Estuaries Vol. 27. No.6. p.

~14 Deoe~r 2004

Comparison of 1982-1983 and 1997. 1998 E Nino Effects on the

A. M. GARCIAl.., P. VIEIRAI, K. O. WINEMILLER2, and AM. GRIMM3

1 FundarM Universidade Federal de Rio Grande (FURG), Departamento de Oceanografia,

LaboratiJrio de Ictiologia, C. P. 474-Rio Grande, RS, Brasil
2 Department of Wildlife and Fisheries Sciences,Texas A&M University, C..allege
Station, Texas
77843-2258
3 Department of Physics, Federlll University of Parana, PR, Brazil

ABSTRACT: Meteorological impacts of E1 Nino events of 1982-1983 and 1997-1998 were observed in locations
dtroughout the world. In southern Brazil, E1 Nino events are associated with increased rainfall and higher freshwater
discharge into Patos Lagoon, a large coastal lagoon that empties into the Atlantic Ocean. Based on interdecadal mete-
orological and biological data sets encompassing the two strongest E1 Nino events of the la.~t 50 yr, we evaluated the
hypothesis that E1 Nino-induced hydrological changes are a major driving force controlling the interannual variation in
the structure and dynamics of fishes in the Patos Lagoon estuary. High rainfall in the drainage basin of the lagoon
coincided with low salinity in the estuarine area during both E1 Nino episodes. Total rainfall in the drainage basin was
higher (767 versus 711 mm) and near-zero salinity conditions in the estuarine area lasted about 3 mo longer during the
1997-1998 E1 Nino event compared with the 1982-1983 event. Hydrological changes triggered by both E1 Nino events
had similar relationships to fish species composition and diversity patterns, but the 1997-1998 event appeared to have
stronger effects on the species a.~emblage. Although shifts in species composition were qualitatively similar during the
two E1 Nino events. distance between E1 Nino and non-EI Nino assemblage multivariate centroids was greater during the
1996-2000 sampling period compared with the 1979-1983 period. We provide a conceptual model of the principal
mechanisms and processes connecting the atmospheric-oceanographic interactions triggered by the E1 Nino phenomena
and their effect on the estuarine fish assemblage.

Introduction stood, Timmermann et al. (1999) showed that, un-

The EI Nino phenomenon refers to a warming der a scenario of increasing greenhouse-gas con-
of the tropical Pacific basin in association with a centrations, EI Nino-like conditions would become
weakening of the trade winds that occur at inter- more frequent. The frequency and intensity of EI
vals of 3-7 yr. El Nino events vary in intensity. The Nino events and their effects may also be influ-
1982-1983 and 1997-1998 EI Nino episodes were enced by the phases of interdecadal modes of sea
comparable in magnitude and represented the surface temperature variation (Mestas-Nunez and
strongest warming observed since 1950 (Glantz Enfield 1999, 2001). If El Nino events increase in
2001). Meteorological changes induced by EI Nino severity and duration in response to global change,
events are felt around the world, with important it will become critically important to understand
implications for the dynamics of aquatic ecosys- how meteorological and hydrological patterns as-
tems (Caviedes 2001; Enfield 2001) ranging from sociated with EI Nino events affect ecosystems.
coral reefs (Glynn et aI. 2001) to upwelling regions Patos Lagoon (3~, 490W) represents an appro-
(Brodeur et aI. 1985;Jimenez et aI. 2001; Roy and priate ecosystem to evaluate how disturbances in-
Reason 2001). duced by EI Nino can affect the dynamics and
Some authors suggest that recent EI Nino events structure of fish estuarine assemblages.EI Nino ep-
were stronger and more frequent than those in the isodes have affected the western coasts of the
past (Tudhope et aI. 2001). Although predictions Americas frequently over the last 5,000 yr (Rollins
cannot be conclusive because they may depend on et al. 1986), which coincides with the geomorpho-
atmospheric processes that are not well under- logic age (5,000-6,000 yr) estimated for the Patos
Lagoon system on the Atlantic coast of southern
Brazil (Knoppers and Kjerfve 1999). High rainfall
. Correspondingauthor: tele: +5553/233-6752;fax: +5555/ in southern Brazil during EI Nino events (Grimm
23~1; e-mail:amgarcia@mikrus.com.br et al. 1998, 2000) increases freshwater discharged
C>2004 Estuarine ReseaIdI Federation 905
906 A. M. Garda et 81.

into Patos Lagoon several-fold (Garcia 1996). This versity reveal similar trends during both EI Nino
lagoon is one of the world's largest coastal aquatic events? Did rainfall and salinity conditions prior to
ecosystems (10,360 km2 with a drainage basin of the onset of the EI Nino events influence the mag-
201,626 km2), and its estuarine zone serves as a nitude of responses? Based on current findings
nursery ground for coastal fishes (Vieira and Cas- and those from previous studies, we developed a
tello 1996), including species that support impor- conceptual diagram showing the main mechanisms
tant fisheries in the southwestern Atlantic Ocean and processes linking atmospheric-oceanographic
(Haimovici et al. 1996; Reis and D'incao 2000). interactions triggered by EI Nino phenomena to
Previous studies showed that EI Nino can affect the fish assemblage structure in Patos Lagoon es-
the primary productivity in the southern Brazilian tuary.
coastal area and the fish assemblage of Patos La-
Materials and Medtods
goon estuary. The moderate 1987-1988 EI Nino
increased freshwater discharge into the estuary STUDY SITE
and coastal marine area, and affected phytoplank- The southern 10% of Patos Lagoon (250 kIn
ton production and biogeochemical cycles in the long and 60 km wide) has the geomorphologic fea-
coastal area (Ciotti et al. 1995; Lima et al. 1996). tures of a bar-built estuary. and its upper limnic
Garcia et al. (2001) demonstrated that the fish as- area (30 kIn wide) gradually narrows into a chan-
semblage had opposite trends in fish abundance nel (700 m wide) connected with the south Atlan-
and diversity between cold (1995-1996 La Nina) tic Ocean. Apart from a navigation channel, about
and warm (1997-1998 EI Nino) ENSO episodes. 80% of the estuarine zone is less than 2 m deep
Based on a larger data set, Garcia et al. (2003a) (Fig. 1). The estuarine ecosystem is a nursery
showed that the 1997-1998 EI Nino-induced ground for several abundant fish and shrimp spe-
changes were not highly persistent, with the fish cies (Vieira and Castello 1996; Reis and D'incao
assemblagereturning to its pre-EI Nino state within 2000).
18 mo following the EI Nino. Differences in sam- Tidal influence in the estuary is minimal (mean
pling protocols, methods of data analysis, and lack tidal amplitude is 0.47 m). Wind and seasonalpuls-
of replication (i.e., more than one EI Nino event es of freshwater inflow influence patterns of water
analyzed simultaneously) hinder efforts to con- circulation and salinity (longitudinally and verti-
struct predictive models of EI Nino effects on the cally). The lagoon has a drainage basin formed by
fish assemblage of Patos Lagoon estuary. rivers characterized by a mid latitude flow regime:
Garcia et al. (2001, 2003a) proposed that the high discharge in late winter and early spring fol-
high rainfall associated with EI Nino events signif- lowed by low to moderate discharge through sum-
icantly increases runoff and freshwater discharge mer and autumn (Moller et al. 2001). During El
in the drainage basin of Patos Lagoon, which, in Nino events, rainfall in southern Brazil (Grimm et
tum, changes salinity patterns within the estuary. al. 2000) is above average, and the mesohaline
This hypothesis was based on rainfall data obtained zone can be displaced to the inner continental
from a single meteorological station located in the shelf (Moller et al. 200 1).
estuarine area (Rio Grande City). Due to the spa-
tial variability of EI Nino effects in southem South FIELD SAMPUNG
America (Grimm et al. 2000), there is uncertainty Fishes were sampled using a 9 X 1.5 m beach
as to whether the rainfall pattern observed at a seine (13 mm bar mesh in the wings and 5 mm in
given station would be an accurate predictor of EI the center 3-m section) that was pulled to cover an
Nino-induced changes in rainfall over the entire area approximately 60 m2. Survey stations were lo-
Patos Lagoon drainage basin. cated in shallow waters of the Patos Lagoon estuary
Based on an interdecadal biological data set en- (Fig. 1). The number of survey stations varied be-
compassing the two strongest EI Nino events of the tween the two sampling periods, with four to six
last 50 yr, and an regional rainfall data set that in- survey stations during 1979-1983 and four during
cludes 81 sample stations scattered around the 1996-2000. Each sample was composed of three
drainage basin, we evaluated the hypothesis that EI beach seine hauls. Seven samples (21 hauls) were
Nino-induced hydrological change is a major driv- made during each season within two periods: win-
ing force controlling interannual variation in the ter 1979 to spring 1983 (the one exception was
distribution and abundance of fish species in Patos summer 1983 with 15 hauls) and winter 1996 to
Lagoon estuary. We addressed the following que~ spring 2000. Samples were grouped by seasons as
tions in the present study. Did EI Nino-induced follows: summer (january-March), autumn
changes in rainfall and salinity occur with the same (April-June), winter (july-September) , and spring
magnitude during both El Nmo events? Did spe- (October-December). Salinity was measured
cies composition, abundance pattems, and fish di- monthly at each station, and precipitation was reo.
 )~ ...
r'
,

( EI Nino Effectson Palos LagoonFish 907

"
'"
-'

2705

I I

I
!
Atlantic
35 Ocean
~OW
"
, '-
'.

~ Fig. 1. Drainage basin of the Patos Lagoon (bold line) with the spatial distribution of the 81 meteorological stations (filled squares),
and the estuarine area showing beach survey stations (dots).
, !
corded daily at 81 meteorological stations located tained with different amounts of effort. Evenness
along the lagoon drainage basin. was analyzed by the Evar index proposed by Smith
" and Wilson (1996), which has equal sensitivity to
DATAANALYSIS rare and abundant species. This index converges
Environmental Parameters to one when species abundances are equal. Species
Meteorological effects of the two EI Nino peri- ?iversity w.as.esti~at~d u~ing .the Shannon-Wiener
ods on the Patos Lagoon estuary were determined mdex. VarIatIon m dIversIty, rlchnes.s,and eve?ness
by comparing individual seasonal rainfall and salin- were analyzed by one-way analysIs of varIance.
ity values (winter 1979 to spring 1983 and winter
1996 to spring 2000) with a long-term database of
seasonal averages. Seasonal values (x) from the
short-term data were contrasted
erages (IJ.) using x -
with seasonal av-
IJ.. Rainfall data were com- i~ 100 0
~
~
pared with a database of22 yr (1979-2000) of pre- :
cipitation in the Patos Lagoon drainage basin. A
mean rainfall series for the basin was calculated for E 1
E
this period from the data of 81 stations (Fig. 1). i
Salinity data were compared with a 10-yr database I
(1978-1983 and 1996-2000) of monthly values ob- ~
tained from 10 samplestationslocated at the Patos !-50
Lagoon estuary. Based on the rainfall and salinity 1
distributions and anomalies in Patos Lagoon, we ~
used the following seasons to set the temporallim- I
its of EI Nino events: spring 1982 to winter 1983 t
(1982-1983 EI Nino) and spring 1997 to winter ;Z-1
1998 (1997-1998 EI Nino).
1979I 1- I 1~ I 1- I 1~ 19$ I 1997 I 1- I 1999I 2000
Diversity Indices . 2. Seasona
FIg. I f) uctuatton
.. m th e two samp1mg., penod s
Species richness was calculated by the rarefac- (1979-1983and 1996-2000)of averagerainfall in the drainage
tion method (Sanders 1968; Hurlbert 1971). This basinof PatosLagoon and averagesalinity in its estuarinearea,
approach provides a standardized measure of spe- rainfall anomalies,and salinity anomalies.Anomaliesrepresent
. . . . . difference betweeneach seasonalvalue from 1979-1983and
Cles d!V~rsIty adJust~d for the effect of sample sI~e, 1996-2000and each seasonallong-term average(rainfall: 22 yr
permIttIng companson between assemblages WIth from 1979 to 2000 and salinity: 10 yr from 1978 to 1983and
different densities of organisms or samples ob- 1996to 2000). The shadedarea representsEl Nino periods.

"""'~ I,~",
908 A. M. Garcia et al.

TABLE 1. Relative importance (based on percent nume-rical abundance and frequency of occurrence) of the most representative
fishes captured during the tWo sampling periods (1979-1983 and 1996-2000). ~nter. s-spring. s-summer. a-autumn. See
legend below and details in the material and methods.

79 1980 1981 1982 1983 96 1997 1998 1999 2000

Species
_w:_-~~ a w s saw s saw s saw s w s saw s saw s saw s saw s
Mugil platanus
Atherinella brasiliensis
Odontesthes argentinensis
Micropogonias fumieri
Jenynsia multidentata
Mugil curema
Brevoortia pectinata
Lycengraulis grossidens
Platanichthys platana
Mugil gaimardianus
Ramnogaster arcusta
-I. !j;
Parapimelodus nigribarbis
Gobione/lus shufeldti
Ulaema lefroyi
. . ~
~~;

Astyanas eigenmanniorum
..
.. .
OIigosarcus jenynsii
Trachinotus marginatus
Anchoa marinii I . .. fj;;,'
Genidens genidens
Harengula c/upeola
Syngnathus fofletti

Legend:
-
- frequent

These data met the assumptions of normality (Kol- divided by the total abundance (N) of all species
mogorov-Smirnov test) and homogeneity of vari- captured (X 100). Fishes showing FO% > 50 by
ances (Cochran test). We used the following peri- season were considered frequent species, whereas
ods to achieve a balanced design without con- fishes showing PN% > the ratio (100/number of
founding seasonal samples: spring 1979 to winter species) were considered abundant. These criteria
1980, spring 1980 to winter 1981, spring 1981 to selected 21 fish species (out of 63) as the most
winter 1982, spring 1982 to winter 1983, spring representative for use as predictor variables in
1996 to winter 1997, spring 1997 to winter 1998, DFA
spring 1998 to winter 1999, and spring 1999 to win- We performed stepwise DFA to select variables
ter 2000. El Nino years were represented by spring (F values of 1 for entry and 0 for removal) that
1982 to winter 1983 and spring 1997 to winter yielded the most significant linear combination for
1998. discrimination between groups. Four groups were
determined a priori as follows: EI Nino and non-
Discriminant Function Analysis (Df:4) EI Nino during the 1979-1983 period, and EI Nino
Species classified as infrequent and less abun- and non-EI Nino during the 1996-2000 period. A
dant were omitted to avoid a disproportionate ef- tolerance value of 0.01 was used as the criterion
fect of rare species on the DFA. We used the fol- for assuming that variables were not redundant (a
lowing procedure to sort out rare species. The fre- tolerance value of 0.01 means that the variable is
quency of occurrence (FO%) for each species was 99% redundant with other variables already includ-
calculated as the ratio of the number of occur- ed in the model). Data (log relative abundance +
rences of a species divided by the total (X 100). 1) were checked for normality (normal probability
The percent numerical abundance (PN%) was cal- plots) and homogeneity of variances (Cochran test
culated as the total abundance (n) of a species, and standard deviation&-means plots) prior to use
 EI Nino Effectson PatosLagoonFish 909

TABLE 2. Indices of diversity (Shannon-Wiener,.H'), equita- 8 '"

bility (Evar), and richness (Rarefaction, E[Sn]; adjusted sample 117S-1- .:-EN
equals 155 individuals) calculated for each season during the
two sampling periods (1979-1983 and 1996-2000). The bold 4
represents El Nino periods.
Year Season H' £var £(So) 2
. .
1979 winter 0.30 0.10 4.42 . .. .. .. 6

1980
.
spnng
summer
0.73
0.46
0.49
0.09
7.92
5.63
0 .
:.:.
~~ ~~~
~~#"";~6. -*~ 6
8

autumn
winter 0.43
0.36 0.13
0.10 7.41
4.93 -2 . - . ~.: -11.:
. ~ ~6
spring 0.69 0.13 8.25 ~ .. 6 A
1981 summer 0.61 0.08 6.02 A
autumn 0.81 0.18 11.61 N -4
winter 0.32 0.09 4.80 ~ -4 -2 0 2 4 8
spring 0.64 0.11 7.48 8 1_2000 0 EN
1982 summer 0.93 0.14 13.16 ._-EN.
autumn
winter
0.76
0.57
0.11
0.08
8.82
5.79 4 . -
spring 0.67 0.20 11.54 .. CJ
1983 summer 0.76 0.13 10.33 ".' ." .. .- CJ
2 CJ
autumn 0.47 0.15 7.45 - . -. '.' CJ CJ
winter 0.93 0.16 11.00 -- -'.
"."30 .-. 0 CJ

spring 0.76 0.11 9.21 0 ~: :...gd'."D4.cPCJ CJ

1996 winter 0.66 0.13 6.99 - .'. . -~~~. D.
spring 0.58 0.14 8.12 .". .1::1='.0
1997 summer 0.44 0.11 7.80 -2 0 CJ

autumn 0.58 0.11 8.19

winter 0.72 0.10 7.20
spring 0.76 0.21 12.60 -4
1998 summer 0.87 0.25 13.29 -4 -2 0 2 4 6
Root1
autumn 0.63 0.18 9.05
winter 0.79 0.31 10.39 Fig. 3. Scatter plot of the individual scores for the two first
spring 0.76 0.13 11.04 discriminant functions (Root 1 and 2) for the two sampling pe-
1999 summer 0.74 0.10 9.62 riods 1979-1983 and 1996-2000. Centroid groups are repre-
.. autumn 0.71 0.10 8.66 sented as (1) Non-El Nino, (2) El Nino during the 1979-1983
winter 0.75 0.12 7.19 period, (3) Non-El Nino, and (4) El Nino in 1996-2000.
spring 0.75 0.11 8.85
2000 summer 0.65 0.10 7.61
a~tumn 0.71 0.10 8.98 linity conditions in the estuarine area lasted about
Wln.ter 0.30 0.16 6.35 3 mo longer during the 1997-1998 El Nino com-
spnng 0.70 0.25 9.04 pared with the 1982-1983 El Nino (Fig. 2). Rainfall

was above its historic average across all seasons in

.. . . both EI Nino periods, being particularly strong
m the analysI~. Only a few spe~Ies dId no~ meet during spring and summer of the 1997-1998 El
these .assumptIons, bu.t w~ retaIned them m ~e Nino. High rainfall anomalies during both El Nino
an~lysis beca~se DFA IS fairly r~bu~t to these Vlo- events were preceded by extended periods of be-
latI?ns,.especiallywhen sample SIzeISlarge and the low-average rainfall (Fig. 2). In contrast with rain-
desIgn IS balanced (Underwood 1997). fall trends, salinity was below its historic average
R ults across all seasons during both EI Nino events, es-
es pecially during summer and autumn of 1982-1983
ENVIRONMENTAL PARAMETERS and autumn of 1997-1998 (Fig. 2).
Comparison of seasonal rainfall and salinity with
long-term seasonalaveragesrevealed similar trends SPECIES COMPOSITION AND DIVERSITY
within both sampling periods (1979-1983 and A total of 87,901 fishes, representing 63 species,
1996-2000). High rainfall in the drainage basin of was captured in 750 beach seine hauls made over
Patos Lagoon coincided with low salinity in the es- the study period. Most species, especially marine
tuarine area, which can be attributed to the strong and freshwater vagrants, had low abundance and
EI Nino events of 1982-1983 and 1997-1998. The frequency in samples. Based on percent numerical
magnitudes of meteorological (rainfall) and hydro- abundance and frequency of occurrence, 21 spe-
logical (salinity) effects differed between the two cies were identified as the most important assem-
EI Nino events. Total rainfall in the drainage basin blage elements (Table 1).
was higher (767 versus 711 mm) and near-zero sa- Dominance by a few species was reflected in low
.

~".= ,,;L4~'"
910 A. M. Garda et al.

TABLE 3. Summary of the discriminant function analysis (DFA) and canonical coefficient of the predictor variables selected by the
stepwise forward method. * p < 0.01. Data were log,o transformed.

Axes t
Summary statistics for ordination axes
Eigenvalues 0.821 0.351 O.197
Cumulative percentage variance 60.0 85.6 100.'0
Wilks' Lambda o.~ 0.618 0.:8-'6
Chi~uare 257.0 114.4 42.8
df 54 34 16
Significance
. *

Standardized canonical coefficient

Jmynsia multidenlala -0.525 -0.088 -o.~
OdonIesthes ~nensis -0.502 -0.105 0.154
-0.161 0.082 -0.117
Mugil platanus
Mugilcu~ -'0.148 0.4gg 0.025
-0.094 -0.411 0.216
Mugil gaimanlianus
BmIOOrlia pectinata -0.069 0.4-'4 -0.344
Ulaema lefroyi -0.016 0.289 0.200
Ramnogaster arcuala -0.008 -0.160 0.320
Astyanax eiglmmanniorum 0.388 0.15g -0.495
0.312 0.075 -0.148
Oligosarcus jmynsii
0.280 O.20g -0.148
Parapimelodus nigribarbis
Atherinella brasiliensis 0.263 0.272 0.324
L ,cengraulis grossidens 0.253 -O.~ -0.033
0.196 -0.090 0.260
Genidensgmidens
0.057 0.220 0.273
Ptatanidlthys platana
0.0-'9 0.015 0.403
MiQoopog1miasfurnieri
0.0-'6 0.246 0.153
Hafmgula dupeoia
Gobionellus shufeJdti 0.004 0.267 0.167

values of assemblage evenness (Evar index usually between 1979-1983 (centroids: -0.43 and -1.00)
less than 0.15) among survey periods. Evenness and 1996-2000 (0.17 and 0.65; Fig. 3).
tended to increase during EI Nino events, with sta- The three species most associated with non-EI
tistically significant higher values observed during Nino conditions (positive canonical coefficients
the 1997-1998 event (n = 14, F = 10.598, P < given in Table 3) were jenynsia multidentata, Odon-
0.01). There was also an increase in the number testhesa~tinensis (both estuarine residents), and
of species (obtained by the rarefaction method) Mugil platanus (estuarine dependent; Fig. 4). The
captured during both EI Nino events, with the three species most associated with EI Nino condi-
1997-1998 episode yielding significantly higher tions (negative canonical coefficients) were the
values (n = 14, F = 4.302,P < 0.05; Table 2). freshwater vagrants Astyanax eigenmanniorum, Oli-
gosarcusjenynsii, and Parapimelodusnigribarbis that
FISH COMPOSmON DURING EL NINo VERSUSNoN- occurred in the estuarine area almost exclusively
EL NINO PERIODS during EI Nino events (Fig. 5).
Stepwise DFA derived linear combinations of 18
Discussion
variables (species) to discriminate among the four
periods (EI Nino and non-EI Nino during 1979- COMPARINGTHE STRENGTHOF EL NiNo EVENTS
1983, and EI Nino and non-El Nino during 1996- Meteorological and hydrological pattems ass0-
2000). Three species (Trachinotus mafR'inatus, An- ciated with the 1982-1983 and 1997-1998 EI Nino
choa marinii, and Syngnathus Jolleth) were redun- events had similar relationships to fish speciescom-
dant (tolerance> 0.01) and eliminated from the position and diversity pattems, but the 1997-1998
data set of 21 dominant species. The two first ca- event appeared to have stronger effects on the fish
nonical functions explained 85.6% of the variance, assemblage. Evenness values tended to increase
but the first root alone account for 60% and had during both episodes; this was caused by decreases
higher discrimination power (Wilks' lambda = in number of dominant species and increases in
0.34; Table 3). The distance between EI Nino and vagrant fishes. Species richness also increased dur-
non-El Nino group centroids along the first axes ing both EI Nino events when several freshwater
was greater during 1996-2000 (2.13 to -0.20) than species expanded their ranges into the estuary, but
1979-1983 (1.00 to -0.675). The second root this difference was only statistically significant dur-
modeled a weak trend of interdecadal differences ing the 1990s.The present interdecadal study, with
 EI NRio Effects 00 Palos Lagooo FIsh 911

--I

;1
!c0..

.-
,.-- -

1--
~
I..~
108

.. ". j, A\1I

1 - 1 -. 1..-'
'82 ,- ~ .. I ,.,
,- j
-.
.-.
,- ' ,.. I .-
Fig. 5. Seasonal fluctuation in relative abundance of Astya-
\ i

nax eigmmanniorum, Oligosarcusjenynsii, and Parapimelodusnigri-

barlJisin the tWo sampling periods (1979-1983 and 1996-2000).
These species had higher positive standardized coefficients in
the first discriminant function (Root I, see Table 3), showing
greater association with EI Nmo periods. The shaded area rep-
resents EI Nino periods.

EI Nino replication and regional rainfall data set, conditions associated with the 1982-1983 event.
seemed to corroborate the hypothesis that positive Such differences in the magnitude of effects ass0-
rainfall anomalies triggered by EI Nino events play ciated with the 1982-1983 and 1997-1998 EI Nino
a major role on the interannual variability of the eventshavealso been reported from studiesof Pa-
fish assemblage of Patos Lagoon estuary. cific coastal systems of South America. Sanchez et
Differences in assemblage structure associated al. (2000) demonstrated that both events had com-
with EI Nino events were even more apparent in parable effects on the coastal marine ecosystem off
results from DFA. The distance between EI Nino Peru, although the 1997-1998 produced more ex-
and non-EI Nino DFA centroids was greater during treme conditions. Arcos et al. (2001) showed that
the 1996-2000 sampling period compared with the the 1982-1983 and 1997-1998 EI Ninos led to
1979-1983 period. Although shifts in species com- changes in diet and habitat use by jack mackerel
position had similar patterns during both EI Nino (Trachurus symmetricus mu1fJhyi) off the coast of
events, the temporal changes in species relative Chile. These pattems were only evident during
abundances were greater during the 1997-1998 ep- moderate to strong EI Nino events and little
isode. Some freshwater fishes (e.g., the freshwater change occurred during weak events.
catfish P. nigribarbis and the characins A. eigl':nman-
nionlm and o. jenynsiz) were more common during A SIMPUFlED CoNCEPTUAL MODEL OF EL NINO
EI Nino periods. Freshwater vagrants were only EFFEcrs ON THE FIsH AssEMBLAGE
caught in the estuarine area during higher fresh- A conceptual model summarizes mechanisms
water outflow trigged by the elevated rainfall of EI and processes associated with EI Nino phenomena
Nino episodes. These species achieved greater rel- and their effects on the ichthyofauna of Patos La-
ative abundance during the 1997-1998 El Nino goon estuary (Fig. 6), and synthesizes information
than the 1982-1983 El Nino. scattered within regional literature (Ciotti et al.
Differences in the magnitudes of 1982-1983 and 1995; Lima et aI. 1996; Grimm et al. 1998; Garcia
1997-1998 El Nmo effects on the fish assemblage et al. 2003a; Costa et al. 2003).
of Patos Lagoon estuary probably result from dif- The warming of the sea surface temperatures in
ferences in the intensity of associated salinity the eastern equatorial Pacific Ocean (shaded area
anomalies. During both EI Nino periods, high rain- in Fig. 6 no. 1) during EI Nino episodes enhances
fall within the drainage basin resulted in unusually the convection and rainfall in this area, producing
low salinity in the estuarine zone, but the 1997- anomalous heating in the troposphere. This gen-
1998 EI Nino had a greater effect. The 1997-1998 erates perturbations in the atmospheric divergent
event was associated with near-zero salinity for a Hadley and Walker circulations, as well as anoma-
period that lasted 3 mo longer than the low-salinity lous Rossby waves that propagate poleward and

~
 ~

912 A. M. Garcia et al.

~~::~~j
~,.,,--- .-

I~

-
Atlantic
@ Ocean

Fig. 6. A simplified conceptual diagram showing the linkages between El Nino events in the Equatorial Pacific Ocean and the
estuarine fish assemblage of Patos Lagoon in southern Brazil (see detailed explanation in the Discussion).

eastward. These atmospheric perturbations affect conditions can last for several months (Costa et al.
rainfall in southern Brazil especially in austral 2003; Garcia et al. 2003b). Freshwater from the
spring and the next winter. They produce a cyclon- middle to upper lagoon expands towards the es-
ic circulation anomaly over southern South Amer- tuarine zone carrying with them several freshwater
ica, with an anticyclonic one northeast of this re- fishes (e.g., P. nigribarbis,A. eigenmanniorum,and O.
gion. These anomalies strengthen the subtropical jenynsiz) that spread throughout the estuary, and
jet (a belt of strong upper-level westerly winds) and in some cases,are even caught in the adjacent ma-
produce a cyclonic vorticity advection over south- rine coastal area. This increase in the number of
ern Brazil. These aspects favor the intensification freshwater fishes is the main factor leading to high
of mesoscale systemsand cyclogenesis over this re- fish diversity usually associated with EI Nino events
gion (Grimm el al. 1998,2000). (Garcia et al. 2003a). High freshwater discharge
The above average rainfall triggered by EI Nino also decreases the relative abundance of those fish-
events over the large drainage basin (201,626 km2) es occurring year-round and breeding in the es-
of Patos Lagoon greatly increases freshwater out- tuary. Two phenomena seem to explain these
flow into the lagoon (10,360 km2; no. 2 in Fig. 6). trends. Vegetation-affiliated fishes (e.g.,]. multiden-
Freshwater outflow in the northern region of Patos tata) have greater accessto flooded vegetated hab-
Lagoon estuary normally ranges from 700 to 3,000 itats in low and mid marshes during periods of
m3 S-I. But during EI Nino events, discharges> high freshwater discharge, and may be less abun-
3,000 m3 S-1 frequently occur and the estuarine dant at shoreline sample stations used for seine
plume can extend several kilometers into the ad- surveys (dilution hypothesis; Garcia et al. 2004).
jacent coastal marine area (Garcia 1996; Moller For other species (e.g., Atherinella brasiliensis),pe-
personal communication). lagic juveniles are transported in large numbers
This elevated freshwater outflow affects hydrol- out of the estuary into adjacent marine coastal wa-
ogy, water quality, and the fish assemblage in the ters (flushing hypothesis; Garcia et al. 2001). In
estuarine area (971 km2; no. 3 in Fig. 6). Water contrast to estuarine resident fishes, marine spe-
levels along the estuarine margin increase, water cies that predictably use estuarine areas asjuveniles
transparency tends to fall, and near-zero salinity (e.g., M. platanus) or occasional marine visitors

: i
 EI Nino Effects on Pam L8gOUI ~ 913

(e.g., A. marinii, UlaemaleJroyz)enter the estuary in of ichthyoplankton in the upwelling zone off Oregon during
anomalous El Nino conditions. F..stuarine Coastal and Shelf Sci-
lower numbers during periods of high freshwater
ence 21:365-378.
discharge. Apparently the high freshwater outflow CAVIEDES. C. N. 2001. EI Nino in History: Storming Through
through the narrow mouth of the estuary (only the Ages. University Press of Florida. Gainesville. Florida.
700 m wide) impairs physical mechanisms that al- CIOTI1. A. M.. C. ODEBRECHT. G. FnLMANN. AND O. O. MoU.ER.
low passive movement of these fishes into the es- 1995. Freshwater outflow and subtropical convergence influ-
ence on phytoplankton biomass on the southem Brazilian
tuary (i.e., net upstream movement of high-salinity continental shelf. Continental and Shelf Research 15:1737-1756.
bottom waters). CoSTA. C. S. B.. J. C. MARANGONI. AND A. M. G. AzEVEDO. 2003.
Recruitment of marine organisms into nursery Plant zonation in irregularly flooded salt marshes: Relative
areas of Patos Lagoon estuary may be severely cur- importance of stress tolerance and biological interactions.
journal of Eroiogy 91:951-965.
tailed during high freshwater outflow indirectly ENFIELD. D. B. 2001. Evolution and historical perspective of the
triggered by El Nino events. Mullet (M. platanus) 1997-1998 EI Nino-Southem Oscillation event. Bulletin of Ma-
and pink shrimp (Farfantepenaeuspaulensis), which rine Science 69:7-25.
together sustain the increasingly impoverished ar- GARCiA, A. M.. M. B. RASEIRA.J. P. VIEIRA. K. O. WINEMIULR.
tisanal estuarine fishery in this region (Reis and AND A. M. GRIMM. 2003b. Spatiotemporal variation in shallow-
water freshwater fish distribution and abundance in a large
D'incao 2000), are among those estuarine-depen- subtropical coastal lagoon. Environmental Biology of Fishes 68:
dent species that could be negatively affected by 215-228.
sustained, high freshwater discharge (see discus- GARCiA, A. M..J. P. VIEIRA. AND K. O. WINEMII.I.F.R. 2001. Dynam-
sion in Garcia et al. 2003a). The degree of associ- ics of the shallow-water fish assemblage of the Patos Lagoon
ation between fishery catches of adult stocks and estuary (Brazil) during cold and warm ENSO episodes. jour-
nal of Fuh Biology 59:1218-1238.
juvenile recruitment into the estuary is currently GARCIA, A. M..J. P. VIEIRA, AND K. O. WINEMIU.ER. 2003a. Effects
unknown, particularly for the mullet. According to of 1997-1998 El Nino on the dynamics of the shallow-water
Tyler (1992), management of commercially ex- fish assemblage of the Patos Lagoon estuary (Brazil). Estua-
ploited fish stocks is severely hampered without rine. Coastal and Shelf .5GinJu 57:489-500.
GARCIA. A. M..J. P. VIEIRA. K. O. WINEMIULR, AND M. B. RASt:lRA.
some understanding of the processes that drive re- 2004. Reproductive cycle and spatiotemporal variation in
cruitment variability on a long-term basis. abundance of the one-sided livebearer jenynsia mullibnlala.
As emphasized by Jaksic (2001), EI Nino phe- in Patos Lagoon. Brazil. Hydrobiologia 515:39-48.
nomena provide natural experiments that reveal GARCiA, C. A. E. 1996. Hydrographic characteristics. p. 18-20.
connections between atmospheric-<>ceanographic In U. Seeliger, C. Odebrecht, and J. P. Castello (eds.), Sub-
tropical Convergence Environments: The Coastal and Sea in
processesand marine biota that may occur on very the Southwestern Atlantic. Volume 1. Springer Verlag, Berlin.
broad geographic scales. The causal linkages be- GLANTZ. M. H. 2001. Currents of Change: Impacts of EI Nino
tween EI Nino phenomena in the South Pacific and La Nina on Climate and Society. C.ambridge University
and the ichthyofauna of Patos Lagoon estuary are Press. Cambridge. U.K.
an example of complex, large-scale biophysical in- GI.'tNN, P. W.. J. L. MArt, A. C. BAKER, AND M. O. CALDERON.
2001. Coral bleaching and mortality in Panama and Ecuador
teractions. Additional ecosystem components (e.g., during the 1997-1998 EI Nino-Southem Oscillation even I;
nutrient dynamics, productivity) need to be added Spatial/temporal patterns and comparisons with the 1982-
to the model. 1983 event. Bulletin of Marine Science 69:79-109.
GRIMM, A. M., V. R. BARRos. AND M. E. DOUL 2000. Climate
ACKNOWLEDGMENTS variability in Southern America associated with EI Nino and
La Nina events. journal of Climate 13:35-58.
We thank researchers and staff from the FURG Ichthyology GRIMM. A. M.. S. E. T. FERRAZ,AND J. GoMES. 1998. Precipitation
Laboratory that collected historical data (1979-1983), Lisiane anomalies in southern Brazil associated with EI Nino and La
A. Ramos for collecting data between August 1996 and August Nina events. journal of Climate 11 :2863-2880.
1997, and numerous colleagues who assisted in the field. This HAIMOVlC1. M., J. P. CAsrELt.O, AND C. M. VOORt:N. 1996. Fish-
study received financial support from the Coordenadoria de eries, p. 183-196. In U. Seeliger. C. Odebrecht, andJ. P. Cas-
Aperfei~oamento de Pessoal de Nivel Superior-CAPES (Bra- tello (em.). Subtropical Convergence Environments: The
zil), the Conselho Nacional de Desenvolvimento Cientifico e C.oastal and Sea in the Southwestern Atlantic. Volume I.
TecnolOgico-CNPq (Brazil), Brazilian Long Term Ecological
Springer Verlag. New York.
Research (LTER), and the Inter American Institute for Global HURLBERT. S. H. 1971. The nonconcept of species diversity: A
Change Research (1Al) through a fellowship granted by the
critique and alternative parameters. EcolOgJ52:577-586.
SACC/Consortium (CRN-O19). Alice Grimm acknowledges the JAKSIc, F. M. 2001. Ecological effects of EI Nino in terrestrial
support of Inter-American Institute for Global Change Research
ecosystems of western South America. EaJgraphJ 24:241-250.
(IAl-CRNO55) and of the Conselho Nacional de Desenvolvimen-
JIMtNEZ. C..J. CoRTEs. A. LEON. AND E. Rulz. 2001. Coral bleach-
to Cientifico e TecnolOgico (CNPq, Brazil). ing and mortality associated with the 1997-98 EI Nino in an
upwelling environment in the eastern Pacific (Gulf of Papa-
LITERATURE CITED gayo. Costa Rica). Bulletin of Marim Science 69:151-169.
AR<X>S,D. F., L A. CUBlu.os, AND S. P. NuNEZ. 2001. The jack KNoPPERS. B. AND B. ~ERFVE. 1999. Coastal lagoons of south-
mackerel fishery and El Nino 1998-98 effects off Chile. Pr0- eastern Brazil: Physical and biogeochemical characteristics. p.
gress in OceanDgmph, 49:597-617. 35-66. In G. M. Perillo. M. C. Picolo. and M. Pino (em.),
BRODEUR, R. D., D. M. GADOMSKI, W. G. PEARCY,H. P. BATCH- Estuaries of South America: Their Geomorphology and Dy-
ELDER, AND C. B. MIUER. 1985. Abundance and distribution namics. Volume I. Springer Verlag. Berlin.
914 A. M. Garda et al.

LIMA, I. D., C. A. E. GARCIA,AND O. O. MoU.ER. 1996. Ocean llMMERMANN. A..J. OBERHUBER. A. BACHER. M. ESCIi. M. LATIF.
surface processes on the southern Brazilian shelf: Character- AND E. RoECKNER. 1999. Increased El Nino frequency in a
ization and seasonal variability. Continental Shelf ~h 16: climate model forced by future greenhouse warming. Natun'
1307-1317. 398:694-697.
MESTAs-NuNEZ, A. M. AND D. B. ENFIELD.1999. Rotated global TUDHOPE. A. W.. c. P. CHILCOlT. M. T. McCUlLOCH. E. R. CooK.
modes of non-ENSO sea surface temperature variability. jour- J. CHAPPEu-- R. M. EUAM. D. W. LEA,J. M. LoUGH. AND G. B.
nal of Climate 12:2735-2746. SHIMMIELD. 2001. Variability in the EI Nino-Southern Oscilla-
MESTAs-NuNEZ, A. M. AND D. B. ENAELD.2001. Eastern Pacific tion through a glacial-interglacial cycle. Science 291:1511-
SST variability: ENSO and non-ENSO components and their 1517.
climatic associations.journal of Climate 14:391-402. Tvr.ER. A. V. 1992. A context for recruitment correlations: Why
MoU.ER, O. O. J., P. CAsTING,J.-C. SALoMON,AND P. LAzURE. fisheries biologist should still look for them. Fisheries Oaan-
2001. The influence of local and non-local forcing effects on ogmphJ 1:97-107.
the subtidal circulation of Patos Lagoon. Estuaries24:297-31 I. UNDERWOOD.A.J. 1997. Experiments in Ecology-Their Logical
REls, E. G. ANDF. D'INCAO.2000. The present statUsof artisanal Design and Interpretation Using Analysis of Variance. Cam-
fisheries of extreme Southern Brazil: An effort towards com- bridge University Press. Cambridge. U.K.
munity-based management. Ocean& CoastalMa~ 43: VIEIRA. J. P. AND J. P. CASTELLO. 1996. Fish fauna. p. 56-61. In
585-595. U. Seeliger, C. Odebrecht, andJ. P. Castello (eds.). Subtrop-
ROI.LINS,H. B., J. B. RICHARDSON. AND D. H. SANDWEISS. 1986. ical Convergence Marine Ecosystem. The Coast and the Sea
The birth of EI Nino: geoarchaeological evidence. Geoarchaeol-
in the Warm Temperate Southwestern Atlantic. Volume I.
ogy 1:3-15.
Springer Verlag. New York.
Roy, C. AND C. REASON.2001. ENSO related modulation of
coastal upwelling in the eastern Atlantic. Progressin Oceanog-
raphy 49:245-255. SoURCE OF UNPUBUSHED MATERIALS
SANCHEZ,G., R. CALIENES,AND S. ZUTA. 2000. The 1997-98 E1
Nino and its effects on the coastal marine ecosystemoff Peru. MoU.ER, O. Personal communication. Laboratorio de Oceano-
California CooperativeOceanicFisheriesIn~tigations Reports41: gratia Fisica, Departamento de Fisica, Universidade do Rio
62~. Grande, CP 474, 96200-200 Rio Grande RS, Brazil.
SANDF.RS, H. L. 1968. Marine benthic diversity: A comparative
stUdy. The AmericanNaturalist 102:243-282. ~ October13, 2003
SMITH, B. AND J. B. W~N. 1996. A consumer's guide to even- &viw, June 3, 2003
ness indices. OlKOS 76:70-g2. A~June8, 2004