Acta Tropica 92 (2004) 245–252

Mathematical modelling of American Cutaneous Leishmaniasis:

incidental hosts and threshold conditions for infection persistence
Luis Fernando Chaves1 , Maria-Josefina Hernandez∗
Laboratorio de Biologı́a Teórica & Postgrado en Ecologı́a, Instituto de Zoologı́a Tropical, Facultad de Ciencias,
Universidad Central de Venezuela, Apartado Postal 47058, Caracas 1041-A, Venezuela

Received 29 April 2004; received in revised form 13 August 2004; accepted 13 August 2004
Available online 16 September 2004

Abstract

We present a model for the dynamics of transmission of American Cutaneous Leishmaniasis (ACL) that includes a population
of incidental hosts for parasites (that is, they act only as sinks of infection), along with species that are reservoir hosts (that is,
both source and sink of infection). Although, there is evidence that suggests the existence of incidental hosts for Leishmania
parasites, there are no mathematical models developed to account for this. Based on this model we obtained expressions that allow
computing the threshold conditions for the persistence of the infection using three different approaches. The three expressions
agree in the parameters involved in the computation of the threshold, and also in the parameters that are excluded from it, which
are those related to the transmission in incidental hosts. We also suggest alternatives for future development in the modelling of
the dynamics of transmission of ACL, according to the needs of a more realistic, precise, or general character of the model.
© 2004 Elsevier B.V. All rights reserved.

Keywords: Zoonosis; Basic reproduction ratio; Reservoir hosts; Incidental hosts; American Cutaneous Leishmaniasis

1. Introduction of the genus Leishmania, transmitted by phlebotomine

American Cutaneous Leishmaniasis (ACL) is a
resurgent disease (Gratz, 1999) caused by parasites
∗ Corresponding author. Tel.: +58 212 6051310;
fax: +58 212 6052136.
E-mail addresses: lfchaves@umich.edu (L.F. Chaves),
mjhernan@strix.ciens.ucv.ve (M.-J. Hernandez).
1 Present address: Department of Ecology and Evolutionary Biol-
ogy, University of Michigan, Ann Arbor, MI 48109, USA.
sandflies (Diptera: Psychodidae) from reservoir hosts,
which are fundamental for the establishment of the dis-
ease (Killick-Kendrick and Ward, 1981). This is a com-
mon pattern in nature (Woolhouse et al., 2001). Levins
et al. (1994) state that from 412 infectious diseases af-
fecting humans in at least 180 there are non-human
animal hosts.
Mathematical models are necessary tools in the
study of diseases; they provide the means to explore
and understand regulatory mechanisms of a system

0001-706X/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.actatropica.2004.08.004
246 L.F. Chaves, M.-J. Hernandez / Acta Tropica 92 (2004) 245–252
(Puccia et al., 1994). In particular, the proposal of dis-
ease control strategies using mathematical models is
mainly based on the diminishing of the basic reproduc-
tion ratio (often also referred to as basic reproduction
number, or basic reproductive rate) of the disease, R0
(Smith, 1994).
The basic reproduction ratio of a disease, R0 , for
microparasite infections, is defined by Diekmann et al.
(1990) as the average number of secondary infections
produced when one infected individual is introduced
into a host population, where everyone is susceptible.
R0 is a useful parameter in the study of a disease as it
sets the threshold for its establishment (R0 > 1) or its
extinction (R0 < 1).
In a recent article, Roberts and Heesterbeek (2003)
distinguish between the basic reproduction ratio (R0 )
and the threshold quantity (T), showing that, although
both have the same threshold properties, they do not
have the same biological interpretations. They state that
the definition of R0 refers to the generation of new in-
fections, while T deals with the infection of secondary
hosts in the population of the primary case. Thus, T
provides a direct measure of the control effort required
for eradication of an infection, taking into account dif-
ferent types of hosts.
The mathematical modelling of the dynamics of
transmission of Leishmaniasis has been poorly devel-
oped if we compare it with other infectious diseases as,
for instance, Malaria (Awerbuch, 1994). Hasibeder et
al. (1992) and Dye (1996) developed models on canine
and zoonotic Visceral Leishmaniasis, respectively, fo-
cusing on the dynamics of infection in dogs. Lysenko
and Beljaev (1987) developed an epidemetric model
on the transmission of Cutaneous Leishmaniasis. How-
ever, the mechanisms that govern the dynamics of the
transmission of Leishmania parasites are not explicitly
stated, therefore, no control strategies for the disease
can be proposed. Recently, Rabinovich and Feliciangeli
(2004) added some realism to this model, by making the
probability of developing cutaneous lessons a function
of the number of infective bites by sand flies. However,
this approach does not allow to investigate on the role of
species as reservoir hosts. Burattini et al. (1998) formu-
lated a model on the dynamics of transmission of Leish-
maniasis including populations of vectors, humans and
animal (dogs) hosts. Assuming that all populations are
both source and sink of Leishmania parasites, they esti-
mated R0 —actually, the infection persistence threshold
condition—from the sum of individual terms for trans-
mission in humans and in dogs.
Some authors consider that the dynamics of trans-
mission of dermotropic Leishmania parasites into hu-
mans, especially in the American continent, occur in an
incidental way (Lysenko and Beljaev, 1987; Ashford,
1997). This implies that humans are not an important
source of infection towards vectors and, therefore, nei-
ther they are towards other humans. The role of source
of infection to vectors is only fulfilled by animal reser-
voirs. This justifies the proposition of alternative mod-
els on the dynamics of transmission of ACL, since it is
a zoonosis where not all the species involved are source
and sink of the parasites. Thus, based on this statement,
in this work we propose a simple mathematical model
that allows the calculation of the threshold conditions
for the persistence of the infection for the American
Cutaneous Leishmaniasis.
2. The model
The classical Ross model for Malaria consists of
two first-order differential equations that represent the
dynamics of infection in vector (mosquitoes) and in
human populations. The model considers all individu-
als as both source and sink of infection (Anderson and
May, 1992).
As stated above, the dynamics of transmission in
ACL is bi-directional between vectors and reservoir
hosts, that is, they are all both source and sink of infec-
tion, whereas humans are incidental hosts, that is, they
are only sink for the infection. Thus, we include a third
first-order differential equation in the model to account
for the dynamics of transmission of incidental hosts. All
other assumptions are the same as for the Ross model,
namely: (i) the dynamics of infection follow the mass
action principle, that is, it is assumed that the net rate
of spread of infection is proportional to the product of
the density of susceptible individuals times the density
of infectious individuals; (ii) the total population of all
hosts and vectors is assumed to be constant; (iii) death
rates of infected hosts are very small in comparison
with recovery rates, thus, the former are of negligible
importance in the loss of infected hosts; (iv) vectors
do not recover from infection, and infection in vector
population is regulated by mortality of its individuals;
(v) recovered individuals (hosts) have no immunity to
 L.F. Chaves, M.-J. Hernandez / Acta Tropica 92 (2004) 245–252 247

reinfection, i.e. they become susceptible again; (for a which can be re-arranged as
general review see e.g. Anderson and May, 1992).
Thus, we present here a simple model to represent AβH (BCβR2 − µγR )
H∗ =
the dynamics of transmission of ACL using three differ- βH (BCβR − µγR ) + γH (BβR2
2 + βR µ)
ential equations for the rates of change in population BCβR2 − µγR
densities of infected incidental hosts, H (t), infected R∗ = (3)
βR (CβR + γR )
reservoir hosts, R (t), and infected vectors, V (t), that is
BCβR2 − µγR
V∗ =
H  (t) = βH V (A − H) − γH H βR (BβR + µ)
R (t) = βR V (B − R) − γR R (1)
The threshold condition for the establishment of the
V  (t) = βR R(C − V ) − µV disease can be calculated from expressions in (3); this
is
Where primes denote the rate of change (d/dt) of
population density, H the population size of infected R∗ > 0, V ∗ > 0 ⇔ BCβR2 > µγR (4)
incidental hosts, R the population size of infected reser-
voir hosts, V the population size of infected vectors, and under this condition it is also assured that H∗ > 0.
A the population size of incidental hosts, B the pop- From (4) we can conclude that the threshold for en-
ulation size of reservoir hosts, C the population size demic persistence is
of vectors, βH the rate of infection, per individual, in
vector-incidental host encounters, per unit time, βR the BCβR2
rate of infection, per individual, in vector-reservoir host >1 (5)
µγR
encounters, per unit time, γ H the recovery rate of in-
cidental hosts per unit time, γ R the recovery rate of
reservoir hosts per unit time; µ the mortality rate of
vectors per unit time. 3.2. Computation of threshold conditions for
infection based on the stability analysis of the
model evaluated at trivial equilibrium
3. Threshold conditions for infection
Stability analysis is a widely used tool to study the
performance of dynamical systems at equilibria (see
3.1. Computation of threshold conditions for
e.g. May, 1973). For continuous time systems we first
infection based on equilibrium values
compute the eigenvalues of the jacobian matrix of the
system evaluated at equilibrium. If these are negative
To use this approach we first calculate the equi-
(or, if complex, the real part is negative) then the equi-
librium solutions for all populations involved in the
librium is stable; if they are positive (or, if complex,
dynamics of transmission of the parasite, and subse-
the real part is positive) it is unstable. Smith (1994)
quently the conditions that assure that these values are
states that the establishment of a disease occurs if
greater than zero (Anderson, 1991).
the trivial equilibrium solution, H∗ = R∗ = V∗ = 0, is
The non-trivial equilibrium solutions for popula-
unstable.
tions in system (1), i.e. H∗ , R∗ , V∗ , all = 0, for H (t) =
Let M be the jacobian matrix of model in (1), that
R (t) = V (t) = 0, are
is
AβH V ∗  
H∗ = ∂H  ∂H  ∂H 
βH V ∗ + γH  ∂H ∂R ∂V 
 
BβR V ∗  ∂R ∂R ∂R 
R∗ = (2) M=   (6)
βR V ∗ + γ R  ∂H ∂R ∂V 
CβR R∗  ∂V  ∂V  ∂V  
V∗ =
βR R∗ + µ ∂H ∂R ∂V
248 L.F. Chaves, M.-J. Hernandez / Acta Tropica 92 (2004) 245–252

where
∂H  ∂H  ∂H 
= −βH V − γH =0 = βH (A − H)
∂H ∂R ∂V
∂R ∂R ∂R
=0 = −βR V − γR = βR (B − R)
∂H ∂R ∂V
∂V  ∂V  ∂V 
=0 = βR (C − V ) = −βR R − ␮
∂H ∂R ∂V
We compute the eigenvalues, λi , of M evaluated at triv-
ial equilibrium H∗ = R∗ = V∗ = 0. These are the roots 3.3. Computation of the basic reproduction ratio
of det(M(0,0,0) − λI) = 0, that is R0 from the dominant eigenvalue of the next
  generation operator
−γH − λ 0 βH A
 
det  0 −γR − λ βR B  = 0 Diekmann et al. (1990) define the basic reproduction
0 βR C −µ − λ ratio (R0 ) of a disease as the dominant eigenvalue of
the next generation operator (NGO). This is a matrix
Solving the determinant above we obtain the polyno- which is the result of the product of two functions:
mial (characteristic equation of M): one that takes into account the population density of
susceptible individuals, and another one that describes
λ3 + λ2 (γH + γR + µ) + λ(γH γR + γH µ the infectivity of an infected individual on a susceptible
+ γR µ − BCβR2 ) + γH (γR µ − BCβR2 ) = 0 (7) population.
López et al. (2002) developed a methodology to ob-
According to the Routh–Hurwitz criteria (May, 1973), tain the NGO (details in Appendix B) from which we
for a polynomial of the type obtain the following result:
 
λ3 + λ2 X + λY + Z = 0 (8) NH→H NR→H NV→H
 
NGO =  NH→R NR→R NV→R 
the equilibrium is stable if X > 0, Z > 0, and XY > Z.
NH→V NR→V NV→V
The X term in (7) is always positive; the condition XY
 
> Z holds (see Appendix A); and the Z term is positive βH A
0 0
under the following condition:  µ 
 
 βR B 
Z > 0 ⇔ µγR > BCβR2 =
0 0  (11)
µ 
(9)
 
 βR C 
This means that the system at the trivial solution is 0 0
unstable if condition (9) does not hold, that is, if γR

BCβR2 where the elements Ni→j represent the number of indi-

>1 (10) viduals of species j infected by an infectious individual
µγR
of species i.
It is important to highlight here that when condition From (11) we can calculate the basic reproduction
(10) holds, the non-trivial equilibrium (i.e. all H∗ , R∗ , ratio from det(NGO − λI) = 0 as follows:
V∗ > 0) is globally stable (since the only other solution  
βH A
would be the unstable trivial equilibrium). This means −λ 0
 µ 
that the system will always attain an endemic equilib-  
 βR B 
rium. On the other hand, if (10) does not hold, the trivial det 
 0 −λ =0
solution is globally stable and the non-trivial unstable;  µ  
 βR C 
thus, the system always converges to a situation where 0 −λ
no infection exists. γR
 L.F. Chaves, M.-J. Hernandez / Acta Tropica 92 (2004) 245–252
Solving, we get the polynomial
 
BCβR2
−λ + λ
3
=0
µγR
The basic reproduction ratio is the dominant eigenvalue
of this matrix (Diekmann et al., 1990), that is,
BC
R0 ≡ λ = βR (12)
µγR
Expression (12) presents the same threshold condition
as expressions (5) and (10). For R0 > 1 an endemic
establishment of the infection occurs, and for R0 < 1 the
situation converges to the extinction of the infection.
4. Discussion
4.1. A first point to be discussed from the results
above is the fact that all parameters are the same in
the two expressions obtained for the threshold condi-
tion for infection persistence and the one for R0 —Eqs.
(5), (10), (12)—and above this, that none of them ex-
plicitly include parameters relating to the dynamics
of transmission in incidental hosts—i.e. those param-
eters included in the first equation of model (1). Al-
though some models consider that infection rates are
inversely related to the sum of the densities of all verte-
brate host species (Woolhouse et al., 2001; López et al.,
2002), these rates are independent of other parameters
of transmission related to the remaining hosts species in
the system. Thus, the endemic establishment threshold
conditions are independent of the dynamics of trans-
mission into incidental hosts. This is to be expected
from the model proposed here because the introduction
of infected incidental hosts will not cause infections in
susceptible vectors, and therefore, will not produce new
infections in susceptible incidental hosts either. In fact,
from (11) we can observe that the element NH→V is
equal to zero; that is, there is no transmission of infec-
tion from incidental hosts (H) to vectors (V). Thus, we
expect the dynamics of infection of the subsystem com-
prising only reservoir hosts and vectors to be indepen-
dent of the dynamics of infection of incidental hosts.
But, conversely, the dynamics of infection of incidental
hosts do depend on the other two populations. This is
a very significant result because it implies that control
249
measures directed towards incidental host populations
will not have consequences on the endemic character
of a focus, but rather depends on the endemicity in the
reservoir hosts.
Additionally, these results also indicate that the defi-
nition of R0 as the number of secondary cases produced
by the introduction of a primary case may be inadequate
in multi-specific systems where not all populations play
the role of both source and sink of parasites. That is,
in host populations that only act as sinks of parasites,
the generation of new infections necessarily depends
on the hosts that act as reservoirs of the parasites.
4.2. The mathematical model presented in this work
is an approximation to the study of the dynamics of
infection in zoonoses, transmitted by vectors, that in-
volves two types of vertebrate hosts: reservoirs and in-
cidentals. Although it allows fulfilling the purpose in-
tended, it can be modified and adapted in different ways
according to new goals. Levins (1966) argues that it is
impossible to build mathematical models that, at the
same time, maximize realism, generality and precision
to satisfy the overlapping but not identical goals of un-
derstanding, predicting and modifying nature. Faced
with this conflict, Levins suggests that during the pro-
cess of modelling an option is to develop two of these
characteristics in detriment of the third.
Under this point of view, we can first recognize
the model proposed in (1) as of a general and precise
kind. It is general, because it is an abstraction of the
phenomenon of the dynamics of transmission of a
zoonosis that is transmitted by vectors, and where not
all the host populations involved are reservoirs of the
pathogen. It is precise, because it allows the calculation
of the basic reproduction ratio as a relationship of the
parameters involved in the process of the dynamics
of transmission. However, it is not very realistic. For
example, some of the parameters are abstractions of
several processes; as is the case of the rates of infection
in host–vector contacts, which depend on the probabil-
ity of encounters between a vector or an infected host
with susceptible individuals of the other population,
and on the biting rate for each host type. For example,
if we assume that biting rates are inversely related
to total host density, then, the greater the number of
incidental hosts the greater the odds that the disease
will not be established, due to the reduction in contacts
between reservoir hosts and vectors. This fact is not
obvious when we use rates of infection in contacts
250 L.F. Chaves, M.-J. Hernandez / Acta Tropica 92 (2004) 245–252
instead of a detailed description of all the processes
abstracted in that parameter, but the inclusion of the
latter could make the model unmanageable, without
gaining any insight on the properties of the model.
To increment realism and generality, given that ACL
is a disease affecting humans, we should include the
effect of human behaviour and social organization on
the emergence of the disease. Wilson (1994) suggests
that within the patterns that explain the emergence
of diseases, changes in human behaviour are part of
the events that promote such process; for instance, the
movement of humans or of the pathogens themselves.
This can be illustrated with the emergence of an ACL
focus in Las Rosas, Venezuela, which according to
Aguilar et al. (1984) could have been caused by the
introduction of infected equines brought from other ar-
eas where ACL was endemic.
To increment precision and realism—in detriment of
generality—of the model in (1), we could consider, on
the one hand, the traditional modifications of inclusion
of time of incubation of the pathogens, superinfections,
and host acquired immunity (Awerbuch, 1994). Addi-
tionally, we can think of particular observed phenom-
ena regarding the biology of vectors of leishmaniasis,
the pathogenicity of ACL in some of its hosts, studies
on other diseases, and spatial considerations:
(i) It is reported that the population densities of
vectors of ACL show large fluctuations, which
are correlated to precipitation (Márquez and
Scorza, 1984; Feliciangeli and Rabinovich,
1998; Salomón et al., 2004). Thus, models that
include the relationship of climatic variables with
pathogenic transmission could predict the effects
of global climatic changes on the magnitude and
geographic distribution of vector-borne diseases
(Martens, 1999).
(ii) There is evidence of the presence of DNA of
Leishmania braziliensis in the blood of patients
after 30 years of recovery of cutaneous lesions
(Guevara et al., 1993). Although this does not nec-
essarily entails the presence of parasites, it leads to
the possibility of chronic infections, or of inappar-
ent infections, that is, those that are not detected
because they never lead to the development of
cutaneous lesions. In human patients there is also
evidence of cutaneous lesions that are caused by
reactivation of infections—a phenomenon known
as recrudescence—and not by new infections
with parasites (Saravia et al., 1990). This factor
needs to be taken into account when modelling
ACL infection dynamics to avoid confusions on
the causes of morbidity of individuals.
(iii) Experimental studies on other diseases have
shown that contact infection rates vary with the
density of susceptible and infected host popula-
tions (Fenton et al., 2002). To deal with the issue
of adequately modelling interactions between
individuals, one can use non-linear functions
that describe the biological mechanisms involved
in the process of infection. Such functions can
be stated from an individual perspective, for
example to describe the effect of risk factors at
an individual level on the dynamics of the popu-
lation (Koopman and Longini, 1994) or based on
mechanisms of interaction at a population level
that lead to multiple stable equilibria (Hernandez,
1998; Hernandez and Barradas, 2003).
(iv) We can also include spatial heterogeneity in
the study of the dynamics of infection, which
can be done either using dynamical systems
represented by ordinary differential equations
(Torres-Sorando and Rodrı́guez, 1997; Rodrı́guez
and Torres-Sorando, 2001) or cellular automata
representations (Kiszewski and Spielman, 1994).
Finally, there is also the possibility of using differ-
ent tools for modelling and simulation, either based
on individual event histories or dynamical networks
(Koopman et al., 2001). The development of models
to represent this disease from different perspectives is
very important. As stated by Levins (1966), the con-
gruence in the results or the robustness of the theorems
is fundamental in the understanding of the patterns
and processes that occur in nature. This allows rec-
ognizing, together with experimental testing or obser-
vations, which of the assumptions on which the mod-
els are based are determinant in the behaviour of the
phenomenon under study. In the case of infectious dis-
eases, this may be the basis for the proposal of effective
control measures.
Acknowledgements
We want to acknowledge Dr. Luis F. López at Uni-
versidade de São Paulo, São Paulo, Brazil, for provid-
 L.F. Chaves, M.-J. Hernandez / Acta Tropica 92 (2004) 245–252 251

ing us with copies of some of his papers. A special
acknowledgement is given to our dear friend and col-
league Diego J. Rodrı́guez at Universidad Central de
Venezuela, for his ever-constructive comments and in-
valuable advice. This research was partly supported by
grant no. PI.03.31.4971.2002 from Consejo de Desar-
rollo Cientı́fico y Humanı́stico, Universidad Central de
Venezuela (MJH); and by Fundación Polar (LC). This
work is part of the M.Sc. thesis of LFC.
Appendix A
trices, thus, (B.1) can be expressed as
y = (T − D)y (B.2)
where T is a non-negative matrix that contains the ele-
ments related to the generation of new infections, and
D is a diagonal non-negative matrix that contains the
elements related to loss of infections. Considering that
T corresponds to the infectivity function of an infected
individual on a susceptible population, and that D−1 is
a diagonal matrix that can be interpreted as the suscep-
tibility of populations, then at trivial equilibrium, H∗ =
R∗ = V∗ = 0, the matrix NGO is

A.1. Proof of condition XY>Z fulfilled in NGO = TD−1

polynomial (7)   1 
0 0 βH A γH 0 0
 

= 0 0 βR B   0 1
0

λ3 + λ2 (γH + γR + µ) + λ(γH γR + γH µ γR
0 βR C 0 0 0 1
+ γR µ − BCβR2 ) + γH (γR µ − BCβR2 ) = 0 µ
 βH A 
Given λ3 + λ2 X + λY + Z, then condition XY > Z is 0 0 µ
 βR B 
(γH + γR + µ)(γH γR + γH µ + γR µ − BCβR2 ) =
0 0 µ

 (B.3)
βR C
0 0
> γH (γR µ − BCβR2 ) (A.1) γR

Performing the change of variable U = (γR µ − BCβR2 )

gives: References
(γH + γR + µ) (γH γR + γH µ + U) > γH U (A.2)
Aguilar, C.M., Fernández, E., Fernández, R., Deane, L.M., 1984.
All parameter values in (A.2) are positive, then, condi- Study of an outbreak of cutaneous Leishmaniasis in Venezuela.
The role of domestic animals. Mem. Inst. Oswaldo Cruz. 79,
tion (A.1) holds because the term in the right hand side 181–195.
in (A.2) is just one of the nine terms in the left hand Anderson, R.M., 1991. Populations and infectious diseases: ecology
side. or epidemiology? J. Anim. Ecol. 60, 1–50.
Anderson, R.M., May, R.M., 1992. Infectious Diseases of Humans.
Oxford University Press.
Ashford, R.W., 1997. The Leishmaniasis as model zoonoses. Ann.
Appendix B Trop. Med. Parasitol. 91, 693–710.
Awerbuch, T., 1994. Evolution of mathematical models of epidemics.
B.1. Computation of the Next Generation Ann. N. Y. Acad. Sci. 740, 232–241.
Operator (NGO) Burattini, M.N., Coutinho, F.A.B., López, L.F., Massad, E., 1998.
Modelling the dynamics of leishmaniasis considering human,
animal host and vector populations. J. Biol. Syst. 6, 337–356.
The system of equations in (1) can be re-written as Diekmann, O., Heesterbeek, J.A.P., Metz, J.A.J., 1990. On the defi-
    
H −γH 0 βH (A − H) H nition and the computation of the basic reproduction ratio R0 in
     models for infectious diseases in heterogeneous populations. J.
R = 0 −γR βR (B − R)   R  Math. Biol. 28, 365–382.
V  0 βR (C − V ) −µ V Dye, C., 1996. The logic of visceral leishmaniasis control. Am. J.
Trop. Med. Hyg. 55, 125–130.
(B.1) Feliciangeli, M.D., Rabinovich, J., 1998. Abundance of Lutzomyia
ovallesi but no Lu.gomezi (Diptera: Psychodidae) correlated with
According to López et al. (2002) the square matrix cutaneous leishmaniasis incidence in north-central Venezuela.
above can be re-written as the subtraction of two ma- Med. Vet. Entomol. 12, 121–131.
252 L.F. Chaves, M.-J. Hernandez / Acta Tropica 92 (2004) 245–252
Fenton, A., Fairbairn, J.P., Norman, R., Hudson, P.J., 2002. Parasite
transmission: reconciling theory and reality. J. Anim. Ecol. 71,
893–905.
Gratz, N.G., 1999. Emerging and resurging vector-borne diseases.
Ann. Rev. Entomol. 44, 51–75.
Guevara, P., Ramı́rez, J.L., Rojas, E., Scorza, J.V., González, N.,
Añez, N., 1993. Leishmania braziliensis in blood 30 years after
cure. Lancet 341, 1341.
Hasibeder, G., Dye, C., Carpenter, J., 1992. Mathematical modelling
and theory for estimating the basic reproduction number of canine
leishmaniasis. Parasitology 105, 43–53.
Hernandez, M.J., 1998. Dynamics of transitions between population
interactions: a nonlinear interaction ␣-function defined. Proc. R.
Soc. Lond. B 265, 1433–1440.
Hernandez, M.J., Barradas, I., 2003. Variation in the outcome of
population interactions: bifurcations and catastrophes. J. Math.
Biol. 46, 571–594.
Killick-Kendrick, R., Ward, R.D., 1981. Ecology of Leishmania. Par-
asitology 82 (Suppl.), 143–152.
Kiszewski, A.E., Spielman, A., 1994. Virulence of vector-borne
pathogens. A stochastic automata model of perpetuation. Ann.
N. Y. Acad. Sci. 740, 249–259.
Koopman, J.S., Jacquez, G., Chick, S.E., 2001. New data and tools for
integrating discrete and continuous population modeling strate-
gies. Ann. N. Y. Acad. Sci. 954, 268–294.
Koopman, J.S., Longini Jr., I.M., 1994. The ecological effects of indi-
vidual exposures and nonlinear disease dynamics in populations.
Am. J. Public Health 84, 836–842.
Levins, R., 1966. The strategy of model building in population biol-
ogy. Am. Sci. 54, 421–431.
Levins, R., Awerbuch, T., Brinkmann, U., Eckardt, I., Epstein, P.,
Makhoul, N., de Possas, C.A., Puccia, C., Spielman, A., Wilson,
M.E., 1994. The emergence of new diseases. Am. Sci. 82, 52–60.
López, L.F., Coutinho, F.A.B., Burattini, M.N., Massad, E., 2002.
Threshold conditions for infection persistence in complex
host–vectors interactions. C. R. Biologies. 325, 1073–1084.
Lysenko, A.J., Beljaev, A.E., 1987. Quantitative approaches to epi-
demiology. In: Peters, W., Killick-Kendrick, R. (Eds.), The Leish-
maniasis in Biology and Medicine, vol. I. Academic Press, Lon-
don, pp. 263–290.
Márquez, J.C., Scorza, J.V., 1984. Dinámica poblacional de Lut-
zomyia townsendi (Ortiz, 1959) (Diptera: Psychodidae) y su pari-
dad en Trujillo, Venezuela. Bol. Dir. Malariol. San. Amb. 24,
8–20.
Martens, P., 1999. How will climate change affect human health?
Am. Sci. 87, 534–541.
May, R.M., 1973. Stability and Complexity in Model Ecosystems.
Princeton University Press, Princeton, USA.
Puccia, C.J., Awerbuch, T., Levins, R., 1994. Models for new and
resurgent diseases. Ann. N. Y. Acad. Sci. 740, 225–231.
Rabinovich, J.E., Feliciangeli, M.D., 2004. Parameters of Leishma-
nia braziliensis transmission by indoor Lutzomyia ovallesi in
Venezuela. Am. J. Trop. Med. Hyg. 70, 373–382.
Roberts, M.G., Heesterbeek, J.A.P., 2003. A new method for esti-
mating the effort required to control an infectious disease. Proc.
R. Soc. Lond. B 270, 1359–1364.
Rodrı́guez, D.J., Torres-Sorando, L., 2001. Models of infectious dis-
eases in spatially heterogeneous environments. Bull. Math. Biol.
63, 547–571.
Salomón, O.D., Wilson, M.L., Munstermann, L.E., Travi, B.L.,
2004. Spatial and temporal patterns of phlebotomine sand flies
(Diptera: Psychodidae) in a cutaneous leishmaniasis focus in
northern Argentina. J. Med. Entomol. 41, 33–39.
Saravia, N.G., Weigle, K., Segura, I., Giannini, S.H., Pacheco, R.,
Labrada, L.A., Goncalves, A., 1990. Recurrent lesions in hu-
man Leishmania braziliensis infection-reactivation or reinfec-
tion. Lancet 336, 398–402.
Smith, G., 1994. Characteristics of host-parasite interactions that pro-
mote parasite persistence. Ann. N. Y. Acad. Sci. 740, 242–248.
Torres-Sorando, L., Rodrı́guez, D.J., 1997. Models of spatio-
temporal dynamics in malaria. Ecol. Model. 104, 231–240.
Wilson, M.L., 1994. Developing paradigms to anticipate emerging
diseases. Transmission cycles and a search for a pattern. Ann. N.
Y. Acad. Sci. 740, 418–422.
Woolhouse, M.E.J., Taylor, L.H., Haydon, D.T., 2001. Population
biology of multihost pathogens. Science 293, 1109–1112.