ICES Journal of Marine Science, 53: 139–146. 1996
Relationships between sea surface temperature, salinity, and
pelagic fish distribution off northern Chile
J. Castillo, M. A. Barbieri, and A. Gonzalez
Castillo, J., Barbieri, M. A., and Gonzalez, A. 1996. Relationships between sea surface
temperature, salinity, and pelagic fish distribution off northern Chile. – ICES Journal
of Marine Science, 53: 139–146
Geographic Information System techniques were used in the analysis of the distribution of three pelagic resources: anchovy, sardine, and jack mackerel. The data were
derived from hydroacoustic estimations obtained during four seasonal cruises in
1993–1994 in the region off northern Chile. The spatial distribution of these resources
was analysed relative to the surface temperature and salinity distributions, the latter
two variables being measured concurrently during these cruises. Results indicated the
occurrence of a stratified distribution of the three species together with a high
variability in the distribution pattern between cruises. The distribution of the three
species was associated with the occurrence and intensity of thermal and haline fronts.
? 1996 International Council for the Exploration of the Sea
Key words: acoustic, geographical distribution, Geographical Information System,
pelagic fishes, salinity, temperature.
Address correspondence to: J. Castillo and M. A. Barbieri, Instituto de Fomento
Pesquero (IFOP), Casilla 8-V, Valparaíso, Chile. M. A. Barbieri and A. Gonzalez,
Escuela de Ciencias del Mar, Universidad Católica de Valparaíso-Chile, Casilla 1020,
Valparaíso, Chile.
Introduction
The Humboldt Current ecosystem, dominating the east
coast of South America, shows environmental variability with important changes at short, medium (such
as the ‘‘El Niño’’ phenomenon; annual seasonality and
daily variations), and long-term time scales (geological
scale) (Guillén et al., 1985; Alheit and Bernal, 1993).
Although El Niño has been widely studied to analyse its
impact on fisheries (Arntz et al., 1985; Sanhueza, 1985;
Martínez et al., 1985), there are few papers describing
the spatial distribution of the resources relative to the
environmental variables (Castillo and Guzmán, 1985;
Swartzman et al., 1994).
One of the advantages of acoustic methods is the
possibility of studying the spatial distribution of
resources (Baussant et al., 1993; Simard et al., 1993) and
the inter-relationships of environmental conditions, for
example the influence of the thermocline in the vertical
sense (Swartzman et al., 1994), or the horizontal fronts
in the horizontal sense, which becomes a powerful tool
for understanding the ecosystem and resource dynamics
(Simard et al., 1993; Boudreau, 1992; Scalabrin and
Massé, 1993).
Geographic Information System (GIS) techniques are
widely used in processing satellite images. However,
1054–3139/96/020139+08 $18.00/0
these techniques, with some exceptions, have had
little application in marine sciences (Barbieri et al.,
1989; Yañez et al., 1994), although information from
hydroacoustic surveys conforms quite well with GIS
requirements.
This paper analyses acoustic information on the
geographical distribution of sardine (Sardinops sagax
Girard), anchovy (Engraulis ringens Jenyns), and jack
mackerel (Trachurus murphyi Nichols), relative to
physical oceanographic conditions, mainly surface
temperature and salinity off northern Chile.
Materials and methods
Surveys
Information was taken from four seasonal acoustic
survey cruises: winter and spring in 1993; summer and
autumn in 1994, conducted in northern Chile, between
18)20*S and 24)00*S latitude, and up to 100 nautical
miles (nmi) off the coast. The first of these cruises was
conducted aboard the RV ‘‘Carlos Porter’’ (27 m length,
500 hp); the stern trawler RV ‘‘Abate Molina’’ (43.2 m
length, 1400 hp), was used for the other cruises.
? 1996 International Council for the Exploration of the Sea
140
J. Castillo et al.
Anchovy
72°
71° W 70°
18°
S
19°
20°
21°
22°
23°
24°
Sardine
Densities ranges (SA)
0001 – 0527
0528 – 1056
1057 – 2110
2111 – >
Coast
Jack mackerel
Winter
Spring
Summer
Autumn
Figure 1. Distributions of anchovy (Engraulis ringens), sardine (Sardinops sagax), and jack mackerel (Trachurus murphyi) in winter
and spring of 1993, summer and autumn of 1994 off northern Chile. (The coast on the right of each map is from 24)S to 18)S.)
Number of cells (thousands)
Sea surface temperature, salinity, and pelagic fish distribution
20
18
16
14
12
10
8
6
4
2
0
141
graphic position (lat., long.) was recorded. The temperature and salinity information between surface and 600 m
was registered by a SEABIRD CTD in winter and a Neil
Brown CTDO during the other cruises. For this paper,
surface level information was used.
Information processing
Winter
Spring
Summer
Information was processed using GIS techniques. The
geographic distribution for the different species was
determined taking into account the Sa per ESDU
assigned to each species and the oceanographic information. Data were interpolated using the inverse-square
distance method. For each cruise, surface temperature
and salinity as well as the species distribution images
were generated by this procedure.
Within and between cruises, species coincidence was
analysed by correlation and cross-classification analysis
between associated images. These images were arranged
to have the same number of cells using the interpolation
process. Interrelationships between species distribution
and oceanographic variables was similarly analysed.
Four categories of densities in species distribution
were established for this procedure. Temperature
categories were defined for every 1)C and salinity every
0.1.
Association between species distribution images and
oceanographic variables was checked by Cramer’s V
correlation coefficient (Larson and Mendenhall, 1983),
which ranges from 0 for no correlation and 1 for perfect
correlation. Cramer’s V correlation coefficient significance was tested by Chi-square at the 95% confidence
level.
Species distributions were analysed by the Kappa
Index of Agreement (KIA) (Rosenfield and Fitzpatrick,
1986). This index also ranges between 0 and 1, and
is interpreted in the same way as the Cramer’s V
coefficient.
Autumn
Figure 2. Numbers of cells where species occurred by surveys.
=Jack mackerel; =sardine; =anchovy.
Acoustic information
Hydroacoustic information was obtained with SIMRAD
EK-500 38 kHz split beam echo-sounders, interconnected to GPS systems. The acoustic equipment was
calibrated at the beginning of each cruise (Foote et al.,
1987).
Echo integration was carried out between the surface
and 250 m depth and the sounder was operated with
1 ms pulse length and 2 KW transmitter. An ESDU of
1 nmi was defined. A systematic sampling design was
applied with parallel transects perpendicular to the
coast, separated by 20 nmi in winter and 35 nmi at other
times.
Species recognition was done by means of purse seine
fishing hauls during the winter survey and by midwater
trawls by the RV ‘‘Abate Molina’’ during other surveys.
Oceanographic information
The oceanographic information was collected at discrete
stations over the acoustic transects of 1, 5, 10, 20, 40, 70,
and 100 nmi off the coast. At each station the geo-
Table 1. Cells of coincidence for the occurrence of pairs of species and between surveys.
Winter
A
Autumn
Summer
Spring
Winter
A
S
J
A
S
J
A
S
J
S
J
S
Spring
J
192
A
S
Summer
J
4010
1438
200
S
J
A
J
3965
942
831
148
10 114
253
1306
2379
1147
4414
2009
11
4042
1051
80
265
403
1171
A
Autumn
4197
4216
698
1199
A=Anchovy; S=Sardine; J=Jack mackerel.
1478
3577
732
601
142
J. Castillo et al.
18°
72°
71° W 70°
Temperature
S
19°
20°
21°
22°
23°
24°
Temp.
Salinity
Coast
Coast
15
34.5
16
34.6
17
34.7
18
34.8
19
34.9
20
35.0
21
35.1
22
35.2
23
35.3
24
35.4
25
35.5
26
35.6
Salinity
Winter
Spring
Summer
Autumn
Figure 3. Sea surface temperature and salinity in winter and spring of 1993, summer and autumn of 1994 off northern Chile. (From
24)S to 18)S.)
Results and discussion
Geographic distribution
Species distribution maps were based on 43 686 cells.
Results indicated (Figs 1 and 2), in general, a highly
aggregated pattern, occupying small sectors with relatively high densities, particularly in the case of anchovies
and sardines.
Table 1 gives the number of coincident cells between
species within a cruise and between cruises. It can be
observed that, within a cruise, species were stratified.
This situation was even more evident in winter and
spring, when anchovies were clearly distinguished
from other species. Thus, in winter, anchovies shared
only 403 cells with sardines and 1171 with jack
mackerels. A greater degree of mixing occurred
between sardines and jack mackerels, with which 4216
cells were found to be shared while in spring anchovies
shared 698 cells with sardines and 1199 with jack
mackerel.
Mixing between anchovy and sardine increased
in summer and autumn; 3577 cells were common in
summer and 3965 in autumn. These results concur with
the observations of Yañez et al. (1994), who also registered stratification of the fishing zones, depending on the
captured species.
Analysis between cruises enables one to perceive the
high dynamism present in the areas occupied by each
species, which are relatively low in coincident sectors.
Only in the case of anchovy, in summer and autumn, is
Sea surface temperature, salinity, and pelagic fish distribution
8000
(a)
Winter
6000
4000
4000
2000
2000
0
0
(c)
Spring
Frequency (No. of cells)
8000
6000
8000
6000
4000
4000
2000
2000
16
18
20
8000
22
6000
4000
4000
2000
2000
0
0
(g)
Spring
6000
4000
4000
2000
2000
34.6
35
(d)
18
20
0
34.2
35.4
Salinity
22
24
Summer
(f)
Autumn
(h)
8000
6000
0
34.2
Autumn
8000
6000
8000
(b)
0
14
16
24
Temperature (°C)
(e)
Winter
Summer
8000
6000
0
14
143
34.6
35
35.4
Figure 4. Numbers of cells where species occurred in relation to temperature (a to d) and salinity (e to h) in winter and spring
of 1993, summer and autumn of 1994. (. . . .)=anchovy; (– – –)=sardine; (——)=Jack mackerel; =thermal front; =haline
front.
it possible to observe a high coincidence in the occupied
areas, with 10 114 cells in common.
These results suggest a very high variability in the
geographic distribution of a species between different
seasons of the year, making it very difficult to establish a
pre-stratification sampling criteria.
Oceanographic conditions and resources
distribution
Figure 3 shows the surface oceanographic conditions of
temperature and salinity for the four cruises under
consideration. Temperatures gradually increased from
144
J. Castillo et al.
Table 2. Cramer’s V correlation of the comparison of the distributions of the different species and
between surveys.
Winter
A
Autumn
Summer
Spring
Winter
A
S
J
A
S
J
A
S
J
S
J
S
Spring
J
0.45
A
S
Summer
J
0.48
A
J
A
J
0.46
0.58
0.58
0.58
0.50
0.44
0.45
0.58
0.46
S
Autumn
0.45
0.58
0.48
0.45
0.45
0.45
0.45
0.46
0.45
0.46
0.45
0.45
0.45
0.46
0.45
0.45
0.46
0.45
0.47
A=Anchovy; S=Sardine; J=Jack mackerel.
Table 3. Kappa coefficient of the comparison of the distributions of the different species and between
surveys.
Winter
A
Autumn
Summer
Spring
Winter
A
S
J
A
S
J
A
S
J
S
J
S
Spring
J
0.39
A
S
Summer
J
0.45
0.44
A
J
A
J
0.39
0.37
0.52
0.45
0.57
0.49
0.33
0.41
S
Autumn
0.43
0.55
0.48
0.45
0.45
0.31
0.34
0.38
0.17
0.29
0.59
0.55
0.50
0.28
0.31
0.37
0.64
0.46
0.55
A=Anchovy; S=Sardine; J=Jack mackerel.
17–18)C in winter to 24–26)C in summer, decreasing in
autumn to 20)C. Along with this process, towards the
coast, an increase in the thermal gradient was observed.
Salinity, too, showed an increment ranging from 34.7–
35.3 in winter to 34.7–35.5 in summer, decreasing in
autumn almost to winter levels. In this case, coastal
gradients did not show greater variations, indicating the
presence of permanent upwelling events all year round.
In general, the temperature pattern that limited the
species distribution was unimodal (Fig. 4a–d), except for
the anchovy in winter, which showed a secondary mode.
An increasing trend in the upper limit of the temperature
ranges in the warmer seasons was also observed, varying
from 14 to 19)C in winter, 15 to 22)C in spring, reaching
24)C in summer, and 15–20)C in autumn.
In all cases, it was observed that the anchovies
remained in temperature and salinity ranges lower than
those of sardines and mackerels, especially in summer
and autumn.
Range limits for salinity in winter varied between 34.6
and 35.3 (Fig. 4e–h), with three modes for the jack
mackerel centred at 34.7, 35.0, and 35.3; bimodal for
anchovy centred in 34.8 and 35.2; and unimodal for
sardine at 35.0. In spring, the lowest salinity range limits
increased to 34.7 with a unique major mode for the three
species centred at 34.9. In summer and autumn, modes
for anchovy were 34.7 and 34.9, whereas sardine was
located at 34.9 and 35.0 modes, repectively. In summer,
the jack mackerel occupied a wider range of salinities
than the other species, with a mode at 35.2.
The spatial distribution of the species was influenced
by the presence and intensity of coastal thermal and
haline fronts. Thus, the anchovy was associated with the
anterior zone of the thermal front, and therefore pushed
towards the coast during the spring and summer seasons
when the front becomes stronger. In winter, the thermal
gradient weakened and the anchovy broadened its
western distribution limit. Sardines were mainly located
Sea surface temperature, salinity, and pelagic fish distribution
145
Table 4. Cramer’s V correlation coefficient of the comparison anchovy distribution (A), sardine (S), and jack mackerel (J) with
temperature (T) and salinity (S) and between surveys.
Winter
T ()C)
S
Spring
Summer
Autumn
A
S
J
A
S
J
A
S
J
A
S
J
0.187
0.193
0.173
0.193
0.169
0.235
0.259
0.201
0.237
0.160
0.254
0.178
0.391
0.374
0.308
0.289
0.298
0.269
0.273
0.2735
0.251
0.253
0.302
0.290
to the west side of the thermal front; their distribution
therefore became wider towards the ocean compared
with that of the anchovies.
Because of the jack mackerel’s broad bathymetric
distribution, the relationship of this species relative to
the surface environmental conditions is not considered
further here.
for their assistance during the cruises; Dr Eleuterio
Yañez and José Blanco for their comments on the data
processing and Professor Iván Sepúlveda for help in
the translation into English. We are also grateful to
the reviewers, whose comments helped improve our
paper.
References
Statistical analysis
The statistical analysis for comparison of the species
distribution indicates that there is enough evidence to
reject the null hypothesis of equal distributions, with
chi-square in all cases greater than 50 000. Cramer’s V
coefficients varied between 0.44 and 0.58 (Table 2),
indicating some degree of correlation between images.
This confirms with what has already been said, since, in
all cases, there is some degree of overlapping in those
cells occupied by more than one species. Kappa indexes
varied between 0.17 and 0.64 (Table 3), being greater in
cases in which there is a higher number of cells shared by
more than one species and near zero when the number of
shared cells of equal category was fewer.
Similarly, the statistical analysis for the association
of species distribution and oceanographic conditions
rejected the null hypothesis of equality between distributions and temperature or salinity conditions, with
chi-squared figures between 5153 and 38 400. Cramer’s
index figures varied between 0.16 and 0.39 (Table 4),
thus indicating a low correlation between species distributions and oceanographic variables. This seems to be
logical if one considers that the process associates each
density category with each category of the oceanographic variable (1)C for temperature and 0.1 for
salinity). Therefore, these results lead one to consider
that the analysis should be done with temperature and
salinity categories such that either thermal or haline
gradients are included, because they are the factors
which may greatly affect the species distribution.
Acknowledgements
This research was supported by the Fondo de Investigación Pesquera (FIP). We thank the Captains and
crews of RV ‘‘Carlos Porter’’ and ‘‘Abate Molina’’
Arntz, W., Landa A., and Tarazona, J. (Eds). 1985. El Niño, su
impacto en la fauna marina. Instituto del Mar del Perú.
Boletín Extraordinario. Callao-Perú 222 pp.
Alheit, J. and Bernal, P. 1993. Effects of physical and biological
changes in the biomass yield of the Humbolt current ecosystem. In Large marine ecosystems. Stress, mitigation and
sustainability, pp. 53–68. Ed. by K. Sherman, L. Alexander,
and B. Gold. AAAS Press Publication, USA.
Barbieri, M. A., Yañez, E., Farías, M., and Aguilera, R. 1989.
Determination of probable fishing areas for albacore
(Thunnus alalunga) in Chile’s central zone. In Quantitative
remote sensing: an economic tool for the nineties, IGARSS
’89, Vancouver, BC, Canada, pp. 2447–2450, 10–14 July
1989.
Baussant, T., Ibanez, F., and Étienne, M. 1993. Numeric
analysis of planktonic spatial patterns revealed by echograms. Aquatic Living Resources, 6: 175–184.
Boudreau, P. 1992. Acoustic observations of patterns of aggregation in haddock (Melanogrammus aeglefinus) and their
significance to production and catch. Canadian Journal of
Fisheries and Aquatic Sciences, 49: 23–31.
Castillo, J. and Guzmán, O. 1985. Variaciones de la distribución espacial de la sardina española y jurel durante El
Niño 1982–83. Investigación Pesquera (Chile), 32: 79–93.
Foote, K., Knutsen, H., Vestnes, G., MacLennan, D., and
Simmonds, J. 1987. Calibration of acoustic instruments for
fish density estimation: a practical guide. Cooperative
Research Report, International Council for the Exploration
of the Sea, No. 144. 57 pp.
Guillén, O., Lostaunau, N., and Jacinto, M. 1985. Características del Fenómeno El Niño 1982–83. In El Niño. Su
impacto en la fauna marina, pp. 11–21. Ed. by W. Arntz, A.
Landa, and J. Tarazona. Instituto del Mar del Perú. Boletín
Extraordinario. Callao-Perú.
Larson, O. and Mendenhall, W. 1983. Statistics: a tool for the
social sciencies. Duxbury Press, Boston.
Martínez, C., Salazar, C., Böhm, G., Mendieta, J., and Estrada,
C. 1985. Efectos del fenómeno El Niño 1982–83 sobre los
principales recursos pelágicos y su pesquería (AricaAntofagasta). Investigación Pesquera (Chile), 32: 129–139.
Rosenfield, G. H. and Fitzpatrick, K. 1986. A coefficient of
agreement as a measure of thematic classification accuracy.
Photogrammetric Engineering and Remote Sensing, 52: 223–
227.
146
J. Castillo et al.
Sanhueza, A. (Ed.). 1985. Taller nacional fenómeno El Niño
1982–83. Investigación Pesquera (Chile), (32). 242 pp.
Scalabrin, C. and Massé, J. 1993. Acoustic detection of the
spatial and temporal distribution of fish shoals in the Bay of
Biscay. Aquatic Living Resources, 6: 269–283.
Simard, Y., Marcotte, D., and Bourgault, G. 1993. Exploration
of geostatistical methods for mapping and estimating
acoustic biomass of pelagic fish in the Gulf of St Lawrence:
size of echo-integration unit and auxiliary environment
variables. Aquatic Living Resources, 6: 185–199.
Swartzman, G., Stuetzle, W., Kulman, K., and Powojowski, M.
1994. Relating the distribution of pollock schools in the
Bering Sea to environmental factors. ICES Journal of
Marine Science, 51: 481–492.
Yañez E., Barbieri, M. A., González, A., Pizarro, O., Bravo,
M., Canales, C., and Salinas, S. 1994. Sea surface thermal
structure associated to the pelagic fish resources distribution
in northern Chile. In Proceedings, Second Thematic
Conference on Remote Sensing for Marine and Coastal
Environments. ERIM, Michigan, USA, Vol II: 149–162.