ORIGINAL ARTICLE
Emergence of Asiatic Vibrio Diseases in South America in
Phase With El Niño
Jaime Martinez-Urtaza,a Blanca Huapaya,b Ronnie G. Gavilan,b Veronica Blanco-Abad,a
Juan Ansede-Bermejo,a Carmen Cadarso-Suarez,c Adolfo Figueiras,d,e and Joaquin Trinanesf
Background: The seventh pandemic of Vibrio cholerae unexpect-
edly reached the coast of Peru in 1991, causing an explosive
emergence of infections throughout the American continents. The
origin and routes of dissemination are as yet unknown. A new Vibrio
epidemic arose in 1997 in South America (northern Chile) when the
pandemic clone of Vibrio parahaemolyticus was for the fist time
detected outside of Asia. These 2 cases were concurrent with 2
episodes of El Niño.
Methods: We carried out a survey of records of V. parahaemolyti-
cus infection and of strains existing in the Instituto Nacional de
Salud of Peru between 1994 and 2005. Association between the El
Niño event and the V. parahaemolyticus disease was analyzed
through generalized additive models applied to time-series data with
negative binomial response, selecting some oceanographic factors
distinctive of the movement of the El Niño waters.
Results: Epidemiologic data and laboratory investigations of the
strains showed that V. parahaemolyticus infections caused by the
pandemic clone emerged in the coasts of Peru linked to the 1997 El
Niño episode. The epidemic dissemination of this clone matched the
expansion and dynamics of the poleward propagation and the
receding of the El Niño waters. This pattern was similar to previ-
ously reported onset of cholera epidemic in 1991.
Conclusions: These findings identify the El Niño episodes as a
reliable vehicle for the introduction and propagation of Vibrio
pathogens in South America. The movement of oceanic waters
Submitted 12 April 2007; accepted 18 March 2008.
From the aInstituto de Acuicultura, Universidad de Santiago de Compostela,
Spain; bInstituto Nacional de Salud, Lima, Peru; cDepartamento de
Estadistica e Investigacion Operativa and dDepartamento de Medicina
Preventiva y Salud Publica, Universidad de Santiago de Compostela,
Spain; eConsortium for Biomedical Research in Epidemiology and Pub-
lic Health (CIBERESP); and fInstituto de Investigaciones Tecnologicas,
Universidad de Santiago de Compostela, Spain.
Supported by grants MTM2005-00818 from the Spanish Ministry of Education,
PGIDIT06PXIC208043PN from the Xunta de Galicia (to C.C.-S.), and
PI021512 from the Fondo de Investigación Sanitaria of the Spanish Ministry
of Health (to A.F.).
Supplemental material for this article is available with the online version
of the journal at www.epidem.com; click on “Article Plus.”
Correspondence: Jaime Martinez Urtaza, Instituto de Acuicultura, Univer-
sidad de Santiago de Compostela, 15782 Santiago de Compostela, Spain.
E-mail: jaime.martinez.urtaza@usc.es
Copyright © 2008 by Lippincott Williams & Wilkins
ISSN: 1044-3983/08/1906-0829
DOI: 10.1097/EDE.0b013e3181883d43
Epidemiology • Volume 19, Number 6, November 2008
seems to be one of the driving forces of the spread of Vibrio
diseases.
(Epidemiology 2008;19: 829 – 837)
V ibrio cholerae and Vibrio parahaemolyticus are the major
human pathogens among Vibrio species. Despite their
distinctive biologic and ecologic features, both pathogens
show some parallelisms in their epidemic dynamics and are
currently involved in 2 pandemic spreads.
The seventh cholera pandemic began in Indonesia in
1961 and was confined to Asia until 1970 when it reached
Africa and Europe. It re-emerged in northern Peru in 1991,
causing more than 1 million cases of cholera and 10,000
deaths in Latin America in just 3 years.1 V. cholerae was the
unique case of pandemic spread among Vibrio pathogens
until 1996, when a dramatic increase in the number of V.
parahaemolyticus illnesses was detected in India.2
Before 1996, V. parahaemolyticus infections were spo-
radically detected in various geographic areas associated with
diverse serotypes.3 However, the emergence of V. parahae-
molyticus infections in Calcutta in February 1996 was dis-
tinctively linked to isolates belonging to the O3:K6 serotype,
with identical genotypes (tdh positive and trh negative) and
profiles indistinguishable by molecular typing techniques.2
The first reported isolate belonging to the new O3:K6 clone
was previously detected in a patient from Indonesia in 1995.2
In 1996 and 1997, diseases caused by this clone spread
throughout most Southeast Asian countries (Fig. 1). The
epidemiology of the O3:K6 infections changed abruptly in
November 1997, when infections caused by this clone were
first detected outside of Asia at a single location on the
northern coast of Chile,4 triggering the first pandemic expan-
sion of this organism. The factors that endow the O3:K6
clone with this pandemic potential are not well understood
because no specific advantages in virulence or enhanced
survival of these isolates have yet been established.5
Until the appearance of Vibrio epidemics in South
America, the infection had spread mostly westward, consis-
tent with the prevailing westward movement of water asso-
ciated with the Indonesian throughflow (a system of currents
flowing from the Pacific Ocean to the Indian Ocean through
829
Martinez-Urtaza et al
FIGURE 1. Distribution of pandemic V. parahaemolyticus 1996 –1997 when infections were detected outside of Asia in northern
Chile.
the Indonesian Sea).6,7 The emergence of V. cholerae in 1991
and V. parahaemolyticus in 1997 on the west coast of South
America was evidence of an eastward drift of Asian endemic
infections, concurrent with the arrival of warm equatorial
waters displaced from Asia to America by the last 2 El Niño
episodes.
The origin of the 1991 cholera epidemic in America
remains uncertain. An oceanic theory for the transPacific
transport of cholera from Asia to America was proposed,8 in
which the exchange of water masses throughout the Strait of
Makassar near Indonesia was suggested as the route of entry
of V. cholerae to the Pacific, and from there transporting
bacteria to Peru by eastward equatorial displacement of
waters of El Niño.8 However, this hypothesis could not be
consistently contrasted because of the rapid dissemination of
infections that spread over 2000 km of the west coast of Peru
in just 2 weeks and reached the tropical forests 1 month
later.9,10,10a Furthermore, the scarcity of oceanic records and
satellite data available in 1991 constrained reliable analysis of
the role of El Niño in the origin of the American cholera
epidemic.
The emergence of the V. parahaemolyticus epidemic in
America concurrently with the 1997 El Niño episode pro-
vided an opportunity to test the hypothesis of the oceanic
spread of Vibrio infections from Asia. Unlike V. cholerae, V.
parahaemolyticus is a strictly halophilic organism with a
habitat restricted to the marine environment. This ecologic
feature limits its distribution to coastal areas, thereby en-
abling accurate tracking of the disease spread. Moreover, the
pandemic clone of V. parahaemolyticus emerged in 1996 and
830
Epidemiology • Volume 19, Number 6, November 2008
was restricted to Asia until its detection in America. Finally,
the 1997 El Niño was the best-documented episode of El
Niño, and numerous studies were undertaken to assess the
oceanographic and biologic changes coupled with this event.
To evaluate the hypothesis of oceanic dispersion of
Vibrio infections, we focused our study on Peru, a country
greatly affected historically by El Niño episodes, and also the
place of emergence of the cholera epidemic in 1991.
METHODS
Epidemiologic Information and Strains of V.
parahaemolyticus
V. parahaemolyticus is not routinely investigated in
hospitals in Peru, and related infections are not reported to the
infectious diseases surveillance system. However, mandatory
investigation of V. cholera in clinical laboratories after the
emergence of cholera epidemic in 1991 enabled parallel
identification of other pathogenic Vibrio.
Vibrio strains isolated from clinical sources in hospitals
and public health laboratories in Peru are shipped to the
Instituto Nacional de Salud (Lima, Peru) for final identifica-
tion and characterization. Information concerning isolates
received by this Institute is registered weekly. We conducted
an extensive survey of their records to establish the number of
V. parahaemolyticus strains received from clinical laborato-
ries between 1994 and 2005. We inferred the rate of V.
parahaemolyticus infections from the number of strains re-
ceived by the Institute in that period. As each strain shipped
was associated with a patient who received medical assis-
© 2008 Lippincott Williams & Wilkins
Epidemiology • Volume 19, Number 6, November 2008
tance, we assumed that the number of isolates shipped was
indicative of the dynamics of V. parahaemolyticus infections
over the period of study. However, considering that V. para-
haemolyticus infections often cause mild illnesses that rarely
need medical care, these values probably represent only a
small fraction of the actual number of infections.
Additionally, 63 V. parahaemolyticus isolates from
the National Culture Strain Collection, deposited in the
Instituto Nacional de Salud, were subjected to serologic
and genetic characterization. Two environmental strains of
V. parahaemolyticus obtained from previous studies were
also characterized.
Laboratory Investigations
Identification of the isolates was confirmed by the
presence of the species-specific Vp-toxR and tlh genes. The
presence of the Vp-toxR gene was investigated by PCR as
described previously.11 Presence of tlh and the virulence-
related genes tdh and trh was determined by multiplex PCR
according to the procedure of Bej et al12 Virulence attributes
were also investigated by single PCR protocols for identify-
ing tdh and trh genes, according to Tada et al13 The pandemic
nature of isolates was investigated by use of the Group-
Specific PCR assay as described previously14; this procedure
allows specific detection of the pandemic clone-specific nu-
cleotides in the toxRS operon of V. parahaemolyticus. Sero-
typing of isolates was performed according to the manufac-
turer’s instructions (Denka-Seiken Ltd., Tokyo, Japan).
Lipopolysaccharide (O) and capsular (K) serotypes were
determined by agglutination tests in which specific antisera
were used.
Pulsed-Field Gel Elecrophoresis
Peruvian V. parahaemolyticus isolates were compared
with a panel of isolates from Asia representing prepandemic,
pandemic spread, and pandemic serovar transition periods.
Isolates from the first emergence of pandemic V. parahae-
molyticus in Antofagasta (Chile) in 1997 were also included.
Asian and Chilean isolates were provided by Mitsuaki Nishi-
buchi (Kyoto University, Japan) and Romilio Espejo (Uni-
versity of Chile, Chile), respectively. Pulsed-field gel elec-
trophoresis was performed according to “One-Day (24 –28
hours) Standardized Laboratory Protocol for Molecular Sub-
typing of Nontyphoidal Salmonella by PFGE”15 following a
previously described method.16 Restriction patterns were
compared by use of BioNumerics software (Applied Maths,
Sint-Martens-Latem, Belgium).
Environmental Data
Association between El Niño waters and the arrival and
dissemination of V. parahaemolyticus epidemic was evalu-
ated by selecting oceanographic factors distinctive of El Niño
phenomena. In addition to surface seawater temperature,
which varied seasonally independently of the El Niño epi-
sodes, we also considered sea height anomaly and the heat
© 2008 Lippincott Williams & Wilkins
Vibrio Epidemics in South America and El Niño
content above 20°C as consistent indicators of the advance of
warm waters displaced by the Kelvin waves.
The thermal structure of ocean impacts on sea surface
topography and the variations of this parameter are detected
by radar altimeters, which allow inference of sea height
anomaly values at an accuracy of 1–2 cm rms.17 This param-
eter is the deviation of the sea height from a climatologic
annual mean. We use the gridded sea height anomaly fields
computed by AVISO (http://www.jason.oceanobs.com/).
The heat content above 20°C is defined as the inte-
grated heat content from the surface to the depth of the 20°C
isotherm, estimated by a 2-layer reduced gravity model.18
This isotherm represents the depth of the thermocline in the
equatorial region19 and serves as a proxy for monitoring the
evolution of El Niño and the propagation of Kelvin waves.
Other variables in this parameter are the monthly relation-
ships among the depth of the isotherms above 20°C and
values of sea surface temperature obtained from ⬃9 km
NOAA/NASA AVHRR Oceans Pathfinder Best-surface sea-
water temperature data product20 and the Tropical Rainfall
Measuring Mission Microwave Imager by Remote Sensing
Systems and sponsored by the NASA Earth Science REASoN
DISCOVER Project (www.remss.com).
Time series of these 3 parameters show the evolution of
El Niño in 1997–1998. Fields were averaged and resampled
weekly into 1° ⫻ 1° grids.
Statistical Analysis
The relationships between V. parahaemolyticus strains
isolated in Peru during the study and the environmental
explanatory variables were analyzed by generalized additive
models (GAMs)21 applied to time-series data.22–24 The vari-
able response was defined as the number of strains isolated
per week in the area, and the explanatory variables consid-
ered were surface seawater temperature, sea height anomaly,
and heat content above 20°C. We also used a smooth function
of time as an independent variable when autocorrelation was
detected in the residuals. Because the isolations of strains
throughout 1 week were not independent events, a variance of
the response variable higher than the mean (overdispersion)
may have resulted. We, therefore, used negative binomial
models rather than Poisson response models to obtain stan-
dard errors corrected for overdispersed parameters. The log-
arithm “log” link function was used to achieve interpretable
epidemiologic measures of the effect. We used thin-plate
regression splines as trend smoothers, according to Wood,25
and automatically selected the optimal degrees of freedom by
using generalized cross validation criterion.26 The use of log
link in our model and the high proportion of zeros encoun-
tered in the response variable may cause problems of con-
vergence in the algorithm for fitting GAMs (because of a
fundamental lack of identifiability of the models). We, there-
fore, applied a ridge regression with penalties to impose
identifiability in the GAMs. Models were estimated by use of
831
Martinez-Urtaza et al
the mgcv package version 1.3–13, in the R environment,
version 2.5.0 (R Development Core Team, 2003).
Cases of V. parahaemolyticus infections in Lima were
analyzed during 3 periods: pre–El Niño (1994 –1996), El
Niño (1997 and 1998), and post–El Niño (1999 –2004) to
evaluate whether similar oceanographic conditions were re-
lated to infections associated with the different serotype-domi-
nance periods (1994 –1996, 1997–1998, and 1999 –2004), or
whether the propagation of O3:K6 infections in Peru was asso-
ciated with distinctive oceanographic factors linked to the move-
ment of El Niño waters.
To elaborate the models, we identified lagged periods
(weeks) for each independent variable that showed lower
values of the Akaike’s Information Criterion. For this, we
constructed a model for each explanatory variable and each
lag, with lags ranging from 1–5 weeks. For each period, we
developed 2 models: the first included surface seawater tem-
perature and sea height anomaly as covariates, and the second
included surface seawater temperature and heat content above
20°C. As the sea height anomaly and heat content above 20°C
variables were used as indicators of the same phenomenon,
they were not simultaneously included in the same model to
prevent potential collinearity. To check nonlinear relation-
ships, all 3 variables were included in the models as a smooth
function of continuous covariates through thin-regression
splines. A linear relationship was assumed when the esti-
mated degrees of freedom was 1 or close to 1, whereas when
they were different from 1, a Bayesian approach to smooth
modeling was used to derive standard errors on predictions
and obtain credible confidence bands for the effects, assum-
ing linearity if a straight line could be delineated between the
boundaries of the 95% confidence bands. When this did not
occur, we categorized the variable by using the change points
observed. We then assessed the simple autocorrelation func-
tion of Pearson residuals. If an elevated autocorrelation was
observed, a smooth function of the time was included in the
model. In those cases still supporting residual autocorrelation,
we included autoregressive terms of the dependent variables
for lags with autocorrelation.
Results were expressed as rate ratios (RRs) with 95%
confidence intervals (CIs), indicating an increase or decrease in
the probability of obtaining a response for an increase of 1 unit
in the independent variable. We also included the interquartile
RR (IqRR) to take into account the distributions of dependent
variables24 and to allow for comparing the magnitudes of the
effects of independent variables measured in different units.
RESULTS
Distribution of V. parahaemolyticus infections in Peru
between 1994 and 2005 showed a pronounced seasonal pat-
tern (Fig. 2) during the entire period. Case records from
1994 –1996 were associated with occasional scattered infec-
tions restricted to the warmest months. This pattern of infec-
832
Epidemiology • Volume 19, Number 6, November 2008
tions was suddenly disrupted in 1997 with the presence of an
anomalous increase in the number of V. parahaemolyticus
cases in the course of the austral winter dispersed from north
to south along coastal regions. The onset of infections was
initially detected in July and lasted for 10 months, with 2
peaks occurring in September and February (Fig. 2). Infec-
tions were observed along the entire coastline of the country,
and spread more than 1500 km in just 4 months (Fig. 3A).
The first record was detected in Chiclayo, Department of
Lambayeque, in the north of the country, in July. From there,
infections spread in a constant southward direction, affecting
the Department of Cajamarca in August, La Libertad in
September, Huaraz in October, and finally reaching the south-
ern Peruvian border in November—the same time as the
emergence of V. parahaemolyticus illnesses in the northern
Chilean city of Antofagasta.4
Laboratory investigations of 64 strains recovered from
the Instituto Nacional de Salud strain collections representa-
tive of the 1994 –2005 period showed an extraordinary shift
in pathogenic V. parahaemolyticus populations associated
with the rise of infections in July 1997 (eFigures 1 and 2 and
eTable; available with the online version of this article).
Serotypes O4:K8, O4:K55, and O4:K28 were identified as the
prevailing serovars among pathogenic strains from environ-
mental and clinical sources in a previous study conducted in
Peru in 1983–1984.27 Serotype O4:K8 was also identified in
strains isolated before 1997 in the present study, whereas
O5:KUntypable (UT) was associated with the large outbreak
of disease in Lima in 1995. The onset of infections in 1997
was associated with a complete change in serotype domi-
nance. All the isolates obtained in the 1997–1998 period
belonged exclusively to the O3:K6 serotype, which remained
the dominant serotype until its total disappearance in 2004.
The O3:K6 isolates showed typical molecular characteristics
of the pandemic clone and were positive for the specific PCR
pandemic test. Molecular typing procedures revealed high
clonal uniformity among O3:K6 specimens. Pandemic iso-
lates obtained in 1997 and 1998 presented highly uniform
DNA restriction patterns, as revealed by pulsed-field gel
electrophoresis analysis with NotI and SfI enzymes, and
shared serotype and indistinguishable genetic properties with
the pandemic O3:K6 strains obtained from India, Bangladesh,
Thailand, and Taiwan, for the initial spread of V. parahae-
molyticus epidemic in 1996. Peruvian O3:K6 isolates were also
identical to the pandemic isolates subsequently detected on the
north coast of Chile in November 1997 (eFigs. 1 and 2).
Conversely, whereas O3:K6 populations from Asia evolved
rapidly in 1997 and showed a dynamic serovar transition with
the generation of new serotypes (O4:K68, O1:KUT),14 Peruvian
pandemic strains maintained a unique serotype and homo-
geneous genetic characteristics until their disappearance in
2004.
© 2008 Lippincott Williams & Wilkins
Epidemiology • Volume 19, Number 6, November 2008 Vibrio Epidemics in South America and El Niño

FIGURE 2. Monthly number of cases of V. parahaemolyticus reported in Peru between 1994 and 2005 according to the data
provided by the Instituto Nacional de Salud in Lima. Sea Surface Temperature anomalies and Multivariate ENSO index data for
the same period were provided by NOAA-CIRES Climate Diagnostic Center (http://www.cdc.noaa.gov/).

The emergence and pattern of dissemination of V.
parahaemolyticus infections along coastal cities associated
with O3:K6 strains showed a close correspondence with the
arrival and propagation of 1997 El Niño along the coast of
South America. This El Niño episode reached northern Peru
by April 1997 and exhibited 2 periods of peak intensity in
June and November 1997 (Fig. 3B). The 2 peaks of maximum
intensity were linked to the arrival of the 2 major sets of
equatorial Kelvin waves, which were observed along the
coast of Peru with different degrees of intensity.28,29 The first
set of Kelvin waves reached Chiclayo (7° S latitude) by the
end of May and showed a strong effect in the northern area.
The intensity of this first anomaly decreased gradually during
its poleward displacement, to become almost imperceptible
south of 12° S latitude (Fig. 3B). This dynamic matched with
the first epidemic spread of V. parahaemolyticus infections,
which emerged in Chiclayo in July and was restricted to the
north and central areas of Peru, whereas no infections were
detected in southern Lima. The second and stronger set of
Kelvin waves was first observed at the equator in August
1997 and subsequently detected in northern Peru in Novem-
ber. This additional anomaly had a more severe and extended
effect on the entire coastline of Peru and was observed in the
north of Chile (23° S latitude). The propagation pattern was
similar to that observed for the second spread of infections
that were detected in Moquegua in November and subse-
quently reached the Antofagasta area. The 1997 El Niño first
began to decline in northern Chile in May 1998 and, from this
point, anomalies entered in a constant northward recession
until its total disappearance from the north of Peru by June
1998, coinciding with the decline of V. parahaemolyticus
infections.
Interactions between V. parahaemolyticus cases and the
environmental variables confirmed the earlier observations
linking the emergence of pandemic cases with the arrival of
El Niño to Peru (Table 1). For the pre–El Niño period, surface
seawater temperature was the dominant factor affecting the
disease dynamics in the zone. The rise of temperature with
the warmer months was the prevalent factor that affected the
appearance of infections associated with nonpandemic V.
parahaemolyticus populations. An increment in surface seawa-
ter temperature from 18.4°C–23.4°C accounted for an increase
of more than 600-fold in the risk of infections (IqRR ⫽ 601;
95% CI ⫽ 128 –2811).
By contrast, the dominant effect of surface seawater
temperature disappeared entirely from the model with the

© 2008 Lippincott Williams & Wilkins 833

Martinez-Urtaza et al Epidemiology • Volume 19, Number 6, November 2008

FIGURE 3. Dissemination of V. parahaemolyticus in Peru and progression of the 1997 El Niño waters. A, Southward dissemination
of cases of V. parahaemolyticus infections in 1997 and progression of the El Niño waters during the same period. The sites where
the first cases of cholera were detected in the 1991 epidemic (reported by Seas et al53) were included in the map. B, Advance of
the 1997 El Niño waters in Chiclayo (7° S latitude), Lima (12° S latitude), and Moquegua (16° S latitude) measured by heat content
above 20° C and number of cases of V. parahaemolyticus infections in Peru.

arrival of the 1997 El Niño episode, and the appearance of
pandemic cases in Peru could then be linked only to the
incoming of El Niño waters. The heat content over 20°C for
this period was associated with an IqRR of 4.7 (2.2–9.8).
Finally, disease dynamic during the post–El Niño pe-
riod could not be consistently explained by the models
constructed with the considered parameters, possibly because
of the complex environmental situation that prevailed during
the long period of ecologic restoration of the Peruvian coastal
areas after the El Niño event.
The pattern of spreading of pandemic V. parahaemo-
lyticus in Peru showed characteristics similar to those ob-

834 © 2008 Lippincott Williams & Wilkins

Epidemiology • Volume 19, Number 6, November 2008 Vibrio Epidemics in South America and El Niño

TABLE 1. Association Between Number of V. parahaemolyticus Infections and Environmental Factors. RR and IqRR With Their
Corresponding Confidence Intervals (95% CI) are Provided for the Three Study Periods
Percentiles
Period Variable Lag 25 50 75 RR (95% CI) IqRR (95% CI)

Pre–El Niño Surface seawater temperature 3 18.39 20.90 23.38 3.60 (2.65–4.91) 601 (128–2810)
Sea height anomalya 4 1.00
⬍4b 0.01 (0.00–11.53)
4-0 14.0 (0.33–601)
⬎0
Surface seawater temperature 3 18.39 20.91 23.38 2.00 (1.74–2.30) 32. (16–64)
Heat content above 20°C 4 0.00 10.65 28.41 1.04 (1.02–1.05) 2.90 (2.00–4.21)
El Niño Surface seawater temperature 1 21.29 23.15 25.13 0.97 (0.80–1.17) 0.88 (0.41–1.85)
Sea height anomaly 2 2.19 2.19 4.71 1.31 (1.15–1.49) 1.97 (1.41–2.74)
Surface seawater temperature 1 21.29 23.15 25.13 0.89 (0.71–1.11) 0.64 (0.27–1.51)
Heat content above 20°C 2 18.10 34.37 47.77 1.05 (1.03–1.08) 4.66 (2.23–9.75)
Post–El Niño Surface seawater temperature 5 18.05 20.32 23.58 1.19 (1.01–1.40) 2.59 (1.05–6.41)
Sea height anomaly 3 ⫺4.05 10.43 ⫺0.13 1.11 (0.95–1.28) 1.49 (0.83–2.67)
Surface seawater temperature 5 18.05 20.32 23.58 1.14 (0.94–1.37) 2.02 (0.70–5.80)
Heat content above 20°C 3 0.00 10.43 22.22 1.03 (0.99–1.07) 1.87 (0.85–4.12)
a
Results for the categorized variable due to the absence of a linear relationship between the predictor and variable response.
b
Reference category.

FIGURE 4. Number of cases of V. cholerae and V. parahaemolyticus infection from the emergence of their respective epidemics in
Peru in 1991 and 1997, according to the data from the Instituto Nacional de Salud. Relative abundance of cases of pandemic V.
parahaemolyticus was inferred from the percentage of pandemic strains detected among the representative strains retrieved from
the INS culture collection for laboratory investigation.

served for the 1991 cholera epidemic. The first cholera case
was detected in Trujillo on 23 October 1990; new cases were
then diagnosed in Chimbote on 11 December, Chancay on 26
December, and in Lima on 13 January (Fig. 3A). This gradual
southward dissemination of cases was similar in sequence
and speed to that observed for the fist spreading of pandemic
V. parahaemolyticus in 1997, which was linked to the weak-
est set of Kelvin waves with southern boundary in Lima. Both
epidemics were initially detected in the same area, Chiclayo
and Trujillo, and shared their southward rate progression,
covering 600 km of distance in just 3 months. Furthermore,
although the dimensions of the 2 diseases were not compa-

© 2008 Lippincott Williams & Wilkins 835

Martinez-Urtaza et al
rable, the 2 epidemics also showed similarities in the dynam-
ics of infection (Fig. 4). Both epidemics showed a character-
istic explosive emergence, a rapid fall in the number of cases,
and a final cyclic seasonal pattern under the influence of
increases in surface seawater temperature, before disappear-
ing almost totally. Surface seawater temperature failed to
account for the extraordinary dimension of emergence of
cases in both epidemics, which were linked with the arrival of
specific El Niño episodes.
DISCUSSION
The cause of the dramatic emergence of the cholera
epidemic in Peru in 1991 has not previously been understood.
The appearance of a new Vibrio disease in the same area has
yielded novel information that allows a better understanding
of the spreading mechanisms and the dynamics of these
waterborne diseases. The absence of environmental informa-
tion about the impact of the arrival of El Niño waters on the
native Vibrio community severely constrains a comprehen-
sive overview of the role of El Niño episodes in the intro-
duction of new Vibrio pathogens in these remote regions.
However, the abundant information about the major biologic
and oceanographic impacts of El Niño in this area, obtained
from oceanographic studies conducted in the Pacific coasts of
South America throughout the El Niño event, could poten-
tially provide new insight into the possible mechanisms
involved in these processes.
One of the main manifestations of the El Niño episodes in
America is the inflow of foreign zooplankton populations trapped in
the El Niño waters,29–32 as has been reported for different areas of
Peru and Chile for the 1997 El Niño.28–31,33,34 The arrival of
warm tropical waters produced a drastic reduction in up-
welling and primary production,30,35 which in turn caused a
dramatic decrease in the abundance of phytoplankton and the
subsequent collapse in zooplankton production. Conversely,
this decline in native zooplankton species did not have a
direct effect on the overall abundance of zooplankton, which
was maintained throughout the El Niño episode by the mas-
sive input of equatorial species of zooplankton—mainly
small copepods.28 –31,33,34
The survival and growth of V. parahaemolyticus in the
marine environment have been strongly associated with its
ability to attach to planktonic organisms, predominantly
copepods.36 –39 A close correlation between the zooplankton
blooms and cholera infections in endemic areas has been
suggested.40,41 The chitin present in copepods represents an
important food source for vibrios and provides protection
under unfavorable environmental conditions, thereby allow-
ing these organisms enhanced survival in the marine envi-
ronment.42– 45 The interaction between Vibrio and chitin has
also been found to provide a selective advantage for the
survival in the host.46 – 48 The biologic association of bacte-
rium and copepod may provide a stable platform for the
836
Epidemiology • Volume 19, Number 6, November 2008
displacement and gathering of Vibrio specimens in the open
ocean which, under certain extraordinary oceanographic cir-
cumstances, may allow for the long-distance displacement of
Vibrio specimens. This could have been the mechanism for
the extraordinary migration of Asian pathogens to America
coupled with El Niño. The arrival of the invasive exogenous
zooplankton populations to the coasts of Peru associated with
the El Niño waters may have been the vehicle of entry and
distribution of foreign Vibrio populations along the coast of
Peru and the subsequent contamination of fish and shellfish in
the coastal zones. The consumption of the contaminated
seafood may have finally been responsible for the emergence
of infections in phase with the evolution of the El Niño
episode described in this study. Due to the dose-response
dependence of both pathogens,49 –52 the explosive emergence
of Vibrio infections in the first stages of the epidemics may be
explained by the ingestion of seafood contaminated with
large amounts of Vibrio. From this point, the 2 diseases
acquired separate dimensions due to their distinctive abilities
to spread among the population. The acquisition of a hyper-
infectious state by V. cholerae resulting from its passage
through humans may have stimulated the rapid progression of
the cholera disease in interior regions associated with the
human-to-human transmission,51 whereas V. parahaemolyti-
cus infections remained restricted to coastal areas because of
their dependence on environment-to-human transmission.
The manifestation of an El Niño episode may imply the
generation of a temporary eastward corridor for the sporadic
displacement of marine organisms to America. This process
may additionally provide a periodic and unique source of new
pathogens in America, with serious implications for the
future spreading of waterborne infections on a global scale.
ACKNOWLEDGMENTS
The authors thank Juan Tarazona for helpful informa-
tion on the impact of El Niño in Peru; Mitsuaki Nishibuchi
and Romilio Espejo for providing some of the strains in-
cluded in this study; Douglas Bartlett, Tomas G. Villa, and
Oscar Garcia for critical advice. The authors are grateful for
the financial support and assistance contributed by the Xunta
de Galicia and the NOAA CoastWatch Program.
REFERENCES
1. Tauxe RV, Mintz ED, Quick RE. Epidemic cholera in the new world:
translating field epidemiology into new prevention strategies. Emerging
Infect Dis. 1995;1:141–146.
2. Okuda J, Ishibashi M, Hayakawa E, et al. Emergence of a unique O3:K6
clone of Vibrio parahaemolyticus in Calcutta, India, and isolation of
strains from the same clonal group from Southeast Asian travelers
arriving in Japan. J Clin Microbiol. 1997;35:3150 –3155.
3. Faruque SM, Nair GB. Epidemiology. In: Thompson FL, Austin B,
Swings J, eds. The Biology of Vibrios. Washington, DC: American
Society for Microbiology; 2006.
4. Gonzalez-Escalona N, Cachicas V, Acevedo C, et al. Vibrio parahae-
molyticus diarrhea, Chile, 1998 and 2004. Emerg Infect Dis. 2005;11:
129 –131.
5. Nair GB, Ramamurthy T, Bhattacharya SK, et al. Global dissemination
© 2008 Lippincott Williams & Wilkins
Epidemiology • Volume 19, Number 6, November 2008
of Vibrio parahaemolyticus serotype O3:K6 and its serovariants. Clin
Microbiol Rev. 2007;20:39 – 48.
6. Gordon AL, Susanto RD, Vranes K. Cool Indonesian throughflow as a
consequence of restricted surface layer flow. Nature. 2003;425:824 –
828.
7. Meyers G. Variation of Indonesian through-flow and ENSO. J Geophys
Res. 1996;101:12,255-12,263.
8. Mourino-Perez RR. Oceanography and the seventh cholera pandemic.
Epidemiology. 1998;9:355–357.
9. Pan-American Health Organization. Cholera in the Americas. Bull Pan
Am Health Org. 1991;25:267–273.
10. Pan-American Health Organization.Cholera in the Americas. Epidemiol
Bull. 1995;16:11–12.
10a.Colwell RR. Global climate and infectious disease: the cholera para-
digm. Science. 1996;274:2025–2031.
11. Kim YB, Okuda J, Matsumoto C, et al. Identification of Vibrio parah-
aemolyticus strains at the species level by PCR targeted to the toxR gene.
J Clin Microbiol. 1999;37:1173–1177.
12. Bej AK, Patterson DP, Brasher CW, et al. Detection of total and
hemolysin-producing Vibrio parahaemolyticus in shellfish using multi-
plex PCR amplification of tl, tdh and trh. J Microbiol Methods. 1999;
36:215–225.
13. Tada J, Ohashi T, Nishimura N, et al. Detection of the thermostable
direct hemolysin gene (tdh) and the thermostable direct hemolysin-
related hemolysin gene (trh) of Vibrio parahaemolyticus by polymerase
chain reaction. Mol Cell Probes. 1992;6:477– 487.
14. Matsumoto C, Okuda J, Ishibashi M, et al. Pandemic spread of an O3:K6
clone of Vibrio parahaemolyticus and emergence of related strains
evidenced by arbitrarily primed PCR and toxRS sequence analyses.
J Clin Microbiol. 2000;38:578 –585.
15. CDC. Standardized molecular subtyping of foodborne bacterial patho-
gens by pulsed-field gel electrophoresis. Atlanta, GA: Centers for
Diseases Control and Prevention; 2002.
16. Martinez-Urtaza J, Lozano-Leon A, DePaola A, et al. Characterization
of pathogenic Vibrio parahaemolyticus isolates from clinical sources in
Spain and comparison with Asian and North American pandemic iso-
lates. J Clin Microbiol. 2004;42:4672– 4678.
17. Cheney R, Miller L, Agreen R, et al. TOPEX/POSEIDON: the 2-cm
solution. J Geophys Res. 1991;99:24555–24564.
18. Shay L, Goni G, Black P. Effects of warm oceanic features on Hurricane
Opal. Mon Weather Rev. 2000;128:1366 –1383.
19. Kessler WS, McPhaden MJ, Weickmann KM. Forcing of intraseasonal
Kelvin waves in the equatorial Pacific. J Geophys Res. 1995;100:10613–
10632.
20. Vazquez J, Perry K, Kilpatrick K. NOAA/NASA AVHRR Oceans
pathfinder. Sea surface temperature data set. User’s reference manual
version 4.0. JPL Publication D-14070, 1998.
21. Wood SN. Generalized additive models: an introduction with R. Boca
Raton, FL: Chapman & Hall/CRC; 2006.
22. Bell ML, Samet JM, Dominici F. Time-series studies of particulate
matter. Ann Rev of Public Health. 2004;25:247–280.
23. Saez M, Ballester F, Barcelo MA, et al. A combined analysis of the
short-term effects of photochemical air pollutants on mortality within the
EMECAM project. Environ Health Perspect. 2002;110:221–228.
24. Lee JT, Kim H, Song HY, et al. Air pollution and asthma among
children in Seoul, Korea. Epidemiology. 2002;13:481– 484.
25. Wood SN. Thin plate regression splines.J Royal Stat Soc Ser B: Stat
Methodol. 2003;65:95–114.
26. Wood SN. Stable and efficient multiple smoothing parameter estimation
for generalized additive models. J Am Stat Assoc. 2004;99:673– 686.
27. Guerra Soto HE. Contribucion al hallazgo de especies de Vibrio de
origen humano y ambiental y su significado epidemiologico en Peru.
关PhD thesis兴. Universidad Nacional Mayor de San Marcos, 1988.
28. Ulloa O, Escribano N, Hormazabal S, et al. Evolution and biological
effects of the 1997–98 El Nino in the upwelling ecosystem off northern
Chile. Geophys Res Lett. 2001;28:1591–1594.
29. Sanchez G, Calienes R, Zuta S. The 1997–98 El Nino and its effects on
the coastal marine ecosystem off Peru. Calif Coop Oceanic Fish Invest
Rep. 2000;41:62– 86.
© 2008 Lippincott Williams & Wilkins
Vibrio Epidemics in South America and El Niño
30. Instituto del Mar de Peru. Informe 130. Crucero de evaluacion
hidroacustica de recursos pelagicos BIC Humboldt 9709-10 entre Mata-
rani y Paita. Callao, Peru: Instituto del Mar de Peru, 1998.
31. Aronés Flores K, Ayón Dejo P. Effects of El Niño 1997–1998 on
copepod community structure at San José Coastal Station (Peru). Inves-
tigaciones marinas. 2002;30:104 –105.
32. Escribano R, Daneri G, Farías L, et al. Biological and chemical conse-
quences of the 1997–1998 El Niño in the Chilean coastal upwelling
system: a synthesis. Deep Sea Res, Part II: Topical Studies in Ocean-
ography. 2004;51:2389 –2411.
33. Ochoa N, Ruillón G. Variabilidad del fitoplancton en la Bahía de Ancón,
Lima, Peru, durante el Niño 1997–98. In: Tarazona J, Arntz W, Castillo
de Maruenda E, eds. El Niño en America Latina. Impactos Biológicos y
Sociales. Lima, Peru: Consejo Nacional de Ciencia y Tecnologia, 2001;
39 – 45.
34. Gonzalez HE, Pages F, Sobarzo M, et al. Effects of the 1997–98 El Niño
on the oceanographic conditions and zooplankton community structure
in the coastal upwelling system off northern Chile. Investigaciones
marinas. 2002;30:112–114.
35. Chavez FP, Strutton PG, Friederich CE, et al. Biological and chemical
response of the equatorial Pacific Ocean to the 1997–98 El Nino.
Science. 1999;286:2126 –2131.
36. Kaneko T, Colwell RR. Ecology of Vibrio parahaemolyticus in Chesa-
peake Bay. J Bacteriol. 1973;113:24 –32.
37. Kaneko T, Colwell RR. Incidence of Vibrio parahaemolyticus in Ches-
apeake Bay. Appl Microbiol. 1975;30:251–257.
38. Kaneko T, Colwell RR. Adsorption of Vibrio parahaemolyticus onto
Chitin and Copepods. Appl Microbiol. 1975;29:269 –274.
39. Nalin DR, Daya V, Reid A, et al. Adsorption and growth of Vibrio
cholerae on Chitin. Infect Immun. 1979;25:768 –770.
40. Lipp EK, Huq A, Colwell RR. Effects of global climate on infectious
disease: the cholera model. Clin Microbiol Rev. 2002;15:757–770.
41. Lobitz B, Beck L, Huq A, Wood B, et al. Climate and infectious disease:
use of remote sensing for detection of Vibrio cholerae by indirect
measurement. Proc Natl Acad Sci USA. 2000;97:1438 –1443.
42. Amako K, Shimodori S, Imoto T, et al. Effects of chitin and its soluble
derivatives on survival of Vibrio cholerae O1 at low-temperature. Appl
Environ Microbiol. 1987;53:603– 605.
43. Dawson MP, Humphrey BA, Marshall KC. Adhesion—a tactic in the
survival strategy of a marine Vibrio during starvation. Curr Microbiol.
1981;6:195–199.
44. Dumontet S, Krovacek K, Baloda SB, et al. Ecological relationship
between Aeromonas and Vibrio spp and planktonic copepods in the
coastal marine environment in southern Italy. Comp Immunol Microbiol
Infect Dis. 1996;19:245–254.
45. Huq A, Small EB, West PA, et al. Ecological relationships between
Vibrio cholerae and planktonic crustacean copepods. Appl Environ
Microbiol. 1983;45:275–283.
46. Kaysner CA, Hill WE. Toxigenic Vibrio cholerae O1 in food and water.
In: Wachsmuth KI, Blake PA, Olsvik O, eds. Vibrio cholerae and
Cholera: Molecular to Global Perspectives. Washington, DC: American
Society for Microbiology, 1994.
47. Reidl J, Klose KE. Vibrio cholerae and cholera: out of the water and into
the host. Fems Microbiol Rev. 2002;26:125–139.
48. Reguera G, Kolter R. Virulence and the environment: a novel role for
Vibrio cholerae toxin-coregulated pili in biofilm formation on chitin.
J Bacteriol. 2005;187:3551–3555.
49. Kaper JB, Morris JG, Levine MM. Cholera. Clinical Microbiology
Reviews. 1995;8:48 – 86.
50. Cash RA, Music SI, Libonati JP, et al. Response of man to infection with
Vibrio cholerae. I. Clinical, serologic, and bacteriologic responses to a
known inoculum. J Infect Dis. 1974;129:45–52.
51. Hartley DM, Morris JG, Smith DL. Hyperinfectivity: a critical element in
the ability of V. cholerae to cause epidemics? PLoS Med. 2006;3:63– 69.
52. Quantitative risk assessment on the public health impact of pathogenic
Vibrio parahaemolyticus in raw oysters. FDA. 2005.
53. Seas C, Miranda J, Gil AI, et al. New insights on the emergence of
cholera in Latin America during 1991: the Peruvian experience. Am J
Trop Med Hyg. 2000;62:513–517.
837