Seasonal variation in survival of larval Engraulis mordax estimated from the age distribution of juveniles

Abstract

Juvenile northern anchovy, Engraulrs mordax, collected during autumn of 1978 and 1979 were aged using daily increments in their otoliths. Neither year class was dominated by individuals born during some short period, but March and April had the highest frequency of births in each year. Monthly ichythoplankton sur- veys indicated that significant spawningoccurred from January through May of each year and peaked in early March. Comparison of the temporal distribution of birth dates with larval abundance indicated that larval survival was similar in the first half of each spawning season and greater during April to May of the 1978 spawning season than the same period in 1979. This difference in seasonal pattern of survival was nearly suf- ficient to account for the observed greater recruitment in 1978 and is consistent with the hypothesis that offshore transport of larvae influences recruitment. One goal of fish population dynamics is to under- stand the processes responsible for annual variation in recruitment. The variation can be more than an

Full text

SEASONAL VARIATION IN SURVIVAL
OF
LARVAL
NORTHERN
ANCHOVY,
ENGRAULlS
MORDAX,
ESTIMATED
FROM THE AGE DISTRIBUTION
OF
JUVENILES
RICHARD
D.
METHOT,
JR.'
ABSTRACT
Juvenile northern anchovy,
Engraulrs
mordax,
collected during autumn of 1978 and 1979 were aged using
daily increments in their otoliths. Neither year class was dominated by individuals born during some short
period, but March and April had the highest frequency of births in each year. Monthly ichythoplankton
sur-
veys indicated that significant spawningoccurred from January through May of each year and peaked in early
March. Comparison of the temporal distribution of birth dates with larval abundance indicated that larval
survival was similar in the first half of each spawning season and greater during April to May of the 1978
spawning season than the same period in 1979. This difference in seasonal pattern
of
survival was nearly suf-
ficient to account for the observed greater recruitment in 1978 and is consistent with the hypothesis that
offshore transport of larvae influences recruitment.
One goal of fish population dynamics is to under-
stand the processes responsible for annual variation
in recruitment. The variation can be more than an
order of magnitude and
is
poorly correlated with
abundance
of
spawners (Cushing and Harris 1973).
The concept of a critical period during the early larval
stage (Hjort 1926; Marr 1956; May 1974) has struc-
tured much of the research. Recent work has focused
on the importance of temporal and spatial coin-
cidence of
first
feeding larvae and concentrations of
prey (Beyer and Laurence 1981; Lasker 1978;
Vlymen 1977). However, transport of larvae away
from juvenile nursery areas can influence recruit-
ment (Nelson et al. 1977; Parrish et al. 1981) and the
role of predation is unknown.
Seasonal variation in factors that cause annual
variation in recruitment probably influences the
average timing of spawning. Support for this
hypothesis comes from the latitudinal correlation
between duration of the spawning season and the
plankton bloom (Wyatt 1980). More direct evidence
is
found within the North Sea where the short spawn-
ing season of each herring population bears a fixed
phase relation to the mean date of the local plankton
bloom (Cushing 1975). The timing of the most favor-
able environmental conditions may not be predict-
able in each year. The match-mismatch hypothesis
(Cushing 1975) suggests that variation in the relative
timing of spawning and the seasonal plankton bloom
contributes to variation in recruitment.
'Southwest Fisheries Center La Jolla Laboratory, National Marine
Fisheries Service, NOAA, 8604 La Jolla Shores Drive,
La
Jolla,
CA 92038.
Manuscript
accepted
February
1983
FISHERY
BULLETIN
VOL
81.
NO
4.
1983
Collection of sufficient years of data to test any re-
cruitment hypothesis is difficult. However, a testable
corollary of Cushing's hypothesis is that, in any year,
larvae born during favorable environmental periods
constitute most of the year class. The age distribu-
tion of juveniles-the survivors of the larval stage-is
a function of the seasonal distribution of spawning
and seasonal changes in larval survival. To test the
match-mismatch hypothesis, the birth dates of
juvenile northern anchovy,
Engraulis
mordax,
were
determined from daily increments in otoliths
(Brothers et aL 1976; Methot andKramer 1979; Pan-
nella 197
1)
and compared with the seasonal distribu-
tion of spawning determined from ichthyoplankton
surveys conducted during the 1978 and 1979 spawn-
ing seasons.
METHODS
Larval
Abundance
The seasonal distributions of the northern an-
chovy larval abundance were estimated from ichthyo-
plankton surveys (Kramer et aL 1972) on the
sampling grid (Fig. 1) of the California Cooperative
Oceanic Fisheries Investigations (CalCOFI). Seven
surveys were conducted between December 1977
and August 1978 and four surveys between January
1979 and May 1979. Only larvae from 2.6 mm (live
standard length at hatch) to
5.1
mm (few days after
yolk absorption and onset of feeding) were used in
the analysis.
Each cruise's larval census was the summed abun-
741

FISHERY
BULLETIN
VOL
81.
NO
4
30 samples of northern anchovy were collected from
bait receivers
at
sport fishing docks from San Diego
to
Pt.
Conception, Calif. A total of 1,101 fish were
measured, but sample size varied from
11
to 86 fish
30
each fish's contribution to the overall size distribu-
tion was weighted by the inverse of sample size. From
15 of the
30
samples, 141 fish were aged using daily
increments in otoliths.
During 1-19 November 1979, specimens were
obtained from samples taken with a 15 m midwater
trawl on a survey conducted by the California
Department of Fish and Game (CFG) to investigate
the abundance of juvenile northern anchovy (Mais
1980). Trawls were taken parallel to the coast be-
tween the 30 and 90 m isobaths
at
7.4
km
coastwise
intervals between Blanca Bay, Baja California, and
Pt.
Conception, Calif.
Past
trawl surveys indicate
that the juveniles are concentrated into the
nearshore zone (Mais 1974).
A
total of 2,356 fish
were measured from the 93 positive trawls; sample
size typically was
25
fish per trawl. In addition, 10 of
the 25 fish per trawl were aged by CFG personnel
using annual growth marks in otoliths (Collins and
Spratt 1969). Juveniles are defined here
as
those fish
with no otolith annuli. The fraction of fish with no
annuli in each
5
mm size interval was calculated
(Table
1).
From
8
of the 93 samples, 129 fish were
aged using daily increments in otoliths.
Size distributions were calculated in each of three
alongshore regions: North of
Pt.
Dume (north), San
Diego to
Pt.
Dume (central), and south of San Diego
(south). These regions were selected on the basis of
the distribution of samples in 1978 and minima in the
100
km
FIGURE
1.-Geographic region inhabited by the central population
of northern anchovy. Ichthyoplankton samples were collected along
CalCOFI survey lines. Location of stations is indicated only
on
line
93.3. Samplesofjuvenile anchovy were collectedalong the mainland
coast. In 1978 these samples were obtained from the bait fishery
from Pt. Conception to the United States-Mexico border. In 1979
samples were collected with a midwater trawl from Pt. Conception to
Blanca Bay.
dance along CalCOFI lines 60-1 10. The catch at each
station was adjusted for volume of water filtered and
weighted by the distance to adjacent stations along
the line. Abundances along unsampled lines were
estimated from abundance in adjacent lines and
cruises. A more complete description and further
analyses of these data are in Hewitt and Methot
(1982).
The mean and variance of date of larval catch were
calculated for each cruise. Each station's contribu-
tion to these statistics was weighted by the distance
to adjacent stations and by the catch of larvae. The
effective duration of each cruise was considered to be
k2
standard deviations of catch date.
Juvenile
Northern
Anchovy
Samples
During the period
28
October-14 December 1978,
742
TABLE
1.-Frequency of northern anchovy with and
without otolith annuli. Specimens were collected
by
trawl during November 1979 (Mais 1980). Results are
stratified by region and
5
mm size interval.
Nonh
of
Sa"
Diego South
of
PI
Dume Pt.
Dume
San
Diego
SIX
interval
imm)
Nannuli
0
>O
0
>O
0
>O
~____
50
1
55 4
9
60
1
7
3
65
0
12
6
70 2 11 18
75
3
28 59
80
8
23 50
85
6
13 41
90
12
8
41
95 4 4
1
26
26
100
5
5 6371
105 3 5423273
110 2 19 1
66
2 45
115
1
33
1
33
18
120 17 21
3
125 22
8
1
130 7
6
135
5
5
140
1
145 4

mrHor
SURVIVAL
OF
LARVAL
EVCHA[ZIS
MORIMX
alongshore distribution of juveniles in 1979 (Methot
1981). The regional breakdown was necessary
because the overlap in size between juveniles and
older fish varied latitudinally (Table
1).
Although
data in Table
1
are entirely from 1979, they were
used to calculate juvenile size distributions from
overall size distributions in both 1978 and 1979. In
1978 few adults were collected and no comparable
samples were obtained from south of San Diego.2
0
toli th Preparation
Thawed specimens were measured to the nearest
1.0
mm standard length Sagittae (largest otoliths)
were removed, cleaned in distilled water, dried, and
mounted on a microscope slide with a clear
methacrylate-based mounting medium. Otoliths of
northern anchovy larger than about
40
mm are too
thick to transmit sufficient light for viewing the
increments. Material was removed from the otolith's
medial surface by applying
5-10'3
HC1
to selected
regions for about
10
s
at
a time. Immersion oil. pe-
troleum jelly,
or
mounting medium were used to mask
the outer edge of the otolith and regions already suf-
ficiently thin. The selectively etched surface
develops high relief but a thin layer of immersion oil
renders this relief nonrefractory and permits
examination of the otolith. After most increments
became visible, the mounting medium was softened
with
80%
ethanol, and the otolith was turned over
and remounted. Etching of the lateral surface con-
tinued until all incrPments were visible within, but
not necessarily at the surface of, the remaining
materiaL
The otoliths in 1978 were prepared by embedding
in polyester casting resin and grinding sagittal sec-
tions on 400 and
600
grit wet sandpaper. Selective
etching was faster and more successful than
grinding.
Age
Determination
Specimens used for this study also were used to
back calculate juvenile growth and a direct count of
all increments in an otolith was rarely made. Instead,
age was determined from numerical integration
of
otolith growth (increment width). Increments were
measured with a video camera mounted on a com-
pound microscope, an electronic device which
'Two
samples from the Ensenada commercial fishe
were
ro-
vided b
G.
Broadhead (Living Marine Resources San%iego). $he
fish hadVa similar size/birth-date relation
to
fish frdm the bait fishery
in
the
U.S.
coastal waters. Because the commercial fishery
is
biased
against small fish, the
size
distribution
of
juveniles in Mexico could
not be estimated.
positioned a cursor in the video image, and a mi-
crocomputer interfaced to the device. All measure-
ments were made along the longest radius of the
otolith (towards the posterior margin). The observer
positioned the cursor at the outer edge of an incre-
ment and keyed in the number of increments be-
tween that point and the previous point while the
computer recorded the radius to that increment.
Increment width usually did not change rapidly
so
2-
10 increments of similar size were entered together.
Data from different regions along the longest radius
were recorded at various stages of the etching
process.
Data from both otoliths and several replicate tran-
sects per otolith were combined in the calculation of
age. Mean increment width was calculated at all
points along the longest radius. Etching errors
occasionally produced a region in which increments
could not be seen. When this occurred, increment
width was interpolated from mean increment width
in adjacent intervals by a linear interpolation of
increment width on radius. Age was calculated from
numerical evaluation of the following expression:
where the following definitions and boundary con-
ditions apply:
r,
=
set of all radial distances where increment
width changed perceptibly
C(r,)
=
average increment width between
r,
,
and
r,
r,,
=
otolith radius at onset of increment forma-
tion (6.5 pm)
r,,
=
maximum otolith radius
G(r,)
=
typical initial increment width
(0.8
pm
per increment)
G(r,,)
=
G(r,z-,)
if
C(r,)
not measurable.
The result converges exactly to the count of
increments if each individual increment
is
measured
once. The age estimate was accepted if
<20%
of
the
age was from interpolated increments. About
3%
of
the fish were rejected by this criterion. The mean per-
centage of interpolated increments for the accepted
fish was 4.7% and the median was 2.5%. Usually an
independent age estimate could be made from each
otolith. When fish were stratified into 50-d age inter-
vals, the coefficient of variation of age between
otoliths within fish averaged
4.6%
in 1978 and
3.5%
in 1979. Thus, the
95%
confidence intervalfora250-
d-old fish was k14 d.
743

FISHERY
BULLETIN VOL
HI.
NO
4
were selected to span a wide size range for an analysis
of seasonal patterns of juvenile growth (Methot
1981). A less biased estimate of the birth-date dis-
tribution was obtained from the size-frequency dis-
tribution of a large sample of juveniles and a
size/birth-date nomograph (Fridriksson 1934;
Kimura 1977). In each year’s nomograph, birth-date
frequencies (by month) were calculated for fish in
each
10
mm size interval. All samples within each
year were combined in that year’s nomograph.
The birth dates calculated in this study are actually
dates of onset of increment formation. The northern
anchovy larvae deposit the
first
increment at about
the end of yolk absorption, the fifth day after hatch-
ing
at
16°C (Brothers et al. 1976). This is close
to
the
mean age of larvae used to estimate larval abun-
dances
so
no constant was added to the juveniles’
ages when calculating birth dates.
The fish in the present study usually had 150-400
increments but the daily deposition of increments in
northern anchovy otoliths has been confirmed only to
100
d in the laboratory (Brothers et al. 1976). The
accuracy of my interpretation of daily increments in
juveniles was tested by comparing birth-date dis-
tributions calculated from early samples with dis-
tributions calculated from late samples. The
distributions should be indistinguishable if the sam-
ples were of the same cohort and mortality during the
period was not age selective. In addition to the
December 1978 samples used in this study, samples
were collected at San Diego in September 1978 and
February 1979 (Table
2).
The three birth-date dis-
tributions were compared by the Smirnov test for dif-
ferences in cumulative probabilities (Conover 197 1,
p. 309). The September and December distributions
were very similar (maximum difference
=
0.105,
P
<
0.2)
and the February distribution was also not
significantly different from September’s (0.243,
P
<
0.02).
This test is sensitive to aging errors of the
same magnitude as the precision of the ages.
If
ages
of February’s fish had been overestimated by
15
d
(one- half of fish in eachmonth shifted to the following
month) the difference between September and Feb-
ruarywould haveincreased to0.376,
P<
0.1. A
1-mo
error in aging the February juveniles would have
made the September 1978 to February 1979 com-
parison highly significant
(P
<
0.01).
I
conclude that
any bias in aging must be less than about
15
d.
TABLE
2.-Birth-date frequency
ofjuvenilenorthern anchovy
col-
lected at San
Diego
between September
1978
and February
1979.
Sepr Oct 1978
Oec
1978
Feb
1979
N
samples
2
2
1
N
fish 28 19 15
Length
(mm)
mean
77 8
76
4 82
1
Month
SO
55 65 59
Jan.
Feb
Mar
Apr.
May
0
2
1
3
1
1
9 4
6
7
0
1
2
15
10
Birth-Date Distribution
The selected specimens produce a biased estimate
of the juvenile birth-date distribution because they
744
RESULTS
Larval Abundance
The temporal distributions of northern anchovy lar-
vae differed between the two years (Table 3). The
maximum abundance occurred in February-Marchof
each year but the peak was greater in 1978. Larvae
were much more abundant during May 1979 than
during May 1978. The average larva in 1979 was in
water 1°C colder than the average larva in 1978 and
was further offshore (Table
3).
Larval production per
30-d
date interval was calculated by numerical
integration of the area under the dashed lines in
Figures
2
and 3. Total larval production during
January-May 1979 was 2.1% greater than during the
same period in 1978
Size and Birth-Date Distributions
In 1978 and 1979 northern anchovy juveniles,
collected north of
Pt.
Dume, were typically larger
(Table
4)
and had been born earlier (Table
5)
than
juveniles collected to the south. The size/birth-date
nomograms (Table
6)
applied
to
the juvenile size dis-
tributions produced birth-date distributions (Table
7) with peaks in March-April for southern fish and
TABLE
3.-Abundance
of
northern anchovy larvae.
Value
in
parentheses
is
the fraction of the abundance that was inter-
polated,
N
samples exclude offshore samples with
no
larvae. Date.
distance offshore. and temperature
at
10
m were weighted
by
larval
catch at each stallon.
Date
1977-78
oec
15
Jan
18
Feb
25
Apr 9
May
26
June
30
Aug 13
Jan
18
Mar
2
Apr 14
May
10
1979
N
samples
42
70
87
74
64
48
20
33
71
34
54
Abundance
448 1008)
558
(000)
2 367
IO
00)
686
10001
143
(0001
26
1000)
26
(0001
448 (0451
1524
1000)
1392 (040)
653
I0001
Olslance
lkml
119
65
65
41
44
72
22
111
65
113
54
Temp
lo
CI
-
16
3
155
14 9
158
152
16
1
18
1
14
0
139
14 5
14 8

METHOT
SURVIVAL
OF
LARVAL
E.VCH;\(ZIS
M0RUA.X
04-
Y)
U
0
s:
I
0
a
03-
2
02-
0
?
01-
00-
TABLE
4.-Size distributions of juvenile northern anchovy calculated from
size distributions of all fish and size-specific juvenile fraction (Table
1).
The
percentage
of
the population composed
of
juveniles was also calculated
(9
juveniles). Area
is
the surface area (square nautical miles) nearshore of the
50 fathom isobath and excludes the shallow area around islands that was
not sampled.
1978 1979
~___~
nonh central nonh Central south
Area
476 484 476 484 1,905
Slle
N
f1.n
429 724
460
757 1,333
N samples
9 21 26 41
66
/mm/
%
iuveniles
820 988 280 40.8 54 4
P,
1978
YEAR
CLASS
I1
11
1,
-
2000
I,
,
,-
I(
1,
I/
f
I-
I
0
,
2
I
U
,
0
I I
-
1000
5
I1
I
I
I
m
I
i
l
1
U
I
-
500
2
0
a
-
1500
\
A
,
I
0
30
0
30
60
90
120
150
180
40-49
00
03
00 00
03
50-59
00
19
00
29 28
60-69 33 171 31 178 39
70-79 183 454 101 317 27
3
80-89 269 273
22
5 269 41
0
90-99 34
3
77 318 113 22 5
t00-109 16 7 84
18
03 248
110
119 05
00
77 10
06
Y)
U
2
03-
-
-
0
tS79
YEAR
CLASS
__
.
J
-._
7
1
0
!
!
-
\
02-
L
Y
--
Y
d
-
--
L
m
-
-
a
0,-
birth-date distributions were similar to the observed
birth-date occurrences in the two regions in 1978; the
-lsOO
calculated differed from the observed in 1979
because selection
of
specimens
was
highly nonran-
dom in 1979.
A
substantial fraction
of
90-110
mm
fish in 1979 were assigned
to
birth dates before
December 1978. Because these fish had no otolith
annulus they are considered part
of
the 1979 year
2000
!
3
1,000
<
2
2
500
U
5
class.'
0
-1977+,9,8-*
FIGURE 2.-Comparison
of
the seasonal distributions of northern
anchovy larval abundance and birth dates of the 1978 year class.
The width of the stippled bars is the effective duration of the
ichthyoplankton survey
(k2
standard deviations of the sample date
where each sample is weighted by its catch
of
northern anchovy lar-
vae). The dashed line was used to interpolate larval abundance per
30-d period. The open histogram indicates the fraction of juvenile's
birth dates occurring per 30 d.
TABLE
5.-Frequency of observed birth dates of juvenile
northern anchovy stratified by year and region and sum-
marized by 10-d date interval.
1978 1979
oate
,ntewai
mnn
central nonh
central souin
<-30
3
0
7
1
2
Dec
-30
2
0
0
0 0
-20
2
0
2
0 0
-10
6
0
2
1
0
Jan
0
1
0
4
1
0
10 5 2
5
0 0
20
3
1
3
0
1
Feb
30
1
4 3
0 0
40
1
0
1
1
0
50
6
0
2 3
1
Mar
60
2 9
0
2 4
70
0
9 7 7 5
80
1
6
3 7 5
100
0
14
0
1
0
110
1
16
0
0 0
130
0
7 2
0
0
140
0
7
0 0
0
June
150
0
2 2
0 0
160
0
2
1
0 0
170
0 0
1
1
0
Apr
90 2 12 2 7 7
May
120
0
10 2
1
1
Julv
>180
0
4
3
13 2

FISHERY
BULLETIN
VOL.
81.
NO.
4
TABLE
6.-
Sidbirth-date nomograms stratified by
10
mm size interval and
30-d date interval (labeled by approximate month).
Sire
Binh
month
(mm)
<Dec.
Dec
Jan.
Feb.
Mar.
Apr
May
June
>June
1977-78
40
12
50 421
60
10
21410
0
1
70 2 7205
1
80 4
4
15
10
90
2 54431
,ntewal
100
1
532
1978-79
30
4
40
6
50 16
60
30432
70 21991
1
80
1
2761781
90
7 3621
100
2
011
TABLE
7
-Birth date distributions calculated from size/birth date nomograms (Table6) and regional
juvenile size distributions (Table
4)
Distributions are presented as %/30-d date intervals (labelled as
approximate months) Comblned birth-date distribution is mean ofregional distributions with weight-
ing factors proportional
to
nearshore shallow area (Table
4)
Mortality correctlon factor accounts for
the greater duration that the early born fish are exposed to juvenile mortality (see text) After multiply-
ing the combined distnbutions by the mortality correction factors, the distributions
are
presented only
for those months with larval abundance data
Weight
<Oec
Dec
Jan
Feb
Mar
Apr
May
June
>June
factor
Regional distributions
1978
centra1 08
21 55 76 237 419 154 20
10
834
nonh
52 168 153 147 215 218
41
05
01
166
soulh
105 55 145 105 349 155 31 22 33 665
Central
95 31
104
93 347 140
76
59 55 169
nonh
285
61
219 154 170 76 19
11
05 166
1979
Combined dirlributionr
1978 15 46 71 a8 234 385 135 17
09
1979 133 52 150
111
320 138 37 27 32
Monalily
correction
factors
Combined distributions
corrected
for
iuvencle mOnal8ty
209 185
164
145 128 113
10
1978 70 96 104 247 359
111
13
1979 244 159 406 155 36
1979 a5 218 142 363 139 32 21
Combining the regional results to produce an
overall juvenile birth-date distribution is prob-
lematic, especially in 1978 when no samples were
collected south of San Diego. Each region’s weight-
ing factor should be proportional to the abundance of
juveniles in the region. Because local abundance of a
pelagic schooling fish is measured crudely by a trawl
survey, the areas of the primary juvenile habitat
(Table
4)
were used as weighting factors. The north
region has only 16.6% of the total area nearshore of
the 90 m
(50
fathom) isobath,
so
contributes little to
the total. Although the north region contributed
nearly
50%
of the total area from which samples were
obtained in 1978 (Table
4),
I
assume that the un-
sampled fish from Baja California had birth dates
similar to those of San Diego-Pt. Dume fish
so
I
use
746
16.6% for the north‘s weighting factor in 1978. The
combined birth-date distributions are in Table 7.
Correction
for
Juvenile Mortality
The birth-date distributions for the northern
anchovy presented above represent the birth dates of
those fish which survived until November. A monthly
cohort’s contribution to the birth-date distribution of
its year class
is
a function
of
the spawningrate during
that month, the mortality rates experienced by that
cohort, and the age of that cohort when sampled
Northern anchovy juveniles which had been born
during January are expected to be less abundant in
November than juveniles born in
May
because the
older juveniles experienced mortality as juveniles for

METHOT
SURVIVAL
OF
LARVAL
ENGRAIJLIS
MORDAX
a longer period. A correction for this difference in age
is necessary before differences between the seasonal
distribution of larval abundance and the resultant
distribution
of
juvenile birth dates can be interpreted
as differences in larval survivaL Few juveniles collect-
ed in November were <5-mo-old
so
the relative
abundance of older juveniles need only be adjusted
by the inverse
of
survival from age
5
mo to age
at
cap-
ture. Age-specific survival rate was assumed to
increase from 64% per month at age 3 mo to
88%
at
10 mo (calculated from preliminary estimates of
juvenile mortality rates in Smith 1981).
If
one
assumes no seasonality in juvenile survival, the resul-
tant birth-date distributions are as
if
all the monthly
cohorts had been sampled at the same age rather
than at variable ages in November. Because most
juveniles were between 6 and 10 mo old, corrected
birth-date distributions are similar to the uncor-
rected distributions (Table 7).
L
a
I
v)
>
v)
w
>
I-
<
A
w
K
a
a
-
Relative Larval Survival
The juveniles’ birth-date distributions, corrected
for juvenile mortality, and the seasonal distributions
of larval production
for
the northern anchovy are pre-
sented in Figures
2
and
3.
The ratio of monthly birth-
date frequency to monthly larval production is an
index of larval survival relative to survival from other
months
in
the same spawning season. Survival tended
to increase within the 1978 season and decrease
within the 1979 season (Fig. 4). In both years the only
anomalies to the trends were low relative survival of
larvae born in February.
DISCUSSION
This study has documented seasonal changes in
survivorship of larval northern anchovy. Both the
magnitude and the timing of changes are important.
The magnitude of the seasonal changes determines
whether annual variation in recruitment could be
caused by short seasonal events. The timing of the
changes in survival relative to environmental events
elucidates the linkage between oceanographic con-
ditions and recruitment.
Temporal Patterns in Larval
Survival
To evaluate the importance
of
seasonal changes in
larval survival
of
the northern anchovy requires
estimates of annual variation in recruitment. The age
composition of the central population of northern
anchovy is monitored through the fishery and trawl
~~
v-
DEC
JAN
FEE
MAR APR MAY JUN
MONTH
FIGURE
4.-Relative survivorship is the ratio
of
the fraction
of
north-
ernanchovyjuvenile’s birthdates to the fractionofannuallarval pro-
duction per
30
d.
To
enable comparison between years the relative
survivorship for
1978
has been scaled by the
ratio
of recruitment
(2.0)
and the ratio
of
annual larval production
(0.98)
between the
2
years.
surveys. Mais (198la) analyzed the age composition
of the commercial fishery in southern California and
suggested the following ranking of recent year
classes: 1974 weak, 1975 weak, 1976 mediocre-
strong, 1977 weak, and 1978 very strong. During the
1978-79 fishing season the 1978 year class
of
the
northern anchovy contributed 65% of the southern
California catch and in the following season the 1979
year class contributed 35% (Mais 198la). In the
spring 1979 trawl survey the 1978 year class con-
tributed62% andinspring 1980 the 1979 yearclass con-
tributed 35% (Mais 1980, 1981b). Thus the fishery
and the survey indicate that the 1978 year class was
about twice as large as the 1979 year class.
This difference in recruitment cannot be explained
by the abundance of young northern anchovy larvae.
Northern anchovy larval production in 1979 was
2.1% greater than in 1978. Thus the larger 1978 year
class resulted from higher survival because larval
abundance was less in 1978 than in 1979.
The critical question is whether the recruitment
variation described above requires more annual
variation in larval survival than is caused by the
seasonal changes described in this study.
I
evaluated
this question by scaling the northern anchovy larval
abundance estimates (Figs.
2,3)
with the ratio of lar-
747

FISHERY
BULLETIN:
VOL.
81,
NO.
4
for northern fish, if currents and eddies did not sub-
stantially redistribute the larvae. In addition, if
juveniles routinely move northward along the coast,
early born fish will be further north by November.
There are no data on the distribution of late larvae
with which to investigate the cause of latitudinal pat-
tern in juvenile birth dates. Geographic pattern in
birth date will contribute to geographic pattern in
size
at
age of adults.
Val abundance between the 2 years (0.979), and the
monthly fractions of juvenile birth dates (Table 7)
were scaled by the ratio of recruitments
(2.0).
The
ratio, (scaled birth date fraction)/( scaled larval abun-
dance fraction), estimates relative survival of a
month's spawn (Fig. 4). Because of the scaling, these
ratios can be compared both between and within
years. Survival of winter spawn in 1979 was similar to
survival of winter spawn in 1978, but survival was
much greater in April-May 1978 than April-May
1979. Thus, the larger 1978 year class was not
necessarily the result of greater survival throughout
the spawning season. The increase in survival during
the last
2
major months of the 1978 spawning season
was sufficient to cause a large increase in
recruitment.
Detection
of
Changes in Survival
Detection of a match between spawning and favor-
able environmental conditions of the northern
anchovy seems more likely than detection of an event
which results in poor survivaL
If
a short-duration
favorable environmental period results in a doubling
of year class abundance, then more than half of the
year class would have been born during that period
Such a concentration of birth dates could have been,
but was not, detected in 1978 and 1979. A particular
match apparently was not necessary for these two
year classes. Conversely, a short- duration unfavor-
able environmental period that destroys all northern
anchovy larvae born during the period may cause
only a small reduction in year class abundance and a
short gap in the birth-date distribution that would be
difficult to detect with small sample sizes.
The effect of other environmental events will be
more difficult to detect. Long-duration events
or
events that affect larvae of a wide age range will have
little effect on the birth-date distribution. Secondly,
environmental events which do not extend over the
geographic range of spawning may not be detected
even though they have important local effects.
Spatial Pattern
This study was designed to study temporal changes
in frequency of juvenile birth dates but a strong spa-
tial pattern also was detected. Northern anchovy
juveniles collected north of Pt. Dume were larger
than southern juveniles because
of
earlier birth
dates. Hewitt and Methot (1982) showed that spawn-
ing contracted towards the San Diego area as the
1978 and 1979 spawning seasons progressed. This
trend would contribute to an earlier mean birth date
748
Relation to Environmental
Conditions
Two oceanographic factors, stability of the upper
water column (Lasker 1978) and offshore transport
(Parrish et aL 1981), have been suggested as factors
in recruitment of northern anchovy. These factors
should have had different effects
on
recruitment in
1978 and 1979.
Lasker (1975, 1978) suggested that northern
anchovy larvae are likely to encounter adequate prey
concentrations only when the upper water column is
stable and prey are aggregated into layers. Of par-
ticular importance
is
the inshore chlorophyll max-
imum layer which may be composed of dino-
flagellates suitable as prey for first feeding larvae.
These layers are homogenized as storms
or
upwelling
events destroy the stratification of the upper tens of
meters ofthe watercolumn. The winter of 1977-1978
was particularly stormy (Lasker 1981) and the
isothermal surface layer was as deep as
50
m until
stratification was restored in March. This hypothesis
correctly predicts lower larval survival in winter 1978
than in spring 1978, but incorrectly predicts a poor
year class in 1978.
Any hypothesis concerning the availability of prey
will predict increasing survival through the major
spawning season (December to June). Zooplankton
biomass increases to its seasonal maximum in June
(Smith and Lasker 1978). Stratification (differences
between temperature at
10
and
30
m) increases
so
the prey of larval fish probably are increasingly
aggregated into layers. Day length increases
so
that
larvae can feed longer per day (Hunter 1972).
A
measurable response of northern anchovy larvae to
the above factors
is
an increase in growth rate from
0.43
mdd in January to
0.55
mm/d in June (Methot
1981) despite trivial changes in mean temperature
(Table
3).
If
food availability is important to larval
survival then survival should consistently increase
through the spawning season.
The second major hypothesis concerns offshore
transport and coastal upwelling caused by the pre-
dominantly northwest winds (Parrish et al. 1981).

METHOT
SURVIVAL
OF
LARVAL
E.V(;R-\[
'1.19'
Mf)H/)AS
Northern anchovy larvae which are transported
offshore may experience higher larval mortality
rates, and the survivors may be unable to return to
the inshore juvenile nursery areas. Monthly indices
of upwelling (Bakun 1973) at lat.
30"
and
33"N,
rela-
tive to long-term monthly means, were exceptionally
low during January-March 1978 (downwelling
occurred) and remained below normal through May
1978 (McLain and Ingraham 1980). Seckel et al.
(1978) suggested that these conditions entrained lar-
vae close to shore and correctly predicted an abun-
dant 1978 year class. However, the results obtained
here suggest that larval survival was higher during
late spring 1978 than during the winter when
downwelling occurred.
Upwelling was as
low
in December 1978 and
January 1979 as in winter 1978, but the storms were
less severe in 1979. Later in 1979, upwellingwasnear
normal. The increased upwelling in 1979 relative to
1978 may have been responsible for the greater
offshore displacement of northern anchovy larvae in
1979 (Table
4).
The transport hypothesis correctly
predicts a poorer year class in 1979 relative to 1978
and decreasing larval survival through the 1979
spawning season.
This brief examination of environmental data does
not completely account for the patterns of recruit-
ment of the northern anchovy in 1978 and 1979.
Indices of offshore transport seem more important
than indices of food availability, but the seasonal pat-
tern of survival in 1978 could not be explained by
transport. It
is
simplistic to assume that only one fac-
tor is involved in recruitment and that the effect of
this factor is linear. One plausible scenario is that the
winter storms of 1978 caused high mortality of early
larvae, but the low upwelling throughout theyear per-
mitted high entrainment of late larvae and resulted in
the good year class. It is also possible that the
extremely low upwelling during winter 1978 did not
have a proportionally greater effect than the low
upwelling of spring 1978. Relatively low upwelling
late in the spawning season may be important
because absolute upwelling and transport typically
increase through the spring (Smith and Lasker 1978).
The spawning season may be timed to avoid low food
availability in winter and high transport in late
spring.
Other evidence indicates that survival
of
early
northern anchovy larvae was nearly constant during
the 1978 and 1979 spawning seasons. Hewitt and
Methot (1982) inferred larval mortality from the
slopes of the larval age-frequency distributions and
found no significant seasonal changes. This evidence
is
consistent with the hypothesis that the significant
change in survival which caused the difference in re-
cruitment occurred after the early larval stage.
Adverse larval drift would not necessarily cause
increased mortality during the age interval examined
by Hewitt and Methot but may affect the fraction of
the surviving larvae which are entrained in the range
of the juvenile habitat.
LITERATURE CITED
BAKLY, A.
1973. Coastal upwelling indices. west coast of North
America. 1946-71.
17.5.
Dep. Commer.. NOAA Tech.
Rep. NMFS SSRF-671.
10:i
p.
BEYER.
J.
E.,
A\LI
G. C.
L.-\txE\rE.
1981.
Aspects of stochasticity in modelling growth and sur-
vival of clupeoid fish larvae. Rapp. P.-\'. Reun Cons. Int.
Explor. Mer 178:17-23
BROTHER^.
E,
B..
C.
P.
MA.I'HKWS.
AUI)
R. LA~KEH.
1976. Daily growth increments in otoliths from larval and
adult fishes. Fish. Bull.,
Y.S.
;4:1-8.
Col.i.Iss. R. A..
.&si)
J.
D.
SPRA
1-1'
1969. Age determination of northern anchovies,
Erigruu/i,\
mordnx,
from otoliths.
In
J.
D. Messersmith (editor). The
northern anchovy
(Engrnuii,,
mordax)
and its fishery
1965-1968, p. 39-55. Calif. Dep. Fish Game. Fish
Bull..
147.
CONOVER,
w.
J.
1971.
Practical nonparametric statistics. Wiley.
N.Y..
462 p.
CllSHING,
D.
H.
1975.
Marine ecology and fisheries. Camb. Univ. Press,
Camb., Engl.,
278
p.
CL'SHISG.
D.
H.,
A~D
J.
C.
K.
HARHIS.
1973.
Stock and recruitment and the problem of density
dependence. Rapp.
P:V.
Reun. Cons. Perm. Int. Explor.
Mer 164:142-155.
F~rr)~rc~ssos.
A.
1934. On the calculation of age-distribution within a stock of
'cod by means of relatively few age determinations as a key
to measurements on a large scale. Rapp.
P.-V.
Reun
Cons. Int. Explor. Mer 86:l-15.
HEWITI',
R.,
A\I)
R.
D. METHOT,
JR.
in
1978
and
19i9.
CalCOFI Rep. 23:226-245.
1982.
Distribution and mortality of northern anchovy larvae
HJORT,
J.
1926. Fluctuations in the year classes of important food
fishes.
J.
Cons. Int. Explor. Mer
1:5-38.
HLINTER.
J.
R.
1972.
Swimming and feeding behavior of larval anchovy
Engrnulis
rnordnx.
Fish Bull..
U.S.
70:821-838.
KIMI'RA, D.
K.
1977. Statistical assessment of the age-length key.
J.
Fish
KRAMER, D., M.
J.
KALIN,
E. G.
STEVENS,
J.
R.
THRAILKILL,
AND
J.
1972.
Collecting and processing data on fish eggs and larvae
in the California Current region
U.S.
Dep. Commer.,
NOAA Tech Rep. NMFS Circ.
370,38
p.
Res. Board Can. 34:317-324.
R.
ZU'EIFEL.
LASKER, R.
1975.
Field criteria for survivalof anchovy larvae: therelation
between inshore chlorophyll maximum layers and suc-
cessful first feeding. Fish. Bull.,
U.S.
73:453-462.
7
49

FlSHERY BULLETIN: VOL
81.
SO.
4
1978. The relation between oceanographic conditions and
larval anchovy food in the California Current: identifica-
tion of factors contributing to recruitment failure. Rapp.
P.-V. Reun. Cons. Int. Explor. Mer 173912-230.
1981. Theroleofastableoceanin larvalfishsurvivaland sub-
sequent recruitment.
In
R.
Lasker (editor), Marine fish
larvae: Morphology, ecology, and relation to fisheries, p.
80-87. Wash Sea Grant Program, Univ. Wash Press.
McLAI\. D. R.,
AND
W.
J.
INCRAHAM, JR.
1980. Marine environmental conditions in the eastern North
Pacific Ocean
In
E. D. Haynes (editor), Marine environ-
mental conditions off the coasts
of
the United States
January 1978
-
March 1979, p. 5-39.
U.S.
Dep. Commer..
NOAA Tech Memo. NMFS-OF-5.
MAIS,
K
F.
1974. Pelagic fish surveys in the California Current. Calif.
Fish Game, Fish Bull. 162, 79 p.
1980. California Department of Fish and Game sea survey
cruises 1979. CalCOFI Data Rep. 2232-87.
198
1
a. Age-composition changes in the anchovy.
Engraulis
rnordax,
central population. CalCOFI Rep. 22:82-87.
1981b. California Department of Fish and Game sea survey
cruises 1980. CalCOFI Data Rep. 3O:l-64.
MARR. J. C.
1956. The "critical period" in the early life history of marine
fishes.
J.
Cons. Int. Explor. Mer 21:160-170.
MAY,
R.
C.
1974. Larval mortality in marine fishes and the critical period
concept.
In
J.
H.
S.
Blaxter (editor), The early life history
of fish, p. 3-20. Springer, Heidelberg.
METHOT,
R.
D., JR.
1981. Growth rates and age distributions of larval and juve-
nile northern anchovy,
Engraulis
rnordax.
with inferences
on larval survival. PhD. Thesis, Univ. California, San
Diego,
200
p.
METHOT.R. D.. JR.,
AND
D. KHAMER.
1979. Growth of northern anchovy,
Engradis
rnordu,
larvae
in the sea. Fish Bull..
U.S.
77:413-423.
NELSON, W.
R.,
M. C. INGHAM,
AND
W.
E. SHAAF.
1977.
Larval transport and year-class strength of Atlantic
menhaden,
Rreuoortia
tyrannus.
Fish Bull.,
U.S.
75:23-
41.
PASNELLA, G.
1971. Fish otoliths: daily growth layers and periodical pat-
terns. Science (Wash., D.C.) 173:1124-1127.
PARRISH,
R. H., C.
S.
NELSON,
AND
A. BAKCN.
1981. Transport mechanisms and reproductive success of
fishes in the California Current. Biol. Oceanogr. 1:175-
203.
SECKEL,
G.
R., A. BAKIK,
AND
R. PARRISH.
1978. Large-scale atmospheric circulationand the west coast
upwelling regime in 1976/77 and
1977l78.
In
Pro-
ceedings of the Third Annual Climate Diagnostics
Workshop: Cooperative Institute for Marine and
Atmospheric Studies, University of Miami, Miami,
Florida. October 31-November
2,
1978.
p.
7-1-7-12.
V.S.
Dep. Commer.. NOAA. Natl. Clim. Program Off..
Rockville. Md.
SMITH,
P.
E.
1981. Sampling to determine anchovy larval mortality in the
sea.
In
R.
Lasker (editor), Marine fish larvae: Morphol-
ogy. ecology, and relation to fisheries, p.
12-17.
Wash.
Sea Grant Program, Univ. Wash Press.
SMITH,
P.
E.,
AND
R. LASKER.
1978 Position of larval fish in an ecosystem. Rapp.
Reun. Cons. Int. Explor. Mer 173:77-84.
VLYMES,
W.
J.
1977. A mathematical model of the relationship between lar-
val anchovy
(Engraulk rnordax)
growth, prey microdis-
tribution, and larval behavior. Environ. BioL Fishes
2:211-233.
WYAW,
T.
1980. The growth season in the sea. J. Plankton Res. 2:81-
97.
150