SCRS/2007/167 Collect. Vol. Sci. Pap. ICCAT, 62(5): 1602-1609 (2008)
STANDARDIZED CPUE OF BLUE SHARK CAUGHT BY SÃO PAULO TUNA
LONGLINERS OPERATING OFF SOUTHERN BRAZIL (1998-2006)
Bruno L. Mourato1, Humberto G. Hazin2, Alberto F. Amorim3,
Carlos A. Arfelli3, Fábio H. V. Hazin2
In the present study, a GLM approach, assuming a quasi-Poisson error distribution and log as link
function, was used with a view to generate a standardized CPUE series for the blue shark caught by the
tuna longliners based in Santos, São Paulo, off southern Brazil, from 1998 to 2006. The response
variable was the CPUE defined as number of fish caught per thousand hooks. The following factors
were tested in the analyses: year (9), month (12), area (5) and target (3). The levels of the factor “area”
were based on the spatial distribution of the fishing sets, while the levels of the factor “target” were
defined by a cluster analysis, as five areas; and cluster C1: other fishes; C2: blue shark; and C3:
swordfish. A stepwise approach was initially used to identify the variables and interactions with a
significant influence on the CPUE. The “target” and “area” were the main factors explaining the
observed CPUE variability (36.5% and 28.5%, respectively). The overall pattern of the standardized
CPUE indicates a relatively high stability, up to 2005. In 2006, however, there was a strong increase of
the CPUE. The reliability of the estimates and their potential usefulness for future stock assessment
were then discussed.
Dans la présente étude, une approche GLM, postulant une distribution d’erreur quasi-Poisson et une
valeur logarithmique comme fonction de lien, a été utilisée afin de générer une série de CPUE
standardisée pour le requin peau bleue capturé par les palangriers thoniers basés à Santos, São Paulo,
au sud du Brésil, de 1998 à 2006. La variable de réponse était la CPUE définie comme le nombre de
poissons capturés par mille hameçons. Les facteurs suivants ont été testés dans les analyses: année (9),
mois (12), zone (5) et cible (3). Les niveaux du facteur “zone” se basaient sur la distribution spatiale
des opérations de pêche, alors que les niveaux du facteur “cible” étaient définis comme cinq zones par
une analyse de regroupement : regroupement C1: autres poissons ; C2: requin peau bleue et C3:
espadon. Une approche pas à pas a tout d’abord été utilisée pour identifier les variables et les
interactions ayant une grande influence sur la CPUE. La “cible” et “zone” étaient les principaux
facteurs expliquant la variabilité observée de la CPUE (36,5% et 28,5%, respectivement). Le schéma
global de la CPUE standardisée indique une stabilité relativement élevée jusqu’en 2005. En 2006, la
CPUE a toutefois fortement augmenté. La fiabilité des estimations et leur utilité potentielle pour la
future évaluation des stocks ont ensuite été discutées.
En el presente estudio se utilizó un enfoque GLM asumiendo una distribución de error quasi-Poisson y
una función de vínculo logarítmica con el fin de generar una serie de CPUE estandarizada para la
tintorera capturada por los palangreros atuneros con base en Santos, São Paulo, al Sur de Brasil,
desde 1998 a 2006. La variable respuesta fue la CPUE definida como el número de peces capturado
por mil anzuelos. En los análisis se probaron los siguientes factores: año (9), mes (12), área (5) y
objetivo (3). Los niveles del factor “área” se basaban en la distribución espacial de los lances
pesqueros, mientras que los niveles del factor “objetivo” fueron definidos por un análisis de
conglomerados como cinco áreas; y conglomerado C1: otros peces; C2: tintorera y C3: pez espada.
Inicialmente se utilizó un enfoque gradual para identificar las variables e interacciones con una
influencia significativa en la CPUE. “Objetivo” y “área” eran los factores principales que explicaban
la variabilidad observada en la CPUE (36,5% y 28,5% respectivamente). El patrón global de la CPUE
estandarizada indica una estabilidad relativamente alta, hasta 2005. En 2006, sin embargo, se produjo
un fuerte aumento de la CPUE. Se discutió la fiabilidad de las estimaciones y su posible utilidad para
futuras evaluaciones de stock.
M.Sc. Student- Instituto de Pesca/ APTA/ SAA, Santos- SP, FAPESP Schorlarship (firstname.lastname@example.org).
Departamento de Pesca e Aqüicultura/ UFRPE (email@example.com, firstname.lastname@example.org), Recife- PE, Brasil.
Instituto de Pesca/ APTA/ SAA, Santos- SP, (email@example.com, firstname.lastname@example.org)
Catch/effort, blue shark, catchability, long lining, tuna fisheries
The tuna longline fishery based in Santos, São Paulo State, southeast Brazil, began in the late 1950s, by
chartered Japanese longliners, later followed by Brazilian boats. After having reached 20 boats in 1998, the fleet
was then reduced to 14 longliners, in 2000 (Amorim et al. 2002). Although the tuna longline fleet based in
Santos regularly catches many shark species (Amorim et al. 1998), the blue shark has always been, by far, the
most abundant one. The blue shark nominal CPUE for this fleet was examined by Amorim (1992) and Amorim
et al. (1998). The Santos fleet operates over a broad geographical area, off Southern Brazil, and have targeted
mainly swordfish and blue shark, for 1971-2006 period (Mourato et al. unpublished).
Information about changes, along time, in the relative abundance of fish stocks is essential to allow a proper
assessment of its condition, in the lack of which the adoption of management and conservation measures,
capable of assuring the long term sustainability of the resource, becomes impossible. To that purpose, the catch
per unit of effort (CPUE) has been extensively employed, although their use as an index of relative abundance
has been questioned by a number of authors (Ricker 1975; Fonteneau 1998; Bigelow et al. 1999; Fréon and
Misund 1999), since the factors influencing the relationship between CPUE, fishing effort and actual abundance
are various and commonly not linearly distributed, easily leading to interpretative errors (Fonteneau 1998;
Maury et al. 2001). In order to compensate for the several factors that directly or indirectly influence CPUE,
such as the spatial distribution of the stock, seasonal fluctuation of abundance, fishing methods and gear, etc, a
common approach has been to use a Generalized Linear Model. One of the main variables influencing the
CPUE, however, is the targeting strategy, a factor that is generally quite subtle and very difficult to be directly
incorporated into the models. In order to overcome such a difficulty, the grouping of sets, based on a cluster
analysis, has been done by several authors (Gaertner et al. 1998; He et al. 1997; Hazin et al. 2007), as an
indication of the targeting strategy. In the present work, this approach was used with a view to generate a
standardized CPUE series for the blue shark caught by the tuna longliners based in Santos, from 1998 to 2006.
The reliability of the estimates and their potential usefulness for future stock assessment were then discussed.
2. Material and methods
The catch and effort data analyzed in the present paper were obtained from the logbooks of tuna longliners based
in São Paulo, made available by the Laboratório de Referência em Controle Estatístico da Produção Pesqueira
Marinha do Instituto de Pesca/ SAA/ SP, through ProPesq® system (Ávila-da-Silva et al. 1999). The logbook
data included a total of 6,486 sets, from 1998 through 2006, containing vessel identification, fishing location,
setting and retrieval time of the longline, number of hooks deployed and number of fish caught by species. The
blue shark was caught in 6,026 sets (93%).
The relative indices of abundance for blue shark were estimated by a generalized linear model (GLM) approach,
assuming a quasi-Poisson error distribution and log as link function. The response variable was the CPUE
defined as number of fish caught per thousand hooks. The following factors were tested in the analyses: year (9),
month (12), area (5) and target (3). The levels of the factor “area” were based on the spatial distribution of the
fishing sets, while the levels of the factor “target” were defined by a cluster analysis (Mourato et al.
unpublished), as follows: Area: 1 > 20°S; 2: 20°- 30°S / <40°W; 3: >30°S / <40°W; 4: 20°- 30°S / >40°W; and
5: >30°S / <40°W (Figure 1); and cluster: C1: other fishes; C2: blue shark; and C3: swordfish. A stepwise
approach was initially used to identify the variables and interactions with a significant influence on the CPUE.
Since all variables were considered significant, only those representing more than 2% of the deviance were
included in the final model. The diagnostic plots described by Ortiz and Arocha (2004) were run to evaluate the
fitness of the models.
3. Results and discussion
The deviance analysis (Table 1) indicated that the “target” and “area” were the main factors explaining the
observed CPUE variability (36.5% and 28.5%, respectively). The high level of influence of the target factor,
inferred by the cluster analysis (see more details in Mourato et al. 2008), found in the present work, was also
observed in other papers for the Brazilian longliner fleet (Hazin et al. 2007; Mourato 2007).
Other factors included in the model were year and month, and the interactions year*month and year*area. All
terms included in the model were significant at p < 0.001 levels (Table 1). The overall deviance explained by the
fit model was 62.2%, indicating a reasonably good adherence to the model, particularly for data that were not
aggregated (i.e. set by set data) (eg. Punt et al., 2000). The distribution of residuals (Figures 2 and 3) also
provided a good fit.
The 95% confidence intervals of the standardized CPUE series were relatively narrow (Table 2, Figure 4),
again, considering the high variance usually associated to CPUE series estimated from set by set data, with no
aggregation. The trend was also very close to the one shown by the nominal CPUE (Figure 5). The overall
pattern of the standardized CPUE indicates a relative high stability, up to 2005. In 2006, however, there was a
strong increase of the CPUE, a trend that can’t be properly explained by the present data. It is likely, however,
that a combination of factors related to the targeting strategy, might be influencing the CPUE, in a way that the
GLM was not capable of detecting.
Mourato et al. (unpublished) showed that both oceanographic conditions, such as the seasonal displacement of
the Subtropical Convergence Zone, off southern Brazil, as well as differences in spatial distribution of the two
main target species of this fisheries, i.e. blue shark and swordfish, accounted for much of the seasonal and yearly
changes of catch composition throughout the studied period, with a clear trend of the fishing fleet, in recent
years, to move more to the southeastward area of the fishing ground (sub-area 5), where the blue sharks are more
abundant. They concluded that the blue shark importance in this fishery has been growing progressively along
the years. Consequently, such a trend, together with local changes in untracked oceanographic conditions, which
might have resulted in a higher availability of blue sharks to Santos longliners, might, at least partly, explain the
sudden increase in blue shark CPUE in 2006. Similarly, it could also explain the sudden drop of the swordfish
CPUE, also detected in 2006, for the same fishery (Hazin et al. 2008). Considering, however, that in both cases
the fishing area and the targeting strategy were included in the GLM, it is not clear why the model failed to
properly compensate those factors. A similar problem, resulting from the interaction between the CPUE of blue
shark and swordfish, in relation to the targeting strategy, has been described by Mejuto and De la Serna (2000).
A comparison between the standardized CPUE found in the present paper with another, longer series, built with
monthly aggregated landing data from the same fishery (Mourato et al., 2007), shows a somewhat similar trend,
as it also does another CPUE series, based on set by set data, from the entire Brazilian tuna longline fleet,
including data from Santos longliners (Hazin et al. 2008) (Figure 6).
The authors wish to thank to Marcelo R. Souza for the data processing and the Fundação de Amparo à Pesquisa
do Estado de São Paulo (FAPESP) for the M.Sc. scholarship. The present work was made possible also by funding
from the Secretaria Especial de Aqüicultura e Pesca da Presidência da Republica do Brasil.
AMORIM, A.F. 1992. Estudo da biologia, pesca e reprodução do cação azul, Prionace glauca, L. 1758,
capturado no sudeste e sul do Brasil. Tese de Doutorado. Universidade Estadual Paulista, Rio Claro, São
Paulo, Brasil. 176 p.
AMORIM, A.F., C.A. Arfelli and L. Fagundes. 1998. Pelagic elasmobranchs caught by longliners off southern
Brazil during 1974-97: An overview. Marine and Freshwater Reserch, Collingwood, 49(7): 621-32.
ARFELLI, C.A., A.F. Amorim and L. Fagundes. 1997. Diagnose das pescarias dos atuneiros de Santos (1971-
1995). In: Congresso Latino-americano sobre Ciências do Mar, 7, 22 a 26 set., Santos, 1997. Resumos
Expandidos... Santos, IOUSP e Associação Latino-americana de Investigadores em Ciências do Mar.p. 48-
ÁVILA-DA-SILVA, A.O., M.H. Carneiro and L. Fagundes. 1999. Sistema gerenciador de banco de dados de
controle estatístico de produção pesqueira marítima - Propesq. IN: Anais do XI CONBEP and I CONLAEP,
Recife, Brazil (17-21/01/1999). 2: 824-832.
BIGELOW, A.K., C.H. Boggs, He, X. 1999. Envirovimental effects on swordfish and blue sharks catch rates in
the U.S. North Pacific longline fishery. Fish. Oceanogr. 8, 178-198.
FONTENEAU, A. 1998. Atlas des pêcheries thonières tropicales: captures mondiales et environnement.
FRÉON, P. and O.A. Misund. 1999. Dynamics of pelagic fish distribution and behaviour: effects on fisheries and
stock assessment. In: Science, B.s (Ed.), Fishing News Books. Oxford, London, p. 348p.
GAERTNER, D., J.C. Gaertner, J. Marcano and M. Pagavino. 1998. Multivariate analysis of relationships
between tuna catches and fishing strategies. Application to the Venezuelan purse seiners in the Caribbean
Sea. ICCAT Tuna symposium. Collect. Vol. Sci. Pap. ICCAT, 50(2): 628–636.
HAZIN, F.H.V., H.G. Hazin, F.C. Carvalho, C. Wor and P. Travassos. 2008. Standardization of CPUE series of
Prionace glauca and Isurus oxyrinchus caught by Brazilian longliners in the Western South Atlantic Ocean,
from 1978 to 2006. Collect. Vol. Sci. Pap. ICCAT, Madrid, 62, in press. (SCRS/2007/090)..
HAZIN, H.G., F.H.V. Hazin, P. Travassos, F.C. Carvalho, K. Erzini. 2007. Fishing strategy and target species of
the Brazilian tuna longline fishery, from 1978 to 2005, inferred from cluster analysis. Collect. Vol. Sci. Pap.
ICCAT, 60(6): 2029-2038.
HAZIN, H.G., B.L. Mourato, A.F. Amorim, C.A. Arfelli and F.H.V. Hazin. 2008. Standardized CPUE of
swordfish (Xiphias gladius) caught by tuna longliners operating off Southern Brazil (1998-2006). Collect.
Vol. Sci. Pap. ICCAT, Madrid, 62, in press (SCRS/2007/158).
HE, X., K.A. Bigelow and C.H. Boggs. 1997. Cluster analysis of longline sets and fishing strategies within the
Hawaii-based fishery. Fish. Res., 31: 147-158.
ICCAT. 2005. Report of the 2004 Inter-sessional Meeting of the ICCAT Sub-Committee on By-catches: Shark
Stock Assessment. Tokyo, Japan, 14-18 de junho, 2004. Collect. Vol. Sci. Pap. ICCAT, 58(3): 799-890.
MAURY, O., D. Gascuel, F. Marsac, A. Fonteneau and A.L. De Rosa. 2001. Hierarchical interpretation of
nonlineay relationships linking yellowfin tuna (Thunnus albacares) distribution to the environment in the
Atlantic Ocean. Can. J. Fish. Aquatic Sci. 58, 458-469.
MEJUTO, J. and J.M. de la Serna. 2000. Standardized catch rates by age and in biomass for the North Atlantic
swordfish (Xiphias gladius) from the Spanish longline fleet for the period 1983-1998 and bias produced by
changes in the fishing strategy. Collect. Vol. Sci. Pap. ICCAT, 51(5): 1387-1410.
MOURATO, B.L. 2007. Padronização da captura por unidade de esforço de espadarte, Xiphias gladius L. 1758 e
de tubarão-azul, Prionace glauca (L., 1758) capturados pela frota atuneira brasileira no oceano Atlântico.
Master Thesis. Programa de Pós-graduação em Aqüicultura e Pesca, Instituto de Pesca/SAA/SP, Brazil, 91p.
MOURATO, B.L., A.F. Amorim and C.A. Arfelli. 2007. Standardized catch rate blue shark (Prionace glauca)
caught by Santos longliners off southern Brazil (1984-2005). Collect. Vol. Sci. Pap. ICCAT, 60(2): 577-587.
MOURATO, B.L., C.A. Arfelli and A.F. Amorim. 2004. Análise da captura de anequim, Isurus oxyrinchus,
pelos atuneiros santistas, em função do direcionamento da pesca. In: Congresso Brasileiro de Pesquisas
Ambientais e Saúde, 4, 18 a 21 jul., Santos, 2004. Anais CBPAS 2004. p. 206-9.
MOURATO, B.L., H.G. Hazin, A.F. Amorim, C.A. Arfelli and F.H.V. Hazin. unpublished. Dynamic and fishing
strategy of Santos longline fleet off the southeastern Brazilian coast (1998-2006).
ORTIZ, M. and F. Arocha. 2004. Alternative error distributions models for standardization of catch rates of non-
target species from a pelagic longline-fishery: billfish species in the Venezuelan tuna longline fishery. Fish.
Res. 70, 275-294.
PUNT, A.E., T.I. Walker, B.L. Taylor and F. Pribac. 2000. Standardization of catch and effort data in a spatially-
structured shark fishery. Fish. Res. 45: 129–145.
RICKER, W.E., 1975. Computation and interpretation of biological statistics of fish population. Bull. Fish. Res.
Bd. Can. 191, 1-382.
Table 1. Deviance analysis of the fitted model for the standardization of blue shark CPUE caught by the Santos
tuna longline fleet, during 1998-2006.
Explained Explained by the
Df Deviance Resid. Df Resid. Dev. Pr (Chi) Deviance Model
NULL 6,446 91,298.5
month 11 6,142 6,435 85,156.3 0.00000 10.8% 6.7%
year 8 6,337 6,427 78,819.6 0.00000 11.2% 13.7%
area 4 16,199 6,423 62,621.0 0.00000 28.5% 31.4%
target 2 20,744 6,421 41,877.2 0.00000 36.5% 54.1%
year*month 88 6,116 6,333 35,761.6 0.00000 10.8% 60.8%
year*area 31 1,283 6,302 34,478.4 0.00000 2.3% 62.2%
Table 2. Standardized CPUE, with the respective standard error, nominal CPUE and scaled values of the fitted
model for the blue shark caught by the Santos tuna longliner during 1998-2006.
Standardized Nominal Scaled Standardized Scaled Nominal
CPUE SE CPUE CPUE CPUE
1998 14.02 1.54 9.44 0.79 0.59
1999 16.18 1.61 7.70 0.91 0.48
2000 11.04 1.52 10.44 0.62 0.65
2001 19.91 1.84 19.96 1.12 1.24
2002 19.72 2.39 17.10 1.11 1.06
2003 16.67 2.32 13.23 0.94 0.82
2004 13.26 1.48 12.85 0.75 0.80
2005 12.98 1.68 15.02 0.73 0.93
2006 36.26 4.03 38.90 2.04 2.42
Figure 1. Division of fishing ground of Santos tuna longliners into 5 sub-areas, following Mourato et al.
(unpublished). 1: > 20°S; 2: 20°- 30°S/ < 40°W; 3: > 30°S/ < 40°W; 4: 20°- 30°S/ > 40°W; 5: > 30°S/ < 40°.
Figure 2. Distribution of residual vs fitted values of the model for the standardization of blue shark CPUE
caught by the Santos tuna longliners, during 1998-2006.
Figure 3. Histogram of residuals of the fitted model for the standardization of blue shark CPUE caught by the
Santos tuna longliners, during 1998-2006.
Figure 4. Standardized CPUE (n°/ 1,000 hooks) of blue shark caught by Santos tuna longliners, between 1998
Figure 5. Scaled CPUE (n°/ 1,000 hooks) of blue shark caught by Santos tuna longliners, between 1998 and
Figure 6. Comparison between the standardized CPUE of blue shark caught by Santos tuna longliners, estimated
from log books (present analysis) and from monthly aggregated landing data (Mourato et al. 2007) and from the
entire Brazilian longline fleet (Hazin et al. 2008).