Padronização do índice de abundância e avaliação do estoque de bonito listrado, (Katsuwonus pelamis Linnaeus, 1758), do Atlântico Ocidental
Abstract
In this paper data on catch, fishing effort and landings from the Brazilian baitboat
fishery, together with information on vessel characteristics were analyzed aiming to:
(a) describe the fishery and characteristics of the fishing fleet, analyze catch
composition and spatial and temporal distribution of catches, fishing effort and catch
rates of skipjack (Katsuwonus pelamys); (b) develop standardized indices of
abundance for skipjack using generalized linear models (GLM); and (c) apply a nonequilibrium
surplus production model for stock assessment of west Atlantic skipjack
through the ASPIC program version 5.0. Skipjack is the most important species
caught from Brazilian tuna fisheries; its catches comprise more then 50% of the total
tuna catches from this fishery. The fishing area is located in the south and southeast
regions of Brazil, from 20oS to 35oS, but fishing operations are carried out mainly
between 28oS and 34oS. The highest catch rates are recorded in the south region
during the first and fourth quarters and the smallest ones in the third quarter. Skipjack
landings are made in Rio de Janeiro, Santa Catarina and Rio Grande do Sul states,
with Santa Catarina being the most important landing place. The highest landings are
recorded during the summer months (February and March) and the smallest ones
during the winter months (August and September). The baitboats shows different
characteristics; length of vessels varies from 15 to 49.5 meters. Skipjack catch rates
from each vessel varies according with its size, which means that fishing power is a
function of vessel size. The frequency distributions of skipjack CPUE are highly
skewed with a relatively large proportion of zero observations. Standardization of
skipjack CPUE (catch per unit of effort) was performed through generalized linear
models, using delta-GLM methods, which involves fitting of two sub-models to the
data. A first sub-model was applied assuming the binomial error distribution for the
proportion of positive catches and a second sub-model was used for the positive
catches assuming a different error distribution. Two alternative distributions were
assumed for the positive catches, the lognormal and the Gamma distribution.
Deviance tables were performed to identify the best set of factors and interactions
that most adequately explained the observed variability in proportion of positive
catches and positive CPUE. Geographical distribution (fishing area) and sea surface
temperature together with the interaction year*GRT were the most important
explanatory effects for the occurrence of a non-zero catch. On the other hand GRT and season, together with year*area, year*season and year*GRT interactions
explained the most variability on the observed CPUE of positive catches. The
standardized indexes were estimated using Generalized Linear Mixed Models, in
which year, area, season, sea surface temperature, vessel length and GRT were
included as main explanatory fixed effect factors and all first order interactions with
year as random components. Delta-lognormal and delta-Gamma models showed a
good fit to the data but narrower confidence intervals and small coefficients of
variation were shown for standardized CPUE estimated by the delta-Gamma model.
Results from these analyses show the importance of the study of CPUE and factors
that have effect on its variations to understand the dynamics of this fishery. However
not all factors that have an effect on variations in skipjack CPUE were considered,
such as, bait species and amount of life bait, because this sort of information was
not present in the majority of data available for analysis. A great amount of the
information collected through logbooks are incomplete and imprecise, implying that
institutions responsible for the implementation of this data collection system are not
aware or do not recognize its importance as an instrument that makes possible to
get information of great value for the management and utilization of fishery
resources. Results of the stock assessment analysis provided an estimate of 26,930
MT for skipjack maximum sustainable yield, which is about 14% higher than catches
taken in 1998. This estimated yield may looks like realistic but other parameter
estimates seems to be unrealistic, suggesting that the west Atlantic skipjack stock is
in an overexploited state, which may not be true. Considering the uncertainties and
limitations about the data some of the parameters estimates may be imprecise.
Therefore, results from this analysis should be cautiously used to make decisions on
management measures for this fishery. A good stock assessment depends not only
on the adequacy of the model available for the analysis but also on the quality of the
data that the model is fitted to. Therefore, in order to have effective stock assessment
for skipjack in future, an efficient system for the collection of data from this fishery
shall be implemented. This data collecting system should include mechanisms for
data verification, such as observer programmes to monitor catch, effort and other
details of the fishing operations.