V. cholerae into Latin America contributed to the
1991 epidemic (Fig. 2). The major epidemic clone
during the 1991 epidemic is represented by the
LAT-1 sublineage, corresponding phylogenetically to the previously described 7PET pandemic
wave 1 isolates that form the West-African South
American (WASA) lineage (26); these were originally typed as serotype Inaba, ribotype 5, and ET 4
(table S4) (12). We identified a frame-shift mutation
at position 165 in the wbe T gene, consistent with
the Inaba serotype in this lineage (table S1).
LAT-1 isolates carry the El Tor variant of ctxB
(ctxB3). The direct ancestors of LAT-1 in our phylogeny are isolates from Western and Central
Africa (Angola, Cote d’Ivoire, Sao Tome) from
the late 1980s. These African isolate genomes
and the LAT-1 sublineage are separated by only
13 nonrecombinant single-nucleotide polymorphisms (SNPs). Further support for an African ancestry is the placement of African isolates (Uganda
1992, Nigeria 1997) within the LAT-1 sublineage.
Several lines of evidence suggest that this introduction occurred close in time to the recorded
start of the epidemic. First, our time-resolved
phylogenies date this introduction to between
1985 and 1989, and the most recent common ancestor of LAT-1 isolates to 1989 (Fig. 2A). Second,
our data show that genetic features that define
the LAT-1 sublineage—the VSP-II gene variants
(insertion between VC_0510 and VC_0516) and the
WASA-1 genomic island—were acquired successively and in Africa during the late 1980s prior to
this introduction (Fig. 2A and fig. S9). Moreover,
Western Africa experienced cholera outbreaks
immediately before the Peruvian epidemic (fig.
S10). Tests for vibriocidal antibodies in stored sera
from Lima, Peru, in 1990 indicate that V. cholerae
was not present at this time (25). LAT-1 isolates
were isolated in Mexico until 2010 (Fig. 2C). The
more recent LAT-1 isolates collected between
2004 and 2010 in Mexico harbored a truncated
CTXϕ duplication and represent localized adaptations of these 7PET strains (27) (Fig. 2A).
The second clone introduced into Latin America
in 1991 was described as serotype Ogawa, ribotype
6a, ET3 (12, 13), and resistant to furazolidone,
sulfisoxazole, and streptomycin (13). This clone
was first detected in a mountainous village near
Mexico City, Mexico, in June 1991 and is believed
to have been imported via coca smugglers using
nearby private airstrips (28). This clone subse-
quently spread throughout Central America (13).
After 1993, this was the major clone circulating
in Mexico, where it persisted until 2000 (Fig. 2C)
(11). Our phylogeny shows that the introduction
of this second sublineage (LAT-2) occurred between
1987 and 1989, making this event concurrent
with that of the LAT-1 introduction (Fig. 2A). The
distinctive drug resistance profile of the LAT-2
sublineage was linked to the presence of a ge-
nomic island (GI-15) (Fig. 2A and figs. S9 and
S11). GI-15 harbors the genes responsible for
streptomycin (aadA) and sulfisoxazole (sul1) re-
sistance. With the exception of a few sporadic
isolates, LAT-1 isolates were pan-susceptible to
antimicrobials, and all lacked GI-15. The most
closely related isolates to those of LAT-2 are those
that were collected in South and Southeast Asia,
Western Asia (Lebanon), and Eastern Europe
(Romania) (Fig. 2A and table S1), many of which
also harbor GI-15. These globally circulating wave
2 isolates, including the LAT-2 sublineage, also
harbor ctxB1 (fig. S9) with CTXϕ integrated into the
smaller chromosome. Thus, the previously iden-
tified 7PET lineage harboring ctxB1 in Mexico (20)
was not a local lineage, but was derived from the
LAT-2 introduction. Our phylogeny indicates that
this lineage originated from South or Southeast
Asia (Fig. 2), from where it radiated globally. How-
ever, we cannot rule out that the introduction into
Mexico came via secondary site(s) and not di-
rectly from Asia (Fig. 2, A and B).
The third introduction (LAT-3) involved the
import of a South Asian strain into Haiti in 2010
and has been well documented (8, 29, 30). The
Haitian clone has been imported into surrounding
countries, including Cuba, the Dominican Republic,
the United States, and Mexico (31, 32) (Fig. 2). An
outbreak in 2013 within the Mexican region of
Hidalgo was suspected to be the result of an im-
port of the Haitian clone (32, 33). Our phylogeny
indicates that these isolates descended from the
Haitian (LAT-3) sublineage (Fig. 2) and share
key genomic features, including the ctxB7 var-
iant and a characteristic deletion within VSP-II
(DVC_0495-VC_0512) (29) (fig. S9).
Combined, we observe that V. cholerae line-
ages are associated with three distinct patterns of
diarrheal disease within Latin America. First, there
are lineages responsible for sporadic cases or
limited outbreaks, in which secondary infections
are rare or nonexistent (Fig. 1C) (11). Second,
lineages that occupy long-term environmental
reservoirs (such as the Gulf Coast lineage) cause
illness over longer periods of time and across
larger geographic areas (Fig. 1C and fig. S4). The
third pattern, caused by pandemic V. cholerae, is
visibly distinct. Pandemic lineages are responsible
for massive, explosive epidemics that occur over
short periods of time. The epidemiological distinc-
tion between local and pandemic lineages is stark—
nearly 20,000 cases per week were seen at the
beginning of the 1991 epidemic in Peru (28), and
more than 250,000 cases were seen over 6 months
at the beginning of the 2010 Haitian epidemic.
By contrast, only 65 infections reported over a
20-year period in the USA were associated with
the Gulf Coast reservoir (34). We expand upon
these definitions in supplementary text note 2.
We show conclusively that both historical cholera
epidemics within Latin America were the result
of intercontinental introductions of globally circu-
lating 7PET lineages and were not derived from
indigenous local lineages. These data (i.e., the
introduction of LAT-1 from Africa) also do not
support the hypothesis that El Niño was respon-
sible for the introduction of cholera in Peru in
1991 by potentiating the long-distance transport
of aquatic pathogens from Asia through a biological
corridor (35) or due to a surge in preexisting local
lineages (36, 37 ) (Figs. 1and 2). Our data are instead
consistent with descriptions of how cholera was
introduced into Haiti in 2010 (i.e., through carriers
or patients from endemic regions) (8, 30). We have
shown that over a 30-year span, several local
lineages are present at relatively constant levels
(Fig. 1C). This underlines that local and pandemic
lineages exhibit different epidemiological behav-
iors, and may occupy different ecological niches in
We show that there are local foci of diverse
V. cholerae lineages that cause sporadic out-
breaks across Latin America. Local lineages share
many characteristics with pandemic clones, such
as being toxigenic and of serogroup O1 (table S3).
Disease caused by these lineages would thus be
defined as cholera by both the World Health
Organization (38) and U.S. Centers for Disease
Control and Prevention (39). However, these local
lineages show markedly different patterns of dis-
ease to that of the 7PET pandemic V. cholerae
lineage. The potential of a V. cholerae isolate to
cause disease is best understood by studying its
genomics, whether by whole-genome sequenc-
ing or a polymerase chain reaction–based typing
scheme, as well as considering clinical symptoms,
epidemiological context, and basic pheno- and
In this study, we have unified previous accounts
of cholera within Latin America into a cohesive
genomic framework that correctly emphasizes the
relative contributions of different bacterial line-
ages to this diarrheal disease. An appreciation of
the differences between pandemic and local line-
ages should inform the design of disease control
strategies in Latin America. Measured and graded
public health responses could be designed based
on an understanding of which lineages are re-
sponsible for outbreaks of cholera. V. cholerae
lineages can be prioritized as public health con-
cerns if they deviate from patterns associated
with local lineages.
REFERENCES AND NOTES
1. World Health Organization, Cholera; www.who.int/
2. D. Ceccarelli, R. R. Colwell, Front. Microbiol. 5, 256 (2014).
3. R. R. Colwell, Science 274, 2025–2031 (1996).
4. E. K. Lipp, A. Huq, R. R. Colwell, Clin. Microbiol. Rev. 15,
5. E. T. Ryan, PLOS Negl. Trop. Dis. 5, e1003 (2011).
6. P. A. Blake et al., N. Engl. J. Med. 302, 305–309 (1980).
7. J. Kumate, J. Sepúlveda, G. Gutiérrez, Bull. Inst. Pasteur 96,
8. L. S. Katz et al., MBio 4, e00398-13 (2013).
9. Pan American Health Organization/ World Health Organization,
Cholera in the Americas - Situation summary. Epidemiol. Update
Cholera (2017); www.paho.org/hq/index.php?option=com_
10. D. N. Cameron, F. M. Khambaty, I. K. Wachsmuth, R. V. Tauxe,
T. J. Barrett, J. Clin. Microbiol. 32, 1685–1690 (1994).
11. M. L. Lizárraga-Partida, M.-L. Quilici, J. Clin. Microbiol. 47,
12. T. Popovic, C. Bopp, O. Olsvik, K. Wachsmuth, J. Clin. Microbiol.
31, 2474–2482 (1993).
13. G. M. Evins et al., J. Infect. Dis. 172, 173–179 (1995).
14. A. Dalsgaard et al., J. Clin. Microbiol. 35, 1151–1156 (1997).
15. Materials and methods are available as supplementary materials.
16. F.-X. Weill et al., Science 358, 785–789 (2017).
17. J. B. Kaper, H. B. Bradford, N. C. Roberts, S. Falkow, J. Clin.
Microbiol. 16, 129–134 (1982).
18. M. Pichel et al., J. Clin. Microbiol. 41, 124–134 (2003).
19. A. Coelho, J. R. Andrade, A. C. Vicente, C. A. Salles, J. Clin.
Microbiol. 33, 114–118 (1995).
20. M. Alam et al., J. Clin. Microbiol. 48, 3666–3674 (2010).
21. M. Alam et al., J. Clin. Microbiol. 50, 2212–2216 (2012).
22. A. Siriphap et al., PLOS ONE 12, e0169324 (2017).
23. P. A. Blake et al., Lancet 2, 912 (1983).
792 10 NOVEMBER 2017 • VOL 358 ISSUE 6364 sciencemag.org SCIENCE
RESEARCH | REPORTS