Page 12 - 10.1.1.498.8230

1310 C. Ravier and J.-M. Fromentin
be used, but 15 years is probably too short to describe
synchrony between long-term fluctuations. Therefore,
correlograms were also computed on time-series having
at least 50 years in common. Original time-series showed
significant spatial autocorrelation over the whole area
studied (Figure 7c). Whatever the distance between them
geographically, time-series were significantly positively
correlated, with mean r of 0.40–0.51 within groups
(traps from either the eastern Mediterranean area, or
from the western Mediterranean/adjacent Atlantic), and
mean r of 0.18–0.42 between groups (traps from
Sardinian-Sicilian-Tunisian group against traps from the
Gibraltar area). The correlogram on detrended time-
series showed that they were only positively correlated
up to 200 km apart (Figure 7d). Therefore, correlograms
on long-term series emphasized the synchrony between
long-term trends and confirmed that short-term
variability was only structured at a scale <200 km.
In conclusion, long-term fluctuations in trap catches
appeared significantly synchronous all around the
western Mediterranean and adjacent North Atlantic,
whereas the year-to-year fluctuations were only
synchronous on a local scale.
Discussion
Data
Mapping of the studied sites reveals strong dis-
parity in the location of traps around the western
Mediterranean and adjacent North Atlantic, which
does not result from incomplete data collection. The
traps constitute important fisheries along the southern
coasts of Portugal and the Atlantic and Mediterranean
coasts of Spain and Morocco, and along the Tunisian,
Sicilian, Sardinian, and western continental Italian
coasts, i.e. along the migration routes of spawning
bluefin tuna (Berthelot, 1869; Farrugio, 1981). Else-
where, along the Algerian, French, and Adriatic coasts,
traps were of minor importance and only caught juv-
eniles and/or sporadically some adults, so such informa-
tion was not used in this study. Graphic representation
of the data set also reveals temporal disparity in the
availability of trap data. This mainly reflects the avail- Doumenge, 1998) highlights the similarity between
ability of historical records and the fact that some of
the catches were not recorded (e.g. in Spain during the
19th century) or were destroyed (e.g. Sicilian data in
the mid-19th century). Nevertheless, we believe that the
data presented in this study are trustworthy and of
good quality, because most historical documents give
extensive information about the catches, including
mean weight of tuna, dates of the ‘‘matanza’’ (collect-
ing tuna from the ‘‘death room’’), environmental con-
ditions, and number of fishers and boats.
Fluctuations in trap catches as a proxy for
fluctuations in abundance
The results of the correlograms showed that long-term
trends in trap catches were synchronous over more than
2000 km, from southwest Portugal to east of Sicily and
from north of Sardinia to south Tunisia. Bluefin tuna
migrate annually from the North Atlantic to reproduce
in two main areas of the western Mediterranean,
hugging the coast on their way to their spawning sites
(Aristotle, 4th century B.C.; Mather et al., 1995). If the
traps did not catch a representative proportion of the
bluefin tuna population, there would have been no
reason for the apparent long-term fluctuations to be
synchronous around the whole western Mediterranean
basin and along the southern coasts of Spain and
Portugal. This result would support the hypothesis of
Fromentin et al. (2000a) that long-term fluctuations
in trap catches are a good proxy for variation in
population abundance.
It may be argued that synchronous historical social
and economic events (wars, plagues, changes in trade
and industry) could explain such a pattern. However,
such events are rarely (if ever) common and/or synchro-
nous between the different countries of the study area,
and those in common were too short to explain the
long-term fluctuations. Moreover, there is no informa-
tion in the literature that indicates the possible influence
of historical social and economic events on trap fisheries
over decades or centuries (Hjort, 1914; Hersart de la
Villemarque´, 1995). Nevertheless, these events can affect
short-term variability, which in turn may be related to
local and/or regional variations in environmental condi-
tions, fish behaviour (tuna migrating farther offshore in
one area than in another in a given year), and fishing
effort (accidental damage to nets, variation in the ability
of the ‘‘Reis’’, the lead fisher). All these short-term
variations, which tend to be on a local scale only,
could explain why the year-to-year fluctuations appear
synchronous within areas <200 km.
The literature (Berthelot, 1869; Pavesi, 1889; de
Braganc¸a, 1899; de Buen, 1928; Thomazi, 1947;
the traps over the whole studied area and the absence
of meaningful technical modification until the
20th century. However, notable changes during the
20th century resulted in increased human perturbation
and crucial changes to fishing gear and technology after
the Second World War. The total number of active traps
varied slightly over time, though most operated from the
16th century. The Spanish bluefin tuna fishery was,
however, less stable than elsewhere. First, beach-seines
were used until the early 19th century and then replaced
by traps. Second, the aristocracy surrendered its rights
on traps around 1815, traps being auctioned and
scattered between numerous owners. From 1804 to