander and DeMartini (2002), for example, surveyed fish biomass (including some TLs >3.5) in North Hawaii under heavily and lightly fished conditions and found a decline of biomass of about 61.5%, with the decline stronger for large apex predators. The decline of predators is predicted to be much lower in the High Seas, which could indicate that they remain relatively pristine. Realistically however, it is probable that my results probably do not capture the decline in the biomass of pelagic predators with TL>4, which make up about half of the total High Seas catch. Production is distributed heterogeneously in the open ocean and many species are highly migratory. This is not represented either in the catch data (for the majority of species, production fronts are not accounted for when defining regions of high catch density) or the model formulation since each cell is assumed to be a closed system, with no dispersal between cells. Also, as before, averaging over extremely large areas masks strong localized decline and this is especially true for the High Seas.  2.4.3  Scenarios of ecosystem response to fishing  The default predictions of biomass made by Ecotroph were re-analyzed in terms of three different scenarios of ecosystem response to fishing. This extra step was necessary because Ecotroph directly relates fishing mortality to catch, with the underlying assumption that declining catches are caused by a reduction in fishing mortality and result in an increase in biomass. There are currently no complete global dataset of fishing mortality and/or effort which could have been used to explain whether declining catches were caused by reduced effort or over-exploitation. However, even if I had had access to such a dataset, fisheries dependent catch per unit effort time-series are rarely 34  a reliable indicator of abundance (Walters and Martell 2004). An alternative would have been to directly include a relationship in the model to describe the ecosystem response to fishing, such that, for example, the biomass associated to a declining catch in 2000 would have been a function of historical catches. However, such a relationship implies the need to define ecosystem resilience to fishing as a function of TL, possibly accounting for local abiotic and biotic factors. My results would thus have been driven by the selected shape of the relationship. Also, defining such a relationship is not straight-forward: the question of single-species resilience to fishing is still one of the most challenging in fisheries science (but see Cheung et al. (2005) on species intrinsic vulnerability to extinction) and it is unclear how population-specific principles scale up at the ecosystem level. For example, it could be that ecosystems become more productive when fished because populations are made of smaller individuals, which turn-over faster (Denney et al. 2002). On the other hand, the ability of energy to be transferred up trophic levels might be hindered by a ‘trophic cul-de-sac’ (defined as the proliferation of some lower-trophic species that are not vulnerable to predation; Bishop et al. 2007; Richardson et al. 2009), such that ecosystems become less efficient at converting energy when under exploitation. Instead of adding an extra layer of complexity to my model based on uncertain principles, I opted to maintain a simpler model structure and adjust the estimates of biomass at equilibrium from a set of clear, straight-forward rules structured within three different scenarios. While the resulting predictions of biomass also depended on how the rules were defined, I still felt that it was more appropriate to keep the simpler, more prudent option than venture in the complicated and uncertain path of defining a function of ecosystem resilience to fisheries. Moreover, the use of time series of catch to infer periods of ‘sustainability’ or overfishing in datapoor situations has been proposed by other authors (for e.g., see MacCall 2009). In the future, the rules driving the scenarios could be fine-tuned to be more representative of real systems. For instance, the recovery function of the intermediate scenario could be parameterized from known intrinsic rates of natural increase vs. weight (Blueweiss et al. 1978) and their relationship to TL. My most conservative scenario (#1) uses the default biomass predictions made by Ecotroph. Not surprisingly, this scenario predicts higher ecosystem biomass since the biomass at a given TL can only be significantly reduced if all of its production is removed either through direct fishing or  35  fishing of lower TLs. Also, this scenario assumes that as soon as fishing pressure is removed, the target TLs are automatically repopulated by the next pulse of production. Figure 2.7 shows that the optimistic scenario predicts small reduction of ecosystem biomass due to fishing but that the main trends are conserved, that is, depletion is strongest for predators and along the coasts. In the intermediate scenario (#2), the intermediate scenario, biomass is assumed to decline proportionally to catch when an overexploitation signal is recorded (see Methods for definition), but is then allowed to recover as a function of TL if the signal stops. In the pessimistic scenario (#3), the pessimistic scenario, biomass is not allowed to recover if overexploitation stops, that is, it either stays constant between decades or declines because of higher catches. These two scenarios gave very similar predictions of biomass especially when looking at the High Seas (Figure 2.7). This is expected because most recorded catch recoveries occurred for species at low trophic level species, which are mostly fished within EEZs. Even then, the largest difference between scenarios 2 and 3 in the predicted proportion of biomass left in 2000 (predators within coasts) was only 2.55% (Figure 2.7).  2.4.4  Issues and possible solutions  In this study, the impact of fishing on ecosystem biomass are only driven by direct catch whereas fishing affects ecosystems in a number of indirect ways, such as habitat disturbance (Watling and Norse 1998), discards (Hall et al. 2000) or ecological effects like trophic cascades (Scheffer et al. 2005). It is therefore important to acknowledge that my results only focus on a subset of the possible ecosystem impacts of fishing. Moreover, the accuracy of Ecotroph’s predictions is driven primarily by the quality of the data, the validity of the parameters used and the model formulation, three factors which I critically discuss below in light of my study’s objectives. Data: The primary production dataset is derived from satellite data, which have a number of recognized issues (Gregg and Casey 2004). The resulting estimates of primary productivity only account for oceanic and coastal phytoplankton, which make up about 94.5% of marine primary production (Duarte and Cebrian 1996). Alternative modes of production, like ice algae in the poles or the detrital pathway, are hard to quantify at a global scale. It is therefore likely that primary production 36  was under-estimated in some regions. Also, I would ideally have used temporal estimates of PP for 1950-2006. However models of past primary production (e.g. Sarmiento et al. 2004) are relatively recent and I felt more confident using mean PP value from a well-established dataset that still captures the main spatial trends of marine productivity. I also showed that the year from which the PP used came from (1998) was representative of the PP since the 1950s (Appendix A). Another important driver to the model was the fisheries catch which came from the global dataset built by the Sea Around Us project. This dataset is based on an algorithm that allocates catch data to individual cells and species based on a set of decision rules. While it is the most comprehensive dataset available, it has some recognized issues. For instance, catches mapped on the High Seas frequently show sharp boundaries between FAO areas (see Figure 2.5 for examples). Also, the historical catch data for a number of small island states is incomplete and unless they have access agreements with the distant-water fleets of other countries, the catch in these countries’ EEZs is assumed to be only that officially recorded by the country itself. This can be seen in Figure 2.5 in the form of ‘belts’ of lower fishing impact around many islands. In my case, however, fine scale attribution of catch is not a serious issue since I am interpreting results at a large spatial resolution. As previously mentioned, it would also have been ideal to include IUU and discards data, but this, again, was not possible due to the lack of adequate spatially-explicit global datasets. Lastly, the smooth parameter that controls how species-specific catches are distributed amongst trophic levels can have a small effect on the predicted biomass for the ecosystem, depending on how catch are distributed (Appendix B). I felt that the value of smooth = 0.075 was a reasonable choice in terms of the within-species TL variation it resulted in for each trophic level (Appendix B, table B.4). Parameters: The most sensitive parameter in Ecotroph is the transfer efficiency (TE), that is, the proportion of production that is retained in transfers between TL. My sensitivity analysis showed that TE had a high impact both on the predicted biomass of an ecosystem and the resulting proportion of biomass removed from fishing (see Appendix B). As transfer efficiency becomes smaller, the decline in biomass from a given level of catch increases. However, the relative spatial trends are not affected 37  by TE for most systems, that is, the impact of fishing on one cell compared to the next remains the same as long as catches do not exceed production for a wide TL range. In terms of my objectives of extracting general spatial and temporal trends, my results are therefore robust to the uncertainty in the TE. This would not necessarily be the case if, as would be expected, TE varies between regions. In such instances, the biomass of systems with higher TE would have been underestimated and the biomass of systems with lower TE overestimated, which would affect the observed relative patterns of biomass decline. There is currently no consensus on trends in TE between systems, which is why I used a constant value for all systems (but see Christensen and Pauly 1993). Additionally, transfer efficiency could change between trophic levels. For instance, higher TL might be less efficient than lower TLs at converting energy because of their larger sizes and slower turn-over rates. Also, transfer efficiency between trophic levels 1 and 2, which is strongly influenced by rates of herbivory, could be very different than that of heterotrophs TL (Cebrian 1999). Lastly, it could be that fishing changes the transfer efficiency of an ecosystem: for example, the removal of a given species allows the appearance of others that are more or less efficient at converting energy or are invulnerable to predators, or if all of the species of higher trophic levels are locally driven to extinction such that the length of the food chain is effectively shortened. In short, knowledge of the biotic or physical factors that drive TE would be an important step towards a better understanding of ecosystem functioning but is currently lacking. Trophic cascades due to fishing have been documented worldwide and it would have been extremely interesting to account for this phenomenon in my model. The parameterization of Ecotroph used in this report assumes that there are no top-down effects. This decision was driven primarily by the lack of scientific consensus on the factors that affect the strength of top-down effects in marine systems (Borer et al. 2005; Gruner et al. 2008; Shurin et al. 2002). I performed a sensitivity analysis (Appendix B) which shows that, as expected, both the absolute value of ecosystem biomass and the distribution of biomass between TLs could be affected by the inclusion of topdown effects if the system is heavily fished. In future work, it could be interesting to look at how different hypotheses about the drivers of top-down effects could change the predicted declines in global marine biomass due to fishing. Also, given that recycling is important in a number of coastal  38  ecosystems (Duarte and Cebrian 1996), I could have included a recycling component to my model such that additional production can be included through the detrital pathway. Again though, the issue was to give spatial weight to the relative importance of recycling between systems and in this first exercise I preferred to assign a constant value for ease of comparison between cells. In general, for any parameter, having an absolute value per se is not what is interesting but rather the spatial variation around the mean value between systems. Whenever robust datasets or ecological understanding were lacking to define those spatial variations, I chose to use the same value for all systems. The kinetics for the unexploited state of the ecosystems were derived from an empirical model of P/B parameterized from the analysis of a large set of published Ecopath models (Gascuel et al. 2008). Most of the Ecopath models included fishing, but in the Methods it is assumed that the parameterized relationship is representative of the unexploited state. However, even if a bias is present, it will be the same for all cells as the kinetics predicted by the empirical model will be evenly over-estimated. Moreover, since the main results I presented are in the form Bf ished /Bunexploited , the kinetics cancelled out from equation 2.1, with only the effects of fishing from the cell’s catch accounting as a factor (compare equations 2.6 and 2.7). Therefore, I am confident that the effects of such a bias on my results are minimal. Model formulation: The Ecotroph model is a very simplified representation of ecosystems, which is justified by its objective to extract general trends of ecosystem structure and function. One of its main premises is that a linear representation of ecosystem is able to capture broad trends in distribution between TL. The underlying assumption here is that only processes of predation (i.e., bottom-up and/or top-down effects) determine the biomass at any trophic levels. In terrestrial systems, it has been shown that production at any TL could be strongly affected by community composition, and that competition between species could have a strong effect on broad ecosystem properties (Tilman et al. 1997). Whether such processes are as important in marine systems is unclear, but it is has been observed that some competing species are considerably more nutritious than others, such that their rate of transfer up the food web is higher (e.g. Atkinson et al. 2004). Also, recruitment 39  feedback is not accounted for in the current formulation of the model, that is, even if the biomass of a trophic level interval is completely depleted by the catch of one year, it will be regenerated in the following year by a new pulse of production. This is probably representative of lower trophic levels that have very fast generation rates and for which fishing only accesses a small proportion of the total biomass. However, this assumption becomes problematic for higher TL where both biomass and diversity (number of species feeding at this TL) decline. An improved formulation of the model could define a relationship between the biomass at a given TL and future recruitment as a function of TL, with the assumption that as TL increases, past biomass has a stronger effect on future recruitment.  2.4.5  Conclusions  By using a modelling approach, my study confirmed three main trends about the effects of global fishing on ecosystems: the impacts of fishing are considerably greater for predators, are concentrated in coastal areas and have gradually expanded from northern to equatorial and southern waters since the 1950s. I showed that industrial fisheries have severely reduced the biomass of predators in most areas of historically exploited oceans and that this trend is spreading rapidly to the non-traditional fishing grounds developed more recently. The advantages of using Ecotroph to model marine ecosystems globally are two-fold. First, by focusing on trophic levels and processes of energy transfer it gives a general overview of ecosystem structure and function. Second, Ecotroph is especially well suited to data-poor situations and can generate biomass estimates from the datasets that are currently globally available. There are many additional factors that would have been relevant to include in this assessment of the impacts of global fisheries. However, it is important to emphasize that this is a first exercise in global marine ecosystem modelling under the effects of fishing and that the factors included have been shown to be important drivers of ecosystem function and response to fishing (Lopez-Urrutia et al. 2006; Chassot et al. 2010). While this was outside the scope of this study, the predictions made by Ecotroph can be easily compared to other ecosystem models or field data by aggregating species by trophic level, and offer the opportunity to highlight which processes of ecosystem functioning are especially important to understand trophic structure. 40  To conclude, in the last decade, many studies have reported the effects of fishing at a local scale, mostly through field studies and the analysis of time-series. My thesis is the first to model directly the effects of fishing at a global scale and brings together many of the observations made in isolation in the literature. The one trend consistently observed is that fishing has removed a considerable portion of the top of the energy pyramid of ecosystems. While my study focused on the effects of fishing only from the stand-point of direct biomass removal, the prediction of generalized predator decline implies widespread and fundamental changes to both the structure and the functioning of global marine communities.  41  Chapter 3  Conclusion 3.1  Summary  Humans have exploited marine ecosystems for millennia but the last 50 years have seen a drastic increase in the intensity and extent of fisheries. Ecological phenomena such as trophic cascades and jellyfish explosions are observed in marine systems worldwide, which suggest that fisheries may have widespread effects on ecosystem functioning. Understanding this impact at multiple scales, from local to regional to global, is important to address the overexploitation of marine populations and is an essential step in the development of efficient management practices. In particular, a global perspective is useful to extract large-scale trends in the effects of fisheries. Global studies are also powerful tools to communicate with both specialists from other disciplines and the general public, an important consideration given that the success of management practices is often driven by factors external to the biology of the exploited system. Global studies of the effects of fisheries have so far been based on either the meta-analysis of data from local systems or indirect inference from data used as proxies (mostly fisheries catch data). The main objective of this Master’s thesis was to evaluate the effects of global fisheries by directly modelling the extent to which fishing decreased the biomass of marine ecosystems. This thesis is the first application of modelling to marine ecosystems that accounts for the effects of fishing at a global scale. I evaluated the impacts of fisheries in terms of the proportion of ecosystem biomass removed compared to the unexploited state. I demonstrated three main trends 42  in the effects of global fisheries: the impacts of fisheries are strongest in coastal areas, impacted areas extend from northern to equatorial and southern waters starting in the 1960-1970s and towards the High Seas in the 1980-1990s, and top-predators are consistently the most affected, with some areas showing very strong declines in the biomass of this group. The results presented in this thesis thus confirm through an independent approach the major trends reported in the scientific literature of the last 15 years. The application of a modelling approach to marine ecosystems at a global scale is what differentiates this work from others. While modelling marine ecosystems is in itself not new, the key is the recognition that the objectives of ecosystem modelling differ at local and global scales and that it is possible to build a simple ecosystem model that can answer questions relevant to a global fisheries context. Here I used a simple ecosystem model called Ecotroph , which represents ecosystems in terms of biomass flow (or energy, a unit common to all living systems), and was able to summarize the effects of global fisheries in terms of space, time and trophic levels. My evaluation of the effects of fisheries focused on the impact of direct catch only and this impact is quantified in terms of a reduction of ecosystem biomass. However, the confirmation that predators worldwide are impacted, often quite severely, hint of more widespread and important ecological consequences than that predicted by my model, and this is without including other effects of fisheries like habitat disruption and discards.  3.2  Challenges  Reflecting back on the process that lead to this Master’s thesis, I identify three main challenges in the production of this work. The first one was expected beforehand and was the most straightforward to address. It consisted in the harmonization of a number of global databases to run the model and obtain temporal estimates of global biomass by trophic levels under fished and unexploited states for about 180,000 independent cells (making up a 0.5x0.5 degree grid covering the world’s oceans). The second challenge is one common to all modelling exercises and consists in model calibration to ensure that the results produced are realistic. In my case the objectives were to estimate global  43  ecosystem biomass under fishing so, by definition, having a dataset of global biomass to calibrate the model would have meant the modelling exercise was moot. However, calibration with local datasets was possible in some instances. Sensitivity analyses run on a selected set of local ecosystems highlighted the fact that the Ecotroph model was especially sensitive to the parameter of transfer efficiency. The absolute values of biomass predicted are thus an estimate only, but the spatial trends derived are robust to the transfer efficiency parameter. While it was a disappointment that the predicted biomass values had to be interpreted in light of the transfer efficiency, the question of how energy is transferred between living units is a fundamental one in ecology and it should be expected that a model based on energy transfer would be sensitive to this parameter. The third and greatest challenge was initially not foreseen and consisted in the realization that the Ecotroph model underestimated the impacts of fishing in instances when catch was declining because of overexploitation. This is most probably due to the absence of a recruitment feedback function. Implementing such a function and applying it at a global scale would have been ideal but was outside the scope of this thesis. To compensate for this behaviour, I created and applied three simple scenarios of ecosystem responses to fishing and interpreted the ecosystem biomass predicted by Ecotroph accordingly. While it was frustrating to have a pragmatic rather than elegant solution to this problem, the process of realizing what the issues were and the ensuing reflection about possible solutions were very valuable. In particular, it highlighted the fact that the effects of fisheries at the ecosystem scale might be understood not only in terms of how fishing directly or indirectly affects population biomass and species interactions, but also in terms of how it affects the rate at which energy is used and converted.  3.3  Future directions  One of the main uses of models is to estimate properties we cannot measure directly and the global biomass of marine ecosystems certainly qualifies under that criterion. Models are also helpful to summarize and evaluate our understanding of a given system. In the case of Ecotroph, there is no doubt that it generates a very simplified version of ecosystem functioning. However the key is not to know that the model fails, but where it does and what additional parameters/responses could  44  improve the quality of its predictions. Comparing the model’s predictions to field data or other ecosystem models is straight-forward if one has data of trophic level by species. Such comparisons could be used to inform which processes might be especially important in specific systems by highlighting where the Ecotroph model in its current formulation succeeds or fails in capturing a defined set of ecosystem properties. Another interesting feature of Ecotroph is that it can include a top-down effect such that the properties of a given trophic level are affected by the biomass of the trophic levels that preys on it. However, while it is clear to the scientific community that top-down effects can play an extremely important role in ecosystem functioning, there is still no consensus about the drivers behind them and how they change between systems. As a result, I was unable to assign a robust value for the top-down effect that covered the world’s marine ecosystems and decided not to include the effect in the analysis. An interesting future direction to this work could be to investigate how different hypotheses about the factors that influence the strength of the top-down effect would affect the predicted spatial, temporal and trophic trends about the effects of fishing.  3.4  Conclusions  This Master’s thesis provides an independent confirmation that fishing has a widespread effect on the global oceans by using an innovative method. It demonstrates a practical use of a modelling approach at a global scale and points to energy as a useful concept to understand the ecosystem effects of fishing. I measured fishing impact in terms of biomass reduction. Such an effect is not necessarily negative, and indeed would be expected even in a well-managed fisheries. The key here is to define how fishing affects the status of desired ecosystem properties (defined either from ecosystems services or intrinsic values) and understand when fishing (along with other anthropogenic stressors) pushes the ecosystem into an ‘undesirable’ zone. Defining the relationship of ecosystem response to fishing is one of the next big steps in ecosystem-based management and it is sobering to realize that we are still struggling in defining this response with individual species. A fundamental question is the degree to which it is possible to extrapolate to the ecosystem scale the type of responses  45  observed with individual species, which is basically the same question community ecologists have been asking for decades. Solving the practical ecological questions brought about by ecosystembased-management will hopefully improve communication between the estranged sister fields of fisheries science and ecology. In the mean time, it might be that energy could serve as a useful unifying concept to consolidate our understanding of ecosystem functioning across different scales of biological organization.  46  References Agnew, D. J., Pearce, J., Pramod, G., Peatman, T., Watson, R., Beddington, J. R., and Pitcher, T. J. (2009). Estimating the worldwide extent of illegal fishing. PLoS ONE, 4(2):e4570. Alder, J. and Sumaila, U. (2004). Western Africa: A Fish Basket of Europe Past and Present. Journal of Environment and Development, 13(2):156–178. Atkinson, A., Siegel, V., Pakhomov, E., and Rothery, P. (2004). Long-term decline in krill stock and increase in salps within the southern ocean. Nature, 432(7013):100–103. Baum, J. and Worm, B. (2009). Cascading top-down effects of changing oceanic predator abundances. Journal of Animal Ecology, 78(4):699–714. Beaugrand, G. (2004). The North Sea regime shift: evidence, causes, mechanisms and consequences. Progress in Oceanography, 60(2-4):245–262. Berkes, F., Hughes, T., Steneck, R., Wilson, J., Bellwood, D., Crona, B., Folke, C., Gunderson, L., Leslie, H., Norberg, J., Nystrom, M., Olsson, P., Osterblom, H., Scheffer, M., and Worm, B. (2006). Globalization, roving bandits, and marine resources. Science, 311(5767):1557–1558. Bishop, M., Kelaher, B., Alquezar, R., York, P., Ralph, P., and Skilbeck, C. (2007). Trophic culde-sac, Pyrazus ebeninus, limits trophic transfer through an estuarine detritus-based food web. Oikos, 116(3):427–438. Blueweiss, L., Fox, H., Kudzma, V., Nakashima, D., Peters, R., and Sams, S. (1978). Relationships between body size and some life history parameters. Oecologia, 37:257–272. Borer, E., Seabloom, E., Shurin, J., Anderson, K., Blanchette, C., Broitman, B., Cooper, S., and Halpern, B. (2005). What determines the strength of a trophic cascade? Ecology, 86(2):528–537. Bouvet, M., Hoepffner, N., and Dowell, M. (2002). Parameterization of a spectral solar irradiance model for the global ocean using multiple satellite sensors. Journal of Geophysical Research Oceans, 107(C12):8.1–8.18. Casini, M., Lovgren, J., Hjelm, J., Cardinale, M., Molinero, J.-C., and Kornilovs, G. (2008). Multilevel trophic cascades in a heavily exploited open marine ecosystem. Proceedings of the Royal Society B: Biological Sciences, 275(1644):1793–1801. Cebrian, J. (1999). Patterns in the fate of production in plant communities. American Naturalist, 154(4):449–468. 47  Chassot, E., Bonhommeau, S., Dulvy, N. K., M´elin, F., Watson, R., Gascuel, D., and Pape, O. L. (2010). Global marine primary production constrains fisheries catches. Ecology Letters, 13(4):495–505. Chavez, F., Ryan, J., Lluch-Cota, S., and Niquen, M. (2003). From anchovies to sardines and back: Multidecadal change in the pacific ocean. Science, 299(5604):217–221. Cheung, W., Pitcher, T., and Pauly, D. (2005). A fuzzy logic expert system to estimate intrinsic extinction vulnerabilities of marine fishes to fishing. Biological Conservation, 124(1):97–111. Christensen, V., Guenette, S., Heymans, J., Walters, C., Watson, R., Zeller, D., and Pauly, D. (2003). Hundred-year decline of north atlantic predatory fishes. Fish and Fisheries, 4(1):1–24. Christensen, V. and Pauly, D. (1993). Flow characteristics of aquatic ecosystems. In ICLARM Conference Proceedings 26, Manila. Clark, C. W. (1990). Mathematical Bioeconomics: The optimal management of renewable resources. Wiley-Interscience, New York. Constable, A., de la Mare, W., Agnew, D., Everson, I., and Miller, D. (2000). Managing fisheries to conserve the antarctic marine ecosystem: practical implementation of the convention on the conservation of antarctic marine living resources (ccamlr). ICES Journal of Marine Sciences, 57(3):778–791. Crowder, L. B., Hazen, E. L., Avissar, N., Bjorkland, R., Latanich, C., and Ogburn, M. B. (2008). The impacts of fisheries on marine ecosystems and the transition to ecosystem-based management. Annual Review of Ecology, Evolution and Systematics, 39:259–278. Croxall, J., Reid, K., and Prince, P. (1999). Diet, provisioning and productivity responses of marine predators to differences in availability of antarctic krill. Marine Ecology Progress Series, 177:115–131. Cury, P. and Shannon, L. (2004). Regime shifts in upwelling ecosystems: observed changes and possible mechanisms in the northern and southern Benguela. Progress in Oceanography, 60(24):223–243. Daskalov, G. M., Grishin, A. N., Rodionov, S., and Mihneva, V. (2007). Trophic cascades triggered by overfishing reveal possible mechanisms of ecosystem regime shifts. Proceedings of the National Academy of Science of the United States of America, 104(25):10518–10523. Denney, N., Jennings, S., and Reynolds, J. (2002). Life-history correlates of maximum population growth rates in marine fishes. Proceedings of the Royal Society B: Biological Sciences, 269:2229– 2237. Duarte, C. and Cebrian, J. (1996). The fate of marine autotrophic production. Limnology and Oceanography, 41(8):1758–1766. Dulvy, N., Sadovy, Y., and Reynolds, J. (2003). Extinction vulnerability in marine populations. Fish and Fisheries, 4(1):25–64.  48  Essington, T., Beaudreau, A., and Wiedenmann, J. (2006). Fishing through marine food webs. Proceedings of the National Academy of Science of the United States of America, 103(9):3171– 3175. FAO (2009). The state of the world fisheries and aquaculture 2008. FAO, Rome, Italy. Frank, K., Petrie, B., Choi, J., and Leggett, W. (2005). Trophic cascades in a formerly coddominated ecosystem. Science, 308(5728):1621–1623. Frank, K. T., Petrie, B., and Shackell, N. L. (2007). The ups and downs of trophic control in continental shelf ecosystems. Trends in Ecology & Evolution, 22(5):236–242. Friedlander, A. and DeMartini, E. (2002). Contrasts in density, size, and biomass of reef fishes between the northwestern and the main hawaiian islands: the effects of fishing down apex predators. Marine Ecology Progress Series, 230:253–264. Fulton, E., Smith, A., and Johnson, C. (2003). Effect of complexity on marine ecosystem models. Marine Ecology Progress Series, 253:1–16. Fulton, E., Smith, A., and Johnson, C. (2004). Biogeochemical marine ecosystem models i: IGBEM - a model of marine bay ecosystems. Ecological Modelling, 174(3):267–307. Gascuel, D. (2005). The trophic-level based model: A theoretical approach of fishing effects on marine ecosystems. Ecological Modelling, 189(3-4):315–332. Gascuel, D., Morissette, L., Palomares, M. L. D., and Christensen, V. (2008). Trophic flow kinetics in marine ecosystems: Toward a theoretical approach to ecosystem functioning. Ecological Modelling, 217(1-2):33–47. Gascuel, D. and Pauly, D. (2009). Ecotroph: modelling marine ecosystem functioning and impact of fishing. Ecological Modelling, 220(21):2885–2898. Gascuel, D., Tremblay-Boyer, L., and Pauly, D. (2009). Ecotroph (ET): a trophic level based software for assessing the impacts of fishing on aquatic ecosystems, Fisheries Centre Research Report 17(1). University of British Columbia, Vancouver. Gordon, S. H. (1954). The economic theory of a common-property resource: the fishery. Journal of Political Economy, 62:124–142. Gregg, W. and Casey, N. (2004). Global and regional evaluation of the seawifs chlorophyll data set. Remote Sensing of the Environment, 93(4):463–479. Gruner, D. S., Smith, J. E., Seabloom, E. W., Sandin, S. A., Ngai, J. T., Hillebrand, H., Harpole, W. S., Elser, J. J., Cleland, E. E., Bracken, M. E. S., Borer, E. T., and Bolker, B. M. (2008). A cross-system synthesis of consumer and nutrient resource control on producer biomass. Ecology Letters, 11(7):740–755. Hall, M., Alverson, D., and Metuzals, K. (2000). By-catch: Problems and solutions. Marine Pollution Bulletin, 41(1-6):204–219.  49  Halpern, B. S., Walbridge, S., Selkoe, K. A., Kappel, C. V., Micheli, F., D’Agrosa, C., Bruno, J. F., Casey, K. S., Ebert, C., Fox, H. E., Fujita, R., Heinemann, D., Lenihan, H. S., Madin, E. M. P., Perry, M. T., Selig, E. R., Spalding, M., Steneck, R., and Watson, R. (2008). A global map of human impact on marine ecosystems. Science, 319(5865):948–952. Heithaus, M. R., Frid, A., Wirsing, A. J., and Worm, B. (2008). Predicting ecological consequences of marine top predator declines. Trends in Ecology & Evolution, 23(4):202–210. Hilborn, R. (2007). Managing fisheries is managing people: what has been learned? Fisheries, 8(4):285–296.  Fish and  Hughes, T. (1994). Catastrophes, phase-shifts, and large-scale degradation of a caribbean coral-reef. Science, 265(5178):1547–1551. Jackson, J. (2001). What was natural in the coastal oceans? Proceedings of the National Academy of Science of the United States of America, 98(10):5411–5418. Jennings, S., M´elin, F., Blanchard, J., and Forster, R. (2008). Global-scale predictions of community and ecosystem properties from simple ecological theory. Proceedings of the Royal Society B: Biological Sciences, 275:1375–1383. Jiao, Y. (2009). Regime shift in marine ecosystems and implications for fisheries management, a review. Reviews in Fish Biology and Fisheries, 19(2):177–191. Kawahara, M., Uye, S., Ohtsu, K., and Izumi, H. (2006). Unusual population explosion of the giant jellyfish Nemopilemia nomurai (Scyphozoa : Rhizostomeae) in East Asian waters. Marine Ecology Progress Series, 307:161–173. Kelleher, K. (2005). Discards in the world’s marine fisheries: an update. FAO, Rome, Italy. Lai, S. (2004). Primary production. Sea Around Us Project, Fisheries Centre, the University of British Columbia, Vancouver. Accessible at: www.seaaroundus.org/saup/_manual.htm/#3. Lopez-Urrutia, A., Martin, E. S., Harris, R., and Irigoien, X. (2006). Scaling the metabolic balance of the oceans. Proceedings of the National Academy of Science of the United States of America, 103(23):8739–8744. Lotze, H. (2007). Rise and fall of fishing and marine resource use in the wadden sea, southern north sea. Fisheries Research, 87(2-3):208–218. MacCall, A. (2009). Depletion-corrected average catch: a simple formula for estimating sustainable yields in data-poor situations. ICES Journal of Marine Sciences, 66:2267–2271. McLeod, K., Lubchenco, J., Palumbi, S., and Rosenberg, A. (2005). Scientific Consensus Statement on Marine Ecosystem-Based Management. Signed by 221 academic scientists and policy experts with relevant expertise and published by the Communication Partnership for Science and the Sea at http://compassonline.org/?q=EBM. Mullon, C., Freon, P., and Cury, P. (2005). The dynamics of collapse in world fisheries. Fish and Fisheries, 6(2):111–120. 50  Myers, R. and Worm, B. (2003). Rapid worldwide depletion of predatory fish communities. Nature, 423:280–283. Pauly, D. (2007). The Sea Around Us Project: Documenting and communicating global fisheries impacts on marine ecosystems. AMBIO, 36(4):290–295. Pauly, D. and Christensen, V. (1995). Primary production required to sustain global fisheries. Nature, 374:255–257. Pauly, D., Christensen, V., Dalsgaard, J., Froese, R., and Torres, F. (1998). Fishing down marine food webs. Science, 279(5352):860–863. ´ (2007). Models for an ecosystem approach to fisheries. FAO Fisheries Technical Paper, Plag´anyi, E. Italy, Rome. Platt, T. and Sathyendranath, S. (1988). Oceanic primary production - estimation by remotesensing at local and regional scales. Science, 241(4873):1613–1620. Pope, J. (1972). An investigation into the accuracy of virtual population analysis. International Commission for the Northwest Atlantic Fisheries Research Bulletin, 9:65–74. Rice, J. (2000). Evaluating fishery impacts using metrics of community structure. ICES Journal of Marine Sciences, 57(3):682–688. Richardson, A., Bakun, A., Hays, G., and Gibbons, M. (2009). The jellyfish joyride: causes, consequences and management responses to a more gelatinous future. Trends in Ecology & Evolution, 24(6):312–322. Roberts, C. (2007). The Unnatural History of the Sea. Island Press, Washington DC. Rochet, M. and Trenkel, V. (2003). Which community indicators can measure the impact of fishing? a review and proposals. Canadian Journal of Fisheries and Aquatic Sciences, 60(1):86–99. Sarmiento, J., Slater, R., Barber, R., Bopp, L., Doney, S., Hirst, A., Kleypas, J., Matear, R., Mikolajewicz, U., Monfray, P., Soldatov, V., Spall, S., and Stouffer, R. (2004). Response of ocean ecosystems to climate warming. Global Biogeochemical Cycles, 18(3):GB3003. Scheffer, M., Carpenter, S., and de Young, B. (2005). Cascading effects of overfishing marine systems. Trends in Ecology & Evolution, 20(11):579–581. Shurin, J., Borer, E., Seabloom, E., Anderson, K., Blanchette, C., Broitman, B., Cooper, S., and Halpern, B. (2002). A cross-ecosystem comparison of the strength of trophic cascades. Ecology Letters, 5(6):785–791. Steneck, R., Graham, M., Bourque, B., Corbett, D., Erlandson, J., Estes, J., and Tegner, M. (2002). Kelp forest ecosystems: biodiversity, stability, resilience and future. Environmental Conservation, 29(4):436–459. Stone, R. (2005). Science in Iran - Attack of the killer jellies. Science, 309(5742):1805–1806. Tilman, D., Knops, J., Wedin, D., Reich, P., Ritchie, M., and Siemann, E. (1997). The influence of functional diversity and composition on ecosystem processes. Science, 277(5330):1300–1302. 51  Vinther, M. (2001). Ad hoc multispecies vpa tuning applied for the baltic and north sea fish stocks. ICES Journal of Marine Sciences, 58(1):311–320. Walters, C. (2003). Folly and fantasy in the analysis of spatial catch rate data. Canadian Journal of Fisheries and Aquatic Sciences, 60(12):1433–1436. Walters, C., Christensen, V., and Pauly, D. (1997). Structuring dynamic models of exploited ecosystems from trophic mass-balance assessments. Reviews in Fish Biology and Fisheries, 7(2):139– 172. Walters, C. and Martell, S. (2004). Fisheries Ecology and Management. Princeton University Press, Princeton, New Jersey. Ward, P. and Myers, R. (2005). Shifts in open-ocean fish communities coinciding with the commencement of commercial fishing. Ecology, 86(4):835–847. Watling, L. and Norse, E. (1998). Disturbance of the seabed by mobile fishing gear: A comparison to forest clearcutting. Conservation Biology, 12(6):1180–1197. Watson, R., Kitchingman, A., Gelchu, A., and Pauly, D. (2004). Mapping global fisheries: sharpening our focus. Fish and Fisheries, 5:168–177. Watson, R. and Pauly, D. (2001). Systematic distortions in world fisheries catch trends. Nature, 414(6863):534–536. Wilson, R., Millero, F., Taylor, J., Walsh, P., Christensen, V., Jennings, S., and Grosell, M. (2009). Contribution of fish to the marine inorganic carbon cycle. Science, 323(5912):359–362. Worm, B., Barbier, E. B., Beaumont, N., Duffy, J. E., Folke, C., Halpern, B. S., Jackson, J. B. C., Lotze, H. K., Micheli, F., Palumbi, S. R., Sala, E., Selkoe, K. A., Stachowicz, J. J., and Watson, R. (2006). Impacts of biodiversity loss on ocean ecosystem services. Science, 314(5800):787–790. Worm, B., Hilborn, R., Baum, J. K., Branch, T. A., Collie, J. S., Costello, C., Fogarty, M. J., Fulton, E. A., Hutchings, J. A., Jennings, S., Jensen, O. P., Lotze, H. K., Mace, P. M., McClanahan, T. R., Minto, C., Palumbi, S. R., Parma, A. M., Ricard, D., Rosenberg, A. A., Watson, R., and Zeller, D. (2009). Rebuilding global fisheries. Science, 325(5940):578–585. Yellen, J., Brooks, A., Cornelissen, E., and Mehlman, M. (1995). A middle stone age worked bone industry from Katanda, Upper Semliki Valley, Zaire. Science, 268(5210):553–556. Zeller, D. and Pauly, D. (2007). Reconstruction of marine fisheries catches for key countries and regions (1950–2005). Fisheries Centre Research Report 15(2). University of British Columbia, Vancouver.  52  Appendix A  Supporting figures In this appendix I have included additional material that might serve in the interpretation of the results of Chapter 2. Section A.1 contains global maps of the primary production and sea surface temperature data used in the model (the two oceanographic inputs) as well as a validation of the use of PP for the year 1998, section A.2 summarizes trends in the global catch by decades and section A.3 includes a global map of the predicted density of unexploited predator biomass.  53  A.1 A.1.1  Inputs to Ecotroph from oceanographical data Global maps of primary production and sea surface temperature  (a) Primary production ! Primary production! (mg m-2 day-1)!  (b) Sea surface temperature! SST (°C)!  Figure A.1: Primary production (top) as compiled by the Sea Around Us Project and sea surface temperature (below) from NOAA’s World Ocean Atlas.  54  A.1.2  Validation of the use of primary production from year 1998  The primary production data compiled by the Sea Around Us Project is for the year 1998. I used a model of global primary production by Sarmiento et al.(unpublished) for the period 1950-2004 in order to ensure that the year 1998 is representative of the primary production in the period 1950-2004. The Sarmiento et al.(unpublished) data were not used in the main results of this study because (1) they are modelled and not actual PP data and (2) there were some worries about the magnitude of PP values predicted in some regions. Since bias in the Sarmiento model should be present in all years, I felt that it was appropriate to use this dataset to compare PP in 1998 to other years. This was done through the following steps: 1. Calculate mean PP and standard deviation over 1950-2004 for each cell i = P P i and SDi ; 2. For each year y, calculate proportion of cells that are within one standard deviation of P P i = Py ; 3. Calculate mean of Py over period 1950-2004 (P 1950−2004 ) and compare to P1998 . This approach allowed to test whether the PP in year 1998 was more different from the mean than the typical year in 1950-2004. For 1998, the proportion of cells that were within one standard deviation of their mean cell PP was 0.740, which is slightly above the mean Py for years 1950-2004 of 0.724 (Figure A.2). It can thus be said that the year 1998 is an adequate representation of the PP in the period 1950-2004.  55  5  10  1998  0  Frequency (number of years)  15  Mean % of cells within 1 SD of mean PP 1950−2004 = 0.724 % of cells within 1 SD in 1998= 0.740  0.55  0.60  0.65  0.70  0.75  0.80  0.85  % of cells within 1 SD of cell mean PP 1950−2004  Figure A.2: Histogram of the proportion of cells in each year that are within one standard deviation of the cell’s mean PP (P P i ) over 1950-2004.  56  prop.pred.y * 100  % world catch  A.2  Trends in reconstructed global catch  100  % of catch with TL >= 3.5  80 60 40  ● ●  ●  ●  ●  ●  ●  ●  20 0 World catch  80 60  ● ●  Average annual catch (million tons)  40 20  ●  ●  0 Catch by ocean: 25  Indian North Atlantic North Pacific South Atlantic South Pacific Antarctic  20 15  ● ●  10  ●  ● ● ●  ● ●  ●  ● ● ● ●  ●  ● ● ●  ● ●  ●  ● ●  ●  5  ● ● ●  ●  0  ● ●  ●  ●  ●  ●  ●  ●  ●  1950  1960  1970  1980  1990  2000  Decade Figure A.3: Proportion of predator species (TL ≥ 3.5) in the world catch (top) and average annual catches over decade (in million tonnes) as reconstructed by the Sea Around Us Project , aggregated for the world (middle) and by oceans (bottom).  57  A.3  Map of unexploited predator biomass Unexploited predator biomass! (tons km-2)!  Figure A.4: Density of unexploited predator biomass for the world  58  Appendix B  Sensitivity analyses Ecotroph has three main parameters for which data is currently unavailable: the transfer efficiency TE, the top-down effect α and the parameter smooth. Here I summarize the results of sensitivity analyses performed on these parameters. Because it was impractical to run the sensitivity analyses at a global scale, they were performed on a set of 0.5x0.5 degree cells belonging to three Large Marine Ecosystems representative of different ecosystem types: the North Sea (heavily exploited temperate system), the Gulf of Mexico (tropical system) and the Guinea Current (upwelling system).  B.1  Transfer efficiency parameter  The transfer efficiency parameter controls how efficiently energy is transferred between trophic levels. The default value, 10%, means that 90% of the production is lost over a transfer of one TL. Ecotroph was run with values of TE between 5% and 15%, which represent a realistic range of TE expected for the world’s marine ecosystems (Pauly and Christensen 1995). I use the decade 1970 as an example below. The predictions of total ecosystem biomass of Ecotroph are very sensitive to the value of the transfer efficiency, as shown in Table B.1. As transfer efficiency increases, total biomass predicted between trophic levels 2 and 5 increases rapidly and the change is even greater for predators. On the other hand, cells generally responded similarly to changes in TE, that is, while the absolute value of biomass is quite sensitive to even small changes of TE value, the relative spatial trends describing  59  Table B.1: Relative change in the 1970s total and predator biomass predicted by Ecotroph for three representative LMEs as a function of the transfer efficiency parameter TE. Biomass was normalized to the biomass predicted under TE=0.10, which is the value used in the main results of this thesis. TE = LME  Subset  Gulf of Mexico  Total Predator Total Predator Total Predator  North Sea Guinea Current  0.05  0.075  0.10  0.125  0.15  0.292 0.059 0.356 0.055 0.326 0.071  0.593 0.328 0.644 0.320 0.614 0.344  1.000 1.000 1.000 1.000 1.000 1.000  1.521 2.169 1.448 2.728 1.500 2.239  2.167 3.951 2.010 6.505 2.132 4.280  the cells most impacted by fishing are very robust to this parameter. This statement holds except in systems that are so heavily fished that observed catches are greater than the available production. As an example, Figure B.1 shows maps of the percentile to which each cell belongs in terms of the decline of 1970 total biomass compared to the unexploited state for a range of TE values. A 5th percentile indicates that the cell is amongst the 5% most impacted in the LME of interest for that decade (where impact is defined as a decline in fished biomass). In both the Gulf of Mexico and the Guinea Current, there is almost no change in the percentile to which each cell belong as a function of TE. In the North Sea however, where catches are often higher than observed production, the percentile of cells changes in some instances by as much as 40% as TE goes from 0.05 to 0.15. This phenomena occurs because the TE also affects the proportion of unexploited ecosystem production occurring in higher trophic level (not shown here): as TE increases, the proportion of ecosystem biomass occuring in high TLs also increases. Therefore, if catches are so high that Pτ goes to 0 for a wide range of high TLs, the % biomass decline observed is greater under high TEs. In summary, the absolute biomass values predicted by Ecotroph are very sensitive to TE but in the majority of cases, different ecosystems (i.e., cells) respond similarly to changes in TE. Catches are rarely high enough to drive production down to 0 and so, in general, the relative spatial trends of biomass decline between cells are robust to the TE parameter.  60  TE = 0.05 30  25  20  TE = 0.075  ●●●●● ● ●●●●●●●●●●● ●● ●●●●●●●● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●● ●  −95  −90  −85  TE = 0.1  ●●●●● ● ●●●●●●●●●●● ●● ●●●●●●●● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●● ●  −80  −95  −90  −85  TE = 0.125  ●●●●● ● ●●●●●●●●●●● ●● ●●●●●●●● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●● ●  −80  −95  −90  −85  TE = 0.15  ●●●●● ● ●●●●●●●●●●● ●● ●●●●●●●● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●● ●  −80  −95  −90  −85  ●●●●● ● ●●●●●●●●●●● ●● ●●●●●●●● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●● ●  −80  −95  −90  −85  −80  65  Latitude  60  55  ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●● ●●●●●●●●●  ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●● ●●●●●●●●●  ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●● ●●●●●●●●●  ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●● ●●●●●●●●●  ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●● ●●●●●●●●●  −5  −5  −5  −5  −5  50  0  5  10  0  5  10  0  5  10  0  5  10  0  5  % percentile  ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●  5% 10 % 15 % 20 % 25 % 30 % 35 % 40 % 45 % 50 % 55 % 60 % 65 % 70 % 75 % 80 % 85 % 90 % 95 % 100 %  10  15 10 5 0 −5  ●● ●●● ●● ●●●●●●●●●●● ● ● ●●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●● ●●●●●●●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●● ● ● ● ● ●● ● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●●●●●● ●●●●●●● ●●●●●●● ● ● ● ●● ● ● ● ●●●●●●●●● ● ● ● ● ● ● ●● ● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●● ●  ●● ●●● ●● ●●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●●●● ● ● ● ● ● ● ● ● ● ● ●● ● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●● ●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●● ● ● ●● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●●●●●● ●●●●●●● ●●●●●●● ●●●●●●●● ● ● ●● ● ● ● ●● ● ● ●● ● ● ● ● ● ●●●●●●●●●● ● ● ● ● ● ● ● ● ●● ●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●● ●  ●● ●●● ●● ●●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●● ●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●●●●●● ●●●●●●● ●●●●●●● ●●●●●●●● ●●●●●●●●● ● ● ●● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●● ●  ●● ●●● ●● ●●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●● ●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ● ●● ● ● ● ● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ● ● ●● ● ● ● ● ●●●●●●● ●●●●●●● ●●●●●●●● ●●●●●●●●● ●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●● ●  ●● ●●● ●● ●●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●● ●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ● ● ●● ● ● ● ● ●●●●●●● ● ● ● ●● ● ● ●●●●●●●● ●●●●●●●●● ● ● ●● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●● ●  −10  −20  −15  −10  −5  0  5  10  −20  −15  −10  −5  0  5  10  −20  −15  −10  −5  0  5  10  −20  −15  −10  −5  0  5  10  −20  −15  −10  −5  0  5  10  Longitude  Figure B.1: Spatial trends in the response of individual cells within three representative LMEs to changes in the transfer efficiency parameter TE. LMEs represented are: Gulf of Mexico (top), North Sea (middle) and Guinea Current (bottom), and the decade was arbitrarily set at 1970. The values of transfer efficiency modelled are 0.05, 0.075, 0.1, 0.125, 0.15 (from left to right). Spatial trends are represented in terms of the percentile to which each cell beyond (each circle on the map stand for one 0.5x0.5 degree cell). A cell belonging in the 5th percentile is amongst the 5% most impacted of the LMEs, where impact is defined as a decline in fished biomass. A rough outline of the coasts is included for clarity as a dark blue line.  B.2  Top-down parameter  A top-down effect can be added to the Ecotroph model such that the biomass of lower TLs is affected by the biomass of their predators. This is implemented by letting the kinetics of a given TL be a function of the predator biomass for that TL as follow:  Kτ = Kτ,unexpl · [1 + α ·  γ γ Bpred − Bpred,unexpl γ Bpred,unexpl  ] + Fτ  (B.1)  As the top-down parameter α increases, the kinetics of τ are increasingly affected by changes in predator biomass compared to the unexploited state. Predators are defined to occur at TLs between τ + 0.8 and τ + 1.3. See Gascuel et al. (2009) for a detailed description of the top-down effect in Ecotroph. The Ecotroph model was run with values of α between 0 and 0.8, arbitrarily selected as a plausible range for the importance of top-down effects. A value of α=1 was deemed unlikely as it would mean that the kinetics of a given TL would be solely driven by the difference in the trophic level’s predator biomass between the fished and unexploited state (see B.1). The shape parameter γ was set at 0.5. Note that the results of the main chapter were produced with α = 1. Table B.2 shows that the value of α can have an effect on the total ecosystem biomass of systems that are intensively exploited like the North Sea, but that the incurred change is small otherwise. Increasing the value of α results in higher predictions of ecosystem biomass because predator release of lower trophic levels increases their standing biomass, and there is intrinsically more biomass at lower than higher trophic levels. For the LMEs sampled here, the α parameter in Ecotroph has more of an effect on the distribution of biomass between TLs than the actual quantity of biomass (not shown here), although not enough that the biomass predicted for predators would change significantly unless α is very high (table B.2). Figure B.2 shows that relative spatial trends in ecosystem biomass can be sensitive to α for cells where high TLs are heavily impacted by fishing. For instance, there are some cells in the north part of the North Sea and in the Guinea Current that appear more affected by α (that is, they belong to different percentiles of fishing impact under different values of α).  62  TD = 0 30  25  20  TD = 0.2  ●●●●● ● ●●●●●●●●●●● ●● ●●●●●●●● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●● ●  −95  −90  −85  TD = 0.4  ●●●●● ● ●●●●●●●●●●● ●● ●●●●●●●● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●● ●  −80  −95  −90  −85  TD = 0.6  ●●●●● ● ●●●●●●●●●●● ●● ●●●●●●●● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●● ●  −80  −95  −90  −85  TD = 0.8  ●●●●● ● ●●●●●●●●●●● ●● ●●●●●●●● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●● ●  −80  −95  −90  −85  ●●●●● ● ●●●●●●●●●●● ●● ●●●●●●●● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●● ●  −80  −95  −90  −85  −80  65  Latitude  60  55  ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●● ●●●●●●●●●  ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●● ●●●●●●●●●  ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●● ●●●●●●●●●  ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●● ●●●●●●●●●  ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●● ●●●●●●●●●  −5  −5  −5  −5  −5  50  0  5  10  0  5  10  0  5  10  0  5  10  0  5  % percentile  ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●  5% 10 % 15 % 20 % 25 % 30 % 35 % 40 % 45 % 50 % 55 % 60 % 65 % 70 % 75 % 80 % 85 % 90 % 95 % 100 %  10  15 10 5 0 −5  ●● ●●● ●● ●●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●●●● ● ● ● ● ● ● ●● ● ● ● ● ● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●● ●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●●●●●● ●●●●●●● ●●●●●●● ●●●●●●●● ●●●●●●●●● ● ● ●● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●● ●  ●● ●●● ●● ●●●●●●●●●●● ●●●●●●●●●●● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●● ●●●●●●●● ●●●●●●●●●●●●●●●●● ● ● ● ● ● ● ●● ● ● ● ● ● ●●●●●●●●●●●●●●●●● ●●●●●●● ●●●●●●●●●●●●●●●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●●●●●● ●●●●●●● ● ● ● ●● ● ● ●●●●●●●● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●● ●  ●● ●●● ●● ●●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●● ●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●●●●●● ●●●●●●● ●●●●●●● ●●●●●●●● ●●●●●●●●● ● ● ●● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●● ●  ●● ●●● ●● ●●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ●●●●●●●●●●●●●●●● ●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●●●●●●●●●●●● ●● ● ● ● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●● ●●●●●●●● ●●●●●●● ●●●●●●● ●●●●●●●● ●●●●●●●●● ●●●●●●●●● ●●●●●●●●●● ● ● ● ●● ● ● ● ● ● ●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ● ● ● ●● ● ● ● ●●●● ●  ●● ●●● ●● ●●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●●●●●●●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●●●●●●●●●●●●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●●●●●●●●●●●● ●● ● ● ● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●● ●●●●●●●● ●●●●●●● ●● ● ● ● ● ● ●●●●●●●● ●●●●●●●●● ●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●● ●  −10  −20  −15  −10  −5  0  5  10  −20  −15  −10  −5  0  5  10  −20  −15  −10  −5  0  5  10  −20  −15  −10  −5  0  5  10  −20  −15  −10  −5  0  5  10  Longitude  Figure B.2: Spatial trends in the response of individual cells within three representative LMEs to changes in the top-down effect parameter α. LMEs represented are: Gulf of Mexico (top), North Sea (middle) and Guinea Current (bottom), and the decade was arbitrarily set at 1970. The values of the top-down effect modelled are 0, 0.2, 0.4, 0.6, 0.8 (from left to right). Spatial trends are represented in terms of the percentile to which each cell beyond (each circle on the map stand for one 0.5x0.5 degree cell). A cell belonging in the 5th percentile is amongst the 5% most impacted of the LMEs, where impact is defined as a decline in fished biomass. A rough outline of the coasts is included for clarity as a dark blue line.  Table B.2: Relative change in the 1970s total and predator biomass predicted by Ecotroph for three representative LMEs as a function of the top down effect parameter α. Biomass was normalized to the biomass predicted under α = 0, which is the value used in the main results of this thesis, and σ was fixed at 0.5. TD = LME  Subset  Gulf of Mexico  Total Predator Total Predator Total Predator  North Sea Guinea Current  B.3  0  0.2  0.4  0.6  0.8  1.000 1.000 1.000 1.000 1.000 1.000  1.017 1.035 1.169 1.039 1.032 1.037  1.032 1.072 1.420 1.092 1.065 1.080  1.046 1.112 1.861 1.180 1.106 1.140  1.072 1.165 3.032 1.410 1.189 1.262  Smooth parameter  In Ecotroph, species-specific catch records are distributed amongst trophic levels by assuming that the distribution of trophic levels amongst the individuals of the same species is log-normal. The mean is defined as the trophic level officially recorded for the species and the standard deviation is set to increase linearly with TL (ref Methods). The underlying assumption is that the intra-specific variation in trophic levels increases for species higher in the foodweb, with species at TL=5 having the greatest variation. The smooth parameter controls the rate of this linear increase as follow: Sτ = smooth · log(τ − 0.05)  (B.2)  The Ecotroph model was run for values of smooth between 0.025 and 0.125. The lower and upper values of this range results in ≈ 95% of the individuals of species at trophic level 5 to feed at a maximum trophic level of 5.3 and 6.5 respectively. These boundaries were chosen to enclose a region of values for the smooth parameter that are realistic from an ecological point of view. The estimates of biomass at the LME scale, both for the whole ecosystem (total) and for the predators, are not very sensitive to the smooth parameter, with the exception of the North Sea which is a heavily fished system (Table B.3). This occurs because as smooth increases more catches are attributed to lower trophic levels and the resulting decline in production affects a higher number 64  Table B.3: Relative change in 1970s total and predator biomass for three LMEs as a function of the smooth parameter. Results are standardized to the biomass predicted under smooth=0.075 (the setting used in the main part of this thesis). smooth = LME  Subset  0.025  0.050  0.075  0.100  0.125  Gulf of Mexico  Total Predator Total Predator Total Predator  1.005 1.015 1.051 1.140 1.003 1.034  1.003 1.008 1.022 1.057 1.000 1.012  1.000 1.000 1.000 1.000 1.000 1.000  0.998 0.992 0.982 0.964 1.000 0.994  0.996 0.985 0.967 0.943 1.000 0.991  North Sea Guinea Current  Table B.4: Range of trophic levels over which 90% of the catch is distributed as a function of the species’ mean TL and under the default value of smooth (0.075). The boundaries are defined by leaving 5% of the distribution in the lower and upper tails. Mean TL  Range of TLs  2.0 3.0 4.0 5.0  2.0-2.2 2.6-3.4 3.4-4.7 4.1-6.1  ∗  *catches are not allowed to be assigned to TLs smaller than 2 during the smoothing procedure  of TLs. This effect is stronger if catches are higher. Additionally, the smooth parameter affects predator biomass more than total biomass because, by definition (see equation B.2), it increases the standard deviation of high TLs by more. As a result, cells with higher proportions of high TL catches are more affected by changes in smooth (Figure B.3). Even so, the sensitivity of Ecotroph’s results to smooth is reasonable if one considers that the strongest effect observed was a ≈ 15% difference in predator biomass when smooth was decreased by a third (from 0.075 to 0.025) for the North Sea (Table B.3), which is the LME that has the heaviest catches recorded in the world. Given this, I feel that the chosen value of smooth = 0.075 was justified both in terms of the sensitivity analysis and the variation in TL it resulted in (Table B.4). 65  smooth = 0.025 30  25  20  smooth = 0.05  ●●●●● ● ●●●●●●●●●●● ●● ●●●●●●●● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●● ●  −95  −90  −85  smooth = 0.075  ●●●●● ● ●●●●●●●●●●● ●● ●●●●●●●● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●● ●  −80  −95  −90  −85  smooth = 0.1  ●●●●● ● ●●●●●●●●●●● ●● ●●●●●●●● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●● ●  −80  −95  −90  −85  smooth = 0.125  ●●●●● ● ●●●●●●●●●●● ●● ●●●●●●●● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●● ●  −80  −95  −90  −85  ●●●●● ● ●●●●●●●●●●● ●● ●●●●●●●● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●●●● ●  −80  −95  −90  −85  −80  65  Latitude  60  55  ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●● ●●●●●●●●●  ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●● ●●●●●●●●●  ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●● ●●●●●●●●●  ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●● ●●●●●●●●●  ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●● ●●●●●●●●●  −5  −5  −5  −5  −5  50  0  5  10  0  5  10  0  5  10  0  5  10  0  5  % percentile  ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●  5% 10 % 15 % 20 % 25 % 30 % 35 % 40 % 45 % 50 % 55 % 60 % 65 % 70 % 75 % 80 % 85 % 90 % 95 % 100 %  10  15 10 5 0 −5  ●● ●●● ●● ●●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●●●● ● ● ● ● ● ● ● ● ● ● ●● ● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●● ● ● ●● ● ● ● ● ●●●●●●●●●●●●●●●●● ● ● ● ● ● ● ● ● ●● ● ● ● ●●●●●●●●●●●●●●●●● ●●●●●●● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ● ● ●● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●● ●● ● ● ● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●● ●●●●●●●● ●●●●●●● ●●●●●●● ●●●●●●●● ●●●●●●●●● ● ● ●● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●● ●  ●● ●●● ●● ●●●●●●●●●●● ●●●●●●●●●●● ● ● ● ● ●● ● ● ● ● ● ● ●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ● ● ●● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●● ● ●● ● ● ● ● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●●●●●● ●●●●●●● ●●●●●●● ●●●●●●●● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●● ●  ●● ●●● ●● ●●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●● ●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ● ●● ● ● ● ● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●●●●●● ●●●●●●● ●●●●●●● ●●●●●●●● ●●●●●●●●● ● ● ●● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●● ●  ●● ● ●● ●● ●●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●●●● ● ● ● ● ● ● ●● ● ● ● ● ● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●● ●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●●●●●● ●●●●●●● ●●●●●●● ●●●●●●●● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●● ●  ●● ●●● ●● ●●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●● ●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●● ●●●●●●●●●●●●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●● ● ●● ● ● ● ● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●●●●●● ●●●●●●● ●●●●●●● ●●●●●●●● ●●●●●●●●● ● ● ●● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ●●●●●●●● ●●●● ●  −10  −20  −15  −10  −5  0  5  10  −20  −15  −10  −5  0  5  10  −20  −15  −10  −5  0  5  10  −20  −15  −10  −5  0  5  10  −20  −15  −10  −5  0  5  10  Longitude  Figure B.3: Spatial trends in the response of individual cells within three representative LMEs to changes in the top-down effect parameter smooth. LMEs represented are: Gulf of Mexico (top), North Sea (middle) and Guinea Current (bottom), and the decade was arbitrarily set at 1970. The values of the smooth parameter modelled are 0.025, 0.05, 0.075, 0.10, 0.125 (from left to right). Spatial trends are represented in terms of the percentile to which each cell beyond (each circle on the map stand for one 0.5x0.5 degree cell). A cell belonging in the 5th percentile is amongst the 5% most impacted of the LMEs, where impact is defined as a decline in fished biomass. A rough outline of the coasts is included for clarity as a dark blue line.  Appendix C  Estimating transfer efficiency from time-series of catch data Transfer efficiency (TE) represents the proportion of the production of lower trophic levels that is transferred to higher trophic levels. TE accounts for individual energy losses due to espiration, digestion, excretion, etc., as well as losses due to non-predatory mortality, e.g. when individuals are not being consumed because they are protected from predation or because consumers are saturated, such as may happen during phytoplankton blooms (Cushing 1973). TE is thus a general measure of an ecosystem’s efficiency at transferring energy from low to high trophic levels. In Ecotroph, the decline of production (P) between trophic levels is modelled as a declining exponential function, where TE represents the proportion of production left over a transfer of one TL unit. In other words:  P (τ + 1) = P (τ )T E  (C.1)  Biomass and production estimates produced by Ecotroph are very sensitive to the value of TE (Gascuel et al. 2009). Also, in marine ecology and in fisheries science in general, TE is an essential parameter involved in answering a number of important questions, such as the fraction of total primary production that is used by fisheries (Pauly and Christensen 1995), the effects of fishing at an ecosystem scale (Libralato et al. 2008), or the energy used by different groups of 67  Figure C.1: Illustration of the rationale behind the estimation of TE from mean TL/catch time-series. If the proportion of production exploited is constant between TLs, catch should decline with increasing TL at a rate proportional to TE, reflecting the concurrent decline in production by TL. fish (Jennings et al. 2008). An average value of TE=10% is obtained for marine ecosystems in general (Pauly and Christensen 1995), which, perhaps surprisingly, also seems to apply to other systems (Morowitz 1992). However, the average value estimated by Pauly and Christensen (1995) masks a great variability between ecosystem types (Jarre-Teichmann and Christensen 1998). For the purpose of the research presented in this thesis, having access to ecosystem-specific TE values would greatly improve the estimates of the biomass by TL as well as highlight how system-specific differences in TE affect their response to fishing. Here, I elaborate on a method previously presented in (Pauly and Palomares 2005) to estimate the transfer efficiency of an ecosystem based on time-series of catch and TL data. This method assumes that, for a given ecosystem, the proportion of production exploited at each trophic level is constant, so that we would expect total catches to decrease with increasing mean TL of the catch at a rate that is proportional to TE (Figure C.1). Assuming that production declines exponentially with trophic levels (as modelled in Ecotroph), the relationship between the mean trophic level of the catch and the log(catch) is thus linear, and the TE can be extracted from the slope as T E = 101/b , where b is the slope. For this method to work optimally, the total catch/mean TL time-series used need to have sufficient contrast in the mean TL of the catch, as well as no major changes in the effort patterns of the underlying fleet. Moreover, the catch must come from the same ecosystem (or a relatively small region); otherwise an observed increase of the catch could reflect a geographic expansion of the fleet (Bhathal and Pauly 2008). If dealing with catch data from larger regions (e.g. FAO areas, as in Pauly and Palomares 2005), this issue can be addressed by computing the Fishing-in-Balance  68  (FiB) indices for the time-series and using only those periods where the FiB index is constant, since they correspond to periods where the fleet was using the same fishing grounds (Bhathal and Pauly 2008). An example of the application of the method to three different types of large marine ecosystems (LMEs) is presented here: the California Current (upwelling), the South China Sea (tropical) and the Sea of Japan (temperate) (Figure C.2). The catch data originate from the Sea Around Us Project global fisheries database (Watson et al. 2004) and the trophic levels from FishBase (www.fishbase.org). I randomly selected three 0.5 degree square cell in each LME from a subset of cells that satisfy the conditions listed above. The mean trophic level of the catch is calculated as:  T Ly =  · Yi,y Y i i,y  i T Li  (C.2)  with Yi,y the catch of species group i in year y and T Li the trophic level of species group i. To reduce the noise induced by abrupt changes in the species targeted by the fishing fleet, years where the mean trophic level of the total catch changes by more than 10% are not used in the linear regression. When applicable, I also removed the years before the one corresponding to the first maximum of the mean TL of the catch, as they reflect a period where the fleet is just starting to develop and the catch is relatively small. Also, I consider half-degree cells to be small enough that an expansion of the fisheries over the cells’ area is a negligible factor. Therefore, unlike Pauly and Palomares (2005), I did not use the FiB index to identify periods where the area of the fishing grounds was constant. I estimate transfer efficiencies of 0.101 for the California Current, 0.062 for the South China Sea and 0.082 for the Sea of Japan (see Figure C.2). These ecosystem-specific TE estimates are generated from the mean TL/total catch data of three cells, and can be made more reliable by using data from more cells in the ecosystem. Despite the fact that they are preliminary, these estimates occur in the region of the expected TE for ecosystems and support the assertion that TE is not constant amongst global marine ecosystems. They demonstrate the potential of the method presented here to derive approximations of ecosystem-specific TE. 69  While the method presented here shows potential in terms of being able to generate worldwide estimates of TE, the results of a sensitivity analysis (not presented here) have demonstrated that it is in fact extremely sensitive to some of its assumptions. The assumption that the proportion of production extracted by fishing is the same for all exploited trophic levels is especially problematic because it is probably rarely met in practice. I investigated without success a number of methods that could account for violations in this assumption and enable us to produce robust estimates of TE. The main challenge is that estimates of the proportion of production exploited are required for each year (as the target of fishing effort changes) and are themselves dependent on the value of TE, with possible issues of circularity in the calculations. It was thus decided that further research on this topic was outside the scope of this thesis, and an average TE value of 10% was used to generate the global estimates of biomass by TL presented in this work.  70  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igure C.2: Estimation of TE from time-series of mean TL of the catch and log(catch) for three LMEs (three cells/LME, one color for each cell). The left and center panels show time series of mean TL and total catch; the right panels show the regression lines fitted through to the plot of mean TL vs. log(total catch). Points that were not used in for the regression (because of high variations in their TL; see text) are shown as open dots.  71  References Bhathal, B. and Pauly, D. (2008). ‘Fishing down marine food webs’ and spatial expansion of coastal fisheries in India, 1950-2000. Fisheries Research, 91(1):26–34. Cushing, D. (1973). Production in the Indian Ocean and the transfer from primary to secondary level. In Zeitzschel, B. and Gerlach, A., editors, The Biology of the Indian Ocean. Springer-Verlag, Berlin. Gascuel, D., Tremblay-Boyer, L., and Pauly, D. (2009). Ecotroph (ET): a trophic level based software for assessing the impacts of fishing on aquatic ecosystems, Fisheries Centre Research Report 17(1). University of British Columbia, Vancouver. Jarre-Teichmann, A. and Christensen, V. (1998). Comparative modelling of trophic flows in four large upwelling ecosystems: global versus local effects. In Durand, M., Cury, P., Mendelssohn, R., Roy, C., Bakun, A., and Pauly, D., editors, Global Versus Local Changes in Upwelling Systems, pages 423–443. Orstom, Paris. Jennings, S., M´elin, F., Blanchard, J., and Forster, R. (2008). Global-scale predictions of community and ecosystem properties from simple ecological theory. Proceedings of the Royal Society B: Biological Sciences, 275:1375–1383. Libralato, S., Coll, M., Tudela, S., Palomera, I., and Pranovi, F. (2008). Novel index for quantification of ecosystem effects of fishing as removal of secondary production. Marine Ecology Progress Series, 355:107–129. Morowitz, H. (1992). The Thermodynamics of Pizza: Essays on Science and Everyday Life. Rutgers University Press, New Brunswick, NJ. Pauly, D. and Christensen, V. (1995). Primary production required to sustain global fisheries. Nature, 374:255–257. Pauly, D. and Palomares, M. (2005). Fishing down marine food web: It is far more pervasive than we thought. Bulletin of Marine Science, 76(2):197–211. Watson, R., Kitchingman, A., Gelchu, A., and Pauly, D. (2004). Mapping global fisheries: sharpening our focus. Fish and Fisheries, 5:168–177.  72  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0071043/manifest

Comment

Related Items