13693780310001656795

7/30/2019 13693780310001656795

1/7

Distribution of paracoccidioidomycosis: determination of

ecologic correlates through spatial analyses

LIGIA BARROZO SIMOES*, SILVIO ALENCAR MARQUES$ & EDUARDO BAGAGLI*

*Departamento de Microbiologia e Imunologia, Instituto de Biociencias, and$Departamento de Dermatologia e

Radioterapia, Instituto de Biociencias, Universidade Estadual Paulista/UNESP, Universidade Estadual Paulista/UNESP, Botucatu,

SP, Brazil

Paracoccidioidomycosis (PCM) is endemic in Latin America and in countries like

Brazil it carries a high mortality rate. The fungus habitat has not been precisely

determined. The present study aims to identify ecologic correlates based on PCM

distribution in a hyper-endemic area in southeastern Brazil. The Geographic

Information System (GIS) and spatial statistics were used to associate environ-

mental attributes, human population density and, PCM distribution. By means of

the Pearson r correlation coefficient, the highest statistically significant associa-

tions with prevalence density were the percent area (by county) of: basaltic rocks

(r0/0.63; PB/0.0001), Podzolic soils (r0/(/0.48; PB/0.001), Latosol soils(r0/0.40; PB/0.01), mean annual precipitation between 1500 and 1600 mm

(r0/0.46; PB/0.001) and, mean precipitation during the wet season between 940

and 1040 mm (r0/(/0.44; PB/0.01). Soil texture and precipitation analyzed

together reached r0/0.61 (PB/0.000002) for fine-textured soils with annual

precipitation above 1400 mm. Environmental correlates indicate that moisture

availability plays an important role in PCM distribution.

Keywords ecological factors, Geographic Information System, Paracoccidioides

brasiliensis, Paracoccidioidomycosis

Introduction

Paracoccidioidomycosis (PCM), caused by the fungus

Paracoccidioides brasiliensis, represents the most im-

portant systemic mycosis in Latin America [1]. PCM is

endemic from Mexico (238N) to Argentina (358S), with

the largest number of occurrences in Brazil, Venezuela

and Colombia [2]. Because PCM is not a notifiable

disease, it is not possible to calculate its real prevalence

and incidence [3]. It is the systemic mycosis with the

highest mortality rate among immunocompetent pa-

tients in Brazil [4]. Overall nationwide mortality per

area during the period 1980/1995 was 3.73/10 000 km

2

with non-homogeneous distribution throughout the

country. The mortality rate is higher than that of

leishmaniasis, and it is the eighth most common cause

of death among the chronic/recurrent infectious and

parasitic diseases in Brazil [4]. Furthermore, the disease

is now emerging in new areas. Recent environmental

and socioeconomic changes in the Brazilian Amazon,

for example, have been associated with new occurrences

of PCM-infection in Amerindian populations [5].

As the mucocutaneous lesions are relatively frequent,

during a certain time it has been claimed that people in

rural areas could have acquired PCM by the common

habit of placing vegetable material in the mouth andchewing it. Nevertheless, there is strong evidence

indicating that the principal route of infection is the

respiratory tract by inhalation of airborne propagules

[6,7], similar to other systemic mycoses caused by the

dimorphic pathogenic fungi Blastomyces dermatitidis,

Coccidioides immitis (C. posadasii) and Histoplasma

capsulatum. Although the infection of some of these

Correspondence: Eduardo Bagagli, Departamento de Microbiologia

e Imunologia, Instituto de Biociencias, Faculdade de Medicina,

Universidade Estadual Paulista/UNESP, Rubiao Jr, s/n 18618-000,

CP 510, Botucatu, SP, Brazil. Tel.: '/55 14 3811 6058; Fax: '/55 14

3815 3744; E-mail: [email protected]

Received 25 July 2003; Accepted 27 November 2003

2004 ISHAM DOI: 10.1080/13693780310001656795

Medical Mycology December 2004, 42, 517/523

7/30/2019 13693780310001656795

2/7

mycoses may be associated with extreme winds and

earthquakes, the prolonged periods of latency and the

lack of outbreaks of PCM do not allow correlating the

infection with sporadic climatic events.

Despite the continuous efforts by several research

groups, little is known about the habitat of P.

brasiliensis [3]. Some authors have suggested that the

fungus has its natural habitat in soil or in the vegetation

present in disease-endemic area [6,8]. However, the

fungus has been rarely isolated from soil and related

materials [9]. Frequent migration of the inhabitants

within the disease-endemic area, among other factors,

increases the difficulties in finding the fungus micro-

niche in nature [8].

Epidemiologic and ecologic studies on PCM have

shown that the disease is widely distributed throughout

an ecologically diverse continental area. Several

authors have pointed out some environment prefer-

ences where PCM is highly endemic, which are:

temperature between 17/248

C; mild winters; meanannual precipitation from 500 to 2500 mm; height

from 500 to 2000 m; proximity of water courses; clayey,

fertile and acid soils [2,6,10/14]. However, most of

previous works have been derived from descriptive

studies [10,12,14/17], others have been supported by

statistically oriented approaches, such as logistical

regression methods [18] and multivariate analyses [19]

or using armadillos as animal sentinels [20].

Due to the spatial character of this epidemiologic

and ecologic issue, fine scale studies are needed to help

to identify hot spots in the disease-endemic areas.

Remote sensing and Geographic Information Systems

(GIS) have been successfully applied to other epide-miologic problems, mainly to vector-borne diseases,

such as malaria [21,22], Rift Valley fever [23], Lyme

disease [24], African trypanosomiasis [25], and infec-

tious diseases such as coccidioidomycosis [26], among

others. Following this approach we present here a study

on ecologic correlates of PCM distribution in its

endemic region in Southeastern Brazil, through GIS

and spatial analyses. Although there are no control

measures that may be applied to prevent this disease

[27], ecologic correlates are important findings to

design epidemiologic risk maps. These maps could

then be used to increase awareness among clinicians,

tourists and the people living in rural areas where PCM

is endemic.

Material and methods

Human cases data and study area

Human PCM data corresponded to the cases seen in

the Dermatologic Sector of the University Hospital

(UH-UNESP), located in Botucatu County, Sao Paulo

State, from 1970 to 1999. Three surveillance measures

were calculated based on the 30 years of data: cases by

county of residence, cases by county per 10 000

inhabitants and prevalence density (cases by county

per area of the county). This approach favors the

spatial distribution of occurrences and the distribution

of the etiologic agent in the environment [4,28]. Average

population density (people/km2) by county for the

period was also calculated based on census data

produced by the Brazilian Institute of Geography and

Statistics and the Foundation for Data Analyses

System of Sao Paulo State. The study area was based

on patients residence (Fig. 1), which largely coincides

with the Botucatu hyper-endemic area of PCM, pre-

viously described [14].

Spatial data

Data were prepared taking into consideration the

boundaries of the 44 studied counties as indicated in

the 1:250 000 scale topographic maps of the region [29].

Elevation data were digitized at the same scale.

Geologic, geomorphologic, and soils data [30/32]

were digitized at a scale of 1:500 000. Carta Linx

software was used to digitize and edit the spatial

databases, which were processed and analyzed using

IDRISI32 GIS software [33].

From the digitized elevation data, it was possible to

interpolate height values and generate a Digital Eleva-

tion Model (DEM) for the study area. Triangulated

Irregular Networks (TIN) modeling was performed to

create the DEM [33].Daily rainfall statistics were obtained by using the

data from 68 rain gauges that was recorded from 1970

through 1999, georeferenced and submitted to geosta-

tistical analyses [34]. Precipitation data were interpo-

lated through ordinary kriging and three maps

subsequently derived: average annual precipitation,

mean precipitation in the wet season (October through

March) and mean precipitation in the dry season (April

through September). Temperature data were calculated

for the same 68 stations using a regression equation [35]

that related elevation, latitude and temperature for Sao

Paulo State.

Results

From 1970 through 1999, 377 confirmed human cases

of PCM were registered at the Department of Derma-

tology of UH-UNESP. Among those, 159 of the cases

came from 33 of the 44 (75%) counties of residence in

the study area. Over the 30 years of data, the

2004 ISHAM, Medical Mycology, 42, 517/523

518 Barrozo Simoes et al.

7/30/2019 13693780310001656795

3/7

distribution of PCM cases by county varied from

0 to 32. Mean incidence rate was 4.27 per 10 000

inhabitants, varying from 0 to 57.9 in each of the 44

counties in the study area. Mean prevalence density

was 8.26 cases/1000 km2, varying from 0 to 41.32 in

each of the 44 counties in the study area. Population

density ranged from 4.92 to 176.28 inhabitants/km2 in

each of the 44 counties in the study area.

Spatial statistics can measure the likelihood that anapparent pattern was produced merely by chance.

Spatial autocorrelation allows for hypotheses testing

about pattern values of a single variable at different

locations [36]. Morans I [37] and Gearys C [38] were

applied to measure spatial autocorrelation [39]. Mor-

ans I ranges from (/1 to '/1, and equals 0 when there

is no spatial autocorrelation (no effect of distance on

the distribution of a variable). Gearys C varies from

0 to 2 and when it equals 1, indicates random

distribution of values. Interpretation of Morans I

and Gearys C, reveal a strong positive autocorrelation

when I/0 and 0B/CB/1 and, a strong negative

autocorrelation when IB/0 and C/1. Standardized

normal variables, Z (I) and Z (C), were derived from

the Morans I and the Gearys C values for each

variable considered to evaluate statistical significance.For a two-tailed test, the critical values are 9/z0.0250/

9/1.96, at the 5% significance level.

Spatial autocorrelation analyses indicated significant

spatial clustering of counties according to population

density, cases by county of residence, cases per 10 000

inhabitants and prevalence density (Table 1).

Surveillance measures indicated spatial autocorrela-

Fig. 1 Prevalence density of PCM by county (cases/thousand km2) in the study area.


Distribution of paracoccidioidomycosis 519

7/30/2019 13693780310001656795

4/7

tion, which makes it appropriate to analyze environ-

mental patterns that appear to overlay the distribution

of PCM.

By means of Pearsons r correlation, prevalence

density by county was associated with the percent

area of each of the environmental variables by county

over the 44 counties.

Geologic and soil features, as well as precipitation

were found to be significant environmental correlates.

The results are shown in Table 2 and below.The Mesozoic basaltic rocks of the Serra Geral

Formation were the only geologic unit to show a

significant correlation with prevalence density. In fact,

this was the highest correlation found amongst any

environmental variable (r0/0.63; PB/0.0001).

Associations with soil types were calculated between

prevalence density and soil order and suborder, with

nine and 57 types, respectively. From the nine main

types of soils in the study area, four different types

showed statistical correlations. Because of their exten-

sions, two of these types are noteworthy. Calculating all

Latosol data (LV and LVA types) r0/0.40 (PB/0.003).

Podzolic soils together (PV and PVA types) did not

change the negative correlation (r0/(/0.48;

PB/0.0003).

Soil texture was also analyzed. The 57 soil types were

grouped according to their texture, based on percent

clay content, as follow: sand (B/15% clay), medium

(15/35% clay), clay (/35% clay). Sand and clay soil

textures correlated statistically with prevalence density

(r0/(/0.40, PB/0.003; r0/0.34, PB/0.02, respec-

tively). Adding medium and clay soil textures, thecorrelation increased to r0/0.40 (PB/0.003).

Statistical analyses of precipitation indexes showed

Table 1 Spatial autocorrelation for PCM distribution, measured by

Morans I and Gearys C coefficients and their derived standard

normal variables, Z(I) and Z(C)

Variable /I /Z (I) /C /Z (C)

Population density 0.53 7.25 0.46 7.01

Ca ses by county of residence 0 .69 9.3 6 0.3 7 8.0 4

Ca ses per 1 0 00 0 inhabitants 0 .48 6.7 3 0.5 1 5.9 1

Prevalence density 0.71 9.6 0.28 9.33

Table 2 Significant statistic correlates of PCM prevalence density by county with percent area of soil (order, suborder and texture),

precipitation, soil texture plus precipitation and geologic types by county (n0/44)

r P-value

Soil order

Red Latosol (LV) 0.40 B/0.003

Red-Yellow Podzolic (PVA) (/0.48 B/0.0003

Lithosol (RL) (/0.28 B/0.05

Red Nitosol (NV) 0.29 B/0.04

Percent area of each soil suborder by county

Red Latosol (LV1) 0.62 B/0.0000008Red Nitosol (NV1) 0.31 B/0.03

Red-Yellow Podzolic (PVA10) (/0.29 B/0.04

Red Nitosol (NV3) 0.35 B/0.02

Red-Yellow Latosol (LVA52) 0.34 B/0.02

Soil texture

Sand (B/15% clay) (/0.40 B/0.003

Clay (/35% clay) 0.34 B/0.02

Medium'/clay (/15% clay) 0.40 B/0.003

Precipitation

Mean annual precipitation (1300/1400 mm) (/0.40 B/0.01

Mean annual precipitation (1500/1600 mm) 0.46 B/0.001

Mean precipitation in the wet season (940/1040 mm) (/0.44 B/0.01

Mea n precipitation in the wet sea son (abov e 10 40 mm) 0 .40 B/0.004

Soil texture'/precipitationSand,B/1400 mm (/0.29 B/0.04

Sand,/1400 mm (/0.27 B/0.05

Medium,B/1500 mm (/0.28 B/0.05

Medium,/1500 mm 0.44 B/0.002

Clay,/1400 mm 0.61 B/0.000002

Medium'/clay,/1400 mm 0.53 B/0.00006

Geology

Mesozoic ba sa ltic rocks of the Serra Gera l Formation 0 .63 B/0.0001



7/30/2019 13693780310001656795

5/7

that mean annual precipitation above 1400 mm and, in

the wet season, a minimum mean precipitation of 1040

mm, increased prevalence density rates. In the study

area, median temperature (from 19.3 to 22.58C),

minimum temperature (from 13.1 to 15.98C) and

maximum temperature (from 25.4 to 29.18C) did not

reveal significant association with prevalence density,

nor did height (varying from 450 to 950 m above sea

level), present significant correlation.

Discussion

Our analyses depended mainly on patients with PCM

admitted to the Dermatologic Sector of the UH-

UNESP, which treats about half of the cases seen in

this center. We assume that the data from this center

were representative of all areas within the region for the

following reasons: (1) this is the oldest and the most

traditional center for PCM diagnosis in the westernportion of the state. The disease is not easily diagnosed,

requiring specialized clinicians and laboratory support

rarely found in other centers in the study area; (2) only

recently has the disease been diagnosed in other centers

and even in these cases, patients are sent to the UH-

UNESP for treatment; and (3) finally, the center offers

its services free of charge.

Considering the difficulties found when attempting

to identify areas where infection with P. brasiliensis

occurs, current surveillance measures based on cases by

county of residence is not sufficiently precise to identify

risk areas of infection. On the other hand, prevalencedensity offered the best spatial clustering, measured by

spatial autocorrelation, and was the unique surveillance

measure that had significant ecologic correlates. As

people are not equally exposed to risk due to the great

variability in rural populations over the counties, and

the infection is not contagious, this approach, standar-

dizing by county area, emphasizes the location of the

occurrences independently of the size of the popula-

tion. As spatial autocorrelation exists, PCM distribu-

tion does not present a random pattern in the study

area. This finding constitutes strong evidence in favor

of some social or ecologic correlates explaining the

presence ofP. brasiliensis in the area. The results of thisstudy show that there are important relationships

among human PCM distribution and population

density, basaltic rocks, Latosol soils and precipitation.

Prevalence density is higher in areas with a larger

percentage of Latosol soils. This can be explained by

Fig. 2 Soil texture, mean annual precipitation and prevalence density by county (cases/thousand km2).



7/30/2019 13693780310001656795

6/7

the fact that these soils, which in general present more

than 35% clay and poor capacity for drainage, retain

moisture. The four Latosol sub-order soils (LV1, NV1,

NV3 and LVA52), which have clay texture and derived

from the basaltic rocks of the Serra Geral Formation,

also correlated positively. High incidence of PCM was

also observed in clayey fertile soils derived from the

Serra Geral Formation cropping out in the Rio Grande

do Sul State, Southern Brazil [12].

Precipitation seems to play a role in the disease-

endemic regions, although its seasonality and geo-

graphic extent have yet not been properly studied.

Mean annual precipitation range previously defined

(500/2500 mm) is too wide to contribute to develop

risk models. Results indicate that precipitation distri-

bution alone cannot completely explain the distribution

of PCM in the study area. When soil texture and

precipitation were analyzed together, the correlation

increased to r0/0.61 (PB/0.000002) for fine-textured

soils (/35% clay) with mean annual precipitationabove 1400 mm. On the other hand, precipitation did

not increase the association with coarse-textured soils

(B/15% clay), which presented r0/(/0.29 (PB/0.04)

for precipitation below 1400 mm and r0/(/0.27

(PB/0.05) when precipitation was above 1400 mm.

Actually, PCM distribution seems to be related to soil-

water interactions, more specifically, to soil surface

storage of water. Riparian forests and vegetation near

watercourses have been strongly associated with in-

fected armadillos [20] and even with human infection

[19]. In the riparian zone, water table is high, favoring

greater soil water availability. The soil water content

varies with particle size, local drainage, and topo-graphic and climatic characteristics. Available soil

water is dependant on precipitation, evaporation rate

and soil characteristics. For the same climatic condi-

tions, fine-textured soils retain more water than coarse-

textured soils due to the greater superficial area by

mass unit of clay particles. Soil water availability in

clayey soils is 1.56 greater than in sand soils [40],

ranging from 75 mm/m of soil profile to 117 mm/m.

At tropical latitudes, annual temperatures are less

variable than in temperate areas and tend to exert a

lesser effect on the development and limits of tolerance

of a species than local hydrologic conditions [41].

Therefore, as temperature did not vary significantly in

the study area, soil water-holding capacity and

precipitation can partially explain PCM distribution

(Fig. 2). Obviously, evaporation and evapotranspira-

tion should also be considered to precisely define the

water budget. In this case, land use has to be carefully

analyzed. Through transpiration, plants have an active

role in soil water use thus strongly conditioning the

water balance [42]. Other conditions that reduce

vegetative cover and compact the soil surface, such as

the overgrazed pasture condition, cause infiltration to

diminish [43] and, consequently, to decrease soil water

availability.

Ecologic correlates of PCM distribution indicate that

moisture availability should be further investigated,

since climate correlates have been extremely useful to

develop models of suitable conditions to parasite

habitats [44].

Prevalence density is higher in fertile clayey soils

where intense agricultural activities generate great

exposure to aerosols. Due to the small size of the clay

particles, these soils are easily spread by the wind. On

the other hand, poor soils are likely to be destined for

grazing activities, being less aerated. It is still premature

to conclude either that the fungus prefers fine-textured

soils or that the high prevalence is associated with

intense agricultural practices. Unfortunately, we still do

not have a clear strategy to prevent future infections.An analyses including land use as a spatial variable

would contribute to the understanding of the relation-

ship among prevalence density, fungus autoecology and

environmental risk factors for PCM.

Acknowledgements

The State of Sao Paulo Research Foundation (FA-

PESP), grants 02/04489-0 (to LBS) and 02/00466-5 (to

EB) supported this research. We thank Celia R. L.

Zimback for her comments, Angela Restrepo for

critically reviewing the manuscript and the two anon-

ymous reviewers for their relevant suggestions.

References

1 Franco M, Lacaz CS, Restrepo-Moreno A, Del Negro G, (eds).

Paracoccidioidomycosis . Boca Raton, FL : C RC Press, 1994.

2 Restrepo A. The ecology ofParacoccidioides brasiliensis: a puzzle

still unsolved. J Med Vet Mycol 1985; 23: 324/334.

3 Restrepo A, McEwen JG, Castaneda E. The habitat of Para-

coccidioides brasiliensis : how far from solving the riddle? Med

Mycol 2001; 39: 233/241.

4 Coutinho ZF, Silva D, L azera M, et al. Paracoccidioidomycosis

mortality in Brazil (1980/1995). Cad Saude Publica, Rio de

Janeiro 200 2; 18: 1441/1454.

5 Coimbra CE Jr, Wanke B, Santos RV, do Valle AC, Costa RL,Zancope-Oliveira RM. Paracoccidiodin and histoplasmin sensi-

tivity in Tupi-Monde Amerindian populations from Brazilian

Amazonia. Ann Trop Med Parasitol 1994; 88: 197/207.

6 Montenegro M, Franco M. Pathology. In: Franco M, Lacaz CS,

Restrepo-Moreno A, Del Negro G (eds). Paracoccidiodomycosis .

Boca Raton, FL: CRC Press, 1994: 131/150.

7 Franco M, Mendes RP, Moscardi-Bacchi M, Rezkallah-Iwasso

M, Montenegro MR. Paracoccidioidomycosis. In: Hay RJ (ed.).

Baillieres Clinical Tropical Medicine and Communicable Diseases,



7/30/2019 13693780310001656795

7/7

Tropical Fungal Infections. London: Bailliere Tindall, 19 89: 1 85/

220.

8 Restrepo A. Ecology of Paracoccidioides brasiliensis. In: Franco

M, Lacaz CS, Restrepo-Moreno A, Del Negro G (eds). Para-

coccidiodomycosis. Boca Raton, FL: CRC Press, 1994: 121/130.

9 Franco M, Bagagli E, Scapolio S, Lacaz CS. A critical analysis of

isolation of Paracoccidioides brasiliensis from soil. Med Mycol

2000; 38: 185/191.

10 Chirife AV, Del-Rio CA. Geopatologia de la blastomicosisSudamericana. Prensa Med Argent 1965; 52: 54/59.

11 Negroni P. El Paracoccidioides brasiliensis vive saprofiticamente

en el suelo Argentino. Prensa Med Argent 1966; 53: 2381/2382.

12 Bopp C, Bernardi CDU. Geopatologia da blastomicose sul-

americana no Rio Grande do Sul. O Hospital 1967; 71: 113/130.

13 Borelli D. Some ecological aspects of paracoccidioidomycosis. In:

Paracoccidioidomycosis . Proceedings of the First Pan American

Symposium, Medellin, 1971. Pan Amer Hlth Org Scient Public

1971; 254: 59/64.

14 Marques SA, Franco MF, Mendes RP, et al. Aspectos epidemio-

log icos da paracoccidioidomicose na area endemica de Botucatu

(Sao Paulo/Brasil). Rev Inst Med trop S Paulo 1983; 25: 87/92.

15 Borelli D. Concepto de reservarea. La reducida reservarea de la

paracoccidioidomicosis. Ver Dermat Venez 1964; 4: 71/77.

16 Londero AT, Ramos CD, Lopes JO. Reservarea de paracocci-dioidomicose no Rio Grande do Sul, Brasil. Rev Inst Med trop S

Paulo 1972; 14: 377/380.

17 Restrepo A, Espinal LS. Algunas consideraciones ecologicas

sobre la paracoccidiodomicosis en Colombia. Antioquia Medica

1968; 18: 433/446.

18 Cadavid D, Restrepo A. Factors associated with Paracoccidioides

brasiliensis infection among permanent residents of three endemic

areas in Colombia. Epidemiol Infect 1993; 111: 121/133.

19 Calle D, Rosero DS, Orozco LC, Camargo D, Castaneda E,

Restrepo A. Paracoccidioidomycosis in Colombia: an ecological

study. Epidemiol Infect 2001; 126: 309/315.

20 Bagagli E, Franco M, Bosco SG, Barbosa FH, Trinca L,

Montenegro MR. High frequency of Paracoccidioides brasiliensis

infection in armadillos (Dasypus novemcintus): an ecological

study. Med Mycol 2003; 41: 217/223.

21 Wood B, Beck L, Washino R. Spectral and spatial characteristics

of rice field mosquito habitat. Int J Remote Sens 1991; 12: 621/

626.

22 Beck LR, Rodriguez MH, Dister SW, et al. Remote sensing as a

landscape epidemiologic tool to identify villages at high risk for

malaria transmission. Am J Trop Med Hyg 1994; 51: 271/280.

23 Linthicum KJ, Bailey CL, Davies FG, Tucker CJ. Detection of

Rift Valley fever viral activity in Kenya by satellite remote sensing

imagery. Science 1987; 235: 1656/1659.

24 Kitron U, Bouseman J, Jones C. Use of ARC/INFO GIS to study

the distribution of Lyme disease ticks in an Illinois county. Prev

Vet Med 1991; 11: 243/248.

25 Rogers D, Randolph S. Mortality rates and population of tsetse

flies correlated with satellite imagery. Nature 1991; 351: 739/741.

26 Centers for Disease Control and Prevention. Increase in

Coccidioidomycosis/Arizona, 1998/2001. MMWR Morb Mortal

Wkly Rep 2003; 52(RR-6): 109/112.

27 Blotta MH, Mamoni RL, Oliveira SJ, et al. Endemic regions of

paracoccidioidomycosis in Brazil: a clinical and epidemiologic

study of 584 cases in the southeast region. Am J Trop Med Hyg

1999; 61: 390/394.

28 Kawa H. Organizacao do espaco e producao da Leishmaniose

Tegumentar Americana no Municpio do Rio de Janeiro. [masters

dissertation]. Rio de Janeiro, RJ: Universidade do Estado do Rio

de Janeiro, Rio de Janeiro, 1998.

29 Instituto Geografico e Geologico do Estado de Sao Paulo. Folhas

Topograficas do Estado de Sao Paulo. Sao Paulo, 1954. Escala

1:250,000.

30 Instituto de Pesquisas Tecnologicas do Estado de Sao Paulo.

Mapa geologico do E stado de Sao Paulo. Sao Paulo: Publicacao

IPT, 198 1. Escala 1:500,000 (serie Monografias 5).

31 Instituto de Pesquisas Tecnologicas do Estado de Sao Paulo.

Mapa geomorfologico do Estado de Sao Paulo . Sao Paulo:

Publicacao IPT, 1981. Escala 1:50 0,000 (serie Monografias 5).

32 Oliveira JB, Camargo MN, Rossi M, Calderano Filho B. Mapa

pedologico do Estado de Sao Paulo. Campinas: IAC, EMBRAPA,

1999. Escala 1:500,000.

33 Eastman JR. IDRISI: a grid-based Geographic Analysis System,

Version 32. Worcester: Clark University Graduate School of

Geography, 1999.34 Isaaks EH, Srivastava RM. Applied Geostatistics. Oxford: Oxford

University Press, 1989.

35 Pedro MJ Jr, Mello MHA, Ortolani AA, Alfonsi RR, Sentelhas

PC. Estimativa das temperaturas medias mensais das max imas e

das mnimas para o Estado de Sao Paulo. Instituto Agronomico de

Campinas, (Bol Tec 142), 1991.

36 Hungerford LL. Use of spatial statistics to identify and test

significance in geographic disease patterns. Prev Vet Med 1991;

11: 237/242.

37 Moran P. Notes on continuous stochastic phenomenon. Biome-

trika 1950; 37: 17/23.

38 Geary R. The contiguity ratio and statistical mapping. Incorp

Stat 1954; 5: 115/145.

39 Sawada M. ROOKCASE: an Excell 97/2000 Visual Basic (VB)

Add-in for Exploring Global and Local Spatial Autocorrelation.Bull Ecol Soc Am 1999; 80: 231/234.

40 Lee R. Forest Hydrology. New York: Columbia University Press,

1980.

41 Bavia ME, Hale LF, Malone JB, Braud DH, Shane SM.

Geographic Information Systems and the environmental risk of

schistosomiasis in Bahia, Brazil. Am J Trop Med Hyg 1999; 60:

566/572.

42 Fernandez-Illescas CP, Porporato A, La io F, Rodriguez-Iturbe I.

The ecohydrological role of texture in a water-limited ecosystem.

Water Resour Res 2001; 37: 2863/2872.

43 Brooks KN, Ffolliott PF, Gregersen HM, DeBano LF. Hydrology

and the management of watersheds. Ames: Iowa State University

Press, 1997: 43/68.

44 Kitron U. Landscape ecology and epidemiology of vector-borne

disease: tools for spatial analysis. J Med Entomol 1998; 35: 435/

445.



Documents

13693780310001656795