JOURNAL OF PLANKTON RESEARCH j VOLUME 30 j NUMBER 4 j PAGES 467 – 481 j 2008

Distribution, growth and survival of

anchovy larvae (Engraulis encrasicolus L.) in
relation to hydrodynamic and trophic
environment in the Bay of Biscay
UNAI COTANO*, XABIER IRIGOIEN, EGOITZ ETXEBESTE, PAULA ÁLVAREZ, LUCÍA ZARAUZ, JULIEN MADER
AND LUIS FERRER
AZTI-TECNALIA, MARINE RESEARCH DIVISION, C/ HERRERA KAIA, PORTUALDEA Z.G., 20110 PASAIA, GIPUZKOA, SPAIN

*CORRESPONDING AUTHOR: [email protected]

Downloaded from http://plankt.oxfordjournals.org/ by guest on August 13, 2015

Received November 6, 2007; accepted in principle January 14, 2008; accepted for publication January 18, 2008; published online January 31, 2008

Corresponding editor: Roger Harris

European anchovy larvae were sampled in June 2004 in the south-eastern part of the Bay of
Biscay (limits: 48W, 468N). Larval distribution was analysed and compared with spawning dis-
tribution in previous weeks. Patches of old larvae were found around the French and Cantabrian
shelf (CS) edges and even in oceanic waters, far from the main spawning areas, whereas the young-
est larvae appeared in the middle French shelf, associated with adult spawning locations. The
analysis of environmental conditions showed significant differences from coastal to oceanic stations
as well as in areas under the influence of river discharges. A principal component analysis carried
out on temperature, stratification, chlorophyll a and plankton concentration distinguished five areas:
Gironde, Northern French Shelf, Southern French Shelf, CS and Oceanic area. Growth and mor-
tality rates were estimated for anchovy larvae in these contrasting areas. Advective processes due to
dominant north-easterly winds during summer influence larval distribution in the Bay of Biscay
after spawning. However higher food concentration on the shelf did not result in significant differ-
ences in the growth rate. Mortality rates seem to be negatively related to larval size, although this
effect may be partially due to advection. This is the first attempt to study the effect of environ-
mental factors on anchovy larval distribution and mortality in the Bay of Biscay.

the “match– mismatch” hypothesis (Cushing, 1975) or

I N T RO D U C T I O N
the larval retention processes (Iles and Sinclair, 1982)
As for any organism reproducing with an r strategy, and (ii) top-down hypotheses based on predation (e.g.
early stage mortality is a key factor in fish population Bailey and Houde, 1989). Actually, bottom-up and
dynamics and one of the main causes of population top-down effects might well be linked, and the match–
fluctuations (Hjort, 1914; Leggett and Deblois, 1994). mismatch hypothesis already recognizes that well-fed
Understanding factors controlling recruitment variabil- larvae are more likely to be able to escape predators
ity of small pelagic fish species is one of the main pro- (Cushing, 1990), and on the other hand, at low food
blems in fisheries research (see review in Leggett and concentrations, larvae will have to be more active to
Deblois, 1994). find food and therefore increase the risk of detection by
Hypotheses proposed to explain variability in larval predators (McNamara and Houston, 1987). In any case,
survival rates and associated recruitment can be separ- whether bottom-up, top-down or mixed, the environ-
ated into two large groups: (i) bottom-up hypotheses mental factors should translate into a spatial variability
such as starvation, the “critical period” (Hjort, 1914), that has been scarcely investigated. Recruitment is

doi:10.1093/plankt/fbn011, available online at www.plankt.oxfordjournals.org

# The Author 2008. Published by Oxford University Press. All rights reserved. For permissions, please email: [email protected]
 JOURNAL OF PLANKTON RESEARCH j VOLUME 30 j NUMBER 4 j PAGES 467 – 481 j 2008

generally measured as a single event for a whole stock, Oceanographic water column data were collected at
where in reality it is most likely to be the sum of a each sampling station using a CTD profiler equipped
number of events distributed spatially and temporally. with a fluorescence sensor. The fluorometer voltage was
The spawning area of anchovy in the Bay of Biscay is converted to chlorophyll a (Chl a) concentration values
well known since an annual evaluation of the biomass is based on standard calibration procedures using
carried out using the daily egg production method extracted Chl a at different stations. A stratification
(DEPM). The peak spawning occurs between May and index was calculated as the difference in density
June in productive-enriched areas such as river plumes, between 0 and 20 m. Current profiles down to a
especially zones under the influence of the large French maximum depth of 100 m were measured throughout
rivers (Gironde and Adour, and secondarily the smaller the cruise, by means of a vessel-mounted RDI 300 kHz
numerous Cantabrian rivers), shelf break fronts and ADCP (acoustic current Doppler profiler). Tidal cur-
oceanic gyres (Motos et al., 1996). Uriarte et al. (Uriarte rents for eight primary harmonic constituents (M2, S2,
et al., 2001) proposed that a significant proportion of the N2, K2, K1, O1, P1, Q1) have been removed using the
larvae hatched in these areas were advected off the shelf OSU TOPEX/Poseidon Global Inverse Solution
since small juveniles are often found off the shelf. But version 5.0 (TPXO.5). So the general water circulation
larval distribution has hardly been studied in the Bay of in the area was obtained with the calculated residual

Downloaded from http://plankt.oxfordjournals.org/ by guest on August 13, 2015

Biscay (Arbault and Lacroix, 1977). Thus, one of the currents. Only upper layers (up to 40 m deep) are pre-
objectives of this study was to map the post-spawning sented because anchovy larvae remain mainly in the
distribution of anchovy larvae in the Bay of Biscay as a upper 30 m layer (Palomera, 1991). Owing to technical
function of environmental factors. problems, current velocities could only be obtained for
Borja et al. (Borja et al., 1998) and Allain et al. (Allain the French shelf area.
et al., 2001) related recruitment of anchovy to physical Ichthyoplankton samples were collected with oblique
processes such as upwelling, turbulence and stratifica- tows to a maximum depth of 70 m, or to 5 m above
tion. Although growth-selective mortality has been the bottom in shallower waters, using a double Bongo
suggested (Allain et al., 2003), studies measuring growth net with 40 cm mouth diameter and 335 mm mesh
rates and survival of larval anchovy in the Bay of Biscay size net. Previously calibrated flowmeters were fitted to
have not been made. Therefore, the main aim of the the nets to determine the filtered volume. The vessel
present study is to determine, through otolith analyses, speed was set at around 2 – 2.5 knots (kn, 1 knot
larval growth rates in different environmental areas of = 0.5 m s21) and retrieving speed was 20 m min21.
the south-eastern Bay of Biscay as a function of environ- The samples were preserved in buffered 80% ethanol
mental conditions in order to understand the factors until analysis.
which control growth and survival. As an initial hypoth- Micro- and mesozooplankton sampling was carried
esis, it could be expected that the retention of larvae in out with 5 L Niskin bottles at a depth of 3, 10 and
high productive areas close to the spawning centres 30 m, and vertical Pairovet net (2-Calvet nets) tows
would enhance growth and survival rates, whereas using 150 mm mesh size, respectively. Pairovet samples
advection of these larvae to oceanic waters would result were also employed to determine anchovy egg abun-
in lower growth rates and increase the mortality of dances. Egg abundance and distribution were compared
anchovy larvae. with that obtained during the peak spawning in the
DEPM survey in May.
The food availability of each sampling station was
estimated by measuring microplankton and mesozoo-
plankton abundances, which were estimated by using a
METHOD
FlowCAM (Flow Cytometer and Microscope) and a
high resolution scanner with image analysis software
Sampling and environmental (Plankton Visual Analyser, PVA), respectively. Besides
characterization total abundances, size structure was also taken into
Sampling was carried out in June 2004 (21st– 30th) account and different size classes were considered:
with the RV Investigador in the south-eastern part of the Microplankton abundances (number of particles per
Bay of Biscay (limited to 3834’W and 45838’N) over a mL) were determined for 25– 75, 75 –125 and 125–
grid of 85 stations distributed in transects perpendicular 250 mm equivalent spherical diameter (ESD) size-
to the coast (Fig. 1). Transects were 15 nautical-miles classes, whereas mesozooplankton was divided into less
(nm) apart and the stations were situated at 9 nm than 0.2 mm ESD, 0.2– 0.5, 0.5 –1, 1 – 1.5 and 1.5–
intervals. 2 mm ESD.

468
 U. COTANO ET AL. j DISTRIBUTION, GROWTH AND SURVIVAL OF ANCHOVY LARVAE

Downloaded from http://plankt.oxfordjournals.org/ by guest on August 13, 2015

Fig. 1. Study area showing the sampling station grid. Figures indicate the station number.

A principal component analysis (PCA) was carried out measured to the nearest 0.1 mm standard length (n .
so as to characterize different areas in relation to 4000). The standard length (LS) was corrected for
environmental features. To run this PCA, mean temp- shrinkage using Theilacker’s function (Theilacker,
erature in first 20 m, stratification, mean Chl a concen- 1980). Additionally, a larval aggregation index (Ia) was
tration in the first 20 m and number of plankton calculated following the formula described by Bez (Bez,
particles per mL on first 10 m were selected. Analysis of 2000) for larval densities. This index measures the prob-
variance (ANOVA) was further carried out to confirm ability that two larvae taken at random in the popu-
significant differences in environmental features between lation appear in the same sample.
the different areas selected by the PCA analysis (Table I).
P 2
i zi
Ia ¼ P 2 ð1Þ
Abundance and distribution S i zi

All anchovy eggs and larvae were sorted, identified and

counted, and abundances were standardized to the where zi is the larval density in ith station and S the
number of larvae per m2. Anchovy larvae were mean sampled surface.

469
 JOURNAL OF PLANKTON RESEARCH j VOLUME 30 j NUMBER 4 j PAGES 467 – 481 j 2008

Table I: Mean values of surface temperature (TS), surface salinity (SS), mean temperature in the upper
20 m (T20m), index of stratification (Istr), mean Chl a concentration in the upper 20 m (F20m), plankton
abundance in the upper 10 m (cell mL21), anchovy egg (egg m22) and larvae abundance (larvae m22)
and standard length (LS) for the five environmental areas (G, Gironde area; O, Oceanic area)
Oceanographic parameters Food parameters

Areas Ts Ss T20m Istr F20m Cell mL21 Eggs m22 Larvae m22 LS

G 17.2 a 33.8 a 12.4 a 1.85 a 2.35 a 12 316 a 63 a 20 a 5.8 a

NFS 19.6 a 34.0 b 17.2 c 0.85 b 1.08 b 8819 b 68 a 166 b 5.6 a
SFS 19.4 a 34.1 b 14.4 b 1.91 a 1.06 b 8788 b 48 a 87 c 6.2 a
CS 20.5 c 34.4 c 14.5 b 2.22 a 0.56 c 3944 c 25 a 14 a 9.5 b
O 20.2 b 35.0 c 18.3 d 0.75 b 0.26 d 2536 d 15 a 13 a 10.6 b

ANOVA, P , 0.01. The same letter denotes the absence of statistically significant differences (ANOVA, P , 0.01).

Growth and mortality regression of a log transformation of the equation.

Downloaded from http://plankt.oxfordjournals.org/ by guest on August 13, 2015

Sagittae otoliths were picked out and prepared for Age – frequency distribution was estimated by transform-
reading following the same methodology used in ing Ls data into age using the age-at-length relationship.
Cotano and Álvarez (Cotano and Álvarez, 2003) for This analysis was carried out for all larvae together and
mackerel larvae. The otoliths were measured and read for the larvae collected in each area. This approach
under a light microscope with immersion oil at a mag- assumes that the egg production for about 20 days (age
nification of 1000, using an image analysis system of the older larvae) was constant. The validity of this
(VISILOG/TNPC 3.2). For the discrimination of sub- assumption is difficult to test, the larvae hatched during
daily rings, the criterion applied by Palomera et al. the maximum spawning period for anchovy in the Bay
(Palomera et al., 1988) was followed. Otolith interpret- of Biscay, and one could assume a more or less constant
ation was carried out by two independent readers, individual egg production but at the same time the
recording the result only if agreement was reached, and population is being exploited, so the population egg
every otolith with a discrepancy exceeding two rings production is likely to decrease. Furthermore, applying
between both readers was immediately rejected. A total this approach to the different areas may produce biased
of 220 larvae were aged and an age-at-length relation- results because of advection from one area to other.
ship was established. This age-at-length relationship was In order to avoid the effect of advection between
further used to transform larval lengths to ages in days. areas on the calculation of mortality rates, the instan-
Larvae were grouped into 5-day groups for the analysis taneous mortality between every two consecutive ages
of larval distribution at age. In order to assess the influ- was calculated from equation (2) as:
ence of environmental conditions on larval growth a  
recent 5-day growth estimate (G5) was calculated as the loge ðNd Þ  loge ðNdþ1 Þ Nd
Zd ¼ ¼ loge ð3Þ
slopes of the length-at-age relationship divided into 5 ðd þ 1Þ  d Ndþ1
days periods.
Daily mortality rate of anchovy larvae was calculated where d is the age, Nd the number of larvae for age d
using the traditional instantaneous mortality method and Nd+1 the number of larvae in age d + 1. It is reason-
described by Ahlstrom (Ahlstrom, 1954) and Smith and able to assume constant egg production for 2 successive
Richardson (Smith and Richardson, 1977). The survival days. Furthermore, although advection might play a
curve (i.e. the descending limb of the age frequency dis- major role in the abundance of larvae in each area
tribution) of anchovy larvae was employed to calculate when considering a large age range, we assume that the
the mortality rate: effect of advection on the abundance of larvae of 2 con-
secutive days in each area is minor with respect to the
N ¼ aeZage ð2Þ processes inside the area. To check this last assumption,
we calculated the daily transport between the different
PCA defined areas using a circulation model (ROMS,
where N is the abundance of larvae (number m22), a is Shchepetkin and McWilliams, 2005). The model
a constant and Z is the instantaneous mortality coeffi- extended in latitude from 41800’N to 48800’N and in
cient (mortality rate). Z is calculated using linear longitude from 13830’W to 0830’W, with a mean

470
 U. COTANO ET AL. j DISTRIBUTION, GROWTH AND SURVIVAL OF ANCHOVY LARVAE

horizontal resolution of 6.6 km. The water column was minimum of 33.2, and the highest in oceanic waters
vertically divided in 32 sigma coordinate levels, concen- with a maximum of 35.5. Mean Chl a concentration in
trated at the surface. The surface forcing used in the the upper 20 m (F20m) ranged between 0.1 and
model was 6 h NCEP/NCAR re-analysis data (www. 2.9 mg L21 with the highest values being in the north-
cdc.noaa.gov) for the same period as the survey dur- eastern part of the study area. The abundance of micro
ation; 100 000 particles were homogeneously distributed and mesozooplankton was higher in the mouth of the
in each area, between 0 and 30 m and the exchange Gironde estuary, although mesozooplankton was also
with the other areas estimated in 24 h periods. Table II abundant along the French and Cantabrian shelf (CS)
presents the average daily exchange between areas for (Fig. 3). On the other hand, an increase in the average
the 20 days corresponding to the period. The average size of the zooplankton was observed in the external
daily contribution from one area to another was never part of the French shelf and in the CS (Fig. 4).
higher than 8% and therefore this is the magnitude of Current fields obtained from ADCP data showed pre-
the error due to advection expected in the abundances dominant south and south-westward currents on the
used to estimate mortality from consecutive days. The inner French shelf (less than 100 m deep), especially in
mortality estimates were not corrected with the trans- surface waters (from 0 to 30 m). A reduction in current
port estimates because circulation models have difficul- intensity was observed in the middle of the French shelf

Downloaded from http://plankt.oxfordjournals.org/ by guest on August 13, 2015

ties in areas with abrupt topographies such as the but again very strong south-westward current was
shelf-break. Therefore, we would be correcting the mor- observed around the shelf break in surface waters.
tality error with a transport estimate of unknown error. Below 30 m deep, weaker north-westward currents were
Furthermore, the exact vertical position and vertical observed over the shelf break (Fig. 5).
migratory behaviour of the larvae is unknown, whereas After running the PCA, the 85 stations were grouped
these factors may play a key role in the trajectories in five different groups according to the first two com-
(Vikebø et al., 2007). ponents which explained 90% of the observed variabil-
The SURFER software package and the kriging ity (Fig. 6). The first component was related to abiotic
interpolation routine were employed to map environ- factors (especially temperature and stratification),
mental variables and larval age distribution. whereas the second component indicated productivity
or food availability. The five groups which were differ-
entiated were: the Gironde area (G), which is under the
influence of the Gironde River’s plume, the Northern
R E S U LT S French Shelf (NFS), the Southern French Shelf (SFS),
the CS and finally an Oceanic area (O) (Fig. 7).
Environmental characterization The environmental conditions were significantly
different between areas (Table I). The Gironde area had
Surface water temperature (Ts) ranged from 16.28C to
lower temperatues and salinity and significantly higher
21.98C (Fig. 2). The lowest values were found close to
Chl a concentration and plankton abundance. Food
the mouth of the Gironde River, whereas the maximum
parameters also showed significantly higher values on
values were those of the oceanic and Cantabrian
the French Shelf (both northern and southern) than on
waters. In the same way, the lowest surface water salinity
CS, and the values observed in the Oceanic area were
(Ss) values were observed near the Gironde, with a
significantly lower than those found in the rest of the
areas. Thus, the principal characteristics of the Gironde
Table II: Mean percentage and standard area were the low temperature and high food avail-
deviation (in brackets) of particles moving each ability; on the NFS, the concentration of Chl a was also
day from one area (source area) to another high but, in contrast, the temperature was significantly
(end area) from 10 to 30 June according to higher; the SFS was characterized by intermediate con-
ditions; the CS was mainly characterized by low food
ROMS circulation model
availability and finally, the Oceanic area showed
End area extreme values of high temperature and low Chl a and
G NFS SFS CS O cell concentration.
Source area G 85.0 (7.5) 1.3 (1.8) 7.5 (5.7) 0.0 (0.0) 0.0 (0.0)
NFS 1.8 (1.0) 91.1 (3.3) 1.6 (1.0) 0.0 (0.0) 3.4 (2.5)
SFS 0.2 (0.4) 2.0 (0.8) 94.5 (2.9) 0.8 (0.3) 2.1 (1.2)
Egg and larvae abundance and distribution
CS 0.0 0.0 3.7 (3.4) 93.8 (3.6) 1.4 (0.8) Comparing with peak spawning distribution, in May,
O 0.0 0.2 (0.1) 0.6 (0.4) 0.5 (0.2) 95.3 (1.5)
anchovy eggs appeared more dispersed and with a

471
 JOURNAL OF PLANKTON RESEARCH j VOLUME 30 j NUMBER 4 j PAGES 467 – 481 j 2008

Downloaded from http://plankt.oxfordjournals.org/ by guest on August 13, 2015

Fig. 2. Surface temperature (TS), surface salinity (SS) and Chl a concentration in the upper 20 m in the study area.

more north-westerly distribution at the end of June LS for anchovy larvae ranged from 2.4 to 20.9 mm,
(Fig. 8a and b, adapted from Santos and Uriarte, which correspond to 0 and 28-day-old larvae, according
2004). The most important patch was found in the to the calculated age-at-length relationship:
northern part of the French shelf. In both cases, very
few eggs were found in oceanic waters and around age ¼ 1:46  LS  3:03; r 2 ¼ 0:89; P , 0:001;
the shelf break. Few eggs were also found in June in ð4Þ
the southern part of the study area. n ¼ 220
Larval distribution matched with that found for eggs,
with maximum values over the French shelf. However, Larval distribution at age showed a high concen-
taking into account the low egg abundance found in tration of larvae younger than 5 days distributed on the
this area, a relative high concentration of larvae middle French shelf, associated with the egg distribution
appeared in the south of the French shelf (Fig. 8b and (Fig. 9). The outer part of the French shelf and the shelf
c). Larval concentration on CS was quite low, but break were mainly dominated by larvae between 6 and
larvae appeared at all stations. 10 days old, whereas older larvae were mainly related

Fig. 3. Microzooplankton (a) and mesozooplankton (b) abundance (ind L21) and distribution in the study area.

472
 U. COTANO ET AL. j DISTRIBUTION, GROWTH AND SURVIVAL OF ANCHOVY LARVAE

Downloaded from http://plankt.oxfordjournals.org/ by guest on August 13, 2015

Fig. 4. Mesozooplankton distribution by mean ESD.

to the break shelf, oceanic waters, the Cantabrian coast Cantabrian area compared with the other areas.
and the southern part of the French shelf. A low con- However, this difference was due to the lack of larvae
centration of old larvae also appeared close to the coast younger than 5 days old. For the rest of areas, the inter-
in the south of the mouth of the Gironde estuary. A cepts ranged between 2.6 and 3.2 mm.
clear age gradient (Fig. 9) was noticed from the middle We did not find significant differences either in the
French shelf to the south-western waters of the study recent growth estimates (slopes in 5 days ranges,
area. Larval abundance and mean LS showed signifi- ANCOVA analysis of slopes, Table III), except for 10–
cant differences between areas (ANOVA, P , 0.01). The 15-day-old larvae on CS which showed significantly
abundance was significantly higher on the NFS, and the higher values than in the rest of the areas.
larvae captured in the CS and O areas were signifi- A common exponential model was fitted to describe
cantly larger (Table I). the decrease in larval abundance-at-age. Young larvae
The aggregation index analysis showed that from were not representatively sampled and they showed very
10 mm in length, an incipient schooling behaviour can low abundances, probably due to extrusion from the
be detected and a significant increase in aggregation net. The model fitted explained 95% of the variation in
index was observed from 15 mm in length onwards larval abundance and predicted an average instan-
(Fig. 10). These larvae were mainly distributed on the taneous mortality coefficient of 0.28 for larvae older
CS, the outer part of the NFS and the oceanic waters. than 3 days, representing a daily mortality rate of
24.5%. Once these results were analysed by areas, some
differences could be appreciated (Fig. 12). Thus, inter-
Growth and mortality mediate mortality rates were found on both NFS (Z =
As derived from equation (4), growth in length was 0.33) and SFS (Z = 0.36), significantly higher mortality
fairly linear. The growth curves were similar for all rates in the Gironde (Z = 0.52) (ANCOVA, P , 0.01)
areas (ANCOVA, P . 0.1). There were no significant and significantly lower rates on the CS (Z = 0.16) and
differences between slopes (mean growth rates), which in oceanic waters (Z = 0.21). These differences were
were over 0.6 mm day21 on average (Fig. 11). Only one related to the average size of the larvae in each area
significant difference was found in the intercept for the (Fig. 13a).

473
 JOURNAL OF PLANKTON RESEARCH j VOLUME 30 j NUMBER 4 j PAGES 467 – 481 j 2008

Downloaded from http://plankt.oxfordjournals.org/ by guest on August 13, 2015

Fig. 5. Residual current fields at depth of 15, 23, 31 and 39 m in the Bay of Biscay, obtained with ADCP measurements and tidal current
modelling between 23 and 26 June 2004.

The analysis of larval mortality between two consecu- areas (Table IV and Fig. 13b). Nevertheless, significant
tives days ( partially excluding advective effects) drasti- differences were again found between CS and Gironde
cally reduced the differences in mortality rates between areas with higher mortality values in the Gironde area,

474
 U. COTANO ET AL. j DISTRIBUTION, GROWTH AND SURVIVAL OF ANCHOVY LARVAE

Downloaded from http://plankt.oxfordjournals.org/ by guest on August 13, 2015

Fig. 6. PCA of the sampling stations and the five environmental areas. G, Gironde area; O, Oceanic area.

although larval size ranges were quite different between larvae are found offshore, far from the spawning areas;
both areas. (ii) there was no clear relation between food concen-
tration and growth and (iii) mortality rates were appar-
ently size-related, although correction for advective
effects resulted in a much higher variability of the mor-
DISCUSSION tality rates.
Our study reports three interesting new aspects of the
ecology of anchovy larvae in the Bay of Biscay: (i) older
Larvae distribution
Early hatched larvae appeared mostly on the middle
French shelf, associated with the spawning area.
Meanwhile, older larvae were found in the south-
eastern area, the CS and oceanic waters of the Bay of
Biscay. This obvious age-gradient could be due to: (i)
advection of larvae born on the NFS to the south or
south-western area, (ii) a preferential survival of larval
anchovy in the aforementioned areas or (iii) both
factors. However, older larvae were found on CS and
Oceanic areas, where spawning activity was very low
(Santos and Uriarte, 2004; this study). Furthermore,
inshore – offshore advective transport corresponds to the
general circulation pattern during the summer
(Koutsikopoulus and Le Cann, 1996), to the measured
instantaneous currents (Fig. 5) and to the model predic-
tions for the study period (Table II). At early stages, fish
larvae can be considered as passive particles.
Badenhorst and Boyd (Badenhorst and Boyd, 1980)
indicated that only anchovy .7 mm began to show
active swimming. Other authors (Hewitt, 1981; Hunter
and Coyne; 1982; Vasconcellos et al., 1998) have also
found, for different anchovy species, that the beginning
Fig. 7. The different environmental areas in the study area according of schooling behaviour, which would indicate a begin-
to the PCA. G, Gironde area; O, Oceanic area. ning of active movement by larvae, takes place between

475
 JOURNAL OF PLANKTON RESEARCH j VOLUME 30 j NUMBER 4 j PAGES 467 – 481 j 2008

Downloaded from http://plankt.oxfordjournals.org/ by guest on August 13, 2015

Fig. 8. Anchovy eggs and larval abundance (number per m2) and distribution. (a) Egg distribution during the peak spawning season (MPDH
survey, 2– 22 May). (b) Egg distribution in June. (c) Larval distribution in June (adapted from Santos and Uriarte, 2004).

10 and 15 mm in length. These results agree with those Growth rates

found in this study where schooling behaviour for larval To our knowledge, daily somatic growth rates for larval
anchovy is obvious only from 15 mm in length. So, anchovy in the Bay of Biscay had never been measured
although a partial active movement can be presumed before. The growth rates calculated in this study are
for larvae sampled in this study, it can be said that most within the range of 0.4 – 1.0 mm day21 (for larvae
of them are still in length ranges which are passively younger than 30 days) reported for this species in
transported. Therefore, transport of anchovy larvae Mediterranean waters (Garcı́a and Palomera, 1996;
from north-eastern waters seems to be the most likely Dulčić, 1997; Garcı́a et al., 2003; see also revision in
explanation for the observed age gradient. Palomera et al., 2007). Interestingly, we did not find

Fig. 9. Anchovy larval abundances (number per m2) at age and the distribution of the mean age.

476
 U. COTANO ET AL. j DISTRIBUTION, GROWTH AND SURVIVAL OF ANCHOVY LARVAE

activity of adult anchovy was higher in open waters

probably related to selective feeding. For anchovy larvae
in the Mediterranean, Conway et al. (Conway et al.,
1998) also found that the feeding success of anchovy
larvae was higher when a marked reduction in potential
food abundance occurred, as a consequence of prey
selection. Schmitt (Schmitt, 1986) observed, under lab-
oratory conditions, that larval anchovy, Engraulis mordax,
larger than 20 mm tended to select the largest prey
available. In the Bay of Biscay, a clear gradient of zoo-
plankton size can be observed from highly productive
coastal waters close to the mouth of Gironde River to
shelf edge waters (Albaina and Irigoien, 2004), with
smaller individuals associated with coastal waters. Thus,
the gradient in large prey concentration for large
anchovy larvae may well not be as marked as the con-

Downloaded from http://plankt.oxfordjournals.org/ by guest on August 13, 2015

centration of total or small zooplankton.
Fig. 10. Aggregation index of anchovy larvae at length.
Furthermore, for organisms such as fish larvae where
size is a major determinant of mortality (McGurk,
significant differences in growth rate associated with the 1986), there may be behavioural adaptations (higher
different environmental conditions, in particular food feeding activity) to maintain maximum growth at the
concentration. This can be due to the average environ- risk of higher mortalities (Fiksen et al., 2007). Food con-
mental history being the same for larvae drifting centration has already been observed to affect the mor-
between areas. However, when considering short term tality rates in E. encrasicolus (Palomera and Lleonart,
growth (Table III), which should better represent the 1989). In our study, although there is not a significant
influence of the environmental conditions in the relation between mortality and food concentration,
capture area, the differences are also minimal. mortality certainly appears to be much more variable
However, the relationship between food and growth is than growth rates as predicted by changes in behaviour
not always clear as higher food concentration does not to maintain maximum growth rates.
necessarily mean higher ingestion. In the same study
area, Plounevez and Champalbert (Plounevez and
Champalbert, 1999) found that despite the higher food Mortality rates
concentration in the water plume of the Gironde Daily mortality rates estimated with the traditional
estuary compared with Oceanic areas, the feeding method fall within the range of 15– 45%, found for the

Fig. 11. Age–length relationship. The line was fitted by: y = 0.61x þ 2.93 (n = 220; r 2 = 0.89, P , 0.001).

477
 JOURNAL OF PLANKTON RESEARCH j VOLUME 30 j NUMBER 4 j PAGES 467 – 481 j 2008

Table III: Slopes and number of individuals (n) of the length-at-age relationship for 5 days larval
groups in each of the sampling areas
Area G NFS SFS CS O

Age Slope n Slope n Slope n Slope n Slope n

1–5 days 0.76 9 0.27 7 0.85 15 — 8 0.75 9
5–10 days 0.25 17 0.61 9 0.59 28 0.38 9 0.58 13
10 –15 days 0.55 13 0.84 11 0.50 20 1.43* 9 0.63 13
15 –20 days 1.47 3 0.86 6 0.40 15 0.19 19 0.64 24
20 –25 days — — — — 0.75 4 1.04 6 0.94 3

*Statistically significant differences (ANCOVA; P , 0.01).

same species in Mediterranean waters (Palomera and Following the consecutive days approach, which
Lleonart, 1989, Dulčić, 1995, Coombs et al., 2003). reduces the effect of size and advection, differences in
Following the traditional method (Fig. 13a), the highest mortality rates between areas disappear. Care should be
mortality rates are found on the shelf (Gironde, NFS taken when interpreting these data because of the high

Downloaded from http://plankt.oxfordjournals.org/ by guest on August 13, 2015

and SFS area), which is also the area with the highest variability of mortality within an area for different age
food concentration. This can be partially explained by groups and the lower amount of data to define the
the lower size of the larvae in that area (spawning area). decrease in abundance which may affect the accuracy
It must also be taken into account that a considerable of the estimation. However, although highly variable,
percentage of small larvae may die before starting mortality remains generally lower on the CS and
exogenous feeding (Browman, 1995), which reduces the Oceanic area than on the French shelf (Fig. 13b).
link between food and mortality, but contributes to Predation could be another factor that would explain
increase the mortality in the spawning areas with the mortality distribution. Capture statistics (www.ices.
younger larvae. However, it is clear that following the dk) indicate that the populations of planktivorous fish
standard procedure, larval advection from the spawning (mackerel, horse, mackerel, sardine, sprat and anchovy
areas to the other areas may have lead to an important itself ) are much larger in the French shelf area than on
overestimation of mortality in Gironde and French shelf the CS and oceanic waters. Larval fish are generally
waters and to underestimate it in the CS and oceanic more susceptible to predation than other planktonic
waters. organisms of comparable size (McGurk, 1986). Thus,

Fig. 12. Mortality curves by areas. G, Gironde area; O, Oceanic area. The thick line represents the general mortality curve calculated for all
larvae together.

478
 U. COTANO ET AL. j DISTRIBUTION, GROWTH AND SURVIVAL OF ANCHOVY LARVAE

with lower food might be due to tradeoffs or develop-

mental niche shifts, where certain size classes take
refuge in less profitable feeding habitats, as proposed by
Fuiman and Magurran (Fuiman and Magurran, 1994).
In summary, advective processes due to dominant
north-easterly winds during summer influence larval
distribution in the Bay of Biscay after spawning.
However, higher food concentration on the shelf did
not result in significant differences in the growth rate.
Mortality rates do not seem to be related to the size,
although this effect may be partially due to advection.
Our observations are apparently contradictory to the
triad concept developed by Bakun (Bakun, 1996) and
with models predicting minimum or no survival in the
Oceanic area (Allain et al., 2007). However, the triad
concept having been developed for upwelling areas,

Downloaded from http://plankt.oxfordjournals.org/ by guest on August 13, 2015

retention probably requires a wider interpretation in the
Bay of Biscay where general circulation allows return to
the shelf (Irigoien et al., 2007).
This study was carried out under historical minimum
abundances of anchovy in the Bay of Biscay (ICES,
2006). Further studies under different population abun-
Fig. 13. Mortality rates in relation to size. (a) Mortality rate dance and climatic conditions are required for a better
estimated with the traditional method (Fig. 12) in relation to the understanding of the factors determining anchovy
average size in each area. (b) Mortality rate estimated from the
abundance of successive ages as a function of size in each area.
growth and mortality in the Bay of Biscay.

Table IV: Mortality rates (Zd) between AC K N OW L E D G E M E N T S

consecutive ages by areas We wish to thank A. Uriarte for comments on an early
Age Zd draft of the manuscript. This paper is contribution no.
G NFS SFS CS O
392 from AZTI-Tecnalia (Marine Research).
3 0.30
4 0.11
5 0.13 FUNDING
6 0.35 0.47 0.11
7 0.16 0.20 0.09 This study was supported in part by a grant from
8 1.24 0.15 0.17
9 0.04 0.60 0.27 0.04 0.12
Agriculture, Fisheries and Food Department of the
10 0.48 0.11 0.49 0.04 0.21 Basque Government and by the Department of Industry,
11 0.92 0.11 0.25 0.07 0.12 Trade and Tourism of the Basque Government.
12 0.36 0.32 0.20 0.02
13 0.50 0.65 0.13 0.38
14 0.90 0.43 0.12 0.26
15 0.01 0.52 0.62 0.70
16 0.65 0.41 0.15 0.29 REFERENCES
17 0.13 0.47 0.15 0.29
18 0.50 0.57 1.25 Ahlstrom, E. H. (1954) Distribution and abundance of egg and larval
19 0.78 populations of the Pacific sardine. Fish. Bull., 56, 83– 140.
Albaina, A. and Irigoien, X. (2004) Relationships between frontal
structures and zooplankton communities along a cross-shelf transect
in the Bay of Biscay (1995 to 2003). Mar. Ecol. Prog. Ser., 284,
65– 75.
predation pressure on anchovy larvae could be higher
Allain, G., Petitgas, P. and Lazure, P. (2001) The influence of
in these more productive areas and increase mortality meso-scale ocean processes on anchovy (Engraulis encrasicolus) recruit-
rates, despite the relatively higher food concentration ment in the bay of Biscay estimated using a 3D hydrodynamic
available in these areas. This lower mortality in areas model. Fish. Oceanogr., 10, 151– 163.

479
 JOURNAL OF PLANKTON RESEARCH j VOLUME 30 j NUMBER 4 j PAGES 467 – 481 j 2008

Allain, G., Petitgas, P., Grellier, P. et al. (2003) The selection process anchovy (SW Mediterranean) during the 2000–2001 spawning
from larval to juvenile stages of anchovy (Engraulis encrasicolus) in the seasons. Sci. Mar., 67, 477– 490.
Bay of Biscay investigated by Lagrangian simulations and compara- Hewitt, R. (1981) Spatial pattern and survival of anchovy larvae:
tive otolith growth. Fish. Oceanogr., 12, 407 –418. implications of adult reproductive strategy. PhD Dissertation,
Allain, G., Petitgas, P. and Lazure, P. (2007) Biophysical modelling of University of California, San Diego, p. 186.
larval drift, growth and survival for the prediction of anchovy Hjort, J. (1914) Fluctuations in the great fisheries of Northern Europe.
(Engraulis encrasicolus) recruitment in the Bay of Biscay (NE Atlantic). Rapp. P.-v. Reun. Cons. Int. Explor. Mer., 20, 1 –28.
Fish. Oceanogr., 16, 489– 505.
Hunter, J. R. and Coyne, K. M. (1982) The onset of schooling in north-
Arbault, S. and Lacroix, N. (1977) Oeufs et larves de Clupeides et ern anchovy larvae Engraulis mordax. CalCOFI Rep., 23, 246–251.
Engraulides dans le Golfe de Gascogne (1969–1973). Distribution
ICES (2006) Report of the Working Group on the Assessment of
des frayères. Relations entre les facteurs de milliey et la reproduc-
tion. Rev. Trav. Inst. Pêches Marit., 41, 227 –254. Mackarel, Horse Mackerel, Sardine and Anchovy (WGMHSA).
Galway, Ireland, September 5 –14, 2006, ICES CM
Badenhorst, A. and Boyd, A. J. (1980) Distributional ecology of the 2006\ACFM, 36, 601 pp..
larvae and juveniles of the anchovy Engraulis capensis Gilchrist in
relation to the hydrological environment off South West Africa, Iles, T. D. and Sinclair, M. (1982) Atlantic herring: stock, discreteness
and abundance. Science, 215, 627– 633.
1978– 79. Fish. Bull. (S. Afr.), 13, 83–106.
Irigoien, X., Fiksen, Ø., Cotano, U. et al. (2007) Could Biscay Bay Anchovy
Bailey, K. M. and Houde, E. D. (1989) Predation on eggs and larvae of
marine fishes and the recruitment problem. Adv. Mar. Biol., 25, 1–8. recruit through a spatial loophole? Prog. Oceanogr., 74, 132–148.

Downloaded from http://plankt.oxfordjournals.org/ by guest on August 13, 2015

Koutsikopoulus, C. and Le Cann, B. (1996) Physical processes and
Bakun, A. (1996) Patterns in the Ocean: Ocean Processes and Marine Population
hydrological structures related to the Bay of Biscay anchovy. Sci.
Dynamics. California Sea Grant College System, La Jolla, p. 323.
Mar., 60, 9 –19.
Bez, N. (2000) On the use of Lloyd’s index of patchiness. Fish.
Leggett, W. C. and Deblois, E. (1994) Recruitment in marine fishes: Is
Oceanogr., 9, 372– 376.
it regulated by starvation and predation in the egg and larval
Borja, A., Uriarte, A., Egaña, J. et al. (1998) Relationships between stages? Neth. J. Sea Res., 32, 119–134.
anchovy (Engraulis encrasicolus) recruitment and environment in the
McGurk, M. D. (1986) Natural mortality of marine pelagic fish eggs and
Bay of Biscay (1967– 1996). Fish. Oceanogr., 7, 375– 380.
larvae—role of spatial patchiness. Mar. Ecol. Prog. Ser., 34, 227–242.
Browman, H. I. (1995) Introduction: commentaries on current
research trends in recruitment studies. Mar. Ecol. Prog. Ser., 128, 305. McNamara, J. M. and Houston, A. I. (1987) Starvation and predation
as factors limiting population size. Ecology, 68, 1515–1519.
Conway, D. V. P., Coombs, S. H. and Smith, C. (1998) Feeding of
anchovy Engraulis encrasicolus larvae in the northwestern Adriatic Sea Motos, L., Uriarte, A. and Valencia, V. (1996) The spawning environ-
ment of the Bay of Biscay anchovy (Engraulis encrasicholus L.). Sci.
in response to changing hydrobiological conditions. Mar. Ecol. Prog.
Mar., 60 (Suppl. 2), 117– 140.
Ser., 175, 35– 49.
Coombs, S. H., Giovanardi, O., Halliday, N. C. et al. (2003) Wind Palomera, I. (1991) Vertical distribution of eggs and larvae of Engraulis
encrasicolus in stratified waters of the western Mediterranean. Mar.
mixing, food availability and mortality of anchovy larvae Engraulis
Biol., 111, 37– 44.
encrasicolus in the northern Adriatic Sea. Mar. Ecol. Prog. Ser., 248,
221–235. Palomera, I., Morales-Nin, B. and Lleonart, J. (1988) Larval growth of
anchovy, Engraulis encrasicolus, in the western Mediterranean. Mar.
Cotano, U. and Álvarez, P. (2003) Growth of young-of-the-year mack-
erel in the Bay of Biscay. J. Fish Biol., 62, 1010–1020. Biol., 111, 37– 44.

Cushing, D. H. (1975) Marine Ecology and Fisheries. Cambridge Palomera, I. and Lleonart, J. (1989) Field mortality estimates of
anchovy larvae, Engraulis encrasicolus, in the western Mediterranean.
University Press, London (UK), pp. 292.
J. Fish Biol., 35 (Suppl. A), 133–138.
Cushing, D. H. (1990) Plankton production and year-class strength in
Palomera, I., Olivar, M. P., Salat, J. et al. (2007) Small pelagic fish in
fish populations: An update of the match-mismatch hypothesis. Adv.
the NW Mediterranean Sea: An ecological review. Prog. Oceanogr.,
Mar. Biol., 26, 249–294.
74, 377– 396.
Dulčić, J. (1995) Spawning of the anchovy (Engraulis encrasicolus) in the
Plounevez, S. and Champalbert, G. (1999) Feeding behaviour and
Northern Adriatic Sea in 1989, the year of intensive blooms. Ann.
trophic environmetn of Engraulis encrasicolus (L.) in the Bay of Biscay.
Istrian Mediterr. Stud., 7, 55–60.
Est. Coast. Shelf Sci., 49, 177–191.
Dulčić, J. (1997) Growth of anchovy, Engraulis encrasicolus (L.), larvae in
Santos, M. and Uriarte, A. (2004) Estimates of Sapwning Stock
the Northern Adriatic Sea. Fish. Res., 31, 189–195.
Biomass of the Bay of Biscay anchovy (Engraulis encrasicolus, L.) in
Fiksen, Ø., Jørgensen, C., Kristiansen, T. et al. (2007) Linking beha- 2004. Working Document to the ICES Working Group on the
vioural ecology and oceanography: how behaviour determines assessment of Mackerel, Horse Mackerel, Sardine and Anchovy,
growth, mortality and dispersal. Mar. Ecol. Prog. Ser., 347, 195–205. Copenhagen, September 6– 19, 2004.
Fuiman, L. A. and Magurran, A. E. (1994) Development of predator Schmitt, P. D. (1986) Prey size selectivity and feeding rate of larvae of
defences in fishes. Rev. Fish Biol. Fish, 4, 145– 183. the northern anchovy, Engraulis mordax Girard. CalCOFI Rep., 27, 153–161.
Garcı́a, A. and Palomera, I. (1996) Anchovy early life history and its Shchepetkin, A. F. and McWilliams, J. C. (2005) The regional oceanic
relation to its surrounding environment in the western modelling system (ROMS): a split-explicit, free-surface, topography-
Mediterranean Basin. Sci. Mar., 60 (Suppl. 2), 155–166. following-coordinate oceanic model. Ocean Model., 9, 347– 404.
Garcı́a, A., Cortés, D., Ramı́rez, T. et al. (2003) Contribution of larval Smith, P. E. and Richardson, S. K. (1977) Standard techniques
growth rate variability to the recruitment of the Bay of Malaga for pelagic egg and larval surveys. FAO Fish. Tech. Pap., 155, 100 pp.

480
 U. COTANO ET AL. j DISTRIBUTION, GROWTH AND SURVIVAL OF ANCHOVY LARVAE

Theilacker, G. H. (1980) Changes in body measurements of larval Vasconcellos, M. C., Freire, K. F. and Castello, J. P. (1998)
northern anchovy, Engraulis mordax, and other fishes due to handling Distribution patterns and feeding success of anchovy, Engraulis
and preservation. Fish. Bull., 76, 685– 692. anchoita, larvae off southern Brazil. Sci. Mar., 62, 385– 392.
Uriarte, A., Sagarminaga, Y., Scalabrin, C. et al. (2001) Ecology of Vikebø, F., Jørgensen, C., Kristiansen, T. et al. (2007) Larval drift,
anchovy juveniles in the Bay of Biscay 4 months after peak spawn- growth and survival emerging from behavioural rules in a general
ing: do they form part of the plankton?. ICES CM 2001/W, 20. circulation model. Mar. Ecol. Prog. Ser., 347, 207– 219.

Downloaded from http://plankt.oxfordjournals.org/ by guest on August 13, 2015

481