Population Dynamics of the Blue Shark, Prionace glauca, in the
North Atlantic Ocean
Alexandre Aires Silva
A dissertation
submitted in partial fulfillment of the
requirements for the degree of
Doctor of Philosophy
University of Washington
2008
Program Authorized to Offer Degree:
School of Aquatic and Fishery Sciences
©Copyright 2008
Alexandre Aires Silva
University of Washington
Graduate School
This is to certify that I have examined this copy of a doctoral dissertation by
Alexandre Aires Silva
and have found that it is complete and satisfactory in all respects,
and that any and all revisions required by the final
examining committee have been made.
Chair of the Supervisory Committee:
____________________________________________________________
Vincent F. Gallucci
Reading Committee:
____________________________________________________________
Vincent F. Gallucci
____________________________________________________________
André E. Punt
____________________________________________________________
John R. Skalski
Date:___________________________
In presenting this dissertation in partial fulfillment of the requirements for the
doctoral degree at the University of Washington, I agree that the Library shall make
its copies freely available for inspection. I further agree that extensive copying of the
dissertation is allowable only for scholarly purposes, consistent with “fair use” as
prescribed in the U.S. Copyright Law. Requests for copying or reproduction of this
dissertation may be referred to ProQuest Information and Learning, 300 North Zeeb
Road, Ann Arbor, MI 48106-1346, 1-800-521-0600, to whom the author has granted
“the right to reproduce and sell (a) copies of the manuscript in microform and/or (b)
printed copies of the manuscript made from microform.”
Signature_______________________________
Date___________________________________
University of Washington
Abstract
Population Dynamics of the Blue Shark, Prionace glauca, in the North Atlantic
Ocean
Alexandre Aires Silva
Chair of the Supervisory Committee:
Professor Vincent F. Gallucci
School of Aquatic and Fishery Sciences
Large numbers of blue sharks are caught as bycatch, and have even become
the target species in pelagic longline fisheries in the North Atlantic Ocean. The status
of the stock is ambiguous due to the limitations of the fishery-dependent and
-independent data. In chapter 2, fishery-independent demographic and risk analyses
are developed for the North Atlantic blue shark. An age-structured matrix population
model in which the life-history parameters are stochastic was build. A risk analysis to
evaluate the future prospects of the population under different harvest rates is
presented and implications for management discussed.
Chapter 3 develops a historic index of abundance for the blue shark in the
western North Atlantic. Pelagic longline catch records from several fishery observer
programs that operated since the late 1970s are linked to observations from historical
fishery-independent longline surveys undertaken in the late 1950s, 1960s and 1970s.
Generalized linear model (GLM) techniques are used to derive standardize catch rates
(catch-per-unit-effort, CPUE). The index of abundance revealed a decline in blue
shark CPUE of approximately 30% in the western North Atlantic from 1957 through
2000. The magnitude of this decline is less than other recently published estimates
and seems reasonable in light of the high productivity of blue shark revealed by the
demographic analyses in Chapter 2.
Chapter 4 analyzed the available tagging data in a spatio-temporal analysis to
derive estimates of movement and fishing mortality rates for the blue shark in the
North Atlantic Ocean. The analyses are based upon the historical blue shark tagrecovery data archived by the U.S.-NMFS Cooperative Shark Tagging Program
(1965-2004). Fishing mortality rates (F) were highly heterogeneous across the North
Atlantic Ocean. Estimates of F for the western North Atlantic were historically lower
than 0.10 yr-1, well below a critical reference point, Fc (=0.20 yr-1), derived from the
risk analysis in Chapter 2. Such low fishing mortality rates are consistent with the low
declines in catch rate estimated for the western North Atlantic in Chapter 3. In
contrast, F estimates over the most recent decade (1990s) in the eastern side of the
Atlantic were found to be rapidly approaching Fc .
TABLE OF CONTENTS
Page
List of Figures ............................................................................................................. iii
List of Tables .............................................................................................................. iv
Chapter 1: Background and Synopsis ............................................................................1
1.1. The Context-of-the-Problem .......................................................................1
1.2. The Case Study – North Atlantic Blue Shark .............................................4
1.2.1. The Single Stock Concept ............................................................4
1.2.2. The Fishery Exploitation Scenario...............................................5
1.2.3. Knowledge About Stock Status ...................................................9
1.3. Dissertation Outline ..................................................................................10
Chapter 2: Demographic and Risk Analyses Applied to Management and
Conservation of the Blue Shark in the North Atlantic Ocean ................................14
2.1. Introduction ...............................................................................................14
2.2. Materials and Methods ..............................................................................18
2.2.1. Age-Structured Model ...............................................................18
2.2.2. Uncertainty of Life-history Parameters .....................................20
2.2.3. Stochastic Demographic Analysis .............................................22
2.3.4. Population Projections and Risk Analysis .................................25
2.3. Results .......................................................................................................27
2.3.1. Survivorship ...............................................................................27
2.3.2. Demographic Analysis ...............................................................28
2.3.3. Risk Analysis .............................................................................30
2.3. Discussion .................................................................................................31
i
Chapter 3: A Historic Index of Abundance for the Blue Shark in the Western
North Atlantic ........................................................................................................47
3.1. Introduction ...............................................................................................47
3.2. Materials and Methods ..............................................................................51
3.2.1. Data Selection ............................................................................51
3.2.2. Catch-Effort Standardization .....................................................53
3.3. Results .......................................................................................................59
3.3.1. Model Selection .........................................................................59
3.3.2. Catch-Effort Standardization .....................................................60
3.4. Discussion .................................................................................................62
Chapter 4: A Spatially-Structured Tagging Model to Estimate Movement and
Fishing Mortality Rates for the Blue Shark in the North Atlantic Ocean ..............80
4.1. Introduction ...............................................................................................80
4.2. Materials and Methods ..............................................................................83
4.2.1. Tagging Data ..............................................................................83
4.2.2. Geographical Stratification and Definition of Fisheries ............84
4.2.3. Model Development...................................................................85
4.2.4. Model Hypotheses – Base Case Model and Sensitivity
Analyses ........................................................................................91
4.3. Results .......................................................................................................93
4.3.1. Model fit to Observed Tag Recoveries ......................................93
4.3.2. Movement Parameters ...............................................................94
4.3.3. Fishing Mortality Rates..............................................................94
4.4. Discussion .................................................................................................97
Bibliography ..............................................................................................................119
Appendix A: Supplementary Information of Chapter 2 ............................................137
Appendix B: Supplementary Information of Chapter 3 .............................................141
Appendix C: Supplementary Information of Chapter 4 .............................................151
ii
LIST OF FIGURES
Figure Number
Page
2.1. Uncertainty in the Life-History Parameters of the North Atlantic blue
Shark ................................................................................................................42
2.2. Statistical Distributions of Demographic Parameters for an Unfished Blue
Shark Population ..............................................................................................43
2.3. a) Isoclines for the Average Response in Population Growth Rate to
Different Harvest Rates Starting at Different Ages-at-First-Capture.
b) Distributions Derived for the Stationary Harvest Rate ................................44
2.4. Example Stochastic Population Projections....................................................45
2.5. Risk Curves for Depleting the Population to Levels Below 50% of its
Pre-Exploited Population Size .........................................................................46
3.1. Geographical Locations of Pelagic Longline Sets for Each Target Species
Fishing Operation.............................................................................................76
3.2. Diagnostic Check Plots for the Different Error Models Fit to the Blue
Shark Longline Reconciled Data Set ...............................................................77
3.3. Delta-Lognormal Model Estimates .................................................................78
3.4. Mean Effects of Selected Explanatory Factors on Blue Shark Catch Rates ...79
4.1. Blue Shark Tagging Data Collected by the Cooperative Shark Tagging
Program (CSTP) in the North Atlantic Ocean (1965-2004). .........................113
4.2. Annual Time Series of Blue Shark Tag Releases and Recoveries in each
Region of the North Atlantic Ocean. .............................................................114
4.3. Movement Hypothesis for the Blue Shark in the North Atlantic Ocean
Assumed in the Tagging Model. ....................................................................115
4.4. Base Case Model Fit to the Blue Shark Tag Recovery Observations for
each Fishery by Geographical Region. ..........................................................116
4.5. Fishery-Specific Instantaneous Fishing Mortality Rates in Each
Geographical Region as Estimated from the Base Case Model. ...................117
4.6. Sensitivity Analyses of the Estimate of Total Instantaneous Fishing
Mortality to Different Scenarios About the Tag Shedding Rate and the Tag
Reporting Rates of Major Fisheries ...............................................................118
iii
LIST OF TABLES
Table Number
Page
2.1. Definition of Mathematical Notation ..............................................................38
2.2. Estimates of Natural Survivorship Produced by the Indirect Methods...........39
2.3. Descriptive Statistics for the Demographic Parameters..................................40
2.4. Risk (Probability) of Obtaining a Population Decline to Levels Below
50% of the Pre-Exploited Levels .....................................................................41
3.1. Fishery-Independent and –Dependent Pelagic Longline Data Sources ..........70
3.2. Summary Statistics of Fishery-Independent and Fishery-Dependent Data ....71
3.3. Annual Time Series of Pelagic Longline Sets for the Different Target
Species .............................................................................................................72
3.4. Categorical Variables Assumed in the Blue Catch-Effort Standardization ....73
3.5. Summary Statistics of the Selected Four Error Models ..................................74
3.6. Delta-Lognormal Standardized CPUE for the Blue Shark .............................75
4.1. Blue Shark Tag Recoveries Reported by Fishing Fleet (Nation or Group)
for Each Geographical Region .......................................................................107
4.2. List of the Fisheries Defined in the Tagging Model .....................................108
4.3. Estimated Parameters of the Tagging Model. ...............................................109
4.4. Model Sensitivity Analysis. ..........................................................................110
4.5. Movement and Fishing Mortality Parameter Estimates and their
Coefficient of Variation (CV) Obtained from the Base Case Model.............111
4.6. Catchability Parameter Estimates Obtained for each Fishery from the
Base Case Model............................................................................................112
iv
ACKNOWLEDGEMENTS
I learned early on along the way that “nobody does it alone” in most important
endeavors of life. This is also valid for the scientific and scholarly process. I am
forever grateful for the encouragement, support, friendship and kindness of many
friends and colleagues with whom I had the privilege to cross my path with during
my voyage as a doctoral student at the University of Washington.
My graduate program was made possible through several sources of financial
support. I am eternally grateful to my main sponsors: the Fulbright Commission and
the Fundação Luso-Americana para o Desenvolvimento in Portugal. I also extend my
gratitude to my following supplementary sponsors at the University of Washington:
the Graduate School, the Claire L. and Evelyn S. Egtvedt Fellowship, the Galen and
Helen Maxfield Fisheries Scholarship, and a Wakefield Scholarship fund. My special
thanks to Jean Whitcomb at the Graduate School for the support over the years.
I wish to express my most sincere gratitude to my major advisor, Dr. Vincent
F. Gallucci for his wise mentorship during my graduate program. By pointing me into
the right direction, Vince’s advice was critical to “convert” me from a field fisheries
biologist into a quantitative fisheries sock assessment scientist. Vince helped to
transform my fascination on the blue shark problem, which I acquired in the waters of
Portugal, into a real dissertation project. By allowing me enough space to explore my
own research interests, Vince helped me to become an independent scientist without
losing focus on the importance of team work. I thank Vince for the countless
overnight hours put into critically reviewing my manuscripts and helping to make
them suitable for publication. Because of that, all the chapters presented in this
dissertation will truly become disseminated, rather than to eternally lie in the archives
of a library. His constant support, advice and guidance played a key role in my
success. I deeply appreciate our strong friendship.
Besides my major advisor, I am deeply grateful to the five additional members
of my supervisory committee. I am particularly indebted to Drs. André Punt and John
Skalski for their permanent support, their critical review of my work and manuscripts,
and for the intellectual challenge along my program. My sincere thanks to Dr. Miles
Logsdon for helping me to find a bit of space for the tools of Geographical
Information Systems (GIS) in my research, and for broadening my interests on spatial
analysis. I thank Miles for his encouragement and patience. Dr. Gregor Cailliet
(MLML-CSU) has inspired my shark research since I was an undergraduate student
in Portugal. It was a privilege to have Greg on my committee. My special thanks to
Greg Cailliet for inspiring me not to loose focus of the real biological problems in the
midst of mathematical sophistication. I am honored to have had Dr. Jerry Franklyn
assigned as the Graduate School Representative to my committee. My hearty
v
gratitude to Dr. Franklin for his prompt availability and for his expressed interest at
committee meetings.
Although not an official member on my supervisory committee, Dr. John
Hoey (NOAA-NEFSC) ex-officio’s role on the committee is totality legitimate. I am
eternally grateful to John Hoey for his constant mentorship, inspiration and guidance.
John made it possible for my research experience to physically spread out from
SAFS-UW, Seattle, WA, into two labs on the east coast (NOAA’s headquarters in
Silver Spring, MD, and then the NMFS-NEFSC Narragansett laboratory,
Narragansett, RI). John showed me that a deep understanding of fisheries data, as
well as the fisheries reality form which it was sampled from, should precede analysis.
Our journey together through the recovery and construction of the longline historical
database was fascinating. I greatly benefited from John’s extensive knowledge about
the pelagic longline fishery in the North Atlantic. I am forever grateful to Maria and
John Hoey for truly embracing me as an additional family member during many trips
to the NMFS Narragansett lab, Rhode Island. I deeply value our friendship.
My fascination for the North Atlantic blue shark program initiated through my
exposure to the tagging results of the U.S. NMFS-Cooperative Shark Tagging
Program back in Portugal. Later on, it was a privilege to get fully immersed into the
program during many working trips to Rhode Island. My profound gratitude for the
support and friendship that I found at the Narragansett laboratory, RI. I am
particularly indebted to John Hoey and Nancy Kohler for allowing me to pursue this
project. My heartfelt thanks go to Patricia Turner, Ruth Briggs, Kelly Kreis for their
invaluable support with the tagging and survey data. I am also grateful to Chris
Orphanides for his support with GIS. I thank Harold “Wes” Pratt, Lisa Natanson,
Karen Tougas, and Anthony Wood for their support. I am immensely grateful to
Cami and Steve McCandless, as well as to Ruth Briggs for kindly receiving me at
their homes. I thank David Martins and Amy VanAtten for their friendship, and the
wonderful family moments under a severe winter blizzard on the east coast. A
summer working trip to NOAA’s headquarters in Silver Spring, MD, was also part of
my research. My thanks to Mark Holliday and Liz Pritchard (NOAA), and to Fred
Moreno for the great hospitality in Silver Spring.
I extend my gratitude to all the SAFS-UW community. A substantial amount
of my training at UW came through interactions with many colleagues and friends at
UW. It was a privilege to share an office with Ian Taylor at SAFS. I am immensely
grateful to Ian for his permanent support, encouragement, friendship and continuous
inter-change of ideas. My hearty gratitude to many other colleagues and friends at the
Shark Lab for several years of camaraderie and sharing of knowledge: Joel Rice,
Muktha Menon, Hans Nesse, Nicci Vega, Brian Langseth, Cindy Tribuzio, Jason
Gasper, Danny Badger and Shannon O’Brien. I thank many colleagues at the Hilborn
and Punt labs for support and many discussions. My special thanks to Jason Cope,
Billy Ernst, Arni Magnusson, Carolina Minte Vera, Juan Valero and Ian Stewart.
vi
My life in Seattle was enriched by a wonderful group of friends with whom I
shared this journey. I am eternally grateful to Manu Esteve and Mari Carmen (and
Mafi), Juan Valero, Billy Ernst, Jesus Jurado Molina e Tina Campos, Alexandre
Zerbini, Carlos Alvarez-Flores, Julian Burgos, Nancy Guarderas, Marina Roldan,
Georgeana Mezerani, Carolina Minte-Vera, Manoel Souza and Nicolas Gutierrez. I
thank Dan Jackson for his kind hospitality upon my arrival in Seattle. My special
gratitude to Allison Claussen.
My sincere gratitude to the staff of the Inter-American Tropical Tuna
Commission (IATTC) in La Jolla, California. My special thanks to Drs. Rick Deriso,
Mark Maunder, Guillermo Compéan and Robin Allen, for giving me enough
flexibility to finish this dissertation while starting a new job at the IATTC. I am
gratefully indebted to Mark Maunder for his support with the tagging analysis and for
his encouragement.
My journey as graduate student in the U.S. would not have started without the
encouragement and support of many friends and colleagues in Portugal. My special
thanks to Karim Erzini (Universidade do Algarve) and Helder Marques da Silva
(Universidade dos Açores) for encouraging me to pursue a graduate program in the
U.S. I extend my gratitude to all colleagues at Departamento de Oceanografia e
Pescas, Universidade dos Açores (DOP-UAç). I thank Pedro Afonso, Telmo Gomes,
Jorge Fontes, Valentina Costa, Ricardo Serrão Santos, Gui Menezes, Mário Rui
Pinho, Ana Martins and Helen Martins. My special thanks to Sara Magalhães.
I am eternally grateful to my family. To my parents, Alice (Lita) and José, to
whom I dedicate this study, for allowing me to pursue my interests and dreams, even
when these went beyond geographical boundaries. I am thankful for their
unconditional support and love. To my brother Frederico (Nico), for his support, and
for his tolerance with my physical absence in the family.
Last, but not the least, I am deeply grateful to Cynthia Dai, my soon-to-be
wife. My journey would not have been the same without Cynthia’s support,
encouragement and love. I thank Cynthia for her patience and for helping me to see
what really matters in the most challenging moments.
vii
DEDICATION
Para a Lita, José e Nico
viii
1
CHAPTER 1
BACKGROUND AND SYNOPSIS
1.1. THE CONTEXT-OF-THE PROBLEM
There is great concern about the population depletions of the apex fish
predators and the impacts that harvest removals can have on marine ecosystems
(Pauly et al. 1998; Stevens et al. 2000; 2003). One study indicated that the biomass of
the large pelagic fish predators in the world’s oceans, mainly tuna and billfish, has
been reduced to 10% of their historic virgin levels (Myers and Worm 2003). These
concerns extended to shark populations in various regions of the globe, producing
studies that likewise indicated serious declines due to fishing pressure (Baum et al.
2003; Baum and Myers 2004; Myers and Worm 2005; Shepherd and Myers 2005;
Ward and Myers 2005).
The validity of these alarming prospects is subject to debate (Walters 2003;
Burgess et al. 2005a; Burgess et al. 2005b; Maunder et al. 2006; Polacheck 2006).
However, there is no doubt about the relatively higher vulnerability to
overexploitation of certain groups of large pelagic fish predators. This is the case of
certain cartilaginous fish (class Chondrichthyes), particularly the elasmobranchs
(subclass Elasmobranchii) which include the sharks and rays (Compagno 1999). In
most cases, sharks lie on the K side of the life-history continuum defined by r/K
selection theory (Musick 1999; Stevens 1999). Their life-history strategy is usually
characterized by slow growth rates over long life spans, a late sexual maturity with
2
production of limited offspring (low fecundities) after long gestation periods
(Branstetter 1990; Stevens et al. 2000).
Despite the lack of empirical data on the nature of the relationship between
stock and recruitment (S-R) for sharks (Walker 1998; Cortés 2004), and based on
their reproductive limitations, the assumption of a virtually direct S-R relationship
was early on taken as a paradigm in elasmobranch fishery science (Holden 1973).
This relationship implies that shark populations decline more rapidly and compensate
less strongly for losses due to harvest, when compared with traditional exploited
teleost fish stocks (Stevens et al. 2000). Overall, the life-strategy of sharks is similar
to that evolved by marine mammals and reptiles, many of which under conservation
threat (Hoenig and Gruber 1990; Musick et al. 2000).
The history of shark fisheries demonstrate the general vulnerability of sharks
and rays to fishery exploitation (Holden 1974; Holden 1977; Anderson 1990; Hoff
and Musick 1990; Walker 1998; Cortés 2004). Typically, the exploitation begins with
a highly productive phase which is shortly followed by a rapid decline in the catch
rates and eventual collapse of the fishery. The following are the “classic” case study
examples of this so-called “boom and bust” shark fishery pattern: the Californian
fisheries for soupfin and thresher sharks (Ripley 1946; Cailliet and Bedford 1983),
the Australian school shark (Olsen 1959), the basking shark of the west coast of
Ireland (Parker and Stott 1965), the spiny dogfish fisheries in eastern North Atlantic
(Holden 1968) and the Pacific coast of North America (Ketchen 1969; Wood et al.
3
1979), and the porbeagle fisheries in the western North Atlantic (Casey et al. 1978;
Campana et al. 2002).
Although the shark reproductive limitations and the numerous cases of
overexploitation are not encouraging, sustainable shark fisheries may well be
possible. In fact, both Holden (1973) and Walker (1998) responded positively to the
question “ can shark resources be harvested sustainably?”. The limited opportunities
seem to be dependent on the biology of the species and capacity for
density-dependent regulation. Walker’s (1998) view goes even further when claiming
that the same life-history traits that make elasmobranchs so vulnerable to exploitation
can be turned into an advantage to achieve stable shark populations and sustainable
exploitation patterns. This results from their high survival rates and low recruitment
variability when compared to teleost species.
Unfortunately, the global scenario for shark fisheries shows little signs of
sustainability. Most of the world’s catches of sharks are taken incidentally (bycatch)
on various fishing gears in a wide variety of non-industrial and industrial non-shark
target fisheries (see overviews by Compagno 1990; Bonfil 1994; Rose 1996; Walker
1998). This non-target peculiarity, in addition to the low economic value of shark
catches, has resulted in a general low priority for data collection and research by
national government bodies (Stevens et al. 2000). Shark fisheries are therefore the
typical example of a data-limited situation in terms of fishery-dependent statistics
(Bonfil 1994). As a result, stock assessment analysis and evaluation of stock status is
a difficult task for sharks. Despite the data shortcomings, preliminary evaluations of
4
the world’s shark fisheries have been made and these indicated severe declines
(Castro et al. 1999).
1.2. THE CASE STUDY - NORTH ATLANTIC BLUE SHARK
1.2.1. The Single Stock Concept
The blue shark, Prionace glauca Linnaeus 1758 (Carcharhinidae) is
considered the most widespread and abundant of the pelagic sharks in the world’s
oceans (Gubanov and Grigor'yev 1975; Compagno 1984; Nakano and Seki 2003).
However, the blue shark stock structure in different oceans of the globe remains
unclear since a comprehensive global genetic study has not been done.
In the North Atlantic, this knowledge gap has been partially narrowed via
several tagging programs (e.g., Stevens 1976; Kohler et al. 2002; Fitzmaurice et al.
2005). Tag-recovery records have shown seasonal latitudinal movements of blue sharks
in both sides of the North Atlantic (e.g., Stevens 1976; Queiroz et al., 2004, for the
eastern side; Casey 1982, 1985; Kohler et al. 2002 for the western side). Transatlantic
migrations of blue sharks in both directions are common, but movements across the
equator are extremely rare. This highly migratory and trans-boundary nature of the
species became the supporting basis for the hypothesis of a single stock of blue sharks in
the North Atlantic Ocean, for stock assessment and management purposes (Casey 1982;
Kohler et al. 2002).
Apparently, blue sharks seem to use the main North Atlantic current systems to
move over the large ocean basin with different segments of the stock apparently
5
segregating by life-stage (size) and sex. While the occurrence of subadult females and
adult males seems to be more prevalent in the waters of the western North Atlantic (Pratt
1979; Casey 1985; Kohler et al. 2002), neonates, juveniles and adult females are
dominant on the eastern side (Castro and Mejuto 1995; Mejuto and García-Cortés 2005;
Litvinov 2006; Aires-da-Silva et al. 2008a).
1.2.2. The Fishery Exploitation Scenario
The blue shark meat has historically been regarded as unpalatable due to its
soft texture and strong odor of ammonia (Walker 1998; Castro et al., 1999; personal
observations of the author in the Azores, Portugal). In addition, its skin was
considered of poor quality for the fish leather tanning industry (Draganik and
Pelczarski 1984). This situation, however, has changed recently as shown by a study
of the U.S trade of fish leather in which blue and tiger sharks were the dominant
species among sharks (Grey et al. 2006). A major threat to blue sharks is a great
demand for their fins for shark fin soup and other dishes in some Asian cultures
(Draganik and Pelczarski 1984). In fact, a recent study showed that the blue shark fins
dominate the world’s largest Hong Kong shark fin market (Clarke et al. 2006a). As
the new technologies of blast-freezing become available to the high seas fisheries,
new markets for blue shark meat as food are also emerging (Fleming and
Papageorgiou 1996; Fordham 2006).
6
In the North Atlantic Ocean, blue sharks are predominantly caught as bycatch
of pelagic longline fisheries targeting tuna and swordfish (Anonymous 2005). The
main fisheries are described below.
The Asian Far Seas Tuna Fisheries – Pelagic sharks are an important bycatch
in the Asian fisheries targeting tuna in the high seas (Bonfil 1994). The major Asian
nations fishing in the Atlantic Ocean are Japan, the Republic of Korea, and Taiwan.
Japanese longliners have fished for tuna in the Atlantic Ocean since the
mid-1950s. The fishery initially developed in the equatorial grounds of the western
Atlantic and expanded to the entire ocean in 1970 (Bonfil 1994). Estimated catch
rates of pelagic sharks caught by Japanese longliners in the Atlantic Ocean ranged
between 1 to 10 sharks/1000 hooks (Bonfil 1994). A preliminary estimate of
approximately 26, 322 metric tons was obtained for the total pelagic shark bycatch
taken by the Japanese longliners in the Atlantic during 1989 (Bonfil 1994). Hoff and
Musick (1990) reported that the blue shark bycatch comprised 85% of the pelagic
shark bycatch taken by Japanese longliners operating in U.S. waters during the mid
1980s. The majority of the sharks caught in the Atlantic are discarded at sea after the
fins are removed (Nakano 1993), since these are highly profitable in the Asian shark
fin markets (Clarke et al. 2006a).
Longliners from the Republic of Korea have targeted tuna in the Atlantic
Ocean since the mid-1960s. The majority of the fishing effort by the Korean fleet has
7
taken place in the equatorial region between 20º N and 20º S, but mostly south of the
equator (Anonymous 2007). Reports on pelagic shark catches by Korea are not
available, but there is an estimate of approximately 190,000 sharks caught by Korean
longliners in the Atlantic during 1989, of which around 45% were blue sharks (Bonfil
1994). It was also reported that around 97% of the pelagic sharks caught by Korean
longliners in the Atlantic were discarded. However, the extent of fining practices by
the Korean fleet is unknown.
The Iberian Swordfish Fisheries - Blue sharks and other pelagic sharks are
taken as bycatch in some European fisheries (see Heessen et al., 2003 for a review).
The swordfish longline fisheries of the Iberian nations (Spain and Portugal) are the
dominant sources of blue shark fishing mortality by the European fisheries. The
Spanish fishery initiated its expansion throughout the waters of the eastern North
Atlantic during the mid-1960s (Garcés and Rey 1988; Rey et al. 1988). This fleet was
confined to the fishing grounds and archipelagos around the Iberian Peninsula and the
coast of Morocco until the 1970s. Garcés and Rey (1983) estimated that six blue
sharks were caught for each swordfish taken during the period from 1973 to 1981. By
the mid 1980s, the fishing grounds used by the Spanish fleet expanded westward to
areas close to the Grand Banks (Hoey et al. 1988; Mejuto and Iglesias 1988; Rey et
al. 1988), and as far south to the Gulf of Guinea along the coast of Africa. Most of the
blue shark bycatch by the Spanish fleet was discarded during the mid to late 1980s
8
(68-81%). However, fining and use of the carcass for bait were common during this
period (Mejuto 1985; Rey et al. 1988).
The swordfish longline fishery of mainland Portugal represents another
important source of blue shark bycatch in the eastern North Atlantic since the mid
1980s (Neves-dos-Santos et al. 2002). This fishery takes place mainly in the waters
around the Azores archipelago but also expands to southern latitudes off the coast of
Africa (Aires-da-Silva and Pereira 1999). On average, blue sharks represented up to
86% of the total catch on a longline set by the Portuguese fishery from 1993 to 1998
(Aires-da-Silva et al. 2008a). Blue sharks are also caught as bycatch of a regional
small-scale swordfish longline fishery in the Azores (Simões 1999).
The North American Fisheries -Swordfish fishing by U.S. and Canadian
longliners initiated in the early 1960s following reports of the incidental capture of
the species by Japanese and Norwegian longliners fishing for tuna and porbeagle
sharks, respectively (Anderson 1985). A westward expansion of the Spanish
swordfish fleet to the fishing grounds exploited by the North American and the
Japanese fisheries occurred in the 1980s (Mejuto and Iglesias 1988).
The Recreational Fisheries - Pelagic sharks and mainly blue sharks are
subject to recreational fishing on both sides of the North Atlantic Ocean. In the
western North Atlantic, blue sharks are the main target species of U.S. and Canadian
recreational fisheries (Kohler et al. 1998; Campana et al. 2005). In the eastern North
9
Atlantic, blue sharks are the main target species of recreational fisheries in waters of
England, Ireland, Spain and Portugal (Heessen 2003; Fitzmaurice et al. 2005).
1.2.3. Knowledge About Stock Status
An evaluation of the stock status for blue sharks is strongly handicapped by
limited fishery statistics as it is for any other bycaught shark species (Bonfil 1994).
Under- or non-reporting of bycatch, unknown discard levels, status (dead or alive) of
discards, and poor knowledge on the extent of fining practices are among the major
reasons for the lack of data.
Data collection and evaluation of the stock status of the blue shark stocks in
the Atlantic Ocean is conducted under the auspices of the International Commission
for the Conservation of Atlantic Tuna (ICCAT). A Sub-Committee on By-catch was
established in 1996 to coordinate these tasks (Kebe et al. 2002). At that time, it was
still not possible to conduct stock assessments for blue shark due to the limitations of
the fishery statistics.
Despite continuing data shortcomings, ICCAT has recently been able to
conduct assessments for the North and South Atlantic blue sharks stocks (Anonymous
2005). All of the analyses indicated that the current fishing mortality rates for the blue
shark in the North and South Atlantic are sustainable, and that current biomass levels
may even be close to the virgin level. These results are similar to those from stock
assessments for blue shark in the North Pacific Ocean (Kleiber et al. 2001; Sibert et
al. 2006). However, ICCAT considered the Atlantic assessments to be “very
10
preliminary” due to the uncertainties of the fishery statistics, and recommended better
use of other sources of data available, including tagging data (Aires-da-Silva et al.
2005).
The use of fishery-independent data, however, provided a different
interpretation about the status of blue shark stocks worldwide. Specifically, a recent
analysis of shark fin trade records indicated that blue sharks globally are being
exploited at levels close to or even above maximum sustainable yield (MSY) (Clarke
et al. 2006b). Unfortunately, a fishery-independent evaluation of the status of the
stocks for each ocean basin is not available yet.
1.3. DISSERTATION OUTLINE
The broad goal of this dissertation is to improve the quality of the scientific
information available for management and conservation of the North Atlantic blue
shark. The limited quality of the fishery statistics available leads to uncertainties
about any blue shark stock assessment (Anonymous 2005). While improvements need
to be made at the fishery-dependent data level, fishery-independent approaches are
particularly needed to provide alternative (and supplemental) information to those
which can be derived from traditional stock assessment techniques.
The title of this dissertation should be interpreted broadly. “Population
dynamics” in traditional fishery science is usually interpreted as the study of the
factors that affect growth, stability, and decline of populations. This dissertation
focuses on the study of three factors that affect the blue shark population dynamics:
11
demography, fishery exploitation, and movement. While independent per se, the
following three chapters address one or more of these aspects and are all related. The
chapters were prepared for direct submission of articles to peer review scientific
journals. Supplemental information (tables and figures) not to be published are
presented in the appendices.
Chapter 2 presents fishery-independent demographic and risk analyses for the
North Atlantic blue shark. While the fishery statistics for the blue shark are limited, it
is also data-rich with respect to the availability of life-history information. An
age-structured matrix population model where the life-history parameters are
stochastic was built. The demography of the blue shark is investigated for both virgin
(unharvested) and harvested scenarios. The results confirm that the blue shark is one
of the most productive of the pelagic sharks. However, an elasticity analysis showed
that this unusual capacity for population growth is dependent on the survival of
juvenile blue sharks. A risk analysis to evaluate the future prospects of the population
under different harvest rates is presented and implications for management discussed.
The results could be used for precautionary management, as a supplement to the
data-limited preliminary stock assessments.
Chapter 3 develops an index of abundance that covers the largest possible
period of the exploitation history of blue sharks in the western North Atlantic. This
chapter aims to gain perspective on the “missing baseline” when stock conditions
should have been close to pristine relative to the levels in more recent decades.
Pelagic longline catch records from several fishery observer programs that operated
12
between the late 1970s and the present are linked to observations from historical
fishery-independent longline surveys undertaken in the late 1950s, 1960s and 1970s.
Generalized linear model (GLM) techniques are used to derive standardize catch rates
(catch-per-unit-effort, CPUE) over four decades of exploitation history (1957-2000).
The index of abundance revealed a decline in blue shark CPUE of approximately
30% in the western North Atlantic from 1957 through 2000. The magnitude of this
decline is less than other recently published estimates and seems reasonable in light
of the high productivity of blue shark revealed by the demography analyses (Chapter
2; Aires-da-Silva and Gallucci 2007). The index of relative abundance could improve
the quality of the tuning process of future stock assessment models.
Chapter 4 analyzed the available tagging data in a spatio-temporal analysis to
derive estimates of movement and fishing mortality rates for the blue shark in the
North Atlantic Ocean. Although severely limited in terms of fishery-dependent
statistics, there is more tagging data for the North Atlantic blue shark than for any
pelagic shark stock. The analyses are based upon the historical blue shark
tag-recovery data archived by the U.S.-NMFS Cooperative Shark Tagging Program
(1965-2004).
Four broad geographical regions (two on each side of the ocean) were defined.
Despite substantial annual mixing of blue sharks among regions, a high proportion of
sharks remain in most regions each year. This spatial structure should be incorporated
into North Atlantic blue shark stock assessments. Fishing mortality rates (F) were
highly heterogeneous across the North Atlantic Ocean. Estimates of F for the western
13
North Atlantic were historically lower than 0.10 yr-1. Such low fishing mortality rates
are consistent with the low declines in catch rates estimated for the western North
Atlantic in Chapter 3. In contrast, F estimates over the most recent decade (1990s) in
the eastern side of the Atlantic were found to be rapidly approaching an estimated
reference point for conservation ( Fc =0.20 yr-1), estimated from the demographic and
risk analyses (Chapter 2; Aires-da-Silva and Gallucci 2007). These results suggest
caution since the juvenile and pregnant female segments of the stock are highly
vulnerable to the fishery in the waters of the eastern North Atlantic.
14
CHAPTER 2
DEMOGRAPHIC AND RISK ANALYSES APPLIED TO
MANAGEMENT AND CONSERVATION OF THE
BLUE SHARK IN THE NORTH ATLANTIC OCEAN
Equation Chapter 2 Section 1
2.1. INTRODUCTION
There is great concern about the population depletions of large pelagic
predatory fishes. A recent estimate indicates that the current biomass of many tuna
and billfish stocks is only at about 10% of their pristine levels in the world’s oceans
(Myers and Worm 2003). Similar alarming concerns have also been extended to shark
populations from many regions of the globe (Baum and Myers 2004; Myers and
Worm 2005; Ward and Myers 2005).
The fin trade is one of the major threats to shark populations. A recent
analysis of fin trade records showed the blue shark, Prionace glauca
(Carcharhinidae), as the dominant species in the world’s largest shark fin market
(Clarke et al. 2006a). The blue shark is an oceanic species with a wide-ranging
distribution throughout the oceans of the globe (Nakano and Seki 2003). A second
analysis of trade records suggests that blue sharks globally are currently exploited at
levels close to or even above maximum sustainable yield (Clarke et al. 2006b).
However, no evaluations of the blue shark stock status were made for individual
ocean basins.
In the North Atlantic, tag-recapture records have shown size and sex-related
latitudinal movements of blue sharks in both the western side (Casey 1985; Kohler et
15
al. 2002) and the eastern side (Stevens 1976; Fitzmaurice et al. 2005; Mejuto et al.
2005). Although long-distance trans-Atlantic migrations are commonly observed in
the different tagging programs, little mixing across the equator has been recorded.
This tagging information, together with catch and biological data, suggest the
existence of a single stock of blue sharks in the North Atlantic Ocean (Casey 1985;
Kohler et al. 2002).
Blue sharks are caught primarily as bycatch in tuna and billfish longline
fisheries by fleets from several nations fishing in the North Atlantic (Heessen 2003;
Anonymous 2005). Unfortunately, many of the logbooks for these various fleets that
are available are incomplete with respect to captures and release status (alive vs.
dead) of blue sharks. Limited numbers of observer programs cover only a small
fraction of the commercial fleets and extend over short temporal spans. This
data-limited situation handicaps a blue shark stock assessment analysis, which would
be difficult enough for this highly migratory species even if complete data were
available. The modeling process is complicated because of the complex sexual and
life-stage segregation patterns of the blue shark (Kohler et al. 2002).
Knowledge of the stock status of the blue shark in the North Atlantic is both
limited and contradictory. There are two opposing views. First, Baum et al. (2003)
reported the “collapse” of shark populations in the Northwest Atlantic based upon
catch rate analysis. Blue sharks, in particular, are noted to have declined dramatically
(by 60%) since the mid eighties. However, the methods and claims of Baum et al.
(2003) have been challenged (Baum et al. 2005; Burgess et al. 2005b; Burgess et al.
16
2005a). In contrast to the view of Baum et al. (2003), an evaluation of the stock status
by the International Commission for Conservation of Atlantic Tuna (ICCAT) reported
that the current exploitation levels are sustainable (Anonymous 2005), and that
biomass levels may even be close to the unfished carrying capacity. The assessment,
however, was considered by ICCAT as “very preliminary” due to the poor quality of
the data.
Although the catch levels of blue sharks appear to be on the rise (Anonymous
2005; Campana et al. 2006), the dichotomous perspectives on the stock status have
contributed to a low priority for management. New markets are emerging and moving
the blue shark trade from one dominated by fins (Clarke et al. 2006a), to one where
meat is also marketed. The new pattern is seen in the longline swordfish fisheries
from Spain and Portugal, at least, where blue shark meat is a large proportion of the
landings (Mejuto et al. 2002; Aires-da-Silva et al. 2008a), entering into some
European markets (Fleming and Papageorgiou 1996; Fordham 2006). Management of
these fisheries in the eastern North Atlantic waters is crucial since these waters
include important nursery grounds where pregnant females and juveniles are
particularly vulnerable (Casey 1985; Castro and Mejuto 1995).
Acquisition and maintenance of a fishery-dependent database that adequately
represents both exploitation by the multi-nation fleet and the spatial complexity of the
stock is difficult (Anonymous 2005). Alternatives to traditional fishery-dependent
stock assessment methods are thus needed (Cortés 1998). Demographic
fishery-independent based analyses are particularly appropriate for the blue shark
17
since it has one of the most complete life-history datasets available for sharks (see
synopsis by Nakano and Seki 2003 for blue sharks, worldwide). Demographic
parameters for the unfished blue shark population are available and are reported in
meta-analyses (Smith et al. 1998; Cortés 2002a; Frisk et al. 2004), one comparative
analysis of pelagic sharks (Cortés 2007), and the gray literature (Heessen 2003;
Campana et al. 2005; Takeuchi et al. 2005). While all the analyses indicate that P.
glauca is a highly productive shark species, knowledge of responses to perturbations
from different fishery scenarios is missing.
An age-structured matrix population model was constructed to investigate
blue shark demography and population dynamics. A stochastic framework that
accounts for the uncertainty in the vital rates was implemented following Cortés
(2002a). The demography of the population in unfished conditions was analyzed, and
elasticity analysis was used to identify the vital rates (fecundity and survival of pups,
juveniles and two adult stages) that contribute most to the rate of population growth
(Heppell et al. 1999). The impacts of a variety of harvesting scenarios on population
growth rates were examined and harvest rates that allow for a stationary population
were estimated.
While demography, and elasticity analysis, in particular, could offer a
“first-cut” knowledge basis for conservation and management (Heppell et al. 1999),
risk assessment analysis provides a framework to quantify risks that account for the
inherent variability of natural systems (Burgman et al. 1993). The risk (probability) of
18
a 50% decline of the population with respect to pre-exploited levels was evaluated
under different fishery scenarios.
2.2. MATERIALS AND METHODS
2.2.1. Age-Structured Model
An age-structured matrix population model (Caswell 2001) was used to
investigate the demography of the blue shark in the North Atlantic (definition of
mathematical notation is given in Table 2.1). The projection model
Ν t+1  MHN t
(2.1)
was used, in which Nt is the vector with numbers at each age in year t. The matrix M
is a Leslie population projection matrix
 f0
s
 0
M0

0
 0
f1
f2
0
0
0
s1
0
0
0
0
0
0
sx 1
fx 
0 
0,

0
0 
(2.2)
in which the sx element is the annual natural survivorship term for age x. The fx
elements represent the age-specific per-capita fecundity rates. A birth-pulse
population and a postbreeding census were assumed (Caswell 2001). Accordingly, the
first age-class (age 0) is represented by the newborn pups and the fecundity (fx) terms
include the probability that a pregnant female survives and delivers the pups at the
19
end of the year (fx=sxmx), in which mx is the average number of female pups per
female). The mx terms were calculated as the product of the number of pups per
female and the female sex ratio of the litters, which was then divided by the length of
the reproductive cycle in years. Thus, this is a female population growth model.
Fishery exploitation was modeled following Lefkovitch (1967) and Caswell
(2001). Gallucci et al. (2006) investigated the reproductive potential lost associated
with alterative exploitation strategies for sharks in the same way. The method
incorporates the use of a diagonal matrix, the harvest (or exploitation) matrix H, in
the projection model. The elements of the harvest matrix, hx, are the proportions of
individuals of age x surviving harvest. The model assumes that harvesting occurs
prior to natural mortality and reproduction. Essentially, the augmented projection
matrix MH has the same form as the original Leslie matrix M but is composed of
reduced terms for both fecundity and survival of the harvested age-classes.
Knowledge about the compensatory mechanisms of density-dependent
population growth for sharks is mainly limited to theoretical hypotheses (Walker
1998). Conceptualizing any hypothesis of regulation based on competition for
resources in a space-limited habitat becomes nothing more than speculation for the
highly migratory blue shark, with a vast distribution throughout the North Atlantic.
The dynamics of density-dependent growth was thus omitted from this research. The
underlying assumptions of the matrix model (2.1) are that the population will grow
exponentially and reach a stable age distribution (hereafter referred to as sad)
(Caswell 2001).
20
2.2.2. Uncertainty of Life-History Parameters
The biology of the blue shark has received substantial scientific attention to
date (see Nakano and Seki 2003 for synopsis), but uncertainty in critical life-history
parameters still exists. Uncertainty was accounted for using a simulation approach as
in Cortés (2002), where statistical distribution functions where defined for each lifehistory parameter, based on published records.
Age and growth - The age and growth of the blue shark in the North Atlantic
is addressed by several authors (Stevens 1975; Aires-da-Silva 1996; Henderson et al.
2001; Skomal and Natanson 2003). Parameter estimates from the latter study were
used since they were estimated from most of the blue shark size-range (Figure 2.1a).
A maximum age (longevity) estimate of 15 years was determined empirically for
females based upon interpretation of vertebral banding patterns (Skomal and
Natanson 2003). The authors also provide theoretical estimates of longevity (21 and
26 years). Twenty one years may be most realistic, and agrees closely with Stevens
(1975) theoretical estimate of 20 years. The empirical (15 years) and theoretical (21
years) values were used as the bounds of a discrete uniform probability mass function
(pmf) for maximum age (Figure 2.1b).
Reproduction - Pratt (1979) reported full sexual maturity at 185 cm FL for
female blue sharks which translates to an age-at-maturity of about 6 years using the
juvenile growth curves of Stevens (1975), Aires-da-Silva (1996) and Henderson et al.
21
(2001). Skomal and Natanson (2003) suggest faster growth and an earlier
age-at-maturity at five years of age. The growth curve inconsistencies were regarded
as a form of uncertainly. Accordingly, a “knife-edge” age-at-maturity was allowed to
vary uniformly between 5 years (Skomal and Natanson’s hypothesis) and 6 years
(other growth studies) (Figure 2.1c).
The length-at-age growth curve for mature females (Skomal and Natanson
2003) was used to obtain a fecundity-at-age curve (Figure 2.1d). A positive linear
relationship between litter size and fork length of the pregnant females for the
Atlantic Ocean (Mejuto and García-Cortés 2005) was used to convert mean
lengths-at-age to mean fecundities-at-age. A mean value of 0.5 was used for the sex
ratio of the litters (Castro and Mejuto 1995; Mejuto and García-Cortés, 2005; A.
Aires-da-Silva, unpublished data from the Azores). The length of the blue shark
reproductive cycle in any region of the globe is yet not fully understood. A
conservative estimate of a 2-year cycle was taken (Pratt 1979).
Survivorship - Natural mortality for the blue shark was quantified using a set
of indirect techniques that rely on relationships between life-history parameters and
natural mortality (Cortés 2002a; Simpfendorfer et al. 2005). Estimates of the
instantaneous rate of natural mortality M (hence survivorship, s=e-M) were generated
by six indirect methods: 1) Hoenig (1983); 2) Pauly (1980); 3) Chen and Watanabe
(1989); 4) Peterson and Wroblewski (1984); 5) Jensen’s (1996) age-at-maturity
method; and, 6) Jensen’s (1996) K growth coefficient method. The mathematical
22
equations for each method are in Appendix A. Blue shark pups are born with a mean
size between 37-47 cm FL (Pratt 1979) and probably become highly vulnerable to
natural sources of mortality during their first year of life (Cortés 2000). Rapid
juvenile growth should decrease and balance a high rate of natural mortality
(Branstetter 1990). Thus, two life-stages were assumed when defining the probability
density functions (pdf) for natural survivorship: 0-1 years and 2+ year old sharks.
The triangular pdf can be used to represent the uncertainty in life-history
parameters that precedes stochastic demographic work (Cortés 2002a; Cortés 2007).
This distribution is particularly convenient because it allows a lower and upper bound
for the parameter, and the assignment of a most likely value between this range. The
lowest and highest estimates of survival derived from the six methods were taken as
the bounds, and the median value was assumed as the most likely value in the
triangular distributions for the two stages (Figure 2.1e,f). Defining a pdf in terms of
natural mortality (M) is an alternative approach. The mean and coefficient of
variation (CV) obtained for M across methods were used as parameter estimates to
define a lognormal distribution. A lognormal error structure for M ensures that the
transformed estimates and resulting pdf of annual survivorship (s=e-M) vary between
0 and 1 (Figure 2.1e,f).
2.2.3. Stochastic Demographic Analysis
Virgin population - The productivity of the blue shark population was first
examined for the unfished (virgin) status. A set of demographic parameters was
23
generated (Caswell 2001). The dominant eigenvalue of the Leslie M matrix is the
finite rate of population increase (λ) at sad. The population doubling time follows as
2
t 2  ln  

(2.3)
Net reproductive rate (R0) was defined as the rate of growth of the population
from one generation to the next (see Caswell 2001 for matrix algebra).
Generation time was calculated as the time T required for the population to
increase by a factor of R0. T was computed to satisfy  T  R0 , hence
T
ln R0
ln 
(2.4)
Elasticity (proportional change) analysis was used to gain insight into the
influence that changes in reproductive and survival rates can have on the population
growth rate, λ (Heppell et al. 1999; Caswell 2001). Elasticities of the matrix elements
(eij) were calculated from
eij 
aij vi w j
 w, v
,
(2.5)
in which aij are the elements of the Leslie matrix M, v is the reproductive value
column vector with elements vi, w is the sad row vector with elements wj, and w, v
is the scalar product between v and w. The life-span was aggregated into four stages
for elasticity analysis: pups (0 years), juveniles (1-4 years) and two adult stages (5-9
24
years and a 10+ years group). Fecundity and survival elasticities for these age-groups
were calculated by aggregating elasticities across elements of interest (elasticities add
up to 1).
A stochastic approach was used to mimic natural variability and quantify the
uncertainty in the demographic parameters as in Cortés (2002a). The Monte Carlo
simulations consisted of randomly drawing values of the life-history parameters from
the statistical distributions described earlier (Figure 2.1). These were used as elements
in the Leslie matrix model and to derive estimates of the demographic quantities of
interest. Statistical distributions for the parameters were based on 10,000 simulations
and the 2.5th and 97.5th percentiles were used as approximate 95% confidence
intervals. All demographic and simulation analyses were coded in the R Language for
Statistical Computing (R Development Core Team, 2006).
The sensitivity of the demographic estimates to the choice of the statistical
distributions for the life-history parameters is an important consideration since there
is an element of subjectivity in the selection process (Cortés 2002a; Cortés 2007).
This is especially the case for survivorship, commonly the least known life-history
parameter for sharks (Simpfendorfer et al. 2005). The sensitivity of the demographic
estimates to the choice among the triangular and lognormal techniques described
early for deriving the pdf for survivorship (Figure 2.1c,d) was examined.
Harvested population - The harvest matrix model approach (1) was used to
investigate the growth response of the population, in terms of λ, in a fishery where the
25
controlling variables are the harvest rate (u) and the age-of-first-capture (tc).
Variability in the vital rates was accounted for by the stochastic framework described
earlier. The triangular pdf was chosen to define survivorship in the harvest analysis
because the results from the sensitivity analysis that follows show similar estimates of
λ for the two different statistical distributions of survivorship. The underlying
assumptions of the harvest analysis were: selectivity is “knife-edge” at tc, and the
fraction harvested annually (u) is applied equally to all selected ages.
The “stationary harvest” (us) was defined as the annual harvest rate u resulting
in a stationary population size (λ= 1). Three different scenarios were considered for tc.
The first scenario consisted of a harvest including the pups (tc=0 years). The second
hypothesis was that the pups are not selected into the fishery, but all remaining
juvenile and adult ages are harvested (tc=1 year). A third scenario assumed that the
whole juvenile segment (0-4 years) is protected and the fishery selects only the adult
segment (5+ years) of the population (tc=5). Estimates of us were determined
iteratively and statistical distributions were produced as described earlier for the
demographic parameters.
2.2.4. Population Projections and Risk Analysis
A risk assessment (Burgman et al. 1993) was conducted to quantify the
probability of a long-term population decline to critical levels. The risk analysis
requires that a critical value be chosen, below which the population is said to be in
risk of overexploitation. Maximum sustainable yield (MSY) for some pelagic shark
26
species has been postulated to be at or above 50% of their virgin biomass (Cortés
2007). In particular, it is close to 50% for the blue shark and shifted further toward
virgin biomass for two of the less fecund Alopias species and shortfin mako (Isurus
oxyrinchus). Accordingly, the 50% seems a reasonable estimate of the critical value
for precautionary management.
The mean stable age distribution was taken as the initial condition to initiate
the population projections for the risk analysis. The sad assumption, however,
represents one of the major shortcomings of traditional demographic analysis because
it assumes that survival and reproductive rates are constant over time (Cortés 1998).
This static view is unrealistic and could result in misleading optimistic interpretations
based on the use of the dominant eigenvalue λ for population projection work
(Tuljapurkar 1990). The sad assumption and dependence upon the dominant
eigenvalue theory where relaxed in the projection work and risk analysis (Burgman et
al. 1993).
Instead of assuming that vital rates are fixed over time, they were allowed to
vary randomly from time step to time step. This corresponds to a time-variant Leslie
matrix (Mt) in the projection model (2.1). Forward population projections were made
over a 15-year period, which is the empirical longevity estimate for female blue
sharks (Skomal and Natanson 2003). This process was repeated 10,000 times using
Monte Carlo simulations to provide a sample of population trajectories from which to
calculate the risk statistic.
27
The harvest scenarios modeled were different combinations of harvest rates
focusing on juvenile (1-4 years) and adult (5+ years) blue sharks. Pups (0 years) were
assumed not to be selected into the longline fishery due to small mean size (37-47 cm
FL; Pratt, 1979). Five levels of harvest were considered (u=0, 0.1, 0.2, 0.3, 0.4).
Harvest rates were assumed constant across the juvenile and adult age-classes to
allow for a tractable number of harvesting combinations. A constant harvest policy
over the projection period was assumed.
The population depletion obtained under each harvest strategy at the end of
the projection period is dependent upon the depletion status in the starting year. Three
hypotheses were considered for the initial population size: a population still at its preexploited levels (N0), at 75% of N0, and at 50% of N0.
2.3. RESULTS
2.3.1. Survivorship
Estimates of natural survivorship (s=e-M) produced for ages 0-1 years by the
six indirect methods averaged 0.73 (median=0.74; Table 2.2). A minimum value of
0.53 was obtained using the technique of Chen and Watanabe (1989), which relies on
the von Bertalanffy growth parameters. The Hoenig (1983) method, assuming a
maximum age of 21 years, provided the maximum estimate (0.82) among all
methods. Survivorship for ages 2+ years averaged 0.78 (median=0.81). The method
of Pauly (1980) used the von Bertalanffy growth parameters and a mean water
temperature value, and generated the minimum estimate of 0.62. A maximum
28
estimate of 0.91 was produced by the method of Peterson and Wroblewski (1984),
based on weight-at-age data.
2.3.2. Demographic Analysis
Estimates of the finite rate of population increase (λ) averaged 1.23 year-1
when the triangular pdf was assumed for survivorship (Figure 2a) and the distribution
for the population doubling times (t2) averaged 3.08 years (Figure 2b). A mean
growth rate of 8.33 per-generation was estimated for the net reproductive rate R0
(Figure 2c), with the estimate of mean generation time (T) being 9.39 (Figure 2d).
The stable age distribution (sad) under virgin conditions showed a dominance of
juveniles (0-4 years) (Figure 2e). This segment represented, in average terms, 91.2%
of the population in numbers. Pups alone accounted for 41.3% of the total number of
animals in the population. In contrast, the adult segment (age-classes 5+) accounted
for a mean proportion of 8.8% of the population at a sad. Age-classes 10+ represented
less than 1% of total population numbers.
Elasticity analysis showed that juvenile survival is the dominant contributor to
population growth (mean aggregated elasticity of 57.7% across ages 0-4; Figure 2f).
Comparatively, adult survival (5+ years) contributed much less to λ (survival
elasticity of 30.7% aggregated across mature ages of 5+ years). The proportional
contribution of fecundity to population increase was quantified as a mean elasticity
value of 11.5%.
29
The demographic and elasticity statistics were rather insensitive to the choice
between defining the pdf of survivorship using the triangular or the lognormal
distribution (Table 2.1, Figure 2.2). The lognormal distribution generally led to a
wider range of values as a result of the larger dispersion represented in the lognormal
pdf (Figures. 2.1e,f). Nevertheless, the measures of central tendency for the analysis
based on the lognormal distribution were very similar to those produced by the
analysis based on the triangular distribution.
The mean growth response of the population in terms of λ when the harvest
rate (u) and age-at-first-capture (tc) are the controlling variables was represented
using isopleths (lines of equal λ; Figure 2.3a). Population declines (λ<1) were found
to be associated with harvesting strategies selecting the juvenile age-classes (0-5
years).
The distributions obtained for the stationary harvest (us) associated with
fisheries focusing on different population segments (defined by different values of tc)
are shown in Figure 2.3b. A mean stationary harvest of 0.19 (median=0.19, 95% CI:
0.07-0.26) was obtained for a fishery starting at age 0 (pups) and including all age
classes. The estimates for us barely increased to a mean value of 0.21 (median=0.22;
95% CI: 0.07-0.29) when pups were protected and harvesting started at age 1. Higher
stationary harvest rates were possible (mean us=0.42; median=0.43, 95% CI: 0.150.63) when the whole juvenile segment was protected (0-4 years) and the exploitation
focused on the mature sharks only (tc set to the age-of-maturity, 5 years).
30
2.3.3. Risk Analysis
The stochastic framework produced a sample of individual population
trajectories under varying harvest scenarios, while accounting for the annual natural
variability in the vital rates (Figure 2.4).
In this analysis, the risk of depleting the population to levels below 50% of the
pre-exploited levels after a 15-year period of constant harvest rate was evaluated. The
risk statistic depends on three factors: the harvest rates for the juveniles, for adults,
and the initial population size relative to the pre-exploited level (N0). The risks
associated with different levels for the juveniles and adults (from 0 to 40%) are
provided in Table 2.4. The corresponding risk curves for the three hypotheses about
the initial population size (N0, 0.75N0, and 0.5N0) are shown in Figure 2.5.
For example, if a 30% harvest rate is considered for both juveniles and adults,
there is a risk of 100% of the population falling below 50% of N0 for all three
hypotheses about the initial population size. If a more conservative strategy is
considered, e.g., a 20% harvest rate applied to juveniles and adults, there is zero risk
of the population falling below 50% of N0, if the initial population sizes were still at
N0 or at 0.75N0. However, there is an approximately 25% risk if the initial population
size was 50% of N0.
It falls to the manager to determine what is an acceptable risk level. The
domain of the risk curves (Figure 2.5) can be narrowed to include all strategies where
the risk is less than some acceptable level, e.g., 25%. Take an extreme case for adult
harvesting as an example. If the strategy is to allow for a harvest rate of 30% on the
31
adults, then a juvenile harvest rate below 10% is required if the risk is not to exceed
the acceptable 25% level for all three hypotheses about initial population size.
2.4. DISCUSSION
This research quantitatively confirms that that the North Atlantic blue shark
stock is highly productive within the shark group. An annual rate of increase of 1.23
year-1 (mean λ) and mean doubling time of 3.1 years (t2) indicate a high capacity for
growth, in the absence of harvesting. The values for λ (=er) calculated in this study
are consistent with the range of estimates obtained for the blue shark in previous
analyses in deterministic (1.03-1.28, Heessen, 2003; 1.43, Campana et al., 2005) and
stochastic analysis: mean=1.40 (95% CI: 1.28-1.53; Cortés, 2002); mean=1.41 (90%
CI: 1.25-1.63; Takeuchi et al., 2005); mean=1.25 (90% CI: 1.15-1.37; Cortés, 2007).
This high capacity for population growth ranked P. glauca third in terms of λ in a
meta-analysis involving 38 sharks species (Cortés 2002a). Another meta-analysis
(Smith et al. 1998) ranked the blue shark seventh among 26 shark species when the
estimates of intrinsic rebound potential (another demographic statistic of
productivity) were compared. These data substantiate that the blue shark is one of the
most productive sharks, especially among the pelagic species.
Elasmobranchs are traditionally placed on the K side of the life-history
continuum defined by r/K selection theory (Musick 1999; Stevens 1999). The group
is usually characterized by slow growth rates, long life spans, late maturity, and
production of limited offspring after long gestation periods (Branstetter 1990; Hoenig
32
and Gruber 1990). This low reproductive output is responsible for the vulnerability of
elasmobranchs to harvest, as shown by many cases of overexploitation (Cortés 2004).
Although certainly K-selected among fish in general, the blue shark seems to
be shifted to the r- side of the life-history spectrum among elasmobranchs. According
to the analysis of life-history patterns by Cortés (2000), this strategy consists of large
production of small offspring that become highly vulnerable to natural mortality.
Such high natural mortality is counterbalanced by one of the highest fecundities
documented among sharks (mean litter size of 37 pups, Mejuto and García-Cortes,
2005; reaching up to 82 pups, Pratt, 1979) and by fast early growth rates (Branstetter
1990) that result in nearly doubling of pup size over the first year (Stevens 1975;
Skomal and Natanson 2003).
Expansion through the pelagic environment may also play a key role in the
successful life-history strategy exhibited by blue sharks. The relatively poor nutrient
supply available in these oceanic oligotrophic waters may be balanced by diminished
competition with other predators in an environment where space should not be a
limiting factor, as opposed to coastal waters. The cosmopolitan distribution of P.
glauca throughout the globe, and the possibility that it may be the most abundant
pelagic shark species (Nakano and Seki 2003), reflect the success of its life-history
strategy.
However, the high productivity of blue sharks could be misleading, and
should not compromise the necessity to manage the current numbers lost to directed
and incidental mortality in the North Atlantic, or in any other region of the globe
33
(Bonfil 1994; Clarke et al. 2006b). The question should still be asked whether there is
any vulnerable aspect of the blue shark population that could put their inherent
productivity at risk and lead to overexploitation.
Elasticity analysis confirmed that survival of the juvenile segment (0-4 years)
is the key factor behind productivity. This pattern is common for long-lived species,
and sharks in particular (Heppell et al. 1999; Cortés 2002a; Cortés 2007). Population
declines (λ<1) could result even if low (0.2-0.3) harvest rates are applied to juveniles.
Protection of the juveniles may actually be the easiest management action to
implement if a sustainable exploitation is desired. This could be attained by sizelimits implemented using regulations on fishing gear. An alternative or supplemental
option is protection on the nursery grounds. Even though the complex sexual and lifestage segregation patterns of the blue shark over the entire North Atlantic are not yet
fully understood, catch and tagging records suggest that the main nursery grounds are
located in the eastern North Atlantic (Casey 1985; Castro and Mejuto 1995).
Management and conservation of the blue shark would therefore benefit from
longline selectivity studies and research focusing on a better delineation and mapping
of the nursery grounds in the North Atlantic.
Implementation of either management policy is handicapped by the fact that
the blue shark is caught primarily as bycatch in commercial longline fisheries for
swordfish and tuna (Heessen 2003; Anonymous 2005). Gear or spatial-temporal
limitations on these fisheries would have therefore socio-economic implications.
Nevertheless, implementation of a seasonal constraint on the longline fisheries
34
overlapping in time and space with the nursery areas appears to be a practical
approach to protect juveniles and to simultaneously minimize the impacts on these
fisheries.
An example is given by the swordfish longline fishery in the waters of the
Azores, eastern North Atlantic. Length-frequency data collected in the Azores during
the spring suggest that this region is an important nursery ground where juvenile blue
sharks congregate during this season, at least (Aires-da-Silva et al. 2008a). Higher
catches of juvenile blue sharks are obtained in the spring, while the high catch season
for the swordfish occurs in the autumn. These two fishing seasons are markedly out
of phase with each other. The extension of this asynchronous pattern from the waters
of the Azores to surrounding nursery areas in the eastern North Atlantic requires
further investigation.
Protection of nurseries falls within the context of large-scale marine reserves
for shark conservation (Baum et al. 2003). The efficiency of open-ocean reserves,
however, is dependent upon the migratory patterns of the juveniles in these regions.
Seasonal latitudinal migrations have been suggested for blue sharks in the eastern
North Atlantic, and these patterns seem to be strongly correlated with variations in
water temperature (Stevens 1976; Queiroz et al. 2005). Thus, quantifying the
movement patterns of the juveniles and their residency times in the nurseries becomes
a critical step in the optimal design of any marine reserve strategy.
The analysis of risk has become a common practice in fisheries stock
assessment (Punt and Hilborn 1997). In this paper, risk analysis lies at the interface
35
between demographic analysis and stock assessment. In the absence of a reliable
fishery-dependent dataset that could form the basis for a stock assessment, the results
from the risk analysis offer supporting information for management and conservation.
First, they help to identify a critical reference point which could guide precautionary
management. In particular, a maximum annual harvest rate of 20% for both juveniles
and adults could be considered (Figure 2.5). Harvest rates below this threshold carry
low risks (<25%) of population declines to levels below half of the pre-exploited size
(N0). These low risks are independent of three hypotheses for the initial population
size at the start of the harvest policy (N0, 0.75N0, and 0.5N0). Increasing the harvest
rate above the 20% threshold for either juvenile or either adults individually, or both,
results in rapidly increasing risk.
An evaluation of the future prospects of the stock could be obtained from the
risk analysis if an estimate of the current harvest rate were available. Unfortunately,
the limitations of the fishery-dependent data introduce uncertainty in the estimation of
this parameter. Nevertheless, a reliable estimate for the harvest rate could well be
obtained in the near future. For this purpose, the large body of mark-recapture theory
(Seber 2002) could be used to capitalize on approximately four decades of extensive
tagging data accumulated by several programs conducted on both sides of the North
Atlantic (Stevens 1976; Kohler et al. 1998; Kohler et al. 2002; Fitzmaurice et al.
2005; Mejuto et al. 2005). Quantitative mark-recapture analyses efforts using these
data have been recently initiated (Aires-da-Silva et al. 2005; Campana et al. 2006).
36
As in any population modeling analysis, the assumptions about the population
are central to the validity of the results. The matrix modeling framework is flexible
and permits the inclusion of different biological hypotheses. The absence of
compensation in the current version of the model implies exponential population
growth while it is possible that population growth rates may be depressed in the real
world leading to more conservative demographic and risk predictions. Therefore,
some form of compensation should be included in subsequent models. This is
difficult, however, given the lack of empirical data on compensation for the sharks of
the open ocean. An alternative to empirical data could follow from simulation
analyses for a variety of hypothetical scenarios of harvest and compensation, such as
density-dependent changes in growth, reproduction, and mortality. An example of a
scenario can be found in the hypothesis that small size blue shark pups may be more
vulnerable to natural mortality (Cortés 2000). Thus, the role of density-dependent
changes in pup survival would be a relevant topic for a simulation study.
The present demographic analyses confirm the importance of juvenile survival
for population growth. Thus, future blue shark population dynamics and stock
assessment models that are not age- or stage-based are of limited value. In addition,
inclusion of spatial and movement components could better describe the complex
spatial segregation and migratory patterns of the population by life-stage and sex, as
well as different harvest scenarios (e.g., exploitation of juveniles in the nurseries).
Finally, the urgency for advances in management of global blue shark
populations is currently an important conservation topic (Clarke et al. 2006b). It
37
seems that the blue shark life-history parameters do not differ greatly over the world’s
oceans (Tanaka et al. 1990; Nakano and Seki 2003). While this paper is concerned
with the North Atlantic region, applications of the demographic and risk analysis
results to blue sharks in other parts of the world appear to be reasonable and are
straightforward to implement.
38
Table 2.1. Definition of mathematical notation.
Nt
M
sx
fx
mx
Η
hx
MH

w
wj
v
vi
t2
R0
T
eij
aij
w, v
us
tc
N0
Vector of numbers at age in year t
Projection (Leslie) matrix
Annual natural survivorship at age x
Fecundity at age x
Mean number of female pups per female at age x
Harvest survival matrix with diagonal elements 1  hx
Fraction surviving harvest at age x
Augmented projection matrix
Dominant eigenvalue of MH (or M in the unexploited case). The annual
rate of increase of the population at SAD.
SAD vector for M , satisfying Μw  w
Element in column j of the SAD vector w
Reproductive value vector for M , satisfying v T Μ   v T
Element in row i of the reproductive value vector v
Population doubling time (in years)
Net reproductive rate at SAD
Mean generation time
Elasticities of the elements i,j in the Leslie matrix ( M )
The elements i,j in the Leslie matrix ( M )
Scalar product between v and w
Sustainable harvest, annual harvest rate (u) resulting in a stationary
population (   1 )
Age at first capture in the fishery
Pre-exploited level of the population
39
Table 2.2. Estimates of natural survivorship produced by the following indirect
methods: 1) Hoenig (1983); 2) Pauly (1980); 3) Chen and Watanabe (1989); 4)
Peterson and Wroblewski (1984); and 5) Jensen’s (1996) age-at-maturity and K
growth coefficient methods.
Age (yr)
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
0-1 years
2+ years
Hoenig
w = 15
w = 21
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
0.76
0.82
mean
0.73
0.78
median
0.74
0.81
Pauly
C&W
0.62
0.62
0.62
0.62
0.62
0.62
0.62
0.62
0.62
0.62
0.62
0.62
0.62
0.62
0.62
0.62
0.62
0.62
0.62
0.62
0.62
0.62
0.53
0.65
0.71
0.75
0.78
0.80
0.82
0.83
0.83
0.84
0.85
0.85
0.86
0.86
0.86
0.86
0.86
0.86
0.87
0.86
0.86
0.86
min
0.53
0.62
max
0.82
0.91
P&W
0.73
0.79
0.83
0.85
0.86
0.87
0.88
0.88
0.89
0.89
0.90
0.90
0.90
0.90
0.90
0.91
0.91
0.91
0.91
0.91
0.91
0.91
Jensen
t mat
0.72
0.72
0.72
0.72
0.72
0.72
0.72
0.72
0.72
0.72
0.72
0.72
0.72
0.72
0.72
0.72
0.72
0.72
0.72
0.72
0.72
0.72
K
0.81
0.81
0.81
0.81
0.81
0.81
0.81
0.81
0.81
0.81
0.81
0.81
0.81
0.81
0.81
0.81
0.81
0.81
0.81
0.81
0.81
0.81
Table 2.3. Descriptive statistics for the demographic parameters: population rate of increase (λ), doubling time (t2), net
reproductive rate (R0), mean generation time (T). Statistics for fecundity and survival elasticities for pups, juveniles (1-4
years), adult state I (5-9 years) and adult stage II (10+ years) are also given.
Elasticities
Method
λ
Demographic statistics
t2
R0
T
Fecundity
Adult I
Adult II
Pups
Survival
Juvenile
Adult I
Adult II
Triangular pdf on s
Mean
Median
Lower 95% CI
Upper 95% CI
1.232
1.242
1.072
1.356
3.080
3.197
2.274
9.874
8.328
7.720
1.883
17.874
9.393
9.392
8.319
10.437
0.094
0.093
0.077
0.110
0.022
0.022
0.015
0.028
0.115
0.116
0.106
0.126
0.462
0.464
0.423
0.502
0.272
0.260
0.226
0.312
0.035
0.034
0.019
0.055
Lognormal pdf on M
Mean
Median
Lower 95% CI
Upper 95% CI
1.241
1.255
0.974
1.424
3.548
2.954
10.891
8.559
0.807
33.679
9.382
9.403
7.999
10.669
0.095
0.093
0.076
0.115
0.021
0.021
0.013
0.030
0.116
0.114
0.105
0.128
0.464
0.457
0.420
0.512
0.270
0.282
0.215
0.314
0.034
0.033
0.017
0.058

15.924
40
41
Table 2.4. Risk (probability) of obtaining a population decline to levels below 50% of
the pre-exploited levels (N0). The management options are different harvest rates for
juveniles (ujuv) and adults (uadult). Three different hypotheses were considered for the
starting population size: a population at N0, at 0.75 N0 , and at 0.50 N0. The shaded
cells highlight risk values above 25%.
Management options
h juv
h adult
No harvest No harvest
0.1
0.2
0.3
0.4
0.1 No harvest
0.1
0.2
0.3
0.4
0.2 No harvest
0.1
0.2
0.3
0.4
0.3 No harvest
0.1
0.2
0.3
0.4
0.4 No harvest
0.1
0.2
0.3
0.4
Hypotheses for starting pop. size (%N 0 )
N0
0.75N 0
0.5N 0
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.03
0.29
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.02
0.25
0.38
0.77
0.99
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.23
0.37
0.79
0.99
1.00
1.00
1.00
0.00
0.00
0.00
0.00
0.01
0.19
0.35
0.80
1.00
1.00
1.00
1.00
1.00
1.00
1.00
0.00
0.00
0.11
0.29
0.74
0.99
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
42
a) VBGFs
Probability mass
Fork length (cm)
300
200
Stevens (1975)
Aires-da-Silva (1996)
100
Skomal and Natanson (2003)
b) Longevity
0.2
0.1
Henderson et al. (2001
0
0.0
0
2
4
6
8
10
12
14
15
16
17
Age (years)
20
21
80
d) Fecundity at age
c) Age at maturity
60
0.50
# of pups
Probability mass
19
Age (years)
0.75
40
0.25
20
0.00
0
3
4
5
6
7
8
5
7
9
11
Age (years)
13
15
17
19
21
Age (years)
10
Probability density
10
Probability density
18
e) Survivorship (0-1 years)
8
triangular s
lognormal M
6
4
2
0
f) Survivorship - 2+ years
8
triangular s
lognormal M
6
4
2
0
0.2
0.3
0.4
0.5
0.6
0.7
Survivorship
0.8
0.9
1.0
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Survivorship
Figure 2.1. Uncertainty in the life-history parameters of the North Atlantic blue shark.
a) von Bertalanffy growth functions (VBGFs) from different studies; b) probability
mass function (pmf) for longevity; c) pmf for age-at-maturity; d) fecundity-at-age
relationship; e) and f) are the probability density functions of natural survivorship (s)
for 0-1 years and 2+ year old sharks.
43
b)
a)
Density
triangular s
lognormal M
0.0
0.4
0.8
1.2
1.6
Population growth rate year
2.0
0
2
1
4
6
8
10
12
Population doubling time (years)
d)
Density
c)
0
10
20
30
40
0
2
Net reproductive rate
6
8
10
12
Generation time (years)
0.8
0.8
e) Stable age distribution (SAD)
Proportion
4
f) Elasticies
0.6
0.6
Fecundity
triangular s
0.4
lognormal M
0.2
Survival
0.4
0.2
0.0
0.0
0
1
2
3
4
5
6
7
8
9
10
Age-class (years)
11
12
13
14 15+
Pup
Juv
Ad1
Ad2
Pup
Juv
Ad1
Ad2
Age-group
Figure 2.2. Statistical distributions of demographic parameters for an unfished blue
shark population: a) finite rate of population increase, λ; b) doubling time, tx; c) net
reproductive rate, R0; and mean generation time, T. e) the stable age distribution, sad;
f) elasticities (of fecundity and survival) for aggregated age-groups (pups – 0 years;
Juv - juveniles, 1-4 years; Ad1 – small-size adults, 5-9 years; Ad2 – large-size adults,
10+ years).
44
Age at first capture (years)
14
a)
12
10
8
b)
6
4
2
0
0.0
0.2
0.4
0.6
0.8
1.0
Annual harvest rate
b)
tc = 0 years
Density
tc = 1 year
tc = 5 years
0.0
0.2
0.4
0.6
0.8
1.0
Stationary harvest rate
Figure 2.3. a) Isoclines for the average response in population growth rate (λ) to
different harvest rates (u) starting at different ages-at-first-capture (tc). The solid line
indicates a stable population (λ=1). Values above and bellow this line represent
population growth (λ>1) and decline (λ<1), respectively. b) Distributions derived for
the stationary harvest rate (us) for fisheries beginning at an age-at-first-capture of 0
years (pups), 1 year (juveniles) and 5 year (adults).
Poulation numbers
45
2
4
6
8
10
12
14
Year
Figure 2.4. Example stochastic population projections. Fifty individual trajectories
over a 15-year harvesting period are presented. The harvest rates used in this example
are the same (u=0.3) for both juveniles and adults.
1.0
46
Nstart
ujuv =0
0.5
N0
0.75N0
0.0
0.50N0
0.1
0.2
0.3
0.4
0.2
0.3
0.4
0.2
0.3
0.4
1.0
0.0
0.0
0.5
ujuv =0.1
0.1
1.0
0.0
0.5
0.0
Risk
ujuv =0.2
0.1
1.0
0.0
0.0
0.5
ujuv =0.3
0.1
0.2
0.3
0.4
1.0
0.0
0.0
0.5
ujuv =0.4
0.0
0.1
0.2
0.3
0.4
Adult harvest rate, uadult
Figure 2.5. Risk curves for depleting the population to levels below 50% of its
pre-exploited population size (N0). The five plots represent different harvest strategies
defined by three factors: harvest of juveniles (ujuv), and adults (uadult), and the initial
population size fraction of N0.
47
CHAPTER 3
A HISTORICAL INDEX OF ABUNDANCE FOR THE
BLUE SHARK IN THE WESTERN NORTH ATLANTIC
Equation Chapter 3 Section 1
3.1. INTRODUCTION
There is concern about the population status of top predatory fishes, the
preferred target species for global fisheries, and the impact that reduced predator
abundance could have on marine ecosystems (Pauly et al. 1998). One meta-analytical
estimate indicated that the current biomass of large predatory fish (mainly tuna and
billfish) is less than 10% of their historic (pristine) levels in the world’s oceans
(Myers and Worm 2003).
Similar alarming concerns were extended to shark populations (Baum et al.
2003; Baum and Myers 2004; Myers and Worm 2005; Shepherd and Myers 2005;
Ward and Myers 2005). Elasmobranchs (sharks and rays) have a life-strategy that
consists of slow growth rates during long life-spans, late ages-of-maturity, and the
production of limited offspring after long gestation periods (Hoenig and Gruber 1990;
Musick et al. 2000). Historically, sharks have been subject to a few directed fisheries
but the dominant source of fishing mortality is bycatch in a wide variety of global
fisheries employing almost every type of fishing gear used (Bonfil 1994; Walker
1998). The world’s unregulated shark fin trade is a major contemporary threat for
sharks (Clarke et al. 2006a; Clarke et al. 2006b). The poor record of sustainable
harvesting of shark populations is cause for concern, considering the many case
studies of overfishing in targeted shark fisheries (Anderson 1990; Cortés 2004).
48
Unfortunately, incomplete historical information on shark fishery statistics and
questions about the fate of discarded catches has resulted in the classification of stock
assessment analysis for sharks as data limited (Bonfil 1994).
The blue shark, Prionace glauca (Carcharhinidae), is among the most wideranging of large-open ocean predators, and it may be the most abundant of all pelagic
sharks in the global oceans (McKenzie and Tibbo 1964; Draganik and Pelczarski
1984; Nakano and Seki 2003). Analyses of trade records indicate that blue shark fins
dominate Hong Kong’s shark fin trade (Clarke et al. 2006a) and that their global
exploitation rate may exceed maximum sustainable yield (Clarke et al. 2006b). The
relative contribution of blue sharks from each ocean basin to the global fin market
remains unknown.
Little is known about the differentiation of blue shark stocks in the world’s
oceans. In the North Atlantic, the hypothesis of a single stock for management and
stock assessment purposes is substantiated by extensive tag-recapture records (Casey
1985; Kohler et al. 1998). These include frequent trans-oceanic basin migrations of
blue sharks, and a very limited number of tag-recapture records crossing the equator.
Blue shark catches in the North Atlantic have historically resulted primarily
from bycatch in multi-national pelagic longline fisheries that target tuna and
swordfish (Heessen 2003; Anonymous 2005). Since the initiation of these fisheries
over half century ago, different exploitation patterns and target fishing practices have
been employed. Japan first deployed pelagic longline gear in its tropical Atlantic tuna
fishery in the mid 1950s (Nakano 1993; Bonfil 1994). In the 1960s, other nations
49
including Spain, the U.S., Canada and the Faroe Islands, utilized longline gear to
target tuna, swordfish, and porbeable sharks, with blue sharks as bycatch. New
exploitation patterns emerged during the last decade in response to reduced
international swordfish quotas. In particular, effort has been diverted to target blue
sharks for both meat (food) production (Mejuto et al. 2002; Aires-da-Silva et al.
2008a) and for the fin trade (Clarke et al. 2006a). The development of markets for
shark meat has been particularly strong in Europe (Fleming and Papageorgiou 1996;
Fordham 2006).
The stock status of blue sharks in the North Atlantic is currently ambiguous.
Baum et al. (2003) reported a general pattern of “collapse” for shark populations in
the western North Atlantic, with blue sharks declining dramatically (by 60%) since
the mid 1980s. The International Commission for the Conservation of Atlantic Tuna
(ICCAT; Anonymous 2005) carried out a preliminary stock assessment and
concluded that current exploitation levels are sustainable and that current biomass
levels are above the level capable of producing maximum sustainable yield (MSY).
Concern was expressed by ICCAT, however, about the limited quality of the fishery
data available. The shortcomings include incomplete landings and fishing effort
statistics, limited data on the status of discards (dead or alive), and on post-release
survival. ICCAT also recommended better fishery-dependent (and -independent)
information such as indices of abundance, to improve the quality of future
assessments (Anonymous 2005).
50
Reliable indices of abundance play a key role in tuning analytical stock
assessments (Maunder and Starr 2003; Maunder and Punt 2004). They are
particularly critical for the North Atlantic blue shark because of the uncertainty about
historical levels of bycatch mortality. Catch rate based standardized indices of
abundance already exist for the blue shark in the western North Atlantic. These
include analyses of U.S., Canadian and Japanese commercial fisheries data (Cramer
2000; Cortés 2002b; Baum et al. 2003; Brooks et al. 2005; Campana et al. 2006) and
recreational fishery data (Babcock et al. 2000; Brown 2000; Skomal et al. 2005;
Campana et al. 2006). A fishery-independent survey index is also available
(Simpfendorfer et al. 2002). Unfortunately, the relative abundance trends reported in
these studies often conflict, ranging from increasing, to stable and to decreasing
trends and usually cover the period from the mid 1980s to the present.
The focus of this research is on the development of an index of abundance for
western North Atlantic blue sharks that covers nearly half-century of exploitation.
The goal is to gain a perspective on the “missing baseline” (Baum and Myers 2004;
Ward and Myers 2005), when stock conditions would have been close to pristine, in
comparison to abundance in recent years after approximately five decades of
exploitation. This is done by linking pelagic longline catch records from recent
fishery observer programs (1978-2000) with historical catch and effort observations
from fishery-independent longline surveys that were undertaken decadely from the
late 1950s to the 1980s. The historical fishery-independent data were obtained from
an extensive data recovery program focusing on the archives of U.S. and Canadian
51
fisheries agencies. Generalized linear model techniques (Maunder and Punt 2004)
were used to develop a standardized catch rate index (1957-2000).
3.2. MATERIALS AND METHODS
3.2.1. Data Selection
Pelagic longline fishery-independent (survey) and –dependent (observer)
records were collected from several sources (Table 3.1). Data from historical longline
survey cruises were coded from original cruise reports, field fishing logs maintained
by scientific personnel, and grant reports. Set specific gear, deployment, retrieval, and
species composition data was coded as consistently as possible with the more detailed
and recent observer series. A detailed description of the different datasets is found in
Appendix B.
A blue shark data subset was chosen to focus the analysis on the geographical
areas with the longest time series, the largest proportion of the total number of blue
shark observations available, and a reduced number of sets that caught no blue sharks
(“zero sets”). The geographical strata (with numerical codes and acronyms) are the
U.S. National Marine Fisheries Service reporting areas for the pelagic longline
fishery (Beerkircher et al. 2003), specifically the Mid Atlantic Bight (5 - MAB),
Northeast Coastal (6 - NEC), and Northeast Distant (7 - NED) strata.
The spatial extent of the analysis is bounded on the south by 35N latitude
(Cape Hatteras, North Carolina U.S.), extends to northern Newfoundland, Canada (up
to 55 N), bounded on the east by 45 W longitude, the general range limit of U.S.
52
vessels. Sets south of 35 N latitude, including areas off Florida, in the Gulf of
Mexico and in the Caribbean, were excluded because they accounted for a small
proportion of the total blue shark catch recorded in the data, and there were a very
limited number of research cruise records south of 35 N available prior to 1978.
Shallow sets in depths less than 50 m north of 35 N were also excluded since
incursions of blue sharks into these inshore waters seem to be heterogeneous along
the U.S. east coast (Casey 1982).
The resulting selection of blue shark fishery-independent (survey) data
consisted of a total of 1,365 longline sets conducted from 1957 to 1994 (Table 3.2a).
These sets targeted tuna, swordfish and pelagic sharks, of which 1,057 (77.4%)
caught blue sharks (“positive sets”). The fishery-dependent (observer) data selection
consisted of tuna and swordfish sets by U.S. vessels (n=1,632; 1985-2000), and tuna
sets by Japanese vessels (n=4,711; 1978-1988) (Table 3.2b). The proportion of blue
shark positive sets was high for both observer series, 86.6% and 83.7%, respectively.
The selected and pooled observer and historical survey data, hereafter referred
to as the “reconciled data”, contain longline observations from fishing operations
targeting different species: tuna, swordfish or pelagic sharks (Table 3.3). Except for
the years of 1973 and 1974, the reconciled data provides continuous annual coverage
from 1957 to 2000. Substantial temporal and spatial overlap exists between the
longline observation time series for different target species (Table 3.3, Figure 3.1).
53
3.3.2. Catch-Effort Standardization
The real changes in blue shark abundance represented in the reconciled data
are confounded with changes over time in fishing practices for different target species
(tuna, swordfish and sharks). “Catch-effort (or catch rate) standardization” consists of
the process used to remove biases and produce a reliable index of abundance (Hinton
and Maunder 2004; Maunder and Punt 2004). Standardization of blue shark catch and
effort data used Generalized Linear Models (GLMs) (McCullagh and Nelder 1989;
Dobson 1991). GLMs provide a single theoretical framework in which the
assumptions of error normality and constancy of variance of traditional linear
regression models are relaxed. The central assumption of a GLM is the existence of a
linear relationship between some function of the expected value of the response
variable and the explanatory variables:
g ( i )  i
(3.1)
in which,
g
is a monotone, differentiable link function between the systematic and
random components of the model;
i
is the expected value of the ith random response variable Yi (the
random component).
The right-hand side of Eq. (3.1) is the linear predictor i , a linear sum of the effects
of the explanatory variables (the systematic component):
54
i  xTi β
(3.2)
in which,
xi
is an  n1 vector of explanatory variables (covariates and dummy
variables for individual levels of factors);
β
an  n1 vector of parameters.
The explanatory variables and the different levels considered for the
catch-effort standardization analysis are shown in Table 3.4. In addition to the set of
main factors, two interactions terms (year:area and year:target) were also included in
the GLMs.
Two types of GLMs can be applied depending upon the choice made for the
response variable. The alternative approaches are models for catch rate or count data.
Both model types were investigated to derive an index of abundance for the blue
shark. While the nominal blue shark catch rates were computed to implement the first
class of models, counts (catch in numbers) were considered for the latter.
The number of hooks deployed and set duration (longline soak times) varied
by data source. Accordingly, fishing effort was quantified in terms of “hook hours”,
the product of total number of hooks deployed (h) and the soak time (s) for each set
(Simpfendorfer et al. 2002). The catch per unit effort (CPUE) was computed as
follows:
55
CPUE 
c
1, 000
hs
(3.3)
in which,
c
is the total number of blue sharks caught in the longline set;
h
the total number of hooks deployed;
s
is the soak time duration of the longline, computed as the time interval
between the start set and end haul times of the set.
The catch-effort standardization process frequently encounters set records
with no catch of the species being investigated (Maunder and Punt 2004). In the blue
shark reconciled dataset, only 11.4% of the total number of sets are “zero sets”.
However, the percentage of zero sets recorded over time for each individual data set,
and among data sets, was variable.
Two statistical approaches that explicitly account for the zero catches were
investigated. The “delta approach” (Lo et al. 1992; Vignaux 1994), the most
common, models the probability of a “positive” (non-zero) set and the mean catch
rate of the positive sets separately. The probability of observing a blue shark positive
set was modeled using a binomial GLM with the logit link (i ):
 i 

 1  i 
i  g ( i )  ln 
(3.4)
56
Two error distributions were used to model the catch rate of the positive sets.
First, a lognormal GLM was used in which the identity link i is:
 i  i
(3.5)
A gamma GLM was also applied with the log link defined by:
i  g (i )  ln(i )
(3.6)
The final index of abundance was produced by multiplying the year effect
generated from the binomial and lognormal (or gamma) models.
The second approach considered models for count (integer) data that explicitly
account for the zeros. A standard Poisson error model is the first logical choice
(Maunder and Punt 2004), since this is the first error assumption traditionally used to
model rare event count data. However, the catch of blue sharks is certainly not a rare
event, as indicated by the large proportion of positive sets observed in the various
target species data sets. In addition, Ortiz and Arocha (2004) presented a
comparative analysis of the performance of different models for standardizing catch
rates of rare non-target species in a longline fishery, and the Poisson model performed
poorly. The negative binomial offers a more flexible error-model for count data given
that it assumes a quadratic relationship between the mean and variance. This error
model is more flexible than assuming that the variance is equal to the mean as in the
Poisson model, which makes it more vulnerable to overdispersion problems (Punt et
al. 2000b). An extensive review of the theory of negative binomial regression models
57
can be found in Hilbe (2007). According to the canonical form derived directly from
the negative binomial PDF, the mean of y, conditioned on  and  is:
1/
 y  1/   1 1 
f  y;  ,    


 1/   1   1   
y
  

 ,
 1   
(3.7)
in which  is the negative binomial ancillary or overdispersion parameter. The
canonical link is given by:
i  g (i )   ln(1/ i  1)
(3.8)
The traditional negative binomial regression model with the log link (Eq 3.4)
was applied (NB-2; Hilber 2007). Effort data (in hook hours) were treated as an offset
explanatory variable in the negative binomial GLMs (Maunder and Punt 2004).
The precision of the relative abundance indices was quantified using a
bootstrap approach (Vignaux 1994). The bootstrap resampling process consisted of
randomly sampling with replacement from the longline datasets and repeating the
GLM analysis for each sample draw (n=1,000). The 2.5th and 97.5th percentiles of the
indices were used as their approximate 95% confidence intervals. All GLM and
bootstrapping were coded in the R Language for Statistical Computing (R
Development Core Team, 2006).
The Akaike Information Criterion (AIC) (Burnham and Anderson 2002) was
used to choose the most parsimonious GLM for each error model. AIC is computed
based on the log-likelihood of the model (l) and number of parameters (p):
AIC  2 ln l  2 p
(3.9)
58
The model which provided the lowest AIC was taken as the most
parsimonious for each of the error assumptions considered. A stepwise algorithm (the
stepAIC function; Venables and Ripley 1997) was used to implement the model
selection. AIC may perform poorly in the number of parameters (p) is low when
compared to the sample size (n). When the ratio n/K is small (<40), a second-order
variant of AIC, AICc, is recommended (Sugiura 1978). In case that n is large with
respect to K, then AIC should perform well. In this study, the ratio n/K is equal to 100
(K for the higher dimensional model is 64; n is equal to 6,413). Therefore, AIC
should perform well (Burnham and Anderson 2002).
Diagnostic plots were used to evaluate the performance of the different error
models selected, and to choose which of the error models (longnormal or gamma)
best fitted the CPUE data of the blue shark positive sets (McCullagh and Nelder
1989; Maunder and Punt 2004; Ortiz and Arocha 2004). For the lognormal error
models, quantile-quantile plots were inspected to check for departures from a
lognormal error distribution. Diagnostic plots for all error distributions included plots
of the standardized residuals against the fitted values to check for systematic
departures from model assumptions. A check of the variance function was also made
by plotting the absolute value of the standardized residuals against the fitted values.
59
3.3. RESULTS
3.3.1. Model Selection
Table 3.5 presents the model selection summary statistics for the GLM fits to
the blue shark reconciled data. The three types of data analyzed were the proportion
of positive (non-zero) sets, the CPUE (catch rates) of the positive sets and the catch
count data. The binomial GLM fit to the data on the proportion of positive sets
explained the smallest fraction (9.3%) of the total deviance observed. With respect to
the CPUE of the positive sets, the lognormal model explained 57% of the total
deviance, whereas the gamma model explained 71.7%. The negative binomial model
fit to the blue shark count data explained 54.7% of the total deviance observed. None
of the interaction terms considered (year:area and year:target) were retained in the
AIC based model selection. Model selection results for GLMs fit separately to each of
the individual target species datasets (Appendix B, Table B.2).
Diagnostic plots were used to ascertain which of the error models best fit the
blue shark observations. Figure 3.2 provides diagnostic plots for the blue shark
reconciled data. The quantile (q-q) plots of the ordered residuals were found to be
consistent with the expected (null) linear pattern of a lognormal distributed error
(Figure 3.2a). The diagnostic on systematic departures from a lognormal error
assumption shows a distribution of residuals consistent with the expected pattern (a
mean of zero and constant variance across a wide range of fitted values) (Figure
3.2b). In contrast, the gamma and negative binomial models show declining trends for
residual means and a non-constant range (variance) of residuals across the observed
60
range of fitted values (Figures 2c,d). These patterns indicate possible overdispersion
problems with the gamma and negative binomial models. The test on the performance
of the variance function for the lognormal model best resembled the expected null
pattern (Figure 3.2e) (an approximate stable trend in the absolute value of the
residuals across fitted values). The gamma and negative binomial diagnostics indicate
non-stable trends (Figures 2f,g).
Diagnostic plots for the three error models fit to the individual target species
datasets area also available (Appendix B, Figures B.1., B.2., B.3). In general, the
residual patterns described above for the reconciled data were also identifiable in the
model fits to the individual target datasets.
3.3.2. Catch-Effort Standardization
Inspection of diagnostic plots supported the choice of a lognormal error
assumption to describe the blue shark catch rate data. A delta-lognormal error model
that accounts for both the zero catch observations and the CPUE of the positive sets
data was chosen to derive an index of abundance for the blue shark. The
delta-lognormal index (Figure 3.3c) was obtained from the product of the yearly
effects from the binomial (Figure 3.3a) and the lognormal models (Figure 3.3b).
The estimated proportion of non-zero sets was generally high (>60%) (Figure
3.3a). However, inter-annual variability is observed over years associated with
different target fishing practices or research cruise series. In particular, the data in the
early 1950-60s show larger proportions of zero sets. This supports the need to account
61
for the zero sets when modeling the blue shark catch rate data, the main objective of
the delta-GLM approach. The nominal CPUE show a few years of unusually high
catch rates (late 1960s and 1970s) (Figure 3.3b). This high variability was greatly
reduced in the catch-effort standardization process. The precision of the
delta-lognormal yearly estimates is generally high in the time series (Table 3.6). More
specifically, the yearly coefficients of variation (CVs) ranged from 0.09 to 0.39 with a
mean CV of 0.18. Precision was noticeable higher during the last two decades as a
result of increased sample sizes from observer programs.
The results suggest two distinct stanzas of catch rates during the exploitation
history. The first stanza is the period of generally higher CPUE values prior to 1980
(mean relative CPUE of 1.2). A second stanza of apparently lower catch rate values
(mean relative CPUE of 0.82) is apparent during the 1980s and 1990s. A comparison
between these two average levels indicates an approximate 30% decline for blue
shark catch rates in the western North Atlantic since the mid 1950s. The higher
variability of the estimates during the first stanza could allow for an interpretation of
a slighter greater or lesser decline. However, the results provide little support for any
interpretation other than that there has not been a pronounced decline of CPUE levels
over the exploitation history. Noticeably, the catch rates for 1971 and 1977 are higher
and more variable when compared with adjacent years.
Delta-lognormal, delta-gamma and negative binomial models are also
available for all individual target species datasets before data reconciliation
(Appendix B, Figures B.4., B.5., B.6). In general, the standardized CPUE for the
62
individual target segments of data showed high variability as well as trends that are
not consistent with the life-history of shark species. These issues, however, were
greatly reduced after data were pooled to form a long time series of longline
observations in what was defined as the reconciled dataset.
The mean effects of selected operational, seasonal and geographical factors on
the blue shark catch rates are shown in Figure 3.4. The probability of catching a blue
shark was much higher for those fishing operations that targeted pelagic sharks when
compared with swordfish and tuna sets (Figure 3.4a). Tuna fishing strategies had the
lowest mean effect on the blue shark catches. Other dominant effects on blue shark
catches for selected operational factors included smaller set sizes (Figure 3.4b), high
proportions of light sticks (Figure 3.4c) and shallow logline sets (Figure 3.4d). The
spring quarter (April-June) had a noticeable higher effect on the blue shark catch rates
(Figure 3.4e). Geographically, longline gear deployed in the Northeast Atlantic
Distant area (NED, Figure 3.4f) and over the continental shelf shallow waters had
high effects (Figure 3.4g).
3.4. DISCUSSION
This study presents standardized catch rates for blue sharks caught in the
western North Atlantic during the past four decades (1957-2000, Figure 3.3). The
time series begins in the mid 1950s when Japan introduced pelagic longline gear into
the tropical Atlantic tuna fisheries (Wise and Le Guen 1969; Nakano 1993; Bonfil
1994). The historical survey information presented here provides perspective on the
63
“missing baseline” (Pauly 1995; Myers and Worm 2003) for blue sharks in the
western North Atlantic. This “baseline” would provide insight into the virgin
population levels. The results support a conclusion that blue shark CPUE in the
region has declined to no more than about 70% of the catch rate levels at a time
immediately after the pristine population levels (mid 1950s). This evaluation assumes
that fishery related exploitation of blue sharks was negligible prior to the expansion
of longline fisheries in the late 1950s and 1960s.
The interpretation of catch rates as an index of abundance must be considered
carefully. Catch rate analyses described in several recent reports provide the basis for
claims that many of the world’s populations of large predatory fishes, including
sharks, are severely depleted and are currently being exploited at unsustainable rates
(Baum et al. 2003; Myers and Worm 2003; Baum and Myers 2004; Myers and Worm
2005; Shepherd and Myers 2005; Ward and Myers 2005). These catch rate analyses
generated almost immediate contrary responses by several fishery scientists (Walters
2003; Burgess et al. 2005a; Burgess et al. 2005b; Hampton et al. 2005; Hilborn 2006;
Maunder et al. 2006; Polacheck 2006; Sibert et al. 2006). It is therefore critical that
blue shark standardized CPUE series are evaluated in terms of their reliability as a
potential index of abundance.
The use of catch rates as an index of abundance has traditionally relied on the
assumption that CPUE is proportional to abundance (Harley et al. 2001). It also
assumes that the catchability (the q coefficient of proportionality) does not change
over time, when in fact violations of this assumption can be caused by several factors
64
(see Maunder et al. 2006 for review). One of these factors is changes in target fishing
practices over time. Unfortunately, the pooling of catch rate data from different target
practices is commonly the only means to increase sample sizes, increase temporal and
spatial coverage, and improve the information available for data-limited species, such
as bycaught shark species (e.g., Baum et al. 2003; Campana et al. 2005; Ward and
Myers 2005).
The assumption of a time-invariant catchability was violated when longline
data targeting different species (tuna, swordfish, and pelagic sharks) were pooled to
build the historical time series of blue shark catch rates. The operational fishing
practices (time variables) varied among different targeting practices. Deployment of
tuna sets generally begins close to sunrise, within the first half of the day. Swordfish
sets are generally deployed after sunset and before midnight. Starting time for pelagic
shark effort was more variable. Gear dimensions (lengths of dropper and gangion
lines) vary, such that different configurations are designed to exploit specific fishing
depths, often relative to the depth of the thermocline or migrating forage species.
These varying fishing strategies generally exploit differences in diel vertical
distributions and feeding behavior between swordfish, tuna and sharks (Carey and
Robison 1981; Carey and Scharold 1990).
Generalized linear models (GLMs) were used to capture and remove the effect
of the factors which bias CPUE as an index of abundance, particularly changes in
catchability as a result of the different target practices (Maunder and Punt 2004;
Maunder et al. 2006). The critical question is whether the GLM could successfully
65
account for the variability in the fishing practices for different target species that
exists in the data available from both the research cruises and commercial trips. This
is a pre-requisite if the standardized CPUE are to be interpreted as a reliable index of
abundance for blue sharks in the western North Atlantic.
The diagnostic plots suggest that the delta-lognormal error model is a
reasonable choice to explain the blue shark catch rate data. This is consistent with the
results of Ortiz and Arocha (2004) in which the same error model assumption
performed best with longline bycatch data, compared to other error models.
The linkage of the historical longline survey data with the more recent
observer records seems to be the main factor contributing to the success of the blue
shark catch-effort standardization. The temporal and spatial overlap between the
different target species datasets provides enough signal and contrast about the main
effects of various factors (operational, seasonal and geographical) on the blue catch
rates (Table 3.3; Figure 3.1) to allow effective standardization.
Longline sets that targeted pelagic sharks had a much stronger effect on the
blue shark catch rates than tuna and swordfish sets (Figure 3.4a). Swordfish sets had
significantly higher blue shark catch rates than tuna sets. The standard configuration
for tuna gear consists of a longline catenary (curve) configured to reach deeper waters
at sunrise when tuna are more vulnerable. Swordfish sets are traditionally deployed at
shallower fishing depths at night after dusk to exploit the negative phototaxis
behavior of the species (Carey and Robison 1981). Blue sharks appear to be more
vulnerable to longline gear set at shallower fishing depths and during crepuscular
66
hours (sunrise and sunset) when they seem to actively feed (Carey and Scharold
1990). The GLM successfully accounted for the significant positive effect of light
sticks on the blue shark CPUE (Figure 3.4c). Walsh and Kleiber (2001) found the
same significant effect with a generalized additive model and tree analyses for the
Hawaii-based commercial longline fishery.
The spring quarter (April-June) has a significantly higher effect on blue shark
catch rates than other seasons (Figure 3.4e), coincident with the northern and inshore
migration of blue sharks along the U.S. shelf (Casey 1982). Geographically, the
northeast distant region (NED) showed a dominant effect on the blue shark catch
rates (Figure 3.4f). This region includes the Grand Banks and dynamic frontal zones
where high blue shark catch rates have been reported in other analysis (Baum et al.
2003; Campana et al. 2006).
Sets on shallow waters of the upper continental shelf had a strong effect on the
blue shark catch rates (50-99 m; Figure 3.4g), which seems inconsistent with the
oceanic habitat preferences of the blue shark (Nakano and Seki 2003). However,
these shallow fishing grounds are where research cruise effort targeted blue sharks for
tagging, particularly the fishery-independent surveys described by Simpfendorfer et
al. (2002). These surveys frequently deployed a smaller number of hooks compared to
commercial operations (J. J. Hoey, NOAA-NMFS, personal observations as scientific
crew member of the R/V Geronimo, 1978-1981). These target fishing practices are
also reflected by the dominant effect on blue shark catch rates for sets categorized in
the smallest set size (< 99 hooks; Figure 3.4b).
67
Baum et al. (2003) report a general pattern of collapse for pelagic shark
populations in the western North Atlantic from 1985 to 2000, and blue sharks in
particular, were estimated to have declined by 60% since the mid-1980s. The present
study suggests an approximate 30% decline in the catch rates since the mid 1950s.
The conclusions of Baum et al (2003) were challenged (Burgess et al. 2005a; Burgess
et al. 2005b) for several reasons, including their reliance on self-reported logbook
data which tend to be subject to under- or non-reporting bias. Logbook data were not
included in the present study to minimize such biases. All records in the reconciled
time series were collected by scientific personnel or trained observers.
A fishery-independent catch rate study of blue sharks in the western North
Atlantic reported that male blue sharks declined by 80% between the mid-1980s and
early 1990s (Simpfendorfer et al. 2002), while female catch rates seemed unchanged.
Such a large decline in male relative abundance is difficult to reconcile with the 30%
decline since the mid 1950s reported in the present study. As recommended by
Simpfendorfer et al. (2002), the index presented herein relies on a number of datasets
covering a broader spatial and temporal range for the blue shark and likely provides a
more representative index of abundance. The R/V Geronimo’s data analyzed by
Simpfendorfer et al. (2002) is one of several pelagic longline survey series
investigated in the present study.
The lower rate of CPUE decline reported for blue sharks in this study when
compared to the previous estimates is not surprising. Life-history demographic
studies show that the blue shark is one of the most productive shark species in the
68
world’s oceans (Smith et al. 1998; Cortés 2002a; Aires-da-Silva and Gallucci 2007).
Stable patterns of blue shark CPUE have been documented for longline fisheries in
the Atlantic (Nakano and Clarke 2005) and Pacific oceans (Nakano and Seki 2003).
Preliminary stock assessments of blue shark stocks in the North and South Atlantic by
ICCAT (Anonymous 2005) reported that the current biomass may not be far from
virgin levels. Similar results were also reported for blue sharks in the North Pacific
Ocean (Kleiber et al. 2001; Sibert et al. 2006).
The CPUE stanzas identified seem correlated with the history of pelagic
longline exploitation in the western North Atlantic. The first stanza is the period of
higher CPUE values prior to 1980 (Figure 3.3c). In this period, blue sharks were
taken as bycatch in the tuna and swordfish fisheries from Japan, U.S. and Canada in
the western North Atlantic. Blue shark markets did not exist and there were economic
incentives for the fleets to actively avoid areas with high blue shark catch rates. A
mercury closure period for the U.S. swordfish fishery was in effect from 1970 to 1978
(Hoey et al. 1989; Gibson 1998). The higher catch rates for the years of 1971 and
1977 may be due to a combined effect of increased abundance and the smaller sample
sizes during the closure period (Figure 3.3c). The second stanza of slightly lower
catch rates is apparent in the 1980s and the 1990s. The 1980s coincides with the rapid
expansion of the U.S. swordfish fishery and the coincident westward expansion of the
Spanish swordfish fleet to fishing grounds in the vicinity of the Grand Banks, where
fishing effort overlapped with the North American and Japanese fleets.
69
There is supporting evidence from tagging studies that blue sharks in the
North Atlantic represent a single stock (Kohler et al. 2002). Robust conclusions about
the North Atlantic stock as a whole cannot be drawn from this study since the index
of abundance is based upon catch and effort observations from the western North
Atlantic only. Similar data are needed from the eastern North Atlantic to cover the
entire distribution of the stock and recent fisheries which target blue sharks for food
(Mejuto et al. 2002; Aires-da-Silva et al. 2008a). A stock assessment model that
estimates the dynamics of the population integrating different sources of data,
including CPUE indices, is needed if a reliable evaluation of the stock status is
desired (Maunder et al. 2006; Sibert et al. 2006).
Finally, recovery of longline historical data should be continued on both sides
of the North Atlantic to improve catch rate indices for blue sharks and pelagic sharks
in general. The catch-effort standardization modeling could be improved if additional
gear (operational) variables are obtained. In particular, the GLM standardization
could be enhanced from more refined evaluations of gear changes over time (e.g.,
changes of monofilament to multifilament, different bait types) within target
strategies. Combining the longline catch and effort data with satellite derived
environmental data (e.g., sea surface temperature, frontal zone data), and climatology
indices (e.g., the North Atlantic oscillation index) could also refine the abundance
indices.
Table 3.1. Fishery-independent (research cruise) and –dependent (observer) pelagic longline data sources used to derive the
blue shark index of abundance.
Data sources
a)
Research cruise programs
BCF-WHOI (1957-1970)
BSFW-SHML (1961-1970)
NARR (1975-1996)
NARR-NE (1971-1973)
NARR-comm (1977-1990)
CN-DFO (1961-2000)
b)
Observer programs
SEFSC-PLOP (1992-2000)
NEFSC-PLOP (1990-1995)
Dom-opport. (1985-1988)
US Obs- JPN (1978-1988)
Description
Cooperative cruises by the Bureau of Commercial Fisheries and the Woods Hole Oceanographic Institution, U.S.
Bureau of Sport Fisheries and Wildlife, Sandy Hook Marine Laboratory, U.S.
National Marine Fisheries Service, Narragansett Laboratory, U.S.
Cooperative cruises between NMFS Narragansett Laboratory and other institutions in the northeast U.S.
Opportunistic deployments of NMFS Narragansett laboratory scientists aboard volunteer commercial vessels, U.S.
Department of Fisheries and Oceans, Canada
Southeast Fisheries Science Center, Pelagic Longline Observer Program
Northeast Fisheries Science Center, Pelagic Longline Observer Program
Opportunistic deployments aboard commercial vessels
U.S. observers deployed aboard Japanese vessels
70
Table 3.2. Summary statistics of (a) fishery-independent and (b) fishery-dependent data sources that could form the basis for
the estimation of blue shark abundance indices.
Data source
a)
Research cruise programs
BCF-WHOI
BSFW-SHML
NARR
NARR-NE
NARR-comm.
CN-DFO
b)
Observer programs
US domestic
SEFSC - PLOP
NEFSC - PLOP
Dom-opport.
US foreign
US Obs. - JPN
Target species
Years
# Sets
# Pos. sets
% Pos. sets
Tuna, swordfish
Coastal & pelagic sharks, swordfish, and tuna.
Pelagic sharks
Swordfish, pelagic sharks
Swordfish
Swordfish , Tuna
1957-1970
1966-1970
1975-1994
1971-1972
1977-1990
1963-1993
1957-1994
305
78
590
15
41
336
1,365
202
57
426
15
38
319
1,057
66.2
73.1
72.2
100.0
92.7
94.9
77.4
Swordfish, Tuna
Swordfish, Tuna
Swordfish
1992-2000
1990-1999
1985
1985-2000
629
984
19
1,632
537
862
15
1,414
85.4
87.6
78.9
86.6
1978-1988
1957-2000
4,711
7,708
3,942
6,413
83.7
83.2
Tuna
Totals
71
72
Table 3.3. Annual time series of pelagic longline sets for the different target species.
The acronyms for the longline sets are: (TUN) tuna, (SWF) swordfish, (PLSHK)
pelagic sharks, (TUN-JPN) Japanese tuna sets, (TUN-US) U.S. tuna sets, (SWO-US)
U.S. swordfish sets.
Research cruises
SWF
PLSHK
Year
TUN
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
35
25
27
21
13
4
23
37
14
2
1
Totals
%
213
2.8
468
6.1
684
8.9
# BSH pos
% BSH pos
126
2.0
431
6.7
500
7.8
9
4
22
20
42
44
46
37
20
26
2
25
55
38
10
29
7
8
21
14
Observer programs
TUN-JPN TUN-US
SWO-US
35
25
27
21
13
4
23
37
14
2
1
13
21
25
13
15
7
9
19
33
64
75
46
49
23
32
24
46
48
3
10
26
10
56
2
10
5
TUN
Reconciled data
SWF
PLSHK
9
4
22
20
42
44
46
37
20
26
2
25
55
38
10
216
255
316
758
843
279
338
496
421
500
289
19
216
255
316
758
843
279
338
496
421
500
289
29
7
8
19
13
21
25
13
15
7
9
19
33
64
75
46
49
23
32
24
46
48
3
10
26
10
56
2
10
5
Totals
35
25
27
21
17
26
43
79
58
61
59
45
48
15
32
57
0
0
47
19
43
280
330
391
814
866
311
370
561
469
503
299
26
41
104
181
348
304
260
23
137
76
102
155
7
14
31
128
121
96
14
65
24
39
25
3
34
148
196
178
164
9
72
52
63
130
7
14
31
128
121
96
14
65
24
39
25
24
34
148
210
178
164
9
72
52
63
130
4,711
61.1
564
7.3
1,068
13.9
5,488
71.2
1,536
19.9
684
8.9
7,708
100.0
3,942
61.5
442
6.9
972
15.2
4,510
70.3
1,403
21.9
500
7.8
6,413
83.2
Table 3.4. Categorical variables (factors) assumed in the blue shark catch-effort standardization. Acronyms are given for each
factor. The different factor levels considered in the modeling are identified.
Factors
Acronyms
Factor levels
Area
Bottom depth
Light stick percentage
Rig depth
Set size
Soak time
Start set time
Target species
Quarter
Year
area
btmD
ltPer
rigD
setSz
soakT
startSetT
tar
quarter
yr
5 -MAB, 6 - NEC, 7 - NED
50-99, 100-199, 200-499, 500-499, > 1000 meters
0 (no sticks) 1-24%, 25-49%, 50-74%, 75-100% of hooks
Shallow - <= 12 hooks bet. floats; Deep - > 12 hooks bt flts. (Myabe, 1992)
< 100; 100-299; 300-499, 500-999, >= 1000 hooks/set
0-8 h, 8-16 h, > 16 hours
Dawn (0-8 h); Mid-day (8-16 h), dusk (17-24 h)
Tuna, Swordfish, Pelagic sharks
Jan-Mar, Apr-Jun, Jul-sep, Oct-Dec.
1957-2000
73
74
Table 3.5. Summary statistics of the selected four error models that were fit to the
blue shark reconciled data. The explanatory factors, and the proportion of the total
deviance explained are presented for each model. The models were a binomial GLM
for the proportion of non-zero sets, a lognormal and gamma model for the catch rates
of the non-zero sets, and a negative binomial model for the blue shark counts. The
full text for the acronyms of the explanatory factors are provided in Table 3.4.
Selected models
% of dev.
Proportion of positve sets - binomial
yr + tar + quarter + area + btmD + setSZ + ltPer + rigD
9.26
CPUE of positive sets
- Lognormal:
yr + tar + quarter + area + btmD + setSZ + ltPer + rigD
57.00
- Gamma:
yr + tar + quarter + area + btmD + setSZ + ltPer + rigD
71.66
Negative binomial - fish count data
yr + tar + quarter + area + btmD + setSZ + ltPer + rigD
54.72
75
Table 3.6. Delta-lognormal standardized CPUE for the blue shark. The coefficient of
variation (CV) is given for each yearly estimate.
Year
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
Std. CPUE
0.98
0.48
1.11
1.18
1.13
1.50
0.70
0.87
1.55
1.27
1.43
1.31
1.96
0.97
1.08
1.93
0.88
0.75
1.82
1.06
0.86
0.83
1.05
0.78
1.01
0.68
0.74
0.48
0.50
0.44
0.80
0.94
1.22
0.63
0.95
0.98
0.73
0.47
1.25
1.16
0.76
0.78
CV
0.17
0.16
0.25
0.38
0.35
0.27
0.25
0.17
0.17
0.23
0.21
0.21
0.22
0.32
0.23
0.21
0.19
0.29
0.20
0.11
0.11
0.09
0.09
0.09
0.10
0.10
0.10
0.09
0.10
0.12
0.39
0.17
0.11
0.10
0.09
0.10
0.10
0.30
0.13
0.15
0.13
0.12
76
Tuna - research cruises
80°
70°
60°
50°
Tuna - U.S., observer
80°
40°
70°
60°
50°
40°
50°
50°
50°
50°
40°
40°
40°
40°
30°
30°
30°
30°
80°
70°
60°
50°
80°
40°
Swordfish - research cruises
80°
70°
60°
50°
70°
60°
50°
40°
Swordfish - U.S. observer
40°
80°
70°
60°
50°
40°
50°
50°
50°
50°
40°
40°
40°
40°
30°
30°
30°
30°
80°
70°
60°
50°
40°
80°
70°
60°
50°
40°
Pelagic sharks - research cruises Tuna - Japan, observer
80°
70°
60°
50°
80°
40°
70°
60°
50°
40°
50°
50°
50°
50°
40°
40°
40°
40°
30°
30°
30°
30°
80°
70°
60°
50°
40°
80°
70°
60°
50°
40°
Figure 3.1. Geographical locations of pelagic longline sets for each target species
fishing operation.
77
Figure 3.2. Diagnostic check plots for the different error models fit to the blue shark
longline reconciled data set: lognormal and gamma models fit to the CPUE data of
the non-zero sets, and negative binomial model fit to the count data. Left column: qqplots; middle column: diagnostics on systematic departures from model assumptions;
right column: diagnostic on the performance of the variance function.
78
% pos. sets
1.0
0.5
a) Binomial
0.0
Year
CPUE of non-zero sets
4
b) Lognormal
2
0
Year
Delta-lognormal
4
c) Delta-lognormal
2
0
1950
1960
1970
1980
1990
2000
Year
Figure 3.3. Delta-lognormal model estimates. a) proportion of positive sets from the
binomial model; b) nominal (◦) and standardized CPUE (•) from the lognormal model
on the CPUE of the positive sets; c) delta-lognormal estimates obtained from the
product of a) and b). The vertical bars are the 95% confidence intervals of the
estimates. Loess line smoothers are shown to help visualizing the average trends.
79
Figure 3.4. Mean effects of selected explanatory factors on blue shark catch rates. a)
target species, b) set size, c) percentage of light sticks, d) rig length which defines the
gear depth, e) geographical area, f) quarter, g) sea bottom depth. The effects are
derived from the lognormal model on the catch rates of the positive sets. The vertical
bars are the 95% confidence intervals of the estimates.
80
CHAPTER 4
A SPATIALLY-STRUCTURED TAGGING MODEL TO
ESTIMATE MOVEMENT AND FISHING MORTALITY
RATES FOR THE BLUE SHARK IN THE NORTH
ATLANTIC OCEAN
Equation Chapter 4 Section 1
4.1. INTRODUCTION
The blue shark, Prionace glauca (Carcharhinidae), is a highly migratory apex
predator which seems to be the most wide-ranging and abundant of all pelagic sharks
in the global oceans (Draganik and Pelczarski 1984; Nakano and Seki 2003). Blue
sharks are subject to fishing mortality caused by a great variety of fishing gear in the
world’s oceans, but they are mostly taken as bycatch on pelagic longlines targeting
tuna and billfish (Bonfil 1994). The global shark fin trade is a major threat to shark
populations worldwide, particularly blue sharks, the dominant species in the shark fin
trade (Clarke et al. 2006a). As other stocks of large pelagic fishes have declined, blue
sharks are no longer discarded and finned exclusively, but are gradually becoming a
source of meat in some markets (Castro et al. 1999; Mejuto et al. 2002; Aires-daSilva et al. 2008a).
Unfortunately, the evaluation of the status of shark stocks worldwide is
handicapped by limitations on fishery-dependent data and questions about the fate of
discarded catches, among other uncertainties (Bonfil 1994). Despite these serious
limitations, there has been progress in the field of shark population modeling (Cortés
2004). For the blue shark, in particular, stock assessments have been produced by
some of the world’s Regional Fisheries Management Organizations (RMFOs). North
81
Pacific blue sharks were found to be exploited at sustainable levels (Kleiber et al.
2001; Sibert et al. 2006). Similar results were found for the South and North Atlantic
blue shark stocks (Anonymous 2005). The strong reliance of these assessments on
limited fishery-dependent data, however, calls for alterative (or supplemental)
fishery-independent approaches (Aires-da-Silva and Gallucci 2007).
Recent progress in the field of fishery-independent methods was the statistical
modeling of shark fin trade records to produce estimates of catch and an evaluation of
the blue sharks stocks globally (Clarke et al. 2006b). Unlike the optimistic results
from the traditional stock assessments based on limited fishery data, the analysis of
fin trade records indicated that the global exploitation levels of blue sharks may
exceed maximum sustainable yields (Clarke et al. 2006b). Narrowing down this
global evaluation to the ocean-basin level has not been made yet.
Alternative (or supplemental) approaches to traditional stock assessments
applied to sharks are life-history based demographic methods (Cortés 1998; Cortés
2002a; Gallucci et al. 2006). Demographic and risk analyses applied to the
management and conservation of the North Atlantic blue shark are available (Airesda-Silva and Gallucci 2007). The analyses provide a means to evaluate the risk of
population declines for a wide range of exploitation scenarios. However, the
acquisition of a direct estimate of the current blue shark exploitation rate is still
needed to evaluate the status of the blue shark stock using a demographic analysis
framework.
82
Fish tagging data has long been used to estimate population parameters
critical for management, such as population size, movement, growth and mortality
rates (Beverton and Holt 1957; Quinn and Deriso 1999; Pine et al. 2003). The North
Atlantic Ocean is particularly data-rich with respect to the availability of tagging data
for the blue shark. These data have been accumulated for over almost a half century
from tagging experiments conducted on both the western and eastern sides of the
North Atlantic Ocean (Kohler et al. 1998; Kohler et al., 2002 for the western side;
Stevens 1975; Stevens, 1990; Fitzmaurice et al. 2005; Mejuto et al. 2005, for the
eastern side). The tagging data were critical to improve the understanding of the
biology and ecology of North Atlantic blue sharks (Kohler et al. 2002; Skomal and
Natanson 2003). In particular, tag-recovery records provided strong evidence in
support of the highly migratory behavior of blue sharks in the North Atlantic and the
trans-boundary nature of their fishery exploitation. This contributed to the hypothesis
of a single North Atlantic blue shark stock for stock assessment and management
purposes (Casey 1985; Anonymous 2005). Despite this important scientific
information, quantitative modeling of tagging data for blue sharks has only been
initiated recently (Aires-da-Silva et al. 2005; Campana et al. 2006).
This study presents a tagging analysis using the blue shark tagging data of the
National Marine Fisheries Cooperative Shark Tagging Program (NMFS-CSTP) to
obtain estimates of movement and fishing mortality rates for the blue shark in the
North Atlantic Ocean. Among all existing tagging programs, the CSPT presents the
largest of the data sets and the broadest in terms of spatial coverage for blue sharks. It
83
is also the largest tagging data set available for any pelagic shark species worldwide.
However, the program has relied mainly on opportunistic tagging by sport fisherman,
which makes it difficult to implement any elements of experimental design. Future
improvements could be made to obtain better quantitative population parameters, and
these are discussed.
4.2. MATERIALS AND METHODS
4.2.1. Tagging Data
Blue shark tagging in the North Atlantic Ocean has been carried out by the
U.S. National Marine Fisheries Service Cooperative Shark Tagging Program
(CSTP-NMFS) since the early 1960s. This tagging program is part of a continuous
research project directed at the study of the biology of large Atlantic sharks (Kohler et
al. 1998). An overview of the tagging methods and results obtained by the CSTP for
pelagic sharks is available (Kohler et al. 1998). The tag recovery results for the blue
shark were also described in detail (Casey 1985; Kohler et al. 2002; Kohler and
Turner 2008). A subset of the blue shark tagging data covering the period from 1965
to 2004 was selected from the CSTP database for the tagging modeling. The data
selection included only “M” dart tags (Kohler et al. 1998), which represent the great
majority of the tags released on blue sharks. The blue shark data subset selected
consisted of 95,301 tag release records and 5,859 tag recoveries over a large spatial
extent throughout the North Atlantic (Figure 4.1).
84
4.2.2. Geographical Stratification and Definition of Fisheries
A geographical stratification was made to account for the spatial distribution
of the tag releases and recoveries (Kohler et al. 2002). Specifically, the North Atlantic
Ocean, defined as north of the Equator, was divided into four geographical regions
(Figure 4.1): Northwest (NW, west of 40oW and north of 25oN); Southwest (SW,
west of 40oW and south of 25oN); Northeast (NE; east of 40oW and north of 25oN);
and, Southeast (SE, east of 40oW and south of 25oN). The annual time series of blue
shark tag releases and recoveries for each region are shown in Figure 4.2 (also see
Appendix C, Table C.1 for sample sizes).
In the North Atlantic, blue sharks are caught by commercial and sport
fisheries (Heessen 2003). Bycatch in longline fisheries targeting tuna and swordfish
from several nations is the predominant source of fishing mortality (Bonfil 1994;
Heessen 2003). Fishing effort data are used to model the dynamics of the tag
recoveries under exploitation by the different fisheries (Hilborn 1990). Total
estimated longline fishing effort for the major longline fisheries (in total number of
hooks by quarter of the year and 5x5º square) was made available by ICCAT
(Anonymous 2007). Unfortunately, there is no fishing effort data available for the
other commercial gears (e.g., purse seines) catching blue sharks. Since these
contribute only to a minor source of the blue shark bycatch, this tagging analysis
deals only with the major longline commercial fisheries, and the sport fisheries.
The different fisheries catching blue sharks in the North Atlantic need to be
defined. This definition was based upon the distribution of the tag recoveries by fisher
85
group in each of the geographical regions (Table 4.1), and the ICCAT fishing effort
data. A fishery was defined as the exploitation by a fisher group (e.g., flag) in a
particular region. Fisher groups with low numbers of tag recoveries were pooled to
increase sample sizes (e.g. “Commercial-others”, and “Sport-others”). Fisheries with
reported fishing effort data but no tag recoveries were also defined in the model (e.g,
“Commercial-Venezuela” in the SE region). Table 4.2 presents a list of the 23
fisheries considered in this study. Tag-recovery maps for each fishery are show in
Figures C.2 to C.11. Fishing effort data was not available for some of the longline
fisheries. In such cases, constant fishing effort (set to 1) was assumed (Hilborn 1990).
The annual time series of fishing effort and tag recoveries for the 23 fisheries are
shown in Figure C.12.
4.2.3. Model Development
The tagging modeling framework described below includes two sub-model
components, a population dynamics model of tagged fish and an observation model,
and a likelihood function is also needed to fit the model to the tag recovery data.
These are described below.
Population dynamics model of tagged blue sharks - The general method of
Hilborn (1990) for analysis of fish movement from tag recovery data was modified to
accommodate the blue shark tagging experiment and the multiple fisheries in the
North Atlantic. The first component of the modeling framework is a population
dynamics model which describes the survival and movement of each tagged group of
86
fish. In this study, a tag release group was defined as a group of blue sharks tagged in
a geographical stratum during a particular year (131 blue shark tag release groups
were modeled; Appendix C, Table C.1). The population dynamics model of Hilborn
(1990) was rewritten to account for instantaneous natural mortality and tag shedding
rates, similar to Xiao (1996):
 ( M   F j ,t ) 
n 
f
Nˆ i ,a ,t 1    Nˆ i , j ,t  Ti , j ,t e
 p j ,a
j 1 




f
(4.1)
in which,
Nˆ i ,a ,t
= expected number of tagged fish of tag group i present in area a at time t;
Ti , j ,t
= number of tags released at the start of the year (“recruitment” of tags
occurring once at the time of release) from tag group i in area j at time t;
Fj f,t
= instantaneous fishing mortality rate by fishery f in area j at time t;
M
= constant instantaneous natural mortality rate;

= constant instantaneous long-term tag shedding rate (Type 2 in traditional
terminology, Beverton and Holt, 1957);
p j ,a
= probability of movement from area j to a;
The instantaneous natural mortality rate (M) was fixed at 0.22 yr-1, an estimate
which was derived from a diverse set of empirical methods applied to the North
Atlantic blue shark (Chapter 2; Aires-da-Silva and Gallucci 2007) . Direct estimates
87
of the shedding rate (  ) of the M-darts released by the CSTP have not been
estimated. As described below, sensitivity analyses were conducted to assess the
impact of different values for this parameter.
Observation model - The second component of the modeling framework is an
observation model which predicts the expected numbers of tag recoveries, which are
then compared to the actual numbers of tags that are captured and reported by each
fishery. The observation model is a Baranov catch equation (Beverton and Holt
1957):

Ft f

f
f ˆ
ˆ
Ri ,t   N i ,a ,t 
f
 M     Ft
f




 M   Ft f  



f


 1  e






(4.2)
in which
Rˆi ,ft = expected number of tag recoveries from tag group i at time t by fishery f; and
f
= constant annual tag reporting rate of fishery f.
The available blue shark size data is not adequate to model the selectivity
patterns of the different fisheries. A simplistic assumption that the blue shark size
ranges caught by the different fisheries are fully selected to the fishing gear was
made. The instantaneous fishing mortality rate of each fishery f at time t, Ft f , was
parameterized as follows:
88
Ft f  q f Et f e t
f
(4.3)
in which,
qf
= time-invariant catchability coefficient of fishery f ,
Et f
= fishing effort (in numbers of hooks) of fishery f at time t,
 tf
= annual deviates in the effort of fishery f at time t, assumed to be normally
distributed,  t
N (0,  2t ) .
The fishing effort data should not be considered to be known as reported,
since its credibility is highly variable among fisheries. Therefore, Equation 4.3
assumes that there is process error in the relationship between the instantaneous
fishing mortality (F) and fishing effort (E). The annual fishery-specific effort deviates
(  t ) represent transient deviations in the effort-fishing mortality relationship
(Fournier et al. 1998; Kleiber et al. 2006).
Model fitting - The third component of the model is an objective function
which is minimized to estimate the parameters of the population dynamics and
observation models. The highly migratory nature of the blue shark and the large-scale
of the fishery present various sources of variability in the tag recoveries. Some
examples are the heterogeneous distributions of the various fleets in the North
Atlantic, incomplete mixing of tags in this vast region, time-variant tag-loss and
89
reporting rates over time and space. The negative binomial likelihood is a logical
choice to model tag-recovery data since it allows for substantial variation among
observations (Cormack and Skalski 1992; Hampton and Fournier 2001; Stewart
2007). Assuming each tag release group to be independent, the full likelihood of the
observed tag recoveries ( Ltags ) in log-space is:
tags
L
=

i
a
t
  f
Rˆi ,fa ,t  
f
log    Ri ,a ,t 
   log   Ri ,a ,t  1

 
(


1)

 


  Rˆ f    R f
 log    i ,a ,t    i ,a ,t log( )  Ri ,fa ,t log(  1)
  (  1)   (  1)

 
(4.4)
in which,
 = the standard gamma function,
 = is the parameter defining the degree of overdispersion, given as the ratio of the
variance to the mean:  =
var( Ri.a ,t )
E( Ri ,a ,t )
.
An important component of the total likelihood function is modeling of the
annual fishing effort deviates as model parameters (Fournier et al. 1998). Ideally,
these random effects would be integrated out analytically and estimates for the
standard deviations of their distributions obtained. Unfortunately, this is very
computationally intensive for non-linear models. An alternative approach is to use a
penalized likelihood framework (i.e., to estimate the random effects as parameters in
90
the model). This is done in a similar way as modeling of recruitment deviations in
catch-at-age models (Maunder and Deriso 2003). The annual effort deviations were
assumed to be normally distributed so that  t
N (0,  2t ) . A penalty was added to the
negative log-likelihood function to constrain the annual effort deviations which are
unknown parameters to be estimated. Ignoring constants, the final objective function
( L ) for minimization becomes:
L  Ltags + 
f
  f2
t  2t 2f
 t



(4.5)
The penalized likelihood framework provides flexibility to account for the
credibility of the fishing effort data by specifying different “weighting factors” of the
tagging data from different fisheries. For this purpose, the fishery-specific standard
deviations of the annual effort deviations (  2 f ) are fixed to different values which
t
guide the specification of the different weights. A value of 0.4 or less indicates that
the effort data is well determined, whereas a value of 0.5 or higher would represent a
higher uncertainty and low reliability (Kleiber et al. 2006). Accordingly, the standard
deviations of the effort residuals were fixed to one of two values:  =0.4 if effort data
was available; or  =1 when effort data was not available and a proxy was taken
(constant effort set at 1.0).
Table 4.3 lists the “free” parameters which are estimated in the model. A total
of 22 out of the 23 fishery-specific catchability parameters (q) were estimated since
one fishery did not report any tags (Venezuela - SE). The catchability of this fishery
91
was shared with the estimated q of a similar fishery (Venezuela - SW). The tagging
model requires that a hypothesis be made about the movement patterns of the blue
shark in the North Atlantic. Based on a preliminary analysis of the tag recovery
distributions over time, it was assumed that blue sharks can stay in the same region or
move to the neighbor regions within the annual time step of the model (Figure 4.3).
This hypothesis corresponds to a total of 12 movement parameters, of which 8 only
are estimated in the model (the other 4 can be derived analytically as one minus the
sum of other parameters for a given region). The remainder parameters in the model
are the 599 annual fishing effort deviations (  ) and the data overdispersion parameter
(  ) of the negative binomial distribution. A quasi-Newton minimization procedure
was implemented (AD Model Builder©, Otter Research Ltd., 2000) to estimate the
model parameters.
4.2.4. Model Hypotheses – Base Case Model and Sensitivity Analyses
Assumptions also need to be made about the processes of tag shedding (  )
and tag reporting (  ). Unfortunately, direct estimates of these parameters are not
available for the blue shark. “Best guess” estimates were defined based on expert
opinion of NMFS-CSTP and ICCAT fishery scientists. Twelve scenarios which
represent different hypotheses about the rates of tag shedding (  ) and reporting (  )
were investigated (Table 4.4). The base case model (scenario 6) assumed the “best
guesses” for  and  , as described below. The remaining scenarios examine
sensitivity analyses to alternative assumptions about  and  .
92
There are biological and operational reasons supporting the hypothesis that the
blue shark tag shedding rates are low in the CSTP. Specifically, the blue shark skin is
very thick along the dorsal musculature below the dorsal fin, particularly for females
which bear severe tooth cuts caused by the aggressive mating behavior of males
(Stevens 1974; Pratt 1979). The thick dermis apparently provides a strong anchor
basis for the M-type tags after it is penetrated by the stainless still dart (N. E. Kohler,
NMS-CSTP, personal field work observations). The dominant tagger group of
volunteer sport fishermen in the CSTP is very experienced with blue shark tagging.
For these reasons, a 10% (  =0.11 yr-1) value was taken as the “best guess” for the
tag shedding rate. No shedding, 0% (  =0) and 20% (  =0.22 yr-1) were taken as
alternative hypotheses for sensitivity analysis.
Tag-induced mortality also needs to be considered when modeling tagging
data (Beverton and Holt 1957). The monitoring of physiological data for blue sharks
during angling time indicated that their resilience against tagging-induced stress is the
highest among several large pelagic species (Skomal 2007). Also, observations of the
rapid healing of tag-induced scars and of the good accommodation of M-tags in the
blue shark musculature are common (N.E. Kohler, NMFS-CSTP, personal
observations). For these reasons, tag-induced mortality was assumed to be negligible
for blue sharks.
Defining the scenarios for the tag reporting rates is more difficult due to the
large number of fisheries in the model (23 fisheries in the tagging model). For
tractability, different hypotheses about  were made for the major fisheries only as
93
these are likely to impact the final results to the greatest extent. A “best guess” of
100% tag reporting (  =1) by the NW sport fishery (“NW-sport”) is a reasonable
assumption considering the high cooperation between the U.S. sport fishery and the
CSTP. For sensitivity analysis, an alternative hypothesis of  =0.5 was defined for the
this sport fishery.
The commercial longline fisheries by Japan (“JPN-comm”) and Venezuela
(“VEN-comm”) are major fisheries taking blue sharks. However, the low and even
non-reporting of tags from regions of high fishing effort by these fisheries indicates
that their reporting rates are likely to be low. Accordingly,  =0.2 was defined as the
“best guess” for “JPN-comm” and “VEN-comm”, and  =0.5 as the alternative
hypothesis for sensitivity analysis. Active tag reporting by Spanish fishery scientists
indicates higher reporting of tag recoveries by the Spanish fisheries when compared
with other commercial fisheries. A “best guess” of  =0.5 was defined for “SPNcomm”. Alternatively,  =0.2 was considered for sensitivity analysis. Non-reporting
of tags by the remaining fisheries was assumed to be negligible (  =1).
4.3. RESULTS
4.3.1. Model Fit to Observed Tag Recoveries
The base case model fit to the blue shark tag-recovery is shown in Figure 4.4.
In general, the annual time series of tag recoveries estimated by the base case model
were consistent with the observed tag recoveries in different fisheries. An estimated
94
value of 1.6 for the overdispersion parameter (  , ratio of the variance to the mean),
indicates that the tag recovery data are slightly overdispersed.
4.3.2. Movement Parameters
There are two types of movement parameters in the tagging model: the
probability of staying in a region, and the probability of moving to a neighboring
region. The movement parameter estimates and their CVs for the base case model are
presented in Table 4.5a. In general, the annual proportions of blue sharks staying
(residency rate) in a particular region exceeded the emigration rates by a factor of
two, at least. The exception to this pattern was the southeast region (SE) for which the
total annual emigration rate (57%) exceeded the residency rate (43%). However, the
precision of the movement parameters in the SE was the lowest (CV range of 0.470.80), which could be due to the lower number of releases in that region (Figure 4.2).
4.3.3. Fishing Mortality Rates
The fishing mortality (F) estimates derived from the base case model are
shown for each region in Figure 4.5. Since the most recent years of the ICCAT’s
fishing effort data are subject to updates, the F values for 2000 are chosen for
comparative purposes (Table 4.5b). The critical reference point assumed for
conservation, Fc=0.2 yr-1, is an estimate obtained from demographic and risk analysis
for the North Atlantic blue shark stock (Chapter 2; Aires-da-Silva and Gallucci 2007).
95
The catchability estimates (q) derived for the different fisheries from the base case
model are given in Table 4.6.
Western North Atlantic (NW and SW regions)
Total fishing mortality in the NW (Figure 4.5a) peaked in 2000 at a level of
NW total
0.10 ( F2000
, CV=0.09), less than the 0.2 critical fishing mortality estimate (Fc). At
the fishery-specific level in the NW, the fishing mortality levels for the sport fishery
NWsport
reached 0.028 in 2000 ( F2000
, CV=0.09). The estimates of F for the commercial
UScomm NW
CNcomm NW
U.S. and Canadian fisheries were lower ( F2000
=0.003 yr-1; F2000
=0.001),
but their precision levels were the lowest among all major fisheries in the NW
(CV=0.29 and 0.55, respectively). The commercial longline fleets from Spain and
SPNcomm  NW
Japan had the highest F estimates in the NW, F2000
=0.049 yr-1 (CV=0.13) and
JPNcomm  NW
F2000
=0.018 yr-1 (CV=0.25), respectively. The F levels associated with the
NW Spanish longline fishery increased since the mid 1990s.
The total levels of blue shark fishing mortality in the SW region were the
NW total
lowest among the four regions ( F2000
=0.032 yr-1, CV=0.24; Figure 4.5b). The
Venezuelan longline fishery contributed the most to fishing mortality in the SW
VEN  SW
( F2000
=0.013 yr-1, CV=0.41), followed by the Japanese and Spanish longline
JPN  SW
SPN  SW
fisheries: F2000
=0.007 yr-1 (CV=0.47) and F2000
=0.004 yr-1 (CV=0.35),
respectively.
96
Eastern North Atlantic (NE and SW regions)
The blue shark fishing mortality rate estimates are heterogeneous across the
North Atlantic (Figure 4.5). The exploitation rates in the eastern regions (NE and SE)
exceeded by a factor of two, at least, the F estimates for the western regions (NW and
SW). Total fishing mortality has gradually increased in the NE and SE regions to
levels above 0.10 during the 1980s and 1990s. The total F estimates in the 1990s
NE total
exceeded Fc=0.20 ( F2000
=0.204 yr-1, CV=0.25).
The Spanish longline fishery accounted for 55% of the blue shark total fishing
SPN  NE
mortality in the NE in 2000 ( F2000
=0.113 yr-1, CV=0.29), followed by 18% and
JPN  NE
8%, by the longline fisheries of Japan ( F2000
PT  NE
( F2000
=0.037 yr-1, CV=0.57) and Portugal
=0.016 yr-1, CV=0.65). The remainder of the total F was taken by a pooled
group of “other” minor commercial fisheries (14%) and a pooled sport fishery group
(5%).
In the SE region, the estimated total F was close to the Fc of 0.2 yr-1 in the late
SE total
1990s ( F2000
=0.182 yr-1), but the true parameter value could be below or above the
critical value (CV=0.42). The Japanese longline fishery accounted for the majority
(approximately 40-60%) of the total blue shark fishing mortality over most of the
study period,
JPN  SE
F2000
=0.111 yr-1 (CV=0.53). The Spanish longline fishery accounted
for 29% of the total F in the SE,
SPN  SE
F2000
= 0.052. However, these interpretations are
97
subject to higher uncertainty considering the higher CVs (of about 0.5) of these two
estimates.
Sensitivity analyses
Twelve different model scenarios were fit to the blue shark tag recovery data
incorporating different hypothesis about the tag shedding rate (  ) and the tag
reporting rates (  ) of the major fleets (see Table 4.4 for model definitions). The
movement parameters showed high sensitivity to the different assumptions about 
and  (see parameter estimates and CVs of sensitivities in Appendix C, Table C.2).
Most noticeably, the parameter estimates showed higher sensitivity to the highest rate
assumed for tag shedding (  =0.2). The blue shark fishing mortality regional
estimates were generally robust to the different scenarios investigated for  and 
(Figure 4.6). A minor exception occurred in the SE region in which noticeably higher
estimates for the total F were obtained at the highest levels of tag shedding rates
(20%). The F estimates and their CVs for the sensitivity analyses are given in
Appendix C, Table C.3.
4.4. DISCUSSION
Knowledge of the stock status of the blue shark in the North Atlantic remains
uncertain due to the major limitations of the fishery statistics. The critical question
that needs to be answered is “what are the current levels of blue shark fishing
mortality (F) in the North Atlantic, and how do they relate to the reference points for
98
conservation?” At this point, the estimate of Fc=0.20 yr-1 for the overall North
Atlantic blue shark stock obtained from demographic and risk analyses is useful for
comparative purposes (Chapter 2; Aires-da-Silva and Gallucci, 2007).
The tagging analysis indicated that the blue shark fishing mortality rates are
heterogeneous across the North Atlantic (Figure 4.5). This result could be due to
changes in the blue shark catchability across regions, and the different spatial
distribution patterns of the various bycatch and target fisheries (Anonymous 2007).
During approximately four decades of exploitation (1965-2000) in the western North
Atlantic (NW and SW regions), the F estimates were consistently lower than 0.1yr-1,
well below the Fc =0.20 yr-1 reference point for conservation.
Low blue shark fishing mortality estimates (F  0.1 yr-1) for the NW region are
consistent with the history of pelagic longline exploitation in this region. Specifically,
blue sharks were taken as bycatch in the tuna and swordfish fisheries from the U.S.,
Canada, and Japan, and there were economic incentives for the fisheries to actively
avoid areas with high blue shark catch rates. The westward expansion of the Spanish
swordfish fleet to fishing grounds exploited by the North American and the Japanese
fisheries occurred in the 1980s. In addition, the subsequent development in the 1990s
of some European markets for blue shark meat stimulated new blue shark targeting
practices by the Spanish fishery (Mejuto et al. 2002). These reasons could explain the
recent increase in blue shark fishing mortality levels in the NW region (Figure 4.5).
Nevertheless, the total blue shark fishing mortality levels in the NW have yet not
exceeded the Fc =0.20 yr-1 reference point. This interpretation is supported by a
99
historical index of abundance which revealed a small decline ( 30%) of blue shark
CPUE over approximately four decades of exploitation (1957-2000) in the NW
region (Chapter 3; Aires-da-Silva et al., 2008b).
The recent fishery-specific estimates of F obtained for the commercial
longline fisheries by the U.S. and Canada in the NW (~0.003 and ~0.001,
respectively) were much lower than those obtained for the sport fishery (~0.09). One
explanation is that while blue sharks are caught as bycatch in the two commercial
fisheries, they are actually the main target of the U.S. sport fishery. The estimated
commercial fishing mortalities could also be negatively biased if the tag reporting
rates assumed are too optimistic (  =1). In addition, the F values for the sport fishery
are overestimates since a large proportion of the tagged blue sharks are caught by
sport fisherman and returned alive to the waters without the tag. On the other hand,
the F estimates for the U.S and Canadian commercial fisheries were estimated to be
about 10% and 5% only of the F estimates for the Japanese and Spanish fisheries,
respectively, which appears low. A non-mixing problem with the tags in the NW
could also explain the low F estimates for the commercial fisheries. This issue is
discussed further below.
In the SW region, blue sharks were mainly caught as bycatch in the
Venezuelan and Japanese pelagic tuna fisheries. The blue shark fishing mortality
estimates for the SW were the lowest among all regions (Figure 4.5). This could be
explained by the blue shark submerge into deeper, cooler waters in tropical and
equatorial habitats (Casey 1982; Kohler et al. 2002). Therefore, the blue shark
100
catchability could be lower to the Venezuelan fishery mainly targeting yellowfin tuna,
Thunnus albacares (Ortiz and Arocha 2004). On the other hand, the same behavioral
pattern could also result in a higher catchability of the blue shark to the Japanese
“deep longline gear” targeting bigeye tuna (Thunnus obsesus) in tropical waters.
However, this is not confirmed by the q parameter estimates obtained for the Japanese
fisheries in the NW and SW regions (Table 4.6). This could be due to regional
differences in the operational fishing practices and the tag reporting rates by Japan.
The majority of the fishing effort of the Japanese tropical tuna fishery in the Atlantic
takes place in the SE and not in the SW region (Nakano 1993).
Higher levels of fishing mortality (0.10  F  0.25) were estimated for the
eastern North Atlantic (NE and SE regions; Figure 4.5). The NE region includes the
traditional fishing grounds exploited by the Spanish longline fishery for swordfish, in
which blue sharks are a dominant component of the bycatch (Mejuto et al. 2002). To
a lesser extent, blue sharks were also subject to bycatch in the swordfish and tuna
fisheries of Portugal and Japan. Reduced abundance of the targeted swordfish and
new European markets for blue shark meat in the 1990s, resulted in a recent shift to
blue sharks as a target species in the Iberian (Spain and Portugal) fisheries (Mejuto et
al. 2002; Aires-da-Silva et al. 2008a). Emerging blue shark targeting practices could
account for the higher F estimates in the NE in the last decade. The F estimates for
the NE during the pre-1990 period, however, seem too high in comparison with later
years, particularly for the Spanish fishery which was still developing and not
targeting blue sharks yet. The F estimates could be refined when a better
101
understanding of potential catchability changes over time is available (e.g., swordfish
vs blue shark target practices).
In the SE region, blue sharks were also subject to bycatch or target fishing
mortality in the Spanish longline fishery. However, the Japanese longline fishery was
estimated to account for most of the blue shark fishing mortality in this southern
region, which is the traditional fishing ground of the Japanese Atlantic tropical tuna
fishery (Nakano 1993). The submergence of the blue shark into cooler deeper waters
in the tropical and equatorial regions could explain higher levels of catchability of the
blue shark to the Japanese deep-water longline gear targeting bigeye tuna in the SE
(Table 4.6). The total levels of blue shark fishing mortality in the SE have reached the
0.2yr-1 reference point in the late 1990s, as similar to the NE.
High levels of blue shark fishing mortality in the eastern waters of the North
Atlantic have implications for management and conservation. While the NE region
includes important blue shark nursery grounds where juvenile sharks are highly
vulnerable (Mejuto and García-Cortés 2005; Aires-da-Silva et al. 2008a), the SE
region seems to account for the highest aggregations of neonates and pregnant
females in North Atlantic Ocean (Castro and Mejuto 1995; Mejuto and García-Cortés
2005). Demographic analysis showed the strong dependence of the blue shark
population growth rate on the survival of the juveniles, and that population declines
could result even if low (0.2yr-1  F  0.3 yr-1) fishing mortality rates are applied to
juvenile (Chapter 2; Aires-da-Silva and Gallucci, 2007). Therefore, the emerging
102
targeting fisheries for blue sharks in the eastern North Atlantic (Mejuto et al. 2002;
Aires-da-Silva et al. 2008a) should be monitored with caution.
Blue shark migrations across thousands of miles of the North Atlantic Ocean
are well known from the conventional tag recovery data (Casey 1985; Kohler et al.
2002). These migratory patterns were captured by the tagging model, as indicated by
significant inter-regional movement rates. However, the model also indicated that the
rates of movement across regions seem to be lower than the proportions of sharks
staying in a region on a yearly basis. This suggests that a spatially-structured stock
assessment is necessary to capture the stock dynamics. Lower rates of movement
across regions represent an opportunity for conservation, as large-scale marine
reserves could potentially be an efficient tool for management (Baum et al. 2003;
Aires-da-Silva and Gallucci 2007).
The large-scale of the blue shark biological scenario and the lack of an
experimental design in the opportunistic tagging program, opens possibilities for
violations of the underlying assumptions of tagging analysis. A major assumption of
the tagging model is that the tagged and untagged fish are randomly mixed in the
population (Brownie et al. 1985). Violations to this assumption are commonly
referred to as the “nonmixing” problem (Hoenig et al. 1998; Latour et al. 2001). The
CSTP seems particularly vulnerable to failures of the random mixing assumption
since the majority of the tags are released opportunistically by U.S. sport fisherman in
the western North Atlantic (Kohler et al. 1998). In addition to this unbalanced spatial
design of the tag releases, the same sport fishery accounts for large numbers of tag
103
recoveries, working against a rapid dispersal of the tagged blue sharks. Therefore, the
CSTP could benefit from an improved spatial design if population modeling work
becomes a main goal.
On the other hand, the highly migratory behavior of the blue shark seems to
favor rapid dispersal of tags out of the U.S. sport fishing grounds. In late summer and
autumn, blue sharks initiate southern movements as well as extensive offshore
migrations (Casey 1982; Casey 1985; Kohler et al. 2002). In addition, the CSTP was
able to expand its cooperative tagging efforts geographically, and substantial numbers
of blue sharks have also been tagged and released in the waters of the eastern North
Atlantic since the early 1970s (see Figure 4.2). Finally, the spatial stratification of the
model decreased its vulnerability to the nonmixing problem.
Incorrect assumptions about the processes of tag shedding and tag reporting
could also result in major biases (Beverton and Holt 1957). The sensitivity analyses
showed that the fishing mortality parameters were generally robust to the various
assumptions made about the tag shedding rate (  ) and the tag reporting rates of the
major fisheries (  ). In contrast, the movement parameter estimates were highly
sensitive to the different assumptions made, particularly to higher levels of tag
shedding (  =0.20). The true values of the movement parameters therefore remain
uncertain. Therefore, it is critical that empirical estimates be obtained for the tag
shedding and tag reporting rates. Double-tagging experiments will help to estimate
the tag shedding rate (Beverton and Holt 1957). A high-reward tagging experiment
seems an appropriate choice to estimate the fishery-specific tag reporting rates
104
(Pollock et al. 2001). The modern tools of satellite-based tracking (Block et al. 2005;
Hulbert et al. 2005) could also be useful to validate the blue shark movement
parameter estimates.
An additional potential source of bias in this study is the assumption that the
tag reporting rates for the different fisheries are constant over time. In the real world,
changes of the tag reporting rates are likely to have occurred over the long historical
periods of the fisheries. Simulation studies could be conducted to evaluate the impact
of violations to the assumption of time-invariant tag reporting rates on the model
results (Xiao 1996).
It should be noted that the geographical strata assumed in the model were
mainly defined based on the spatial distribution of the tag releases and recoveries
(Kohler et al. 2002), which largely reflect the fishing effort patterns of the
cooperative taggers, and the different fisheries catching blue sharks. Alternative
geographical strata should be considered in the future, particularly region
designations that account for biological (e.g., sex- and size segregation patterns of the
stock) and oceanographic conditions. Geostatisical methods could be explored for this
purpose (Isaaks and Srivastava 1989).
The negative binomial likelihood was chosen to deal with the overdispersion
of the tag-recovery data, an issue that is commonly encountered with tagging data of
highly migratory species (Hampton and Fournier 2001). However, it is possible that
additional overdispersion in the data still exists which results from violations to the
model assumptions. One example is the assumption that the tags and the fishing effort
105
are randomly distributed in the four geographical strata considered, and this may not
to be the case for some fisheries. Therefore, further corrections for overdispersion
could potentially improve the parameter estimates (Hilbe 2007).
This study indicates that there is potential in the tools of tagging analysis to
improve the limited scientific information available for management and conservation
of the North Atlantic blue shark. Although data-limited in terms of fishery statistics,
this stock is the data-richest, in terms of tagging data, among the world’s stocks of
highly migratory sharks. Unfortunately, the historical blue shark tagging programs in
the North Atlantic (Stevens 1976; Kohler et al. 1998; Fitzmaurice et al. 2005; Mejuto
et al. 2005), all suffer from unbalanced spatial designs as a result of their
opportunistic nature. While the ideal requirements of experimental design are not
implemented, the pooling of individual data sets across the ocean could greatly
improve the quality of the model parameters. Increased sample sizes would allow for
size- and sex- specific modeling, and to estimate size selectivity from the tagging data
(Anganuzzi et al. 1994).
Despite the high volume of tagging data available, the blue shark case study
remains a data- and modeling-limited situation in terms of fishery statistics and even
essential tagging parameters such as the rates of tag shedding and reporting. To
compensate for these uncertainties, several assumptions had to be made in this
analysis. One example is the assumption of annual fishery-specific effort deviates in
the effort-fishing mortality relationship, as in MULTIFAN-CL (Fournier et al. 1998;
Kleiber et al. 2006). This assumption was needed to account for varying levels of
106
credibility of the fishing effort data from the different fisheries. An important
philosophical issue that should be noted is that there is a balance between data quality
and model complexity. Therefore, improved quality and a better understanding of the
longline fishing effort data could improve future blue shark tagging analyses.
Finally, it remains to integrate the blue shark tagging model with a blue shark
stock assessment model (Punt et al. 2000a; Hampton and Fournier 2001). The stock
assessment uncertainties associated with the fishery statistics could be reduced from
integrated tagging and population dynamics modeling (Maunder 2001). Furthermore,
the integrated approach would provide a framework in which the unknown rates of
tag shedding and reporting could be estimated.
107
Table 4.1. Blue shark tag recoveries reported by fishing fleet (nation or group) for
each geographical region. The four geographical regions are: Northwest (NW),
Southwest (SW), Northeast (NE) and Southeast (SE). (X) identifies a fishery for
which there is reported fishing effort but no tag recoveries.
Fisher group
NW
Geographic region
SW
NE
SE
Totals
Sport
Canada 13
US 3,008
North America 3,021
13
3,008
3,021
Commercial
United States 510
Canada 213
Japan 228
Spain 1,005
Venezuela
Portugal 22
1,978
19
7
35
84
145
5
300
34
339
13
117
X
130
529
213
253
1,457
84
56
2,592
Scientific cruises
United States
Commercial others
Other Asian nations - commercial
Other Caribean nations - commercial
Other Eastern Atlantic nations - commercial
Other South American nations -commercial
Not reported
Pooled
Sport others
Carribean - sport
Eastern Atlantic - sport
69
1
24
3
5
11
67
70
1
9
4
3
6
40
73
20
1
10
13
143
1
9
41
79
25
Pooled
1
1
Totals 5,110
25
7
7
250
356
25
8
33
143
5,859
108
Table 4.2. List of the fisheries defined in the tagging model.
Fishery
Sport
Region
Tag recoveries
US and Canada
NW
3,021
Others pooled
SW
NE
25
7
US
NW
SW
510
19
Canada
NW
213
Spain
NW
SW
NE
SE
1,005
35
300
117
Japan
NW
SW
NE
SE
228
7
5
13
Portugal
NW
NE
22
34
Venezuela
SW
SE
84
0
Scientific - US
NW
69
Commercial
Others pooled
Totals
NW
SW
NE
SE
Tag recoveries
Fisheries
41
79
10
13
5,857
23
109
Table 4.3. Estimated parameters of the tagging model.
Parameter type
# of params.
Fishery-specific catchabilities
22
Movement probabilities
8
Fishing effort deviations
599
Overdispersion
1
Total
630
Table 4.4. Model sensitivity analysis. Each model case defines a different hypothesis about the instantaneous tag shedding rate
(  ) and the annual tag reporting rates (  ) of the major fleets.
Model
1
2
3
4
Tag shedding rate, 
(year-1)
0 (0%)
NW sport
1
1
1
0.5
Tag reporting rate,  (%)
Spain comm.
Japan comm.
1
0.2
0.5
0.2
0.5
0.5
0.5
0.2
Venezuela comm.
0.2
0.2
0.5
0.2
5
6 - base case
7
8
0.11 (10%)
1
1
1
0.5
1
0.5
0.5
0.5
0.2
0.2
0.5
0.2
0.2
0.2
0.5
0.2
9
10
11
12
0.22 (20%)
1
1
1
0.5
1
0.5
0.5
0.5
0.2
0.2
0.5
0.2
0.2
0.2
0.5
0.2
11
0
111
Table 4.5. Movement and fishing mortality parameter estimates and their coefficient
of variation (CV) obtained from the base case (model 6 in Table 4.4).
Parameter
Value
CV
a) Movement rates
NW to NW
to SW
to NE
0.683
0.296
0.021
0.024
0.056
0.156
SW to SW
to NW
to SE
0.824
0.067
0.108
0.060
0.266
0.416
NE to NE
to SE
to NW
0.708
0.142
0.150
0.073
0.292
0.269
SE to SE
to SW
to NE
0.430
0.295
0.274
0.474
0.800
0.514
0.028
0.003
0.001
0.049
0.018
0.000
0.001
0.001
0.110
0.290
0.550
0.130
0.250
0.660
1.010
0.100
0.090
0.002
0.004
0.007
0.013
0.002
0.003
0.480
0.350
0.470
0.410
1.050
0.570
0.032
0.240
0.113
0.037
0.016
0.010
0.029
0.290
0.570
0.650
1.090
0.910
0.204
0.250
0.052
0.111
0.000
0.019
0.500
0.530
0.280
0.690
0.182
0.420
1.641
0.066
b) Fishing mortality (F) in 2000
Northwest (NW)
NW sport
U.S. comm..
Canada comm.
Spain comm.
Japan comm.
Portugal comm.
U.S. scientific
Pooled comm.
NW total
F2000
Southwest (SW)
U.S. comm.
Spain comm.
Japan comm.
Venezuela comm.
Pooled sport
Pooled comm.
SW  total
F2000
Northeast (NE)
Spain comm.
Japan comm.
Portugal comm.
Pooled sport
Pooled comm.
NE total
F2000
Southeast (SE)
Spain comm.
Japan comm.
Venezuela comm.
Pooled comm.
SE  total
F2000
c) Overdispersion - negative binomial
112
Table 4.6. Catchability (q) parameter estimates obtained for each fishery from the
base case (model 6 in Table 4.4).
Fishery
Sport
Region
Catchability (q)
US and Canada
NW
0.0160
Others pooled
SW
NE
0.0016
0.0051
US
NW
SW
0.0023
0.0003
Canada
NW
0.0017
Spain
NW
SW
NE
SE
0.0061
0.0038
0.0349
0.0024
Japan
NW
SW
NE
SE
0.0090
0.0082
0.0199
0.0278
Portugal
NW
NE
0.0004
0.0145
Venezuela
SW
SE
0.0132
0.0132
Scientific - US
NW
0.0008
Others pooled
NW
SW
NE
SE
0.0010
0.0030
0.0160
0.0127
Commercial
113
a)
NW
NE
SW
SE
b)
Figure 4.1. Blue shark tagging data collected by the Cooperative Shark Tagging
Program (CSTP) in the North Atlantic Ocean (1965-2004). (a) tag releases and b) tag
recoveries. The four geographical regions defined in the tagging model are:
Northwest (NW), Southwest (SW), Northeast (NE) and Southeast (SE).
1980
1990
5
0
1970
1980
year
year
SW Atlantic
SE Atlantic
1990
2000
1990
2000
year
1990
2000
Recaptures
20
15
0
5
10
200
100
0
Releases
5
Recaptures
15
1980
0
10
1970
300
20
80
60
40
0
20
Releases
Recaptures
10 15 20 25
100
0
2000
25
1970
50
Releases
0
2000
100 200 300
4000
Releases
Recaptures
Recaptures
400
150
NE Atlantic
0
Releases
6000
NW Atlantic
1970
1980
year
11
4
Figure 4.2. Annual time series of blue shark tag releases and recoveries in each region of the North Atlantic Ocean.
115
NW
NE
SW
SE
Figure 4.3. Movement hypothesis for the blue shark in the North Atlantic Ocean
assumed in the tagging model. Blue sharks can stay in the same region or move to the
neighboring regions within the annual time step of the model. The four geographical
regions are: Northwest (NW), Southwest (SW), Northeast (NE) and Southeast (SE).
400
60
Sport-N. America (1)
200
NW
20
0
10
0
Tag recoveries by fishery
10
0
0
10
20
Com. - Portugal (15)
25
125
30
15
0
0
0
0
10
Com. - Spain (6)
5
10
20
Com. - Japan (11)
5
10
Com. - Venez. (17)
10
5
0
0
0
0
0
10
10
10
10
25
5
Com. - Portugal (16)
5
Sport - others (21)
5
0
0
0
0
0
10
20
10
Com. - Spain (8)
15
Com. - Japan (13)
5
0
0
1980
2000
1960
Com. - Venez. (18)
10
2000
1960
Com. - others (26)
5
0
1980
Com. - others (25)
5
30
Comm. - others (24)
5
0
Com. - Japan (12)
10
Sport - others (20)
50
1960
30
Com. - Japan (10)
5
Com. - Spain (7)
SE
60
Com. - Spain (5)
Com. - others (23)
Com. - US (3)
NE
250
Com. - Canada (4)
30
Survey - US (19)
SW
50
Com. - US (2)
0
1980
2000
1960
1980
2000
Years
11
6
Figure 4.4. Base case model fit (solid lines) to the blue shark tag recovery observations (solid circles) for each fishery by
geographical region. The four geographical regions are: Northwest (NW), Southwest (SW), Northeast (NE) and Southeast
(SE).
117
0.5
0.4
NW
0.3
0.2
0.1
0.0
1970
1980
1990
2000
1970
1980
1990
2000
1970
1980
1990
2000
1970
1980
1990
2000
Fishing mortality rate by fishery (F)
0.5
0.4
SW
0.3
0.2
0.1
0.0
0.5
0.4
NE
0.3
0.2
0.1
0.0
0.5
0.4
SE
0.3
0.2
0.1
0.0
US & Canada - sport
Portugal - comm
US - comm
Venezuela - comm
Canada - comm
US - surveys
Spain - comm
Pooled - sport
Japan - comm
Pooled - comm
Years
Figure 4.5. Fishery-specific instantaneous fishing mortality rates (F) in each
geographical region as estimated from the base case model. The four geographical
regions are: Northwest (NW), Southwest (SW), Northeast (NE) and Southeast (SE).
A list of the different fisheries in each region is provided in Table 4.2. The dashed
horizontal lines identify the reference point for conservation (Fc=0.20)
118
0% tag shedding
10% tag shedding
20% tag shedding
0.3
NW
NW
NW
SW
SW
SW
NE
NE
NE
SE
SE
SE
0.2
0.1
0.0
0.3
Total fishing mortality rate (F)
0.2
0.1
0.0
0.3
0.2
0.1
0.0
0.3
0.2
0.1
0.0
1970
1980
1990
2000
1970
1980
1990
2000
1970
1980
1990
2000
NW-sport: 100%; SPN-comm: 100%; JPN-comm: 20%; VEN-comm: 20%
NW-sport: 100%; SPN-comm: 50%; JPN-comm: 20%; VEN-comm: 20%
NW-sport: 100%; SPN-comm: 50%; JPN-comm: 50%; VEN-comm: 50%
NW-sport: 50%; SPN-comm: 50%; JPN-comm: 20%; VEN-comm: 20%
Years
Figure 4.6. Sensitivity analyses of the estimate of total instantaneous fishing mortality
(F) to different scenarios about the tag shedding rate (  ) and the tag reporting rates
(  ) of major fisheries. The different scenarios investigated are identified in Table 4.4
and in the bottom caption of the figure. The dashed horizontal lines identify the
fishing mortality reference point assumed for conservation (Fc=0.2).
119
BIBLIOGRAPHY
Aires-da-Silva A. (1996). Contribution to the knowledge of the age and growth of the
blue shark, Prionace glauca (Carcharhinidae), in the North Atlantic.
Undergraduate Thesis of the Marine Biology and Fisheries Licenciatura
Degree. Universidade do Algarve, Portugal.
Aires-da-Silva A., Pereira J. G. (1999). Catch rates for pelagic sharks taken by the
Portuguese swordfish fishery in the waters around the Azores, 1993-1997.
International Commission for the Conservation of Atlantic Tunas, Collective
Volume of Scientific Papers 49, 300-305.
Aires-da-Silva A., Taylor I., Punt A. E., Gallucci V. F., Kohler N. E., Turner P. A.,
Briggs R., Hoey J. J. (2005). A framework for estimating movement and
fishing mortality rates of the blue shark, Prionace glauca, in the North
Atlantic Ocean from tag-recapture data. International Commission for the
Conservation of Atlantic Tunas, Collective Volume of Scientific Papers 58,
1073-1086.
Aires-da-Silva A., Gallucci V. F. (2007). Demographic and risk analyses applied to
management and conservation of the blue shark (Prionace glauca) in the
North Atlantic Ocean. Marine and Freshwater Research 58, 570-580.
Aires-da-Silva A., Ferreira R. L., Pereira J. G. (2008a). Blue shark catch rate patterns
from the Portuguese longline fishery in the Azores. In 'Sharks of the Open
Ocean'. (Eds M. Camhi, E. K. Pikitch and E. A. Babcock.) pp. 230-234.
(Blackwell Publishing: Oxford.)
Aires-da-Silva A. M., Hoey J. J., Gallucci V. F. (2008b). A historical index of
abundace for the blue shark (Prionace glauca) in the western North Atlantic.
Fisheries Research 92, 41-52.
Anderson E. (1985). Analysis of various sources of pelagic shark catches in the
Northwest and Western Central Atlantic Ocean and Gulf of Mexico with
comments on catches of other large pelagics. In 'Shark catches from selected
fisheries off the U.S. east coast'. (Eds E. D. Anderson, J. G. Casey, J. J. Hoey
and W. W. N.) pp. 1-14. (U.S. Department of Commerce: Washington, DC.)
120
Anderson E. D. (1990). Fisheries models as applied to elasmobranch fisheries. In
'Elasmobranchs as Living Resources: Advances in the Biology, Ecology,
Systematics, and the Status of the Fisheries'. (Eds H. L. Pratt, S. H. Gruber
and T. Taniuchi.) pp. 473-484. (U.S. Department of Commerce: Washington,
DC.)
Anganuzzi A., Hilborn R., Skalski J. R. (1994). Estimation of size-selectivity and
movement rates from mark-recovery data. Canadian Journal of Fisheries and
Aquatic Sciences 51, 734-742.
Anonymous (2005). Report of the 2004 inter-sessional meeting of the ICCAT subcommittee on by-catches: shark stock assessment. International Commission
for the Conservation of Atlantic Tunas, Collective Volume of Scientific Papers
58, 799-890.
Anonymous (2007). Report of the 2007 meeting of the sub-committee on ecosystems.
International Commission for the Consevervation of Atlantic Tunas. Subcommittee on Ecosystems. Madrid, Spain - February 19-23, 2007.
Babcock E. A., Pikitch E. K., McAllister M. K. (2000). Catch rates of blue shark
(Prionace glauca) in the U.S. Atlantic recreational fishery. International
Commission for the Conservation of Atlantic Tunas, Collective Volume of
Scientific Papers 51, 1850-1858.
Baum J. K., Myers R. A., Kehler D. G., Worm B., Harley S. J., Doherty P. A. (2003).
Collapse and conservation of shark populations in the Northwest Atlantic.
Science 299, 389-392.
Baum J. K., Myers R. A. (2004). Shifting baselines and the decline of pelagic sharks
in the Gulf of Mexico. Ecology Letters 7, 135-145.
Baum J. K., Kehler D., Myers R. A. (2005). Robust estimates of decline for pelagic
shark populations in the northwest Atlantic and Gulf of Mexico. Fisheries 30,
27-30.
Beerkircher L. R., Brown C. J., Lee D. W. (2003). Overview of the SEFSC pelagic
observer program in the Northwest Atlantic from 1992-2000. International
Commission for the Conservation of Atlantic Tunas, Collective volume of
scientific papers 55, 331-349.
121
Beverton R. J. H., Holt S. J. (1957). 'On the Dynamics of Exploited Fish Populations.'
(Chapman and Hall: London.)
Block B. A., Teo S. L. H., Walli A., Boustany A., Stokesbury M. J. W., Farwell C. J.,
Weng K. C., Dewar H., Williams T. D. (2005). Electronic tagging and
population structure of Atlantic bluefin tuna. Nature 434, 1121-1127.
Bonfil R. (1994). Overview of world elasmobranch fisheries. FAO Fisheries
Technical Paper 341. FAO, Rome.
Branstetter S. (1990). Early life-history implications of selected carcharhinoid and
lamnoid sharks of the northwest Atlantic. In 'Elasmobranchs as living
resources: advances in the biology, ecology, systematics, and the status of the
fisheries'. (Eds H. L. Pratt, S. H. Gruber and T. Taniuchi.) pp. 17–28. (U.S.
Department of Commerce: Washington, DC.)
Brooks E. N., Ortiz M., Beerkircher L. R., Apostolaki Y., Scott G. P. (2005).
Standardized catch rates for blue shark and shortfin mako shark from the U.S.
pelagic logbook and U.S. pelagic observer program, and U.S. weighout
landings. International Commission for the Conservation of Atlantic Tunas,
Collective Volume of Scientific Papers 58, 1054-1072.
Brown C. A. (2000). Standardized catch rates of four shark species in the VirginiaMassachusetts (U.S.) rod and reel fishery. International Commission for the
Conservation of Atlantic Tunas, Collective Volume of Scientific Papers 51,
1812-1821.
Brownie C., Anderson D. R., Burnham K. P., S. R. D. (1985). 'Statistical inference
from band recovery data - a handbook.' (United States Fish and Wildlife
Service, Resource Publication 156: Washington, DC.)
Burgess G. H., Beerkircher L. R., Cailliet G. M., Carlson J. K., Cortés E., Goldman
K. J., Grubbs R. D., Musick J. A., Musyl M. K., Simpfendorfer C. A. (2005a).
Is the collapse of shark populations in the Northwest Atlantic Ocean and Gulf
of Mexico real? Fisheries 30, 19-26.
Burgess G. H., Beerkircher L. R., Cailliet G. M., Carlson J. K., Cortés E., Goldman
K. J., Grubbs R. D., Musick J. A., Musyl M. K., Simpfendorfer C. A. (2005b).
Reply to "Robust estimates of decline for pelagic shark populations in the
Northwest Atlantic and Gulf of Mexico". Fisheries 30, 30-31.
122
Burgman M. A., Ferson S., Akçakaya H. R. (1993). 'Risk Assessment in
Conservation Biology. Population and Community Biology Series 12.'
(Chapman & Hall: London.)
Burnham K. P., Anderson D. R. (2002). 'Model Selection and Multimodel Inference:
A Practical Information-theoretic Approach.' (Springer-Verlag: New York.)
Cailliet G. M., Bedford D. W. (1983). The biology of three pelagic sharks from
California waters, and their emerging fisheries: a review. CalCOFI Report 24,
57-69.
Campana S., Marks L., Joyce W., Kohler N. (2005). Catch, by-catch and indices of
population status of blue shark (Prionace glauca) in the Canadian Atlantic.
International Commission for the Conservation of Atlantic Tunas, Collective
Volume of Scientific Papers 58, 891-934.
Campana S. E., Joyce W., Marks L., Natanson L. J., Kohler N. E., Jensen C. F., Mello
J. J., Pratt H. L., Myklevoll S. (2002). Population dynamics of the porbeagle
in the northwest Atlantic Ocean. North American Journal of Fisheries
Management 22, 106-121.
Campana S. E., Marks L., Joyce W., Kohler N. E. (2006). Effects of recreational and
commercial fishing on blue sharks (Prionace glauca) in Atlantic Canada, with
inferences on the North Atlantic population. Canadian Journal of Fisheries
and Aquatic Sciences 63, 670-682.
Carey F. G., Robison B. H. (1981). Daily patterns in the activities of swordfish,
Xiphias gladius, observed by acoustic telemetry. Fishery Bulletin 79, 277-292.
Carey F. G., Scharold J. V. (1990). Movements of blue sharks (Prionace glauca) in
depth and course. Marine Biology 106, 329-342.
Casey J. G., Mather F. J., Mason Jr. J. M., Hoenig J. (1978). Offshore fisheries of the
middle Atlantic Bight. Marine Recreational Fisheries 3, 107-129.
Casey J. G. (1982). Blue shark, Prionace glauca. Species synopsis. In 'Ecology of the
Middle Atlantic Bight Fish and Shellfish, Monograph 15, Fish Distribution'.
(Eds M. D. Grosslein and T. Azarovitz.) pp. 45-48. (MESA New York Bight
Project, New York Sea Grant Institute: Albany, New York.)
123
Casey J. G. (1985). Transatlantic migrations of the blue shark: a case history of
cooperative shark tagging. In 'Proceedings of the First World Angling
Conference'. Cap d'Agde, France. (Ed. R. H. Stroud) pp. 253-268.
(International Game Fish Association: Fort Lauderdale, Florida)
Castro J. A., Mejuto J. (1995). Reproductive parameters of blue shark, Prionace
glauca, and other sharks in the Gulf of Guinea. Marine and Freshwater
Research 46, 967-973.
Castro J. I., Woodley C. M., Brudek R. L. (1999). A preliminary evaluation of the
status of shark species. Fisheries Technical Paper 380. FAO, Rome.
Caswell H. (2001). 'Matrix population models. Construction, analysis, and
interpretation.' (Sinauer Associates, Inc. Publishers: Sutherland,
Massachusetts.)
Chen S., Watanabe S. (1989). Age dependence of natural mortality coefficient in fish
population dynamics. Nippon Suisan Gakkaishi 55, 205-208.
Clarke S. C., Magnussen J. E., Abercrombie D. L., McAllister M. K., Shivji M. S.
(2006a). Identification of shark species composition and proportion in the
Hong Kong shark fin market based on molecular genetics and trade records.
Conservation Biology 20, 201-211.
Clarke S. C., McAllister M. K., Milner-Gulland E. J., Kirkwood G. P., Michielsens C.
G. J., Agnew D. J., Pikitch E. K., Nakano H., Shivji M. S. (2006b). Global
estimates of shark catches using trade records from commercial markets.
Ecology Letters 9, 1115-1126.
Compagno L. J. V. (1984). 'FAO species catalogue. Vol. 4. Sharks of the world. An
annotated and illustrated catalogue of shark species known to date. Part 1 Hexanchiformes to Lamniformes.' (FAO: Rome.)
Compagno L. J. V. (1999). Systematics and body form. In 'Sharks, Skates, and Rays:
The Biology of Elasmobranch Fishes'. (Ed. W. C. Hamlett.) pp. 515. (The
Johns Hopkins University Press: Baltimore.)
Cormack R. M., Skalski J. R. (1992). Analysis of Coded Wire Tag Returns from
Commercial Catches. Canadian Journal of Fisheries and Aquatic Sciences 49,
1816-1825.
124
Cortés E. (1998). Demographic analysis as an aid in shark stock assessment and
management. Fisheries Research 39, 199-208.
Cortés E. (2000). Life history patterns and correlations in sharks. Reviews in
Fisheries Science 8, 299-344.
Cortés E. (2002a). Incorporating uncertainty into demographic modeling: Application
to shark populations and their conservation. Conservation Biology 16, 10481062.
Cortés E. (2002b). Catches and catch rates of pelagic sharks from the northwestern
Atlantic, Gulf of Mexico, and Caribbean. International Commission for the
Conservation of Atlantic Tunas, Collective volume of scientific papers 54,
1164-1181.
Cortés E. (2004). Life-history patterns, demography, and population dynamics. In
'Biology of Sharks and their Relatives'. (Eds J. C. Carrier, J. A. Musick and H.
R. Heithaus.) pp. 449–470. (CCR Press: Boca Renton, Florida.)
Cortés E. (2007). Comparative life-history and demography of pelagic sharks. In
'Sharks of the Open Ocean'. (Eds E. K. Pikitch and M. Camhi). (Blackwell
Scientific: Oxford)
Cramer J. (2000). Large pelagic logbook catch rates for sharks. International
Commission for the Conservation of Atlantic Tunas, Collective Volume of
Scientific Papers 51, 1842-1849.
Dobson A. J. (1991). 'An Introduction to Generalized Linear Models.' (Chapman &
Hall: London.)
Draganik B., Pelczarski W. (1984). The occurrence of the blue shark, Prionace
glauca (L.), in the North Atlantic. Reports of the Sea Fisheries Institute 19,
61-75.
Fitzmaurice P., Green P., Kierse G., Kenny M., Clarke M. (2005). Stock
discrimination of the blue shark, based on Irish tagging data. International
Commission for the Conservation of Atlantic Tunas, Collective volume of
scientific papers 58, 1171-1178.
Fleming E. H., Papageorgiou P. A. (1996). 'Shark fisheries and trade in Europe.'
(TRAFFIC Europe: Brussels.)
125
Fordham S. V. (2006). Shark alert: Revealing Europe's impact on shark populations.
The Shark Alliance, Brussels.
Fournier D. A., Hampton J., Sibert J. R. (1998). MULTIFAN-CL: a length-based,
age-structured model for fisheries stock assessment, with application to South
Pacific albacore, Thunnus alalunga. Canadian Journal of Fisheries and
Aquatic Sciences 55, 2105-2116.
Frisk M. G., Miller T. J., Dulvy N. K. (2004). Life histories and vulnerability to
exploitation of elasmobranchs: Inferences from elasticity, perturbation and
phylogenetic analyses. Journal of Northwest Atlantic Fishery Science 35, 2745.
Gallucci V. F., Taylor I. G., Erzini K. (2006). Conservation and management of
exploited shark populations based on reproductive value. Canadian Journal of
Fisheries and Aquatic Sciences 63, 931-942.
Garcés A. G., Rey J. C. (1983). Analisis de la pesqueria española de pez espada,
Xiphias gladius, entre los años 1973 y 1981. International Commission for the
Conservation of Atlantic Tunas, Collective Volume of Scientific Papers 18,
622-628.
Garcés A. G., Rey J. C. (1988). La pesqueria española de pez espada (Xiphias
gladius), 1973-1982. International Commission for the Conservation of
Atlantic Tunas, Collective Volume of Scientific Papers 20, 419-427.
Gibson C. D. (1998). 'The Broadbill Swordfishery of the Northwestern Atlantic: The
Fishery's Economic and Natural History (early 1800s through 1995).' (Ensign
Press: Camden, Maine.)
Grey M., Blais A. M., Hunt B., Vincent A. C. J. (2006). The USA's international
trade in fish leather, from a conservation perspective. Environmental
Conservation 33, 100-108.
Gubanov Y. P., Grigor'yev V. N. (1975). Observations on the distribution and biology
of the blue shark Prionace glauca (Carcharhinidae) of the Indian Ocean.
Journal of Icthyology 15, 37-43.
126
Hampton J., Fournier D. A. (2001). A spatially disaggregated, length-based, agestructured population model of yellowfin tuna (Thunnus albacares) in the
western and central Pacific Ocean. Marine and Freshwater Research 52, 937963.
Hampton J., Sibert J. R., Kleiber P., Maunder M. N., Harley S. J. (2005). Decline of
Pacific tuna populations exaggerated? Nature 434, E1-E2.
Harley S. J., Myers R. A., Dunn A. (2001). Is catch-per-unit-effort proportional to
abundance? Canadian Journal of Fisheries and Aquatic Sciences 58, 17601772.
Heessen H. J. L. (2003). Development of elasmobranch assessments DELASS. Final
report of the DG Fish Study Contract 99/055. Netherlands Institute for
Fisheries Research (RIVO), IJmuiden, The Netherlands.
Henderson A. C., Flannery K., Dunne J. (2001). Observations on the biology and
ecology of the blue shark in the North-east Atlantic. Journal of Fish Biology
58, 1347-1358.
Heppell S. S., Crowder L. B., Menzel T. R. (1999). Life table analysis of long-lived
marine species with implications for conservation and management. In 'Life in
the Slow Lane: Ecology and Conservation of Long-Lived Marine Animals'.
(Ed. J. A. Musick.) pp. 137-148. (American Fisheries Society: Bethesda,
Maryland.)
Hilbe J. M. (2007). 'Negative Binomial Regression.' (Cambridge University Press:
Cambridge.)
Hilborn R. (1990). Determination of fish movement patterns from tag recoveries
using maximum-likelihood estimators. Canadian Journal of Fisheries and
Aquatic Sciences 47, 635-643.
Hilborn R. (2006). Faith-based fisheries. Fisheries 31, 554-555.
Hinton M. G., Maunder M. N. (2004). Methods for standardizing CPUE and how to
select among them. International Commission for the Conservation of Atlantic
Tunas, Collective volume of scientific papers 56, 169-177.
Hoenig J. M. (1983). Empirical use of longevity data to estimate mortality rates.
Fishery Bulletin 81, 898-903.
127
Hoenig J. M., Gruber S. H. (1990). Life-history patterns in the elasmobranchs:
implications for fisheries management. In 'Elasmobranchs as Living
Resources: Advances in the Biology, Ecology, Systematics, and the Status of
the Fisheries'. (Eds H. L. Pratt, S. H. Gruber and T. Taniuchi.) pp. 1–16 (U.S.
Department of Commerce: Washington, DC.)
Hoenig J. M., Barrowman N. J., Pollock K. H., Brooks E. N., Hearn W. S., Polacheck
T. (1998). Models for tagging data that allow for incomplete mixing of newly
tagged animals. Canadian Journal of Fisheries and Aquatic Sciences 55,
1477-1483.
Hoey J., Mejuto J., Iglesias S., Conser R. (1988). A comparative study of the United
States and Spanish longline fleets targeting swordfish in the Atlantic Ocean
targeting swordfish North of 40ºN latitude. International Commission for the
Conservation of Atlantic Tunas, Collective Volume of Scientific Papers 27,
230-239.
Hoey J. J., Conser R. J., Bertolino A. R. (1989). The western North Atlantic
swordfish. In 'Audubon Wildlife Report 1989/1990'. (Ed. W. Chandler.) pp.
457-477. (Academic Press: San Diego.)
Hoff T. B., Musick J. A. (1990). Western North Atlantic shark fishery management
problems and informational requirements. In 'Elasmobranchs as Living
Resources: Advances in the Biology, Ecology, Systematics, and the Status of
the Fisheries'. (Eds H. L. Pratt, S. H. Gruber and T. Taniuchi.) pp. 455-472.
(U.S. Department of Commerce: Washington, DC.)
Holden M. J. (1968). The rational exploitation of the Scottish-Norwegian stocks of
spurdogs (Squalus acanthias L.). Fish. Investig. Ministry of Agriculture, Fish.
Food., London, U.K.
Holden M. J. (1973). Are long-term sustainable fisheries for elasmobranchs possible?
Rapports et Proces-Verbaux des Reunions, Counseil International pour
l'Exploracion de la Mer 164, 330-367.
Holden M. J. (1974). Problems in the rational exploitation of elasmobranch
populations and some suggested solutions. In 'Sea Fisheries Research'. (Ed. F.
R. Harden Jones.) pp. 117-137. (Halsted Press, John Wiley & Sons Ltd.: New
York.)
128
Holden M. J. (1977). Elasmobranchs. In 'Fish Population Dynamics'. (Ed. J. A.
Gulland.) pp. 187-215. (John Wiley & Sons Ltd.: New Work.)
Hulbert L. B., Aires-da-Silva A. M., Gallucci V. F., Rice J. S. (2005). Seasonal
foraging movements and migratory patterns of female Lamna ditropis tagged
in Prince William Sound, Alaska. Journal of Fish Biology 67, 490-509.
Isaaks E. H., Srivastava R. M. (1989). 'An introduction to applied geostatistics.'
(Oxford University Press)
Jensen A. L. (1996). Beverton and Holt life history invariants result from optimal
trade-off of reproduction and survival. Canadian Journal of Fisheries and
Aquatic Sciences 53, 820-822.
Kebe P., Restrepo V., Palma C. (2002). An overview of shark data collection by
ICCAT. International Commission for the Conservation of Atlantic Tunas,
Collective Volume of Scientific Papers 54, 1107-1122.
Ketchen K. S. (1969). A review of the dogfish problem off the western coast of
Canada. Journa of the Fisheries Research Board of Canada. MS Rep. Ser.
1048.
Kleiber P., Takeuchi Y., Nakano H. (2001). Calculation of plausible maximum
sustainable yield (MSY) for blue sharks (Prionace glauca) in the North
Pacific. SWFSC Admin. Rep. H-01-02 and Dept. of Commerce.
Kleiber P., Hampton J., Fournier D. A. (2006). MULTIFAN-CL User's Guide. 123 p.
Kohler N. E., Casey J. G., Turner P. A. (1998). NMFS Cooperative Shark Tagging
Program, 1962-93: An atlas of shark tag and recapture data. Marine Fisheries
Review 60, 1-87.
Kohler N. E., Turner P. A., Hoey J. J., Natanson L. J., Briggs R. (2002). Tag and
recapture data for three pelagic shark species, blue shark (Prionace glauca),
shortfin mako (Isurus oxyrinchus), and porbeagle (Lamna nasus) in the North
Atlantic Ocean. International Commission for the Conservation of Atlantic
Tunas, Collective volume of scientific papers 54, 1231-1260.
129
Kohler N. E., Turner P. A. (2008). The use of pelagic fish tagging data to determine
stock structure with emphasis on the blue shark (Prionace glauca) in the
North Atlantic Ocean. In 'Sharks of the Open Ocean'. (Eds M. Camhi, E. K.
Pikitch and E. A. Babcock.). (Blackwell Publishing: Oxford.)
Latour R. J., Hoenig J. M., Olney J. E., Pollock K. H. (2001). A simple test for
nonmixing in multiyear tagging studies: Application to striped bass tagged in
the Rappahannock River, Virginia. Transactions of the American Fisheries
Society 130, 848-856.
Lee D. W., Brown C. J., Jordan T. L. (1996). Overview of the SEFSC Pelagic
Observer Program in the Northwest Atlantic from 1992-1994 (SCRS/95/103).
International Commission for the Conservation of Atlantic Tunas, Collective
volume of scientific papers 45, 355-365.
Lefkovitch L. P. (1967). A theoretical evaluation of population growth after removing
individuals from some age groups. Bulletin of Entomological Research 57,
437-445.
Litvinov F. F. (2006). On the role of dense aggregations of males and juveniles in the
functional structure of the range of the blue shark Prionace glauca. Journal of
Icthyology 46, 613-624.
Lo N. C. H., Jacobson L. D., Squire J. L. (1992). Indexes of relative abundance from
fish spotter data based on delta-lognormal models. Canadian Journal of
Fisheries and Aquatic Sciences 49, 2515-2526.
Lopez A. M., McClellan D. B., Bertolino A. R., Lange M. D. (1979). Japanese
Longline Fishery in the Gulf of Mexico, 1978. Marine Fisheries Review 41,
23-28.
Maunder M. N. (2001). Integrated Tagging and Catch-At-Age Analysis (ITCAAN):
Model Development and Simulation Testing. . In 'Spatial Processes and
Management of Marine Populations'. (Eds G. H. Kruse, N. Bez, A. Booth, M.
W. Dorn, S. Hills, R. N. Lipicus, D. Pelletier, C. Roy, S. J. Smith and D.
Witherell.) pp. 123-146. (Alaska Sea Grant Program.)
Maunder M. N., Deriso R. B. (2003). Estimation of recruitment in catch-at-age
models. Canadian Journal of Fisheries and Aquatic Sciences 60, 1204-1216.
130
Maunder M. N., Starr P. J. (2003). Fitting fisheries models to standardised CPUE
abundance indices. Fisheries Research 63, 43-50.
Maunder M. N., Punt A. E. (2004). Standardizing catch and effort data: a review of
recent approaches. Fisheries Research 70, 141-159.
Maunder M. N., Sibert J. R., Fonteneau A., Hampton J., Kleiber P., Harley S. J.
(2006). Interpreting catch per unit effort data to assess the status of individual
stocks and communities. Ices Journal of Marine Science 63, 1373-1385.
McCullagh P., Nelder J. A. (1989). 'Generalized Linear Models.' (Chapman& Hall:
London.)
McKenzie R. A., Tibbo S. N. (1964). A morphometric description of blue shark
(Prionace glauca) from Canadian Atlantic waters. Journal of the Fisheries
Research Board of Canada 21, 865-866.
Mejuto J. (1985). Associated catches of sharks, Prionace glauca, Isurus oxyrinchus,
and Lamna nasus, with NW and N Spanish swordfish fishery, in 1984. ICES
CM 1985/H:42.
Mejuto J., Iglesias S. (1988). Campaña comercial de prospeccion de abundancia de
pez espada, Xiphias gladius L., y especies asociadas, en areas proximas a
Grand Banks. International Commission for the Conservation of Atlantic
Tunas, Collective Volume of Scientific Papers 27, 155-163.
Mejuto J., García-Cortés B., de la Serna J. M. (2002). Preliminary scientific
estimations of by-catches landed by the Spanish surface longline fleet in 1999
in the Atlantic Ocean and Mediterranean Sea. International Commission for
the Conservation of Atlantic Tunas, Collective Volume of Scientific Papers 54,
1150-1163.
Mejuto J., García-Cortés B. (2005). Reproductive and distribution parameters of the
blue shark Prionace glauca, on the basis of on-board observations at sea in the
Atlantic, India and Pacific Oceans. International Commission for the
Conservation of Atlantic Tunas, Collective Volume of Scientific Papers 58,
951-973.
131
Mejuto J., García-Cortés B., Ramos-Cartelle A. (2005). Tagging-recapture activities
of large pelagic sharks carried out by Spain or in collaboration with the
tagging programs of other countries. International Commission for the
Conservation of Atlantic Tunas, Collective Volume of Scientific Papers 58,
974-1000.
Miyabe N. (1992). Trend in CPUE for swordfish caught by the Japanese longline
fishery in the Atlantic Ocean. International Commission for the Conservation
of Atlantic Tunas, Collective Volume of Scientific Papers 39, 484-496.
Musick J. A. (1999). Ecology and conservation of long-lived marine animals. In 'Life
in the Slow Lane: Ecology and Conservation of Long-Lived Marine Animals'.
(Ed. J. A. Musick.) pp. 1-10. (American Fisheries Society: Bethesda,
Maryland.)
Musick J. A., Burgess G., Cailliet G., Camhi M., Fordham S. (2000). Management of
sharks and their relatives (Elasmobranchii). Fisheries 25, 9-13.
Myers R. A., Worm B. (2003). Rapid worldwide depletion of predatory fish
communities. Nature 423, 280-283.
Myers R. A., Worm B. (2005). Extinction, survival or recovery of large predatory
fishes. Philosophical Transactions of the Royal Society B-Biological Sciences
360, 13-20.
Nakano H. (1993). A review of the Japanese fishery and research on sharks in the
Atlantic Ocean. International Commission for the Conservation of Atlantic
Tunas, Collective volume of scientific papers 40, 409-412.
Nakano H., Seki M. P. (2003). Synopsis of biological data on the blue shark,
Prionace glauca Linnaeus. Bulletin of Fisheries Research Agency 6, 18-55.
Nakano H., Clarke S. C. (2005). Standardized CPUE for blue sharks caught by the
Japanese longline fishery in the Atlantic Ocean, 1971-2003. International
Commission for the Conservation of Atlantic Tunas, Collective Volume of
Scientific Papers 58, 1127-1134
132
Neves-dos-Santos M., Garcia A., Pereira J. G. (2002). A historical review of the
bycatch from the Portuguese surface long-line swordfish fishery: observations
on blue shark (Prionace glauca) and short-fin mako (Isurus oxyrinchus).
International Commission for the Conservation of Atlantic Tunas, Collective
Volume of Scientific Papers 54, 1333-1340.
Olsen A. M. (1959). Status of the school shark fishery in southeastern fishery in
southeastern Australian waters. Australian Journal of Marine and Freshwater
Research 10, 150-176.
Ortiz M., Arocha F. (2004). Alternative error distribution models for standardization
of catch rates of non-target species from a pelagic longline fishery: billfish
species in the Venezuelan tuna longline fishery. Fisheries Research 70, 275297.
Parker H. W., Stott F. C. (1965). Age, size and vertebral calcification in the basking
shark, Cetorhinus maximus (Gunnerus). Zool. Meded. (Leiden) 40, 305–319.
Pauly D. (1980). On the interrelationships between natural mortality, growth
parameters, and mean environmental temperature in 175 fish stocks. Journal
du Conseil International pour l'Exploration de la Mer 39, 175-192.
Pauly D. (1995). Anecdotes and the shifting baseline syndrome of fisheries. Trends in
Ecology & Evolution 10, 430-430.
Pauly D., Christensen V., Dalsgaard J., Froese R., Torres F. (1998). Fishing down
marine food webs. Science 279, 860-863.
Peterson I., Wroblewski J. S. (1984). Mortality rate of fishes in the pelagic
ecosystem. Canadian Journal of Fisheries and Aquatic Sciences 41, 11171120.
Pine W. E., Pollock K. H., Hightower J. E., Kwak T. J., Rice J. A. (2003). A review
of tagging methods for estimating fish population size and components of
mortality. Fisheries 28, 10-23.
Polacheck T. (2006). Tuna longline catch rates in the Indian Ocean: Did industrial
fishing result in a 90% rapid decline in the abundance of large predatory
species? Marine Policy 30, 470-482.
133
Pollock K. H., Hoenig J. M., Hearn W. S., Calingaert B. (2001). Tag reporting rate
estimation: 1. An evaluation of the high-reward tagging method. North
American Journal of Fisheries Management 21, 521-532.
Pratt H. L. (1979). Reproduction in the blue shark, Prionace glauca. Fishery Bulletin
77, 445-470.
Punt A. E., Hilborn R. (1997). Fisheries stock assessment and decision analysis: The
Bayesian approach. Reviews in Fish Biology and Fisheries 7, 35-63.
Punt A. E., Pribac F., Walker T. I., Taylor B. L., Prince J. D. (2000a). Stock
assessment of school shark, Galeorhinus galeus, based on a spatially explicit
population dynamics model. Marine and Freshwater Research 51, 205-220.
Punt A. E., Walker T. I., Taylor B. L., Pribac F. (2000b). Standardization of catch and
effort data in a spatially-structured shark fishery. Fisheries Research 45, 129145.
Queiroz N., Lima F. P., Maia A., Ribeiro P. A., Correia J. P., Santos A. A. (2005).
Movement of blue shark, Prionace glauca, in the north-east Atlantic based on
mark-recapture data. Journal of the Marine Biological Association of the
United Kingdom 85, 1107-1112.
Quinn T. J., Deriso R. B. (1999). 'Quantitative fish dynamics.' (Oxford University
Press: New York.)
R (2006). R: A language and environment for statistical computing. R Foundation for
Statistical Computing. Vienna, Austria.
Rey J. C., Mejuto J., Iglesias S. (1988). Evolucion historica y situacion actual de la
pesqueria española de pez espada (Xiphias gladius). International
Commission for the Conservation of Atlantic Tunas, Collective Volume of
Scientific Papers 27, 202-213.
Ripley W. E. (1946). The soup-fin shark and the fishery. Fishery Bulletin 64, 7–37.
Seber G. A. F. (2002). 'The estimation of animal abundance and related parameters.'
(The Blackburn Press: Caldwell, New Jersey.)
134
Shepherd T. D., Myers R. A. (2005). Direct and indirect fishery effects on small
coastal elasmobranchs in the northern Gulf of Mexico. Ecology Letters 8,
1095-1104.
Sibert J., Hampton J., Kleiber P., Maunder M. (2006). Biomass, size, and trophic
status of top predators in the Pacific Ocean. Science 314, 1773-1776.
Simões P. R. (1999). By-catch of swordfish fishery in the Azores from 1987-96; an
annotation on shortfin mako shark and blue shark. International Commission
for the Conservation of Atlantic Tunas, Collective Volume of Scientific Papers
49, 283-287.
Simpfendorfer C. A., Hueter R. E., Bergman U., Connett S. M. H. (2002). Results of
a fishery-independent survey for pelagic sharks in the western North Atlantic,
1977-1994. Fisheries Research 55, 175-192.
Simpfendorfer C. A., Bonfil R., Latour R. J. (2005). Mortality estimation. In
'Management Techniques for Elasmobranch Fisheries'. (Eds J. A. Musick and
R. Bonfil.) pp. 165-185 (FAO: Rome.)
Skomal G., Babcock E. A., Pikitch E. K. (2005). Indices of blue and mako shark
abundance derived from U.S Atlantic recreational fishery data. International
Commission for the Conservation of Atlantic Tunas, Collective Volume of
Scientific Papers 58, 1034-1043.
Skomal G. B., Natanson L. J. (2003). Age and growth of the blue shark (Prionace
glauca) in the North Atlantic Ocean. Fishery Bulletin 101, 627-639.
Skomal G. B. (2007). Evaluating the physiological and physical consequences of
capture on post-release survivorship in large pelagic fishes. Fisheries
Management and Ecology 14, 81-89.
Smith S. E., Au D. W., Show C. (1998). Intrinsic rebound potentials of 26 species of
Pacific sharks. Marine and Freshwater Research 49, 663-678.
Stevens J. D. (1974). The Occurrence and Significance of Tooth Cuts on Blue Shark
(Prionace glauca L) from British Waters. Journal of the Marine Biological
Association of the United Kingdom 54, 373-378.
135
Stevens J. D. (1975). Vertebral rings as a means of age determination in blue shark
(Prionace glauca L). Journal of the Marine Biological Association of the
United Kingdom 55, 657-665.
Stevens J. D. (1976). First results of shark tagging in the north-east Atlantic, 19721975. Journal of the Marine Biological Association of the United Kingdom
56, 929-937.
Stevens J. D. (1990). Further Results from a Tagging Study of Pelagic Sharks in the
North-East Atlantic. Journal of the Marine Biological Association of the
United Kingdom 70, 707-720.
Stevens J. D. (1999). Variable resilience to fishing pressure in two sharks: the
significance of different ecological and life history parameters. In 'Life in the
Slow Lane: Ecology and Conservation of Long-Lived Marine Animals'. (Ed.
J. A. Musick.) pp. 11-15. (American Fisheries Society: Bethesda, Maryland.)
Stevens J. D., Bonfil R., Dulvy N. K., Walker P. A. (2000). The effects of fishing on
sharks, rays, and chimaeras (chondrichthyans), and the implications for
marine ecosystems. Ices Journal of Marine Science 57, 476-494.
Stewart I. J. (2007). Defining plausible migration rates based on historical tagging
data: a Bayesian mark-recapture model applied to English sole (Parophrys
vetulus). Fishery Bulletin 105, 470-484.
Sugiura N. (1978). Further analysis of the data by Akaike's information criterion and
the finite corrections. Communications in Statistics, Theory and Methods A7,
13-26.
Takeuchi Y., Senba Y., Nakano H. (2005). Demographic analysis on Atlantic blue
and shortfin mako sharks. International Commission for the Conservation of
Atlantic Tunas, Collective Volume of Scientific Papers 54, 1231-1260.
Tanaka S., Cailliet G. M., Yudin K. G. (1990). Differences in growth of the blue
shark, Prionace glauca: technique or population? In 'Elasmobranchs as Living
Resources: Advances in the Biology, Ecology, Systematics, and the Status of
the Fisheries'. (Eds H. L. Pratt, S. H. Gruber and T. Taniuchi.) pp. 177-187.
(U.S. Department of Commerce: Washington, DC.)
Tuljapurkar S. (1990). 'Population dynamics in variable environments.' (SpringerVerlag: New York.)
136
Venables W. N., Ripley B. D. (1997). 'Modern Applied Atatistics with S-plus.'
(Springer: New York.)
Vignaux M. (1994). Catch per unit effort (CPUE) analysis of west coast South Island
Cook Strait spawning hoki fisheries, 1987–93. NZ Fisheries Association
Research Document 94/11.
Walker T. I. (1998). Can shark resources be harvested sustainably? A question
revisited with a review of shark fisheries. Marine and Freshwater Research
49, 553-572.
Walsh W. A., Kleiber P. (2001). Generalized additive model and regression tree
analyses of blue shark (Prionace glauca) catch rates by the Hawaii-based
commercial longline fishery. Fisheries Research 53, 115-131.
Walters C. (2003). Folly and fantasy in the analysis of spatial catch rate data.
Canadian Journal of Fisheries and Aquatic Sciences 60, 1433-1436.
Ward P., Myers R. A. (2005). Shifts in open-ocean fish communities coinciding with
the commencement of commercial fishing. Ecology 86, 835-847.
Wise J. P., Le Guen J. C. (1969). The Japanese Atlantic Long - Line Fishery, 19561963. In 'Symposium on the Oceanography and Fisheries Resources of the
Tropical Atlantic'. Paris. (UNESCO)
Wood C. C., Ketchen K. S., Beamish R. J. (1979). Population dynamics of spiny
dogfish (Squalus acanthias) in British Columbia waters. Journal of the
Fisheries Research Board of Canada 36, 647-656.
Xiao Y. S. (1996). A framework for evaluating experimental designs for estimating
rates of fish movement from tag recoveries. Canadian Journal of Fisheries
and Aquatic Sciences 53, 1272-1280.
137
APPENDIX A. Supplementary information of Chapter 2.
Equations of the indirect methods used to estimate the instantaneous coefficient
of natural mortality M (hence natural survivorship, s) for the blue shark,
Prionace glauca.
1) Hoenig (1983),
ln(M )  1.46  1.01(ln tmax )
where M is the coefficient of instantaneous natural mortality and tmax is the maximum
age.
2) Pauly (1980)
log(M )  0.0066  0.279log( L )  0.6543log( K )  0.4634log(T )
where L and K are the asymptotic length and the growth coefficient parameters of
the von Bertalanffy growth function, respectively, and T is the mean water
temperature (in degrees Celsius).
3) Chen and Watanabe (1989)
M (t ) 
K
1 e
 K ( t t0 )
, t  tm
138
M (t ) 
K
, t  tm
a0  a1 (t  tm )  a2 (t  tm ) 2
where,
a0  1  e K (tm t0 )
a1  Ke  K (tm t0 )
1
a2   K 2 e  K (tm t0 )
2
and,
tm  
1
ln(1  e K (t0 ) )  t0
K
K and t 0 are parameters of the von Bertalanffy growth function.
4) Peterson and Wroblewski (1984)
M w  1.92 w0.25
where w is weight in kilograms.
5) Jensen (1996) – age at maturity method
M
1.65
xm
where xm is the age at first maturity.
6) Jensen (1996) – growth coefficient method
M  1.65( K )
139
where K is the growth parameter of the von Bertalanffy growth function.
Matrix algebra used to determine the net reproductive rate, R0
Net reproductive rate (R0) was determined following the matrix algebra steps
presented in Caswell (2001). The parameter represents the mean number of offspring
by which a pup will be replaced by the end of its life. In other words, it is the rate by
which the population increases from one generation to the next.
R0 is the dominant eigenvalue of the R matrix, defined as
R = F*N
in which,
F* is the fertility matrix giving the expected number of offspring produced
per time step (denoted with a superscripted asterisk to avoid confusion with F, the
coefficient of fishing mortality). It is obtained from decomposition of the Leslie
matrix ( M ). It has the same dimensions and the first row elements describing
reproduction in M .
N is the fundamental matrix giving the expected number of time steps spent
in each transient state (age-class), defined as
N  (I - T)1
140
in which,
I is the identity matrix with the same dimensions as M
T is the matrix describing the survival transitions resulting from
decomposition of the Leslie matrix. It has the same dimensions and survival terms as
M
141
APPENDIX B – Supplementary information of Chapter 3
Description of data sources
Fishery-dependent (observer) data
Data from commercial pelagic longline fisheries operating in the western
North Atlantic are available from national fishery observer programs. A chronogram
which shows the temporal coverage of U.S. observer programs is shown in Table
B.1a. U.S. observers were deployed aboard Japanese tuna longliners fishing in the
U.S. Economic Exclusive Zone from 1978 to 1988 (Lopez et al. 1979). Between 1985
and 1990, observers and research scientists were opportunistically deployed on a
small number of fishing trips aboard U.S. vessels. After 1990, observers were
deployed on permitted U.S. swordfish and tuna vessels randomly selected by the U.S.
Pelagic Observer Program of the Southeast Fisheries Science Center of the National
Marine Fisheries Service (Lee et al. 1996; Beerkircher et al. 2003).
Fishery- independent (survey) data
Information was recovered from fishery-independent longline surveys
conducted in the western North Atlantic under various research programs (Table
B.1b). Between 1957 and 1970, the U.S. Bureau of Commercial Fisheries (BCF)
coordinated joint research cruises with the Woods Hole Oceanographic Institute
(WHOI) to investigate tuna and swordfish longline fishery potential in the western
North Atlantic (Wilson and Barlett 1967). In the late fifties, the swordfish harpoon
142
fishery was dominated by Canadian vessels. The Canadian Department of Fisheries
and Oceans (DFO) initiated longline research cruises in 1961, continued them
annually through 1972, and occasionally after that until 1990. At the same time
(1961), the U.S. Bureau of Sport Fisheries and Wildlife (BSFW), initiated longline
cruises out of the Sandy Hook Marine Laboratory in New Jersey to gather life-history
information on coastal sharks in response to public concerns about shark attacks on
humans (Casey 1985). This survey and related life-history oriented shark research
shifted to offshore stations by the mid 1960s, where it complemented the offshore
tuna and swordfish surveys (BCF, WHOI and DFO). In 1970, the National Marine
Fisheries Service (NMFS) was established under the U.S. Department of Commerce
and the shark survey and life-history research continued at the NMFS’s Narragansett
Laboratory, Rhode Island, with a primary emphasis on investigation blue and mako
sharks migration patterns and biological studies. A fixed station grid survey design
utilizing bottom longline gear was implemented in the late 1980’s in response to
management concern about the status of large coastal sharks and the expansion of
coastal shark targeted fishing effort.
Table B.1. Chronogram of the pelagic longline data sources investigated: a) fishery-independent, historical research cruise
sources; and, b) fishery-dependent, observer programs on commercial fisheries.
Year
Data sources
57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00
a)
Observer programs
NEFSC-PLOP
10 48 ## ## ## ##
SEFSC-PLOP
25 80 18 15 23 ## 76 ## ##
US-JPN
## ## ## ## ## ## ## ## ## ## ##
b)
Reseach cruises
BCF-WHOI
35 25 27 21 17 26 31 57 24 12
BSFW-SHML
CN-DFO
6
10 14
13 21 25 18
12 22 34 36 32 20 20
1
25 49
NARR
38
29
5
9 19 33 64 75 46 49 23 32 24 46 48
NARR-comm
10
NARR-NE
7
7
3 10 26 10 56
8 19
14
2 10
5
16
8
Legend:
Data sources
Description
a)
Observer programs
SEFSC-PLOP
Southeast Fisheries Science Center, Pelagic Longline Observer Program
NEFSC-PLOP
Northeast Fisheries Science Center, Pelagic Longline Observer Program
Dom-opport.
Opportunistic deployments aboard commercial vessels
US Obs- JPN
U.S. observers deployed aboard Japanese vessels
b)
Research cruise programs
BCF-WHOI
Cooperative cruises by the Bureau of Commercial Fisheries and the Woods Hole Oceanographic Institution, U.S.
BSFW-SHML
Bureau of Sport Fisheries and Wildlife, Sandy Hook Marine Laboratory, U.S.
NARR
National Marine Fisheries Service, Narragansett Laboratory, U.S.
NARR-NE
Cooperative cruises between NMFS Narragansett Laboratory and other institutions in the northeast U.S.
NARR-comm
Opportunistic deployments of NMFS Narragansett laboratory scientists aboard volunteer commercial vessels, U.S.
CN-DFO
Department of Fisheries and Oceans, Canada
14
3
144
Table B.2. Model selection tables. a) binomial error GLM for the proportion of the
non-zero sets. Two error assumptions were investigated for the mean catch rate
(CPUE) of the positive sets: b) lognormal, c) gamma. d) Negative binomial for the
count data. The AIC statistic and the proportion of the total deviance are presented for
the models selected for each individual target data set and the reconciled data.
a) Proportion of positive sets - binomial
Target species
Researh cruises
TUN
SWO
PLSHK - all data
PLSHK - without Geronimo
PLSHK - Geronimo
Observer programs
TUN (JPN)
TUN (US)
SWO (US)
Reconciled
Final model
% of dev.
year + trimester
year + setSzCat1
year + areaNMFS
year
year + areaNMFS + bottomDepthCat1
7.17
27.57
13.34
14.49
22.12
year + trimester + areaNMFS + rigDepthCat
year + trimester + setSzCat1 + lStickCat1
year + trimester + areaNMFS
6.59
11.65
17.75
year + targetID + trimester + areaNMFS + bottomDepthCat1 +
setSzCat1 + lStickCat1 + rigDepthCat
9.26
b ) CPUE of positive sets - lognormal
Target species
Researh cruises
TUN
SWO
PLSHK - all data
PLSHK - without Geronimo
PLSHK - Geronimo
Observer programs
TUN (JPN)
TUN (US)
SWO (US)
Reconciled
Final model
% of dev.
year + startSetCat1 + setSzCat1
year + trimester + areaNMFS + bottomDepthCat1 + setSzCat1
year + trimester + areaNMFS + bottomDepthCat1 + setSzCat1
year + areaNMFS + bottomDepthCat1 + startSetCat1 + setSzCat1
year + areaNMFS + bottomDepthCat1
41.10
50.61
54.68
42.3244
37.41
year + trimester + areaNMFS + bottomDepthCat1 +
startSetCat1 + setSzCat1 + rigDepthCat
year + trimester + areaNMFS + startSetCat1 + setSzCat1
year + trimester + areaNMFS + startSetCat1 + setSzCat1 + lStickCat1
28.79
40.3846
year + targetID + trimester + areaNMFS + bottomDepthCat1 +
setSzCat1 + lStickCat1 + rigDepthCat
6.62
57.00
c) CPUE of positive sets - gamma
Target species
Researh cruises
TUN
SWO
PLSHK - all data
PLSHK - without Geronimo
PLSHK - Geronimo
Observer programs
TUN (JPN)
TUN (US)
SWO (US)
Reconciled
Final model
% of dev.
year + startSetCat1 + setSzCat1
year + trimester + areaNMFS + bottomDepthCat1 + setSzCat1
year + trimester + areaNMFS + bottomDepthCat1 + startSetCat1 + setSzCat1
year + trimester + areaNMFS + bottomDepthCat1 + startSetCat1 + setSzCat1
year + trimester + areaNMFS + bottomDepthCat1
42.41
58.62
53.15
46.24
37.91
year + trimester + areaNMFS + setSzCat1 + rigDepthCat
year + trimester + areaNMFS + startSetCat1 + setSzCat1 + lStickCat1
year + trimester + areaNMFS + startSetCat1 + setSzCat1 + lStickCat1
9.92
36.86
38.22
year + targetID + trimester + areaNMFS + bottomDepthCat1 +
setSzCat1 + lStickCat1 + rigDepthCat
71.66
d) Fish counts - negative binomial
Target species
Researh cruises
TUN
SWO
PLSHK - all data
PLSHK - without Geronimo
PLSHK - Geronimo
Observer programs
TUN (JPN)
TUN (US)
SWO (US)
Reconciled
Final model
% of dev.
year + startSetCat1 + setSzCat1
year + trimester + bottomDepthCat1 + setSzCat1
year + trimester + areaNMFS + bottomDepthCat1 + startSetCat1 + setSzCat1
year + areaNMFS + bottomDepthCat1 + setSzCat1
year + trimester + areaNMFS + bottomDepthCat1
11.98
32.41
29.20
11.30
21.32
year + trimester + bottomDepthCat1 + setSzCat1 + rigDepthCat
year + trimester + areaNMFS + bottomDepthCat1 + startSetCat1
year + trimester + areaNMFS + bottomDepthCat1 + setSzCat1 + lStickCat1
4.39
9.34
23.44
year + targetID + trimester + areaNMFS + bottomDepthCat1 +
setSzCat1 + lStickCat1 + rigDepthCat
54.72
145
Figure B.1. Diagnostic plots for the lognormal model on the catch rates (CPUE) of
the positive sets for each of the target species dataset. Left: qq-plots; Middle:
diagnostics on systematic departures from model assumptions; Right: diagnostic on
the performance of the variance function.
146
Figure B.2. Diagnostic plots for the gamma model on the catch rates (CPUE) of the
positive sets for each of the target species dataset. Left: diagnostics on systematic
departures from model assumptions; Right: diagnostic on the performance of the
variance function.
147
Figure B.3. Diagnostic tests for the negative binomial model on the blue shark count
data for each of the target species dataset. Left: diagnostics on systematic departures
from model assumptions; Right: diagnostic on the performance of the variance
function.
148
Figure B.4. Delta-lognormal indices for the individual target data sets and reconciled
data. The nominal CPUE are also presented for the catch rates of the positive sets. a)
Binomial models on the proportion of the noz-zero sets; b) Lognormal models on the
CPUE of the noz-zero sets; c) Delta-lognormal indices (product between a and b).
149
Figure B.5. Delta-gamma indices for the individual target data sets and reconciled
data. The nominal CPUE are also presented for the catch rates of the positive sets. a)
Binomial models on the proportion of the noz-zero sets; b) Gamma models on the
CPUE of the noz-zero sets; c) Delta-lognormal indices (product between a and b).
150
Figure B.6. Negative binomial indices based on the count data for each of the
individual target data sets and reconciled data.
151
APPENDIX C – Supplementary information of Chapter 4.
Table C.1. Time series of annual tag releases and recoveries by the Cooperative Shark
Tagging Program (CSTP) in different geographical regions of the North Atlantic
Ocean: Northwest (NW), Southwest (SW), Northeast (NE) and the Southeast (SE).
Year
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
Totals
%
Tag releases by geographical region
NW
SW
NE
SE
140
0
0
0
273
0
0
0
58
0
0
0
19
0
0
0
6
0
0
0
316
0
0
0
344
0
42
0
742
1
48
0
213
0
75
0
418
0
54
0
567
1
6
0
994
0
27
0
2511
0
14
0
2681
1
8
0
3154
0
11
0
2732
4
7
0
2494
6
42
214
2038
0
1
375
3233
0
2
76
1606
0
8
277
4243
53
2
84
2636
0
109
0
2375
8
62
0
2984
0
16
0
2328
2
3
0
2375
1
17
0
3699
5
70
0
5046
49
69
4
4293
86
96
1
4099
38
107
14
4820
51
91
4
5916
19
92
3
5801
24
38
1
3511
11
23
0
3016
3
25
1
2419
11
91
0
1863
9
208
0
1921
15
110
0
1796
1
496
2
1974
4
114
4
91,654
403
2,184
1,060
0.962
0.004
0.023
0.011
All
140
273
58
19
6
316
386
791
288
472
574
1,021
2,525
2,690
3,165
2,743
2,756
2,414
3,311
1,891
4,382
2,745
2,445
3,000
2,333
2,393
3,774
5,168
4,476
4,258
4,966
6,030
5,864
3,545
3,045
2,521
2,080
2,046
2,295
2,096
95,301
1.000
Tag recoveries by geographical region
NW
SW
NE
SE
1
0
0
0
7
0
0
0
4
0
0
0
0
0
0
0
0
1
0
0
3
0
0
0
8
1
0
0
13
0
0
0
17
0
0
0
12
1
0
1
23
1
1
0
41
0
0
0
85
2
0
0
147
6
2
1
102
0
1
0
77
7
0
2
90
4
4
5
45
8
2
7
67
2
1
1
39
4
4
0
140
5
7
2
108
4
4
3
52
5
6
1
109
4
8
0
107
8
4
1
101
9
12
0
152
7
7
1
210
12
9
0
260
14
11
4
269
9
21
5
238
14
16
6
359
8
13
10
417
19
25
17
419
26
23
19
321
18
27
20
338
21
33
13
222
14
34
7
197
13
25
6
168
1
15
5
142
2
41
6
5,110
250
356
143
0.872
0.043
0.061
0.024
All
1
7
4
0
1
3
9
13
17
14
25
41
87
156
103
86
103
62
71
47
154
119
64
121
120
122
167
231
289
304
274
390
478
487
386
405
277
241
189
191
5,859
1.000
Table C.2. Sensitivity analyses of the movement parameters to different scenarios about the tag shedding rate (  ) and the tag
reporting rates (  ) of major fisheries. The different model cases investigated are defined in Table 4.4.
from
to
Model case
cv6
M7
M1
cv1
M2
cv2
M3
cv3
M4
cv4
M5
cv5
M6
cv7
M8
cv8
M9
cv9
M10
cv10
M11
cv11
M12
cv12
NW
SW
NE
0.60
0.38
0.02
0.02
0.04
0.15
0.61
0.38
0.02
0.02
0.04
0.15
0.60
0.38
0.02
0.02
0.04
0.15
0.62
0.36
0.02
0.02
0.04
0.15
0.62
0.36
0.02
0.02
0.04
0.15
0.68
0.30
0.02
0.02
0.06
0.16
0.68
0.30
0.02
0.02
0.05
0.16
0.71
0.27
0.02
0.02
0.06
0.16
0.79
0.19
0.02
0.02
0.10
0.16
0.80
0.18
0.02
0.02
0.11
0.16
0.79
0.19
0.02
0.02
0.10
0.16
0.83
0.15
0.03
0.02
0.13
0.16
SW
NW
SE
0.90
0.03
0.07
0.03
0.25
0.45
0.90
0.04
0.07
0.04
0.25
0.45
0.90
0.03
0.06
0.03
0.25
0.45
0.88
0.04
0.08
0.04
0.25
0.45
0.88
0.04
0.08
0.04
0.25
0.45
0.82
0.07
0.11
0.06
0.27
0.42
0.83
0.06
0.10
0.06
0.26
0.42
0.80
0.08
0.12
0.07
0.27
0.41
0.70
0.11
0.18
0.11
0.33
0.36
0.68
0.12
0.20
0.11
0.33
0.35
0.69
0.12
0.19
0.11
0.32
0.36
0.63
0.13
0.24
0.13
0.36
0.34
NE
SE
NW
0.68
0.15
0.17
0.07
0.28
0.26
0.69
0.15
0.16
0.07
0.29
0.26
0.69
0.15
0.16
0.07
0.28
0.26
0.69
0.15
0.16
0.07
0.29
0.26
0.69
0.15
0.16
0.07
0.29
0.26
0.71
0.14
0.15
0.07
0.29
0.27
0.71
0.14
0.15
0.07
0.29
0.27
0.71
0.14
0.15
0.07
0.29
0.27
0.72
0.14
0.15
0.07
0.29
0.27
0.72
0.13
0.14
0.07
0.30
0.28
0.72
0.14
0.14
0.07
0.30
0.28
0.73
0.13
0.14
0.07
0.30
0.28
SE
SW
NE
0.34
0.46
0.20
0.55
0.44
0.55
0.36
0.41
0.23
0.54
0.52
0.55
0.33
0.45
0.22
0.56
0.44
0.55
0.36
0.42
0.22
0.54
0.51
0.56
0.36
0.42
0.22
0.54
0.51
0.56
0.43
0.30
0.27
0.47
0.80
0.51
0.40
0.34
0.26
0.49
0.65
0.51
0.42
0.31
0.27
0.48
0.73
0.51
0.42
0.27
0.30
0.47
0.82
0.46
0.43
0.23
0.34
0.46
1.00
0.44
0.41
0.27
0.32
0.47
0.81
0.44
0.40
0.27
0.33
0.49
0.81
0.44
NW
SW
NE
SE
15
2
Table C.3. Sensitivity analyses of the instantaneous fishing mortality rates (F) to different scenarios about the tag shedding rate
(  ) and the tag reporting rates (  ) of major fisheries. The different scenarios investigated are identified in Table 4.4.
Model case
cv6
M7
cv7
M8
cv8
M9
cv9
M10
cv10
M11
cv11
M12
cv12
0.028
0.003
0.001
0.049
0.018
0.000
0.001
0.001
0.100
0.11
0.29
0.55
0.13
0.25
0.66
1.01
0.08
0.03
0.00
0.00
0.05
0.01
0.00
0.00
0.00
0.09
0.11
0.29
0.55
0.13
0.25
0.66
1.01
0.08
0.05
0.01
0.00
0.05
0.02
0.00
0.00
0.00
0.13
0.11
0.29
0.55
0.13
0.25
0.66
1.01
0.08
0.03
0.00
0.00
0.03
0.02
0.00
0.00
0.00
0.08
0.11
0.29
0.54
0.12
0.25
0.66
1.01
0.09
0.03
0.00
0.00
0.05
0.02
0.00
0.00
0.00
0.10
0.11
0.29
0.55
0.12
0.25
0.66
1.01
0.09
0.03
0.00
0.00
0.05
0.01
0.00
0.00
0.00
0.09
0.11
0.29
0.55
0.12
0.25
0.66
1.01
0.08
0.06
0.01
0.00
0.05
0.02
0.00
0.00
0.00
0.14
0.11
0.29
0.54
0.12
0.25
0.66
1.00
0.08
0.48
0.35
0.47
0.40
1.05
0.57
0.22
0.002
0.004
0.007
0.013
0.002
0.003
0.032
0.48
0.35
0.47
0.40
1.05
0.57
0.22
0.00
0.00
0.00
0.01
0.00
0.00
0.02
0.48
0.35
0.47
0.40
1.04
0.57
0.23
0.00
0.00
0.01
0.02
0.00
0.00
0.04
0.49
0.35
0.47
0.41
1.04
0.57
0.23
0.01
0.00
0.01
0.03
0.00
0.00
0.06
0.50
0.36
0.47
0.42
1.04
0.57
0.26
0.01
0.01
0.01
0.03
0.01
0.01
0.07
0.51
0.36
0.47
0.42
1.04
0.57
0.26
0.01
0.01
0.01
0.01
0.00
0.01
0.04
0.50
0.36
0.48
0.42
1.04
0.57
0.24
0.02
0.01
0.02
0.04
0.01
0.01
0.10
0.52
0.37
0.48
0.43
1.04
0.58
0.27
0.11
0.04
0.02
0.01
0.03
0.19
0.29
0.57
0.65
1.10
0.91
0.25
0.113
0.037
0.016
0.010
0.029
0.204
0.29
0.57
0.65
1.10
0.91
0.25
0.11
0.02
0.02
0.01
0.03
0.19
0.29
0.57
0.65
1.09
0.91
0.25
0.11
0.04
0.02
0.01
0.03
0.20
0.29
0.57
0.65
1.09
0.91
0.25
0.07
0.04
0.02
0.01
0.03
0.16
0.28
0.57
0.64
1.09
0.90
0.27
0.13
0.04
0.02
0.01
0.03
0.22
0.28
0.57
0.65
1.09
0.90
0.25
0.13
0.02
0.02
0.01
0.03
0.21
0.28
0.57
0.65
1.09
0.90
0.25
0.13
0.03
0.02
0.01
0.03
0.22
0.28
0.57
0.65
1.09
0.90
0.25
0.04
0.11
0.00
0.02
0.17
0.53
0.54
0.26
0.71
0.44
0.052
0.111
0.000
0.019
0.182
0.53
0.54
0.26
0.71
0.44
0.05
0.05
0.00
0.02
0.12
0.50
0.54
0.27
0.69
0.40
0.05
0.11
0.00
0.02
0.18
0.49
0.53
0.27
0.69
0.41
0.04
0.12
0.00
0.02
0.18
0.44
0.52
0.28
0.67
0.40
0.08
0.12
0.00
0.03
0.22
0.44
0.52
0.29
0.67
0.37
0.07
0.05
0.00
0.02
0.15
0.44
0.52
0.28
0.67
0.35
0.09
0.12
0.00
0.03
0.23
0.43
0.52
0.30
0.66
0.36
Fishery
M1
cv1
M2
cv2
M3
cv3
M4
cv4
M5
cv5
M6
NA_sport_NWNA
US_comm_NWNA
Canada_comm_NWNA
Spain_comm_NWNA
Japan_comm_NWNA
Portugal_comm_NWNA
US_sci - NWNA
Pooled_comm_NWNA
mean_NWNA
0.03
0.00
0.00
0.02
0.02
0.00
0.00
0.00
0.07
0.11
0.29
0.55
0.13
0.25
0.66
1.01
0.09
0.03
0.00
0.00
0.05
0.02
0.00
0.00
0.00
0.09
0.11
0.29
0.55
0.13
0.25
0.66
1.01
0.09
0.03
0.00
0.00
0.05
0.01
0.00
0.00
0.00
0.08
0.11
0.29
0.55
0.13
0.25
0.66
1.01
0.09
0.05
0.00
0.00
0.05
0.02
0.00
0.00
0.00
0.13
0.11
0.29
0.55
0.13
0.25
0.66
1.01
0.08
0.05
0.00
0.00
0.05
0.02
0.00
0.00
0.00
0.13
0.11
0.29
0.55
0.13
0.25
0.66
1.01
0.08
US_comm_SWNA
Spain_comm_SWNA
Japan_comm_SWNA
Venezuela_comm_SWNA
Pooled_sport_SWNA
Pooled_comm_SWNA
mean_SWNA
0.00
0.00
0.00
0.01
0.00
0.00
0.02
0.47
0.34
0.47
0.40
1.05
0.56
0.24
0.00
0.00
0.00
0.01
0.00
0.00
0.02
0.47
0.35
0.47
0.40
1.04
0.57
0.23
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.47
0.34
0.47
0.40
1.05
0.57
0.23
0.00
0.00
0.01
0.01
0.00
0.00
0.02
0.48
0.35
0.47
0.40
1.05
0.57
0.22
0.00
0.00
0.01
0.01
0.00
0.00
0.02
Spain_comm_NENA
Japan_comm_NENA
Portugal_comm_NENA
Pooled_sport_NENA
Pooled_comm_NENA
mean_NENA
0.05
0.04
0.01
0.01
0.03
0.15
0.28
0.57
0.65
1.10
0.91
0.28
0.11
0.04
0.01
0.01
0.03
0.19
0.29
0.57
0.65
1.10
0.91
0.25
0.11
0.02
0.02
0.01
0.03
0.18
0.29
0.57
0.65
1.10
0.91
0.26
0.11
0.04
0.02
0.01
0.03
0.19
0.29
0.57
0.65
1.10
0.91
0.25
Spain_comm_SENA
Japan_comm_SENA
Venezuela_comm_SENA
Pooled_comm_SENA
Mean_SENA
0.02
0.11
0.00
0.02
0.15
0.53
0.54
0.27
0.70
0.46
0.04
0.11
0.00
0.02
0.17
0.53
0.54
0.27
0.70
0.44
0.04
0.04
0.00
0.02
0.11
0.53
0.55
0.30
0.70
0.42
0.04
0.11
0.00
0.02
0.17
0.53
0.54
0.26
0.71
0.44
15
3
154
Figure C.1. Inter-regional movements of blue sharks observed by the Cooperative
Shark Tagging Project (CSTP) in the North Atlantic Ocean.
155
100°
90°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
60°
50°
50°
40°
40°
30°
30°
20°
20°
10°
0°
Ü
90°
10°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
20°
Figure C.2. Blue shark tag-recoveries by the U.S.-sport fishery, archived by the
Cooperative Shark Tagging Program (CSTP) in the North Atlantic Ocean (19652004).
156
100°
90°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
60°
50°
50°
40°
40°
30°
30°
20°
20°
10°
0°
Ü
90°
10°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
20°
Figure C.3. Blue shark tag-recoveries by the U.S.-commercial fishery, archived by the
Cooperative Shark Tagging Program (CSTP) in the North Atlantic Ocean (19652004).
157
100°
90°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
60°
50°
50°
40°
40°
30°
30°
20°
20°
10°
0°
Ü
90°
10°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
20°
Figure C.4. Blue shark tag-recoveries by the Canada-commercial fishery, archived by
the Cooperative Shark Tagging Program (CSTP) in the North Atlantic Ocean (19652004).
158
100°
90°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
60°
50°
50°
40°
40°
30°
30°
20°
20°
10°
0°
Ü
90°
10°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
20°
Figure C.5. Blue shark tag-recoveries by the Japan-commercial fishery, archived by
the Cooperative Shark Tagging Program (CSTP) in the North Atlantic Ocean (19652004).
159
100°
90°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
60°
50°
50°
40°
40°
30°
30°
20°
20°
10°
0°
Ü
90°
10°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
20°
Figure C.6. Blue shark tag-recoveries by the Spain-commercial fishery, archived by
the Cooperative Shark Tagging Program (CSTP) in the North Atlantic Ocean (19652004).
160
100°
90°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
60°
50°
50°
40°
40°
30°
30°
20°
20°
10°
0°
Ü
90°
10°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
20°
Figure C.7. Blue shark tag-recoveries by the Venezuela-commercial fishery, archived
by the Cooperative Shark Tagging Program (CSTP) in the North Atlantic Ocean
(1965-2004).
161
100°
90°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
60°
50°
50°
40°
40°
30°
30°
20°
20°
10°
0°
Ü
90°
10°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
20°
Figure C.8. Blue shark tag-recoveries by the Portugal-commercial fishery, archived
by the Cooperative Shark Tagging Program (CSTP) in the North Atlantic Ocean
(1965-2004).
162
100°
90°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
60°
50°
50°
40°
40°
30°
30°
20°
20°
10°
0°
Ü
90°
10°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
20°
Figure C.9. Blue shark tag-recoveries by U.S.-scientific cruises, archived by the
Cooperative Shark Tagging Program (CSTP) in the North Atlantic Ocean (19652004).
163
100°
90°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
60°
50°
50°
40°
40°
30°
30°
20°
20°
10°
0°
Ü
90°
10°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
20°
Figure C.10. Blue shark tag-recoveries by the others -commercial fishery, archived by
the Cooperative Shark Tagging Program (CSTP) in the North Atlantic Ocean (19652004).
164
100°
90°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
60°
50°
50°
40°
40°
30°
30°
20°
20°
10°
0°
Ü
90°
10°
80°
70°
60°
50°
40°
30°
20°
10°
0°
10°
20°
Figure C.11. Blue shark tag-recoveries by the others-sport fishery, archived by the
Cooperative Shark Tagging Program (CSTP) in the North Atlantic Ocean (19652004)
4
4
Sport-N. America (1)
2
NW
0
4
0
4
Relative fishing effort
2
0
0
4
Com. - US (3)
Com. - Portugal (15)
2
2
2
0
0
0
0
4
4
Com. - Japan (11)
4
Com. - Venez. (17)
4
Sport - others (20)
Comm. - others (24)
4
2
2
2
2
2
0
0
0
0
0
0
8
4
4
4
4
Com. - Japan (12)
Com. - Portugal (16)
Sport - others (21)
Com. - others (25)
4
2
2
2
2
0
0
0
0
0
40
8
4
4
Com. - Spain (8)
SE
4
Com. - Japan (10)
2
Com. - Spain (6)
Com. - Spain (7)
NE
4
Com. - Spain (5)
Com. - others (23)
2
8
4
Com. - Canada (4)
2
Survey - US (19)
SW
4
Com. - US (2)
20
4
0
1960
Com. - Japan (13)
2
0
1980
2000
1960
Com. - Venez. (18)
2
0
1980
2000
1960
Com. - others (26)
0
1980
2000
1960
1980
2000
Years
Figure C.12. Relative fishing effort for each fishery.
16
5
166
VITA
Alexandre Aires-da-Silva was born in Maputo, Mozambique. He grew up in the city
of Lisbon, Portugal, his country of citizenship. He earned a Licenciatura (Bachelor)
degree in Marine Biology of Fisheries at Universidade do Algarve, Faro, Portugal
(1996). After that, he worked as a fishery scientist at Departamento de Oceanografia
e Pescas, Univerisdade dos Açores, in the Azores, Portugal (1996-2000). In 2000,
Alexandre enrolled as a Fulbright graduate student at the School of Aquatic and
Fishery Sciences, University of Washington, Seattle, Washington, U.S. He now takes
a Senior Scientist position with the Inter-American Tropical Tuna Commission, in La
Jolla, California, U.S. In 2008, Alex graduated from the University of Washington
with a Ph.D. from the School of Aquatic and Fishery Sciences.