Embed Size (px)
Citation preview
INSTITUTO NACIONAL DE PESQUISAS DA AMAZÔNIA – INPA
PROGRAMA DE PÓS-GRADUAÇÃO EM ECOLOGIA
DINÂMICA DE OCORRÊNCIA DE Aedes aegypti e Aedes albopictus
EM RESIDÊNCIAS URBANAS DE MANAUS, BRASIL
SAMAEL DAVID PADILLA TORRES
Manaus, Amazonas
Maio, 2012
SAMAEL DAVID PADILLA TORRES
DINÂMICA DE OCORRÊNCIA DE Aedes aegypti e Aedes albopictus
EM RESIDÊNCIAS URBANAS DE MANAUS, BRASIL
ORIENTADOR: Dr. FERNANDO ABAD-FRANCH
CO-ORIENTADOR: Dr. Gonçalo Ferraz
Manaus, Amazonas
Maio, 2012
Dissertação apresentada ao Instituto Nacional de Pesquisas da Amazônia como parte dos requisitos para a obtenção do título de Mestre em Biologia (Ecologia)
II
BANCA EXAMINDORA DO TRABALHO ESCRITO:
Nome (instituição) Parecer
Steven A. Juliano (Illinois State University) Aprovado
Larissa Bailey (Colorado State University) Aprovado
Ricardo E. Gürtler (Universidad de Buenos Aires) Aprovado
BANCA EXAMINDORA DA DEFESA PÚBLICA DA DISSERTAÇÃO:
Nome (instituição) Parecer
Elizabeth Franklin Chilson (INPA) Aprovado
Paulo E. D. Bobrowiec (INPA) Aprovado
Ricardo A. dos Passos (FVS/AM) Aprovado
III
P123 Padilla Torres, Samael David Dinâmica de ocorrência de Aedes aegypti e Aedes albopictus em residências urbanas de Manaus, Brasil /Samael David Padilla Torres.--- Manaus : [s.n.], 2012. vii, 44 f. : il. color. Dissertação (mestrado) --- INPA, Manaus, 2012 Orientador : Fernando Abad-Franch Co-orientador : Gonçalo Ferraz Área de concentração : Ecologia
1. Ecologia de populações. 2. Controle biológico. 3. Vetor. 4. Vigilância entomológica. 5. Dengue – Manaus (AM). I. Título. CDD 19. ed. 595.7
Sinopse:
Modelou-se a dinâmica de ocorrência dos mosquitos vetores da dengue para quantificar os efeitos das intervenções oficiais de controle sobre estas espécies em um bairro da cidade de Manaus, Amazonas.
Palavras-chave: Ecologia, população, vetor, vigilância entomológica.
IV
AGRADECIMENTOS
Ao Instituto Nacional de Pesquisas da Amazônia por fornecer apoio para a realização da
minha dissertação.
Ao Programa de Pós-Graduação em Ecologia do INPA pela ajuda prestada ao longo do
mestrado.
Ao Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) pela concessão
da bolsa de mestrado através do seu programa de cooperação internacional PEC-PG, processo
190023/10-4.
Ao Instituto Leônidas e Marie Deane (ILMD – Fiocruz Amazônia) por fornecer apoio e
logística para a realização da minha dissertação através do Programa de Pesquisa em Ecologia
de Doenças Transmissíveis na Amazônia (PP-EDTA).
Aos moradores do bairro Tancredo Neves que sempre foram tão prestativos e pacientes ao
longo do monitoramento.
A Ricardo Mota, Diego Leite e Alexis Barbosa pela ajuda em campo e por compartilhar seus
conhecimentos práticos.
À Elvira Zamora-Perea pela imensa ajuda prestada na identificação e processamento do
material no laboratório e por facilitar a minha pesquisa com a sua paixão pelo trabalho bem
feito.
A Roberto Sena, Claudia Ríos-Velásquez, Felipe Pessoa, Sérgio Luz e Sylvain Desmoulière
pelas críticas construtivas e informação facilitada para a elaboração desta dissertação.
Ao Gonçalo Ferraz pelas suas críticas e sua ajuda com os modelos, que melhoraram
significativamente a qualidade da dissertação.
Ao Fernando Abad-Franch por aceitar-me como orientado, por compartir generosamente sua
experiência e conhecimentos, pelas estimulantes conversas sobre pesquisa após o expediente e
pela sua dedicação e esforço.
À Luciana pelo seu apoio incondicional, sua alegria, sua criatividade e carinho.
V
RESUMO
Aedes aegypti e Ae. albopictus são os vetores da dengue; ambos estão adaptados a ambientes antropizados e encontram-se distribuídos pela faixa tropical-subtropical, constituindo um grave problema de saúde pública. Em ausência de tratamentos ou vacinas, a prevenção da dengue baseia-se no controle do vetor. Isto requer um conhecimento detalhado da ecologia populacional destas duas espécies. Porém, os estudos realizados até o presente pressupõem que estes vetores podem ser detectados de forma perfeita; esta suposição é, muito provavelmente, errada. Neste trabalho foram quantificados, usando uma abordagem que considera explicitamente a detecção imperfeita, os efeitos das intervenções de controle executadas pelas agências locais de saúde pública sobre a dinâmica de ocorrência dos vetores da dengue. Os dados incluíram 38 meses de monitoramento entomológico, executado em parceria com o Instituto Leônidas e Marie Deane desde o ano 2008, no bairro Tancredo Neves da cidade de Manaus. Covariáveis meteorológicas e relativas às residências também foram consideradas. As estimativas da infestação domiciliar por Ae. aegypti (~0.9) superaram em uma ordem de grandeza o índice de infestação predial reportado pelo sistema de vigilância de rotina; o monitoramento com ovitrampas foi mais confiável (infestação domiciliar observada ~0.7). As probabilidades de ocorrência dos vetores flutuaram sazonalmente, principalmente devido aos efeitos negativos das altas temperaturas entre junho e setembro. As intervenções de controle coordenadas pela Prefeitura somente tiveram um pequeno efeito negativo, não distinguível de zero, sobre as taxas de infestação domiciliar por vetores de dengue. Os resultados mostraram que tanto a vigilância entomológica de rotina quanto os sistemas de controle de vetores da dengue devem ser melhorados. A gestão dos programas de controle vetorial precisa de alternativas mais efetivas, incluindo métodos que permitam uma avaliação fiável das taxas de infestação domiciliar. A abordagem utilizada neste trabalho, combinando ovitrampas com modelos que incorporam detecção imperfeita, poderia contribuir para o desenvolvimento de tais alternativas.
VI
Dynamics of site-occupancy by Aedes aegypti and Aedes albopictus in urban residences of Manaus, Brazil
ABSTRACT
Aedes aegypti and Ae. albopictus are the vectors of dengue, the most important arboviral disease of humans. Dengue prevention heavily relies upon vector control, and this requires detailed knowledge of the population ecology of these species. Yet, all reports on Aedes ecology published to date assume that the vectors are truly absent from sites where they are not detected; since no perfect detection method exists, this assumption is questionable. Imperfect detection may bias estimates of key vector surveillance/control parameters, including site-occupancy (infestation) rates and control intervention effects. Here, we used a modeling approach that explicitly accounts for imperfect detection to model the effects of regular (municipal) Aedes control interventions on site-occupancy dynamics. The data, collected in partnership with the Leônidas and Marie Deane Institute, encompass 38 months of vector monitoring at 55 sites in the Tancredo Neves neighborhood, Manaus. Meteorological and dwelling-level covariates were also considered. Ae. aegypti site-occupancy was estimated as ~0.9, one order of magnitude higher than the house infestation index reported by routine surveillance (based on ‘rapid larval surveys’) and moderately higher than ascertained with simple oviposition traps (~0.7). Regular vector control interventions, based on breeding-site destruction, had small negative effects, indistinguishable from zero, on the probabilities of dwelling infestation by dengue vectors. Site-occupancy fluctuated seasonally, mainly due to the negative effects of high temperatures in June-September. Rainfall and dwelling-level covariates were poor predictors of infestation. Our results show that regular dengue vector surveillance/control municipal systems perform surprisingly poorly. Both the results of ‘rapid larval surveys’ and the ineffectiveness of control campaigns suggest that Aedes breeding sites are often overlooked by vector control agents. Better alternatives are urgently needed, particularly for the reliable assessment of infestation rates in the context of control program management. The approach we present here, combining oviposition traps and site-occupancy models, could greatly contribute to the development and testing of such alternatives.
VII
SUMÁRIO
RESUMO V
ABSTRACT VI
INTRODUÇÃO 1
OBJETIVOS 3
Capítulo 1 – Artigo: Modeling dengue vector dynamics under imperfect
detection: Aedes aegypti and Aedes albopictus site-occupancy over three years
of monitoring in urban Amazonia.
4
CONCLUSÕES 33
REFERÊNCIAS BIBLIOGRÁFICAS 34
APÊNDICE 37
ANEXOS 39
1
INTRODUCÃO
A dengue é a doença viral transmitida por vetores mais comum em humanos (WHO,
2006; 2012; Guzmán et al., 2010). Aproximadamente 50 milhões de pessoas contraem dengue
cada ano, e 22000 infectados morrem por formas graves da doença (Guzmán e Istúriz, 2010;
WHO, 2012). O vírus da dengue é transmitido por mosquitos do gênero Aedes,
particularmente Aedes aegypti e Ae. albopictus (Gubler, 1998). Devido à ausência de
tratamento etiológico ou vacinas, a prevenção da dengue e de suas formas graves baseia-se
quase que exclusivamente sobre o controle das populações do vetor. Porém, apesar de um
grande investimento e alguns resultados alentadores, nem as populações dos vetores nem a
transmissão da dengue estão sob controle; de fato, ambas estão expandindo-se rapidamente no
mundo (Heintze et al., 2007; Kyle e Harris, 2008; Ballenger-Browning e Elder, 2009; Esu et
al., 2010; Gürtler et al., 2009; WHO, 2012). Dados da América do Sul mostram um
incremento de 4,6 vezes na incidência de casos de dengue reportada nos últimos 30 anos (San
Martín et al., 2010).
Aedes aegypti adaptou-se com sucesso a ambientes urbanos e prefere ovipôr em
criadouros artificiais (onde seus ovos desidratados podem permanecer viáveis por meses),
descansar dentro das casas e se alimentar de sangue humano (Reiter, 2007). Estas
características favoreceram a sua dispersão acidental por humanos nos trópicos (Gubler,
2002a; Gonçalves da Silva et al., 2012) e, junto com a sua capacidade para transmitir
efetivamente o vírus da dengue, transformaram o Ae. aegypti em uma ameaça para a saúde
pública (Gubler, 2002b). Aedes albopictus é mais eclético: encontra-se em habitats urbanos e
rurais do trópico e sub-trópico, pode ovipôr em criadouros artificiais ou naturais e alimenta-se
de sangue humano ou de outros vertebrados (Gratz, 2004). Apesar de ser o principal vetor no
sudeste asiático, Ae. albopictus é considerado um mosquito menos eficiente que Ae. aegypti
na transmissão do vírus (Lambrechts et al., 2010).
Manaus, capital do estado do Amazonas, foi re-infestada por Ae. aegypti no final da
década de 1990 (Figueiredo et al., 2004) e colonizada mais tarde por Ae. albopictus (Fé et al.,
2003); atualmente, ambas as espécies estão amplamente distribuídas (Rios-Velásquez et al.,
2007) e a transmissão da dengue é endêmica na cidade, com epidemias recorrentes e a
presença dos quatro sorotipos conhecidos do vírus (Figueiredo et al., 2008).
Como na maioria de áreas infestadas (Heintze et al., 2007; Ballenger-Browning e
Elder, 2009; Esu et al., 2010), o controle dos vetores da dengue em Manaus baseia-se em
visitas às residências por parte de agentes municipais ou estaduais, que eliminam criadouros
2
manualmente ou usando larvicidas; a borrifação ambiental de adulticidas é utilizada quando
são detectados surtos de dengue (FUNASA, 2001). Os agentes de controle vetorial também
realizam monitoramentos periódicos de infestação predial em uma amostra aleatória de
residências de cada bairro (Ministério da Saúde, 2005). Os resultados destes “Levantamentos
do Índice Rápido de Infestação por Aedes aegypti” (LIRAa) são utilizados para estabelecer
prioridades e tomar decisões sobre as intervenções de controle, enviando equipes de controle
vetorial a um bairro quando o índice de infestação ultrapassa 2%. Oficialmente, a diretiva do
Programa Nacional de Controle da Dengue é manter os índices de infestação abaixo de 1%
(FUNASA, 2002).
O desenho, a implementação e a avaliação destas estratégias de vigilância e controle
requerem, obviamente, um conhecimento detalhado da ecologia das populações locais dos
vetores; por sua vez, este conhecimento depende criticamente da avaliação das taxas de
infestação de residências. Em geral, as intervenções de controle devem ter um efeito negativo
sobre a ocorrência de Ae. aegypti e Ae. albopictus na escala local. A medida desse efeito
requer métodos confiáveis de detecção de infestações; contudo, e como para a maioria das
espécies animais (MacKenzie, 2005), a detecção de insetos vetores nunca é, provavelmente,
perfeita (Abad-Franch et al., 2010). Neste estudo, foi utilizada uma abordagem de modelagem
ecológica para analisar a dinâmica de ocorrência de Ae. aegypti e Ae. albopictus em
residências do bairro Tancredo Neves, Manaus. Levando em conta as falhas na detecção dos
vetores, foram testados os efeitos das intervenções oficiais de controle vetorial da Prefeitura e
de variáveis ambientais selecionadas sobre o principal indicador utilizado nos programas de
controle vetorial: os índices de infestação predial.
3
OBJETIVOS
Objetivo geral
Examinar, utilizando modelos que incorporam a detecção imperfeita de forma
explícita, a dinâmica de ocupação de residências por Aedes aegypti e Ae. albopictus em um
bairro de Manaus, Brasil.
Objetivos específicos
• Estimar os efeitos das intervenções rotineiras de controle vetorial sobre a
dinâmica de ocorrência de ambas as espécies.
• Estimar os efeitos de variáveis meteorológicas selecionadas (pluviosidade e
temperatura) sobre a dinâmica de ocorrência de ambas as espécies.
• Estimar os efeitos de variáveis residenciais selecionadas (estado de casas e
áreas peridomésticas) sobre a dinâmica de ocorrência das espécies.
• Derivar recomendações para a melhora dos sistemas de vigilância e controle
dos vetores da dengue.
Capítulo 1
Padilla-Torres, S.; Ferraz, G.; Luz, S.L.B.; Zamora-Perea,
E.; Abad-Franch, F. Modeling dengue vector dynamics
under imperfect detection: Aedes aegypti and Aedes
albopictus site-occupancy over three years of
monitoring in urban Amazonia. Manuscrito em
preparação para PLoS Neglected Tropical Diseases
7
Modeling dengue vector dynamics under imperfect detection: Aedes aegypti and Aedes 1
albopictus site-occupancy over three years of monitoring in urban Amazonia 2
3
Samael D Padilla-Torres1, Gonçalo Ferraz1,2, Sérgio LB Luz3, Elvira Zamora-Perea3, 4
Fernando Abad-Franch3* 5
6
1 Graduate Program in Ecology, Instituto Nacional de Pesquisas da Amazônia, Manaus, 7
Amazonas, Brazil 8
2 Biological Dynamics of Forest Fragments Project, Smithsonian Tropical Research Institute / 9
Instituto Nacional de Pesquisas da Amazônia, Manaus, Amazonas, Brazil 10
3 Instituto Leônidas e Maria Deane – Fiocruz Amazônia, Manaus, Amazonas, Brazil 11
12
*E-mail: [email protected] 13
14
Funding. PTSP-Dengue Program (Fiocruz-CNPq), Instituto Leônidas e Maria Deane, 15
Fiocruz-Fapeam agreement, Brazilian National Research Council (CNPq) 16
17
18
8
Abstract 19
Background: Aedes aegypti and Ae. albopictus are the vectors of dengue, the most important 20
arboviral disease of humans. Dengue prevention heavily relies upon vector control, and this 21
requires detailed knowledge of the population ecology of these species. To date, however, 22
Aedes ecology studies have assumed that the vectors are truly absent from sites where they are 23
not detected; since no perfect detection method exists, this assumption is questionable. 24
Imperfect detection may bias estimates of key vector surveillance/control parameters, 25
including site-occupancy (infestation) rates and control intervention effects. 26
Methodology/Principal Findings: We used a modeling approach that explicitly accounts for 27
imperfect detection and a 38-month, 55-site presence/absence dataset to measure the effects of 28
regular (municipal) Aedes control interventions on site-occupancy dynamics, considering also 29
meteorological and dwelling-level covariates. Ae. aegypti site-occupancy was estimated as 30
~0.9, one order of magnitude higher than reported by routine surveillance (based on ‘rapid 31
larval surveys’) and moderately higher than ascertained with simple oviposition traps (~0.7). 32
Regular vector control interventions, based on breeding-site elimination, had small negative 33
effects, indistinguishable from zero, on the probabilities of dwelling infestation by dengue 34
vectors. Site-occupancy fluctuated seasonally, mainly due to the negative effects of high 35
maximum (Ae. aegypti) and minimum (Ae. albopictus) summer temperatures (June-October). 36
Rainfall and dwelling-level covariates were poor predictors of infestation. 37
Conclusions/Significance: Our results show that regular dengue vector surveillance/control 38
systems perform surprisingly poorly. Both the results of ‘rapid larval surveys’ and the 39
ineffectiveness of control campaigns suggest that Aedes breeding sites are often overlooked 40
by vector control agents. Better alternatives are urgently needed, particularly for the reliable 41
assessment of infestation rates in the context of control program management. The approach 42
we present here, combining oviposition traps and site-occupancy models, could greatly 43
contribute to the development and testing of such alternatives. 44
45
46
47
48
49
50
51
52
9
Author summary 53
Dengue is a mosquito-transmitted viral disease that affects 50-100 million people annually. 54
Dengue prevention depends on the control of its two main vectors, Aedes aegypti and Ae. 55
albopictus. This requires identifying infested dwellings and eliminating mosquito breeding 56
sites, which is expected to reduce vector populations and, consequently, dwelling infestation 57
rates. We investigated the effects of regular control interventions on Ae. aegypti and Ae. 58
albopictus populations in a central Amazon city (Manaus, Brazil) over three years. Our 59
analyses take into account the fact that, since no vector surveillance system works perfectly, 60
mosquitoes may go undetected in an infested site. Dwelling infestation rates were about 25 61
times higher than reported by control agents, and decreased only slightly in the hottest months 62
of the year. Our analyses provide no evidence that vector control interventions reduced 63
dwelling infestation rates significantly. These results suggest that, in their current form, 64
surveillance-control systems grossly underestimate infestation rates and have no discernible 65
effects on dengue vector populations. Better alternatives are urgently needed to help contain 66
dengue epidemics, and we provide methodological guidance that can foster their 67
development. 68
69
70
10
Introduction 71
Dengue is the most common arboviral disease of humans [1–3]. About 50 million people 72
contract dengue annually, and an estimated 22,000 die from severe forms of the disease [3,4]. 73
Dengue virus is transmitted by mosquitoes of the genus Aedes, particularly Aedes aegypti and 74
Ae. albopictus [5]. In the absence of effective drugs or vaccines, prevention of dengue 75
infections and severe dengue forms heavily relies upon vector control. However, despite 76
massive spending and some encouraging results (e.g., [6–9]), neither vector populations nor, 77
consequently, dengue transmission are currently under control; on the contrary, they are both 78
overtly expanding [2,10]. In South America, dengue cases per 100,000 population increased 79
from ~16 in the 1980s to ~72 in 2000-2007 [11]. 80
Aedes aegypti, a species native to Africa, has successfully adapted to urban environments and 81
preferentially breeds in artificial containers (where desiccated eggs can remain viable for 82
months), rests within houses, and feeds on human blood [12,13]. These traits have favored its 83
man-mediated dispersal throughout the tropics [14,15], and, together with its capacity to 84
transmit dengue virus, have transformed Ae. aegypti in a major public health concern [16]. Ae. 85
albopictus is more eclectic: it exploits both urban and rural tropical-subtropical habitats, 86
makes use of natural and artificial breeding sites, and feeds on either humans or non-human 87
vertebrates [17,18]. Although it is the main dengue vector in South-East Asia and other 88
discrete locations, Ae. albopictus is overall less efficient than Ae. aegypti at transmitting the 89
virus [18]. 90
Dengue vector control is largely based on a combination of strategies aimed at eliminating 91
artificial breeding sites (either physically or by means of larvicides) and reducing adult 92
mosquito populations (through environmental insecticide application) [6–9]. The design, 93
implementation, and assessment of such strategies require detailed knowledge of vector 94
population ecology, including the estimation of dwelling infestation rates [19,20]. In general, 95
vector control interventions are expected to have a negative effect on infestation by Ae. 96
aegypti and Ae. albopictus at the local scale. Measuring such an effect requires reliable 97
methods for ascertaining infestation; yet, detection of most animal species, including disease 98
vectors, is rarely, if ever, perfect [21,22]. Here, we treat infestation as the probability that a 99
dwelling is occupied by vectors (i.e. site-occupancy) and use a robust modeling approach to 100
analyze the dynamics of site-occupancy by Ae. aegypti and Ae. albopictus. Our analysis is 101
based on three-years of oviposition trap data from a central-Amazon urban setting. Taking 102
imperfect detection into account, we quantify the effects of routine control interventions and 103
11
selected environmental variables on the main indicator used in vector control program 104
management – dwelling infestation rates. 105
Materials and methods 106
Study setting. With a population of about 1.8 million, Manaus (3°6’S, 60°1’W) is the largest 107
urban center of the Amazon basin (Fig. 1). The city lies on the north bank of the Negro river 108
and is surrounded by rainforest. The climate is tropical, warm and humid, with a relatively 109
strong seasonality of rainfall and, to a lesser extent, temperature (Fig. 2). After being declared 110
eradicated from Brazil in the 1950s [15], Ae. aegypti reinfested Manaus in the late 1990s [23] 111
and is currently widespread across all its neighborhoods [24]. Ae. albopictus was first 112
recorded in 2002 [25], and is now also widespread [24]. Dengue transmission is endemic in 113
the city, with recurrent epidemics and records of all known dengue virus serotypes [26]. As in 114
other settings, dengue control in Manaus relies on dwelling visits by municipal or state agents, 115
who physically eliminate breeding sites or treat them with larvicides; in “emergency” 116
situations (in practice, when dengue cases begin to soar), environmental insecticide spraying 117
aimed at reducing adult mosquito density is also used [27]. Vector control agents also conduct 118
regular infestation surveys on a random sample of dwellings in each neighborhood (see details 119
in ref. [28]). The results of these ‘rapid larval surveys’ are used to set priorities and make 120
decisions about control interventions, with control teams usually deployed to a neighborhood 121
when infestation rises above 2%; officially, the Brazilian control program aims to keep 122
infestation rates below 1% [20]. 123
Sampling strategy. We selected an area of ~250,000 m2 within the Manaus neighborhood of 124
Tancredo Neves for long-term monitoring (Fig. 1). This neighborhood is frequently infested 125
by both target mosquito species and has been reported as a common location of dengue cases 126
(refs. [24,29,30], and unpublished Municipal Health Department data). The typical Tancredo 127
Neves dwelling – our unit of occupancy analysis – consists of a brick-walled house with a 128
courtyard in a ~10x20m plot. In 2008, we randomly selected 50 dwellings for monthly 129
sampling, and in 2010 we added five more dwellings, which were also sampled on a monthly 130
basis. During the first 25 months, we used a combination of 2-4 ovitraps and 0-2 Adultraps® 131
[31]; afterwards, only the more sensitive ovitraps (3 per dwelling and month) were used. 132
Traps were baited with hay infusion [32] and operated for six days/month. In total, our 133
analyses make use of data from nearly 5800 trap-weeks. Each month, mosquito larvae were 134
identified to species, with the result of each individual trap recorded separately. Thus, for each 135
dwelling and month between September 2008 and October 2011, we have a ‘detection 136
12
history’ consisting of a series of binary results (present = 1 and absent = 0) for each trap and 137
mosquito species. 138
Covariates. In order to model the relation between environmental variables and infestation 139
we obtained daily data on total rainfall, as well as on maximum, mean, and minimum 140
temperature from the Brazilian National Meteorological Institute (INMET). We chose these 141
environmental metrics, or covariates, because we considered them potentially relevant for the 142
spatial-temporal distribution of our two target species [33–36]. Because we had no prior 143
information on possible time lags between meteorological changes and their effects on local 144
mosquito populations, we decided to relate meteorological information and each month’s 145
occupancy in three different ways: (i) looking at meteorological covariates measured, for each 146
month, during the six sampling days and the previous week (denoted 0-lag below); (ii) 147
looking at covariates measured during sampling days and the previous two weeks (0.5-lag); 148
and (iii) looking at covariates measured during the month before sampling (1-lag). All 149
meteorological measurements were standardized to mean zero and standard deviation one 150
before entering the analyses. 151
Apart from rainfall and temperature, we also registered dwelling-level traits throughout the 152
last 13 months of monitoring. Following criteria from Tun-lin et al. [36] adapted to our 153
setting, we separately assessed houses and courtyards; for each of these, we defined a 154
covariate with values of 1 (poor overall maintenance, garbage accumulation, and, for 155
courtyards, overgrown vegetation) or 0 (well-maintained houses and courtyards). Finally, we 156
noted whether routine control interventions were or were not performed in our study area in 157
each of the last 13 months of monitoring. These interventions were carried out by 158
municipal/state agents and military staff, and involved breeding-site elimination – physically 159
or with larvicides. We did not record any environmental insecticide spraying against adult 160
vectors during the period of our surveys. 161
Data analyses. Analyses are based on detection/non-detection data for the two target species. 162
Our analytical approach involved two main steps. First, we used descriptive statistics, tables, 163
and graphs to explore the data [37], and calculated naïve infestation rates (i.e., rates that 164
assume perfect detection of vectors) for later comparison with model-derived estimates (see 165
below). Second, we implemented a set of hierarchical models of occupancy dynamics. These 166
models explicitly account for imperfect detection, providing estimates of detection 167
probability, conditioned on occurrence (denoted p), and treat temporal changes in occupancy 168
(denoted ψ) as a first-order Markov process [38–40]. Models were fit by likelihood 169
maximization, and ranked according to the Akaike information criterion corrected for small 170
13
sample size (AICc) [41]. Model fitting and ranking were carried out with the freely-available 171
software PRESENCE 4.0 [42]. To avoid repetition, further details on model specification, 172
comparison, and selection are presented in the Results section, Tables, and Supporting Table 173
S1. 174
We fit occupancy dynamic models separately to (i) the 13-month subset of data for which we 175
recorded vector control interventions and house/courtyard covariates, and (ii) the full 38-176
month dataset. This resulted in a two-stage analysis. On the first stage, we focused on 177
modeling the effects of control interventions, both in the same month and one month later 178
(lagged effect), on site-occupancy probabilities (ψ). These models also consider 179
meteorological and dwelling conditions. Since two teams were involved in vector monitoring 180
during this period, we modeled detection probability (p) as a function of the observer team to 181
account for possible differences in performance [39,40]. 182
On the second stage, we set aside control interventions and focused on estimating time-183
dependent occupancy for the whole 38-month dataset, along with local (dwelling-level) 184
extinction probabilities and a probability of local recolonization. This second set of models 185
also considered the effects of meteorological covariates on occupancy, albeit with a larger 186
amount of data. Since we used two trapping devices during the first phase of monitoring, 187
detection probabilities were modeled as a function of trap type, and, once again, as a function 188
of the observer team. We also assessed the amount of bias present in naïve vs. model-derived 189
infestation rate estimates (bias = 1 − [naïve / model-derived values]). 190
Results 191
Descriptive results: observed infestation. Both vector species were detected in a high 192
proportion of dwellings throughout the study period (Fig. 2), with harmonic mean values of 193
0.68 for Ae. aegypti (range, 0.50-0.91) and 0.61 for Ae. albopictus (range, 0.28-0.86). There 194
was an apparent relationship with weather seasonal patterns. The particularly hot and dry 195
period of June-September 2009 coincided with a sharp decrease of Ae. albopictus infestation: 196
observed values fell from ~0.70-0.80 to ~0.30-0.50. A less marked decline was also apparent 197
for Ae. aegypti. Both species, however, quickly recovered with the onset of the rainy season. 198
Infestation index values reported by routine surveillance for our study neighborhood, based on 199
13 ‘rapid larval surveys’ [28] carried out between October 2008 and October 2011 (Figs. 3 200
and 4), yielded a harmonic mean of just 0.033 (range, 0.015-0.089). These results are broadly 201
suggestive of a relationship between dwelling infestation and weather, and of an absence of 202
any such relationship with control interventions. However, they rely on the assumption that 203
vectors were absent from those sites at which they were not observed; since no perfect vector-204
14
detection method is available, this assumption is questionable. The modeling results 205
summarized in the next section deal with this key limitation. 206
Modeling results: effects of control interventions. On the first stage of our analysis we 207
modeled the effects of vector control interventions carried out by local health authorities on 208
site-occupancy by Ae. aegypti and Ae. albopictus. These models use data from 55 dwellings 209
monitored from October 2010 to October 2011 with up to three ovitraps per dwelling and 210
month. Overall, the data encompass results from 1907 ovitraps, of which 849 detected Ae. 211
aegypti and 828 detected Ae. albopictus. 212
Aedes aegypti detection/non-detection data are best explained by a model with just one 213
covariate on ψ, the average of maximum daily temperatures measured with 0.5-lag (tmax-0.5-214
lag), which had a negative effect on site-occupancy (Table 1). The second-ranking model is 215
also substantially supported by the data (∆AICc = 0.73); it includes the additive effects of tmax-216
0.5-lag and control interventions carried out during the same month (cont0-lag) on ψ. The effect 217
of temperature was again negative; this model resulted in a negative point estimate of the 218
effect of control on site-occupancy, but uncertainty about this estimate is large and the 95% 219
confidence interval overlaps zero (Table 1). Among candidate models including dwelling 220
covariates, the one with the lowest AICc estimates a weak, positive effect of poor house 221
condition on infestation, but, again, the estimate of this effect is too uncertain to draw any 222
strong conclusions (Table 1). 223
The best-ranking model for Ae. albopictus estimates a negative effect of 0-lag minimum 224
temperatures (tmin-0-lag) on infestation; in addition, the model suggests that houses in poor 225
condition had higher infestation probability, albeit the estimated effect is uncertain and its 226
95% confidence interval includes zero (Table 1). Adding control interventions carried out the 227
month before (cont1-lag) resulted in a model that fits reasonably well (∆AICc < 1). For this 228
second model, the negative effect of cont1-lag on ψ is nevertheless small and imprecise, with 229
confidence intervals including zero (Table 1). Finally, a model with tmin-0-lag, house condition, 230
and cont0-lag as covariates also performed reasonably well; it estimates a small positive effect, 231
again not different from zero, of control interventions (Table 1). The model without any 232
covariates and those models exploring the effects of courtyard covariates all had ∆AICc ≥ 3 233
(see Supporting Table S1). 234
Modeling results: long-term site-occupancy dynamics. The results in the previous section 235
show that modeling time-specific occupancy as a function of control interventions or 236
dwelling-level covariates did not improve the fit of the models. Therefore, we felt justified to 237
15
extend modeling to the full dataset, including periods for which we had no information on 238
vector control activities – and hence without accounting for the effect of such activities. More 239
specifically, we fit one set of occupancy-dynamic models for each Aedes species in order to 240
examine the effects of meteorological covariates on site-occupancy. These models make use 241
of the full 38-month dataset including individual results of 5799 trap-weeks, which detected 242
Ae. aegypti on 2641 occasions and Ae. albopictus on 2538 occasions. Due to numerical 243
convergence problems, local recolonization probability (denoted γ) was constrained to be 244
constant across months, while monthly local (dwelling-level) extinction probabilities (ε), of 245
primary interest in the context of vector control, were derived from ψ and γ estimates as 246
described in MacKenzie et al. [39,40]. 247
Ae. aegypti data were best explained by a model including the 0.5-lag average of daily 248
maximum temperatures (tmax-0.5-lag), which had a negative effect on site-occupancy 249
probabilities (Table 2). The model with an effect of tmax-0-lag on ψ also fitted the data well and 250
estimated a similar effect to that of tmax-0.5-lag (Table 2). The remaining models that we 251
examined, including a null model without any covariates, performed substantially worse than 252
these two top-ranking models (see Supporting Table S1). Among models that included rainfall 253
covariates, the best-performing one had a ∆AICc = 6.29 and estimated a positive effect of 254
total rainfall (r0-lag) on ψ (Table 2). 255
Figure 3A shows monthly site-occupancy estimates for Ae. aegypti derived from the lowest-256
AICc model. With few exceptions, point estimates were consistently >90% (harmonic mean 257
0.91; range, 0.79-0.97), and showed a weak seasonal pattern apparently independent of 258
routine control interventions (arrows in Fig. 3A). Model-based infestation estimates are about 259
30% higher than observed values (median bias, 0.29) (Fig. 4). The estimated average 260
sensitivity of ovitraps at detecting infestation by Ae. aegypti varied from p = 0.48 (SE = 261
0.015) to p = 0.65 (SE = 0.01), depending on which field team performed monitoring (details 262
not shown). Local extinction probability estimates were overall very low (harmonic mean ε = 263
0.04; range, 0.01-0.18), reaching higher values in hotter months (Fig. 5A); mean site-264
recolonization probabilities were estimated as γ = 0.66 (SE = 0.06) over the study period. 265
The best-ranking Ae. albopictus model included only one site-occupancy covariate, tmin-0-lag, 266
which had a negative effect on ψ (Table 2). The remaining models performed substantially 267
worse (∆AICc > 20), but several of the candidate specifications we tested had convergence 268
problems. The only model with a rain covariate estimates a positive effect of 1-lag rainfall on 269
site-occupancy by Ae. albopictus (Table 2). Site-occupancy estimates derived from the best-270
16
ranking model are presented in Fig. 3B. As with Ae. aegypti, monthly ψ values were always 271
high (harmonic mean 0.83; range, 0.66-0.94), with minimum ψ = 0.66 (SE 0.03) in October 272
2011. Monthly Ae. albopictus ψ estimates were more unstable than those of Ae. aegypti, with 273
relatively strong fluctuations after the dry-hot summer of 2009 (Fig. 3B). Observed infestation 274
rates were also biased downwards (by ~26%) in our Ae. albopictus data (Fig. 4); again, 275
ovitraps were fairly sensitive at detecting Ae. albopictus (p = 0.63, SE = 0.01). Monthly local 276
extinction probabilities were low: harmonic mean ε = 0.07, range 0.02-0.32, with the 277
maximum value estimated for October 2011 (Fig. 5B). Mean dwelling recolonization 278
probability was estimated as γ = 0.59 (SE = 0.04). 279
Discussion 280
Reliable dwelling infestation estimates are critical for decision-making in the context of 281
dengue vector surveillance and control. The definition of programmatic goals, the 282
management of resources, and the assessment of interventions all rely heavily upon such 283
estimates. Using a large dataset and a modern analytical approach we have shown that routine 284
vector surveillance and control both can perform disturbingly poorly: at least in our study 285
setting, surveillance missed most instances of dwelling infestation and control had an overall 286
negligible effect on dwelling infestation rates. Our results suggest that combining ovitrap-287
based surveillance [e.g., 43–45] with analytical methods that account for imperfect detection 288
[e.g., 21,22,38–40] would help objectively assess, and likely enhance, dengue control 289
programs. We also suspect that, by applying this approach in other settings, many situations 290
similar to the one we describe here, with grossly underestimated dwelling infestation rates, 291
would be revealed. Given this negative bias in infestation rate estimates, which can reach one 292
order of magnitude for the ‘rapid larval surveys’ used in routine surveillance, our results even 293
suggest that the apparent effectiveness of some Aedes control campaigns might be just a 294
sampling artifact. 295
Before discussing our findings any further, we identify several study limitations to keep in 296
mind when interpreting the results. Importantly, we simplified our analyses in several ways. 297
Thus, we used detection/non-detection data, ignoring variations in vector abundance (but see, 298
e.g., ref. [46]), and measured only, and coarsely, a small number of covariates known to be 299
important for our target species; e.g. [33–36,47]. Our ‘control’ covariate included control 300
interventions in just three out of 13 months of assessment, and this clearly lowered the 301
precision of effect-size estimates: it seems possible that with more interventions we might be 302
able to detect a small but more significant effect. (i.e. one where the 95% confidence intervals 303
17
exclude zero). Yet, since ~70-90% of dwellings remained infested despite control 304
interventions, ‘statistical significance’ would in this case be of no practical importance. 305
Acknowledging these caveats, we feel nonetheless confident that our models adequately 306
estimate infestation rates as well as some of the major determinants of those rates in our study 307
area. The main difference between our approach and previous attempts to assess infestation 308
by dengue vectors is that we go beyond measuring indirect indexes of infestation to produce 309
statistical estimates of the probability that our study units (i.e. the dwellings) are occupied by 310
the target vector species. 311
We found little evidence that dwelling infestation rates decreased measurably as a result of 312
the vector control campaigns carried out by local health authorities in our study 313
neighborhood. These campaigns involved the elimination/treatment of thousands of artificial 314
breeding containers [29,30], and were therefore supposed to have larger effects on Ae. 315
aegypti, which unlike Ae. albopictus rarely breeds in natural water collections [5,17]. Our 316
results show, indeed, a larger effect of control interventions on Ae. aegypti than on Ae. 317
albopictus (Table 1); however, females of both species consistently continued to lay eggs, and 318
probably forage, in most of the dwellings we surveyed, irrespective of whether control 319
interventions had or had not taken place in the neighborhood. In addition, we found no 320
evidence suggesting that the weak effects of interventions persisted beyond a few weeks; Ae. 321
aegypti models assessing one-month-lagged control effects estimate a positive coefficient, but 322
the SE could not be computed. Our models suggest that this lack of effect could be related to 323
the fact that interventions are usually planned to coincide with the wet-cool season, which is 324
when local extinction probabilities drop to their lowest values (Fig. 5). Summer interventions 325
could perhaps be more effective [48], since they could work in synergy with the negative 326
effects of high temperatures on Ae. albopictus and Ae. aegypti detected by our models and in 327
previous studies (e.g., [47–50]). 328
One clear, practical implication of our findings is that Aedes breeding sites are often 329
overlooked by vector control agents, both during active surveillance and in control 330
campaigns. This suggests a key drawback to be addressed in the development of novel Aedes 331
control strategies, which should not heavily depend on the ability of control agents to detect 332
breeding sites while inspecting premises. Two major candidate strategies address this problem 333
from very different, but complementary, perspectives: (i) the use of adult mosquitoes to 334
transfer potent larvicidal particles from contaminated ‘dissemination stations’ to clean 335
breeding sites [51], and (ii) the release of mosquitoes carrying transgenes [52,53] or specific 336
18
Wolbachia strains [54] that impair reproduction and/or reduce competence to transmit dengue 337
virus. 338
Conclusions. Dengue vector surveillance is in urgent need of improvement [55], and our 339
results suggest two promising ways forward. First, simple hay infusion-baited ovitraps should 340
be preferred to ‘larval surveys’ for vector surveillance, particularly when a measure of 341
infestation is required (see also, e.g., refs. [43–45,56]); second, the repeated-sampling 342
approach we present considerably improves infestation rate estimates by explicitly taking 343
imperfect detection into account. Enhanced entomological surveillance systems and data 344
analyses that explicitly account for the detection process would, in turn, allow for reliably 345
assessing the effects of control interventions, irrespective of the specific tactics they employ. 346
Without such an assessment, the grounds on which massive public spending is directed 347
towards dengue vector control (e.g., [57]) remain questionable. 348
Finally, our results on dwelling infestation ascertainment bias also suggest that the findings of 349
most dengue vector ecology studies must be interpreted with caution. Even ovitraps, which 350
perform relatively well, yield naïve infestation rates that are consistently biased downwards. 351
The methods we used here incorporate this sampling-process uncertainty, and could therefore 352
substantially contribute to this field of inquiry. In dengue and in other areas of disease vector 353
research, the uncritical use of observed occurrence data as if they were obtained through 354
perfect-detection methods should perhaps be regarded, paraphrasing Breiman [58], as quite a 355
scandal. 356
357
Acknowledgments 358
We thank RM Mota, DLN Leite, and AC Barbosa for field assistance. RS Rocha and the 359
Manaus Municipal Health Department provided data on routine control-surveillance 360
activities. We particularly thank the residents who participated in the study – and patiently let 361
us keep running our long-term vector monitoring system. SA Juliano, LL Bailey, RE Gürtler, 362
and BW Nelson provided insightful comments on earlier versions of this manuscript. This 363
work is contribution number 17 of the Research Program on Infectious Disease Ecology in the 364
Amazon (RP-IDEA) of the Instituto Leônidas e Maria Deane – Fiocruz Amazônia. 365
366
Supporting Information 367
Table S1. The complete sets of site-occupancy dynamic models. 368
369
370
19
References 371
1. WHO (2006) Scientific Working Group Report on Dengue. Available: 372
http://www.who.int/tdr/publications/documents/swg_dengue_2.pdf. Accessed 2012 Feb 373
14. 374
2. WHO (2012) Global Alert and Response: Impact of Dengue. Available: 375
http://www.who.int/csr/disease/dengue/impact/en/. Accessed 2012 Feb 14. 376
3. Guzmán MG, Halstead SB, Artsob H, Buchy P, Farrar J, et al. (2010) Dengue: a 377
continuing global threat. Nat Rev Microbiol 8(12 Suppl.): S7–S16. 378
4. Guzmán A, Istúriz RE (2010) Update on the global spread of dengue. Int J Antimicrob 379
Agents 36(Suppl. 1): S40–S42. 380
5. Gubler DJ (1998) Dengue and dengue hemorrhagic fever. Clin Microbiol Rev 11(3): 381
480–496. 382
6. Heintze C, Garrido MV, Kroeger A (2007) What do community-based dengue control 383
programmes achieve? A systematic review of published evaluations. Trans R Soc Trop 384
Med Hyg 101(4): 317–325. 385
7. Ballenger-Browning KK, Elder JP (2009) Multi-modal Aedes aegypti mosquito reduction 386
interventions and dengue fever prevention. Trop Med Int Health 14(12): 1542–1551. 387
8. Gürtler RE, Garelli FM, Coto HD (2009) Effects of a five-year citywide intervention 388
program to control Aedes aegypti and prevent dengue outbreaks in northern Argentina. 389
PLoS Negl Trop Dis 3(4): e427. 390
9. Esu E, Lenhart A, Smith L, Horstick O (2010) Effectiveness of peridomestic space 391
spraying with insecticide on dengue transmission; systematic review. Trop Med Int 392
Health 15(5): 619–631. 393
10. Kyle JL, Harris E (2008) Global spread and persistence of dengue. Annu Rev Microbiol 394
62: 71–92. 395
11. San Martín JL, Brathwaite O, Zambrano B, Solórzano JO, Bouckenooghe A, et al. (2010) 396
The epidemiology of dengue in the Americas over the last three decades: a worrisome 397
reality. Am J Trop Med Hyg 82(1): 128–135. 398
12. Gubler DJ (1988) Dengue. In: Monath TP, editor. Epidemiology of Arthropod-Borne 399
Viral Diseases. Boca Raton: CRC Press, Inc. pp. 223–260. 400
13. Reiter P (2007) Oviposition, dispersal, and survival in Aedes aegypti: implications for the 401
efficacy of control strategies. Vector Borne Zoonotic Dis 7(2): 261–273. 402
14. Gubler DJ (2002) Epidemic dengue/dengue hemorrhagic fever as a public health, social 403
and economic problem in the 21st century. Trends Microbiol 10(2): 100–103. 404
20
15. Gonçalves da Silva A, Cunha ICL, Santos WS, Luz SLB, Ribolla PEM, et al. (2012) 405
Gene flow networks among American Aedes aegypti populations. Evol Appl In press. 406
doi:10.1111/j.1752-4571.2012.00244.x. 407
16. Gubler DJ (2002) The global emergence/resurgence of arboviral diseases as public health 408
problems. Arch Med Res 33(4): 330–342. 409
17. Gratz NG (2004) Critical review of the vector status of Aedes albopictus. Med Vet 410
Entomol 18(3): 215–227. 411
18. Lambrechts L, Scott TW, Gubler DJ (2010) Consequences of the expanding global 412
distribution of Aedes albopictus for dengue virus transmission. PLoS Negl Trop Dis 4(5): 413
e646. 414
19. Reiter P, Gubler DJ (1997) Surveillance and control of urban dengue vectors. In: Gubler 415
DJ, Kuno G, editors. Dengue and Dengue Hemorrhagic Fever. London: CAB 416
International. pp. 425–462. 417
20. FUNASA (2002) Programa Nacional de Controle da Dengue. Brasília: Ministério da 418
Saúde/Fundação Nacional de Saúde. 419
21. Mackenzie DI (2005) Was it there? Dealing with imperfect detection for species 420
presence/absence data. Aust NZ J Stat 47(1): 65–74. 421
22. Abad-Franch F, Ferraz G, Campos C, Palomeque FS, Grijalva MJ, et al. (2010) Modeling 422
disease vector occurrence when detection is imperfect: infestation of Amazonian palm 423
trees by triatomine bugs at three spatial scales. PLoS Negl Trop Dis 4(3): e620. 424
23. Figueiredo RMP, Thatcher BD, Lima ML, Almeida TC, Alecrim WD, et al. (2004) 425
Doenças exantemáticas e primeira epidemia de dengue ocorrida em Manaus, Amazonas, 426
no período de 1998-1999. Rev Soc Bras Med Trop 37(6): 476–479. 427
24. Ríos-Velásquez CM, Codeço CT, Honório NA, Sabroza PS, Moresco M, et al. (2007) 428
Distribution of dengue vectors in neighborhoods with different urbanization types of 429
Manaus, state of Amazonas, Brazil. Mem Inst Oswaldo Cruz 102(5): 617–623. 430
25. Fé NF, Barbosa MGV, Alecrim WD, Guerra MVF (2003) Registro da ocorrência de 431
Aedes albopictus em área urbana do município de Manaus, Amazonas. Rev Saude 432
Publica 37(5): 674–675. 433
26. Figueiredo RMP, Naveca FG, Souza MB, Melo M, Viana SS, et al. (2008) Dengue virus 434
type 4, Manaus, Brazil. Emerg Infect Dis 14(4): 667–669. 435
27. FUNASA (2001) Dengue: Instruções para Pessoal de Combate ao Vetor. Manual de 436
Normas Técnicas, 3a ed. Brasília: Ministério da Saúde/Fundação Nacional de Saúde. 437
21
28. Ministério da Saúde (2005) Diagnóstico Rápido nos Municípios para Vigilância 438
Entomológica do Aedes aegypti no Brasil – LIRAa. Metodologia para Avaliação dos 439
Índices Breteau e Predial. Brasília: Ministério da Saúde. 440
29. Fundação de Vigilância em Saúde do Estado do Amazonas (2008) Relatório Final: 441
Operação Impacto – Controle da Dengue. Manaus: Fundação de Vigilância em Saúde do 442
Estado do Amazonas. 443
30. Fundação de Vigilância em Saúde do Estado do Amazonas (2009) Relatório Final: 444
Operação Impacto II – Controle da Dengue. Manaus: Fundação de Vigilância em Saúde 445
do Estado do Amazonas. 446
31. Donatti JE, Gomes AC (2007) Adultrap: descrição de armadilha para adultos de Aedes 447
aegypti (Diptera, Culicidae). Rev Bras Entomol 51(2): 255–256. 448
32. Reiter P, Amador MA, Colon N (1991) Enhancement of the CDC ovitrap with hay 449
infusions for daily monitoring of Aedes aegypti populations. J Am Mosquito Contr Assoc 450
7(1): 52–55. 451
33. Scott TW, Morrison AC, Lorenz LH, Clark GG, Strickman D, et al. (2000) Longitudinal 452
studies of Aedes aegypti (Diptera: Culicidae) in Thailand and Puerto Rico: population 453
dynamics. J Med Entomol 37(1): 77–88. 454
34. Alto BW, Juliano SA (2001) Temperature effects on the dynamics of Aedes albopictus 455
(Diptera: Culicidae) populations in the laboratory. J Med Entomol 38(4): 548–556. 456
35. Favier C, Degallier N, Vilarinhos PDTR, de Carvalho MDSL, Yoshizawa MAC, et al. 457
(2006) Effects of climate and different management strategies on Aedes aegypti breeding 458
sites: a longitudinal survey in Brasília (DF, Brazil). Trop Med Int Health 11(7): 1104–459
1118. 460
36. Tun-Lin W, Kay BH, Barnes A (1995) The premise condition index: a tool for 461
streamlining surveys of Aedes aegypti. Am J Trop Med Hyg 53(6): 591–594. 462
37. Tukey JW (1977) Exploratory Data Analysis. Reading: Addison-Wesley. 463
38. MacKenzie DI, Nichols JD, Lachman GB, Droege S, Royle JA, et al. (2002) Estimating 464
site occupancy rates when detection probabilities are less than one. Ecology 83(8): 2248–465
2255. 466
39. MacKenzie DI, Nichols JD, Hines JE, Knutson MG, Franklin AB (2003) Estimating site 467
occupancy, colonization, and local extinction when a species is detected imperfectly. 468
Ecology 84(8): 2200–2207. 469
22
40. MacKenzie DI, Nichols JD, Royle JA, Pollock KH, Bailey LL, et al. (2006) Occupancy 470
Estimation and Modeling: Inferring Patterns and Dynamics of Species Occurrence. San 471
Diego: Elsevier Academic Press. 472
41. Burnham KP, Anderson DR (2001) Kullback-Leibler information as a basis for strong 473
inference in ecological studies. Wildlife Res 28(2): 111–119. 474
42. Hines JE (2006) PRESENCE 4.0 Software to estimate patch occupancy and related 475
parameters. USGS-PWRC. http://www.mbr-pwrc.usgs.gov/software/presence.shtml 476
43. Furlow BM, Young WW (1970) Larval surveys compared to ovitrap surveys for 477
detecting Aedes aegypti and Aedes triseriatus. Mosq News 30(3): 468–470. 478
44. Chadee DD (1986) A comparison of three Aedes aegypti sampling methods in Trinidad, 479
West-Indies. Cah ORSTOM Ser Ent Med Parasitol 24(3): 199–205. 480
45. Lourenço-de-Oliveira R, Lima JBP, Peres R, Alves FDC, Eiras AE, et al. (2008) 481
Comparison of different uses of adult traps and ovitraps for assessing dengue vector 482
infestation in endemic areas. J Am Mosquito Contr Assoc 24(3): 387–392. 483
46. Mogi M, Choochote W, Khamboonruang C, Suwanpanit P (1990) Applicability of 484
presence-absence and sequential sampling for ovitrap surveillance of Aedes (Diptera: 485
Culicidae) in Chiang Mai, Northern Thailand. J Med Entomol 27(4): 509–514. 486
47. Barrera R, Amador M, Mackay A J (2011) Population dynamics of Aedes aegypti and 487
dengue as influenced by weather and human behavior in San Juan, Puerto Rico. PLoS 488
Negl Trop Dis 5(12): e1378. 489
48. Chadee DD (2009) Impact of pre-seasonal focal treatment on population densities of the 490
mosquito Aedes aegypti in Trinidad, West Indies: A preliminary study. Acta Trop 109(3): 491
236–240. 492
49. Juliano SA, O’Meara GF, Morrill JR, Cutwa MM (2002) Desiccation and thermal 493
tolerance of eggs and the coexistence of competing mosquitoes. Oecologia 130(3): 458–494
469. 495
50. Leisnham PT, Juliano SA (2009) Spatial and temporal patterns of coexistence between 496
competing Aedes mosquitoes in urban Florida. Oecologia 160(2): 343–352. 497
51. Devine GJ, Zamora Perea E, Killeen GF, Stancil JD, Clark SJ, et al. (2009) Using adult 498
mosquitoes to transfer insecticides to Aedes aegypti larval habitats. Proc Natl Acad Sci 499
USA 106(28): 11530–11534. 500
52. Wise de Valdez MR, Nimmo D, Betz J, Gong HF, James AA, et al. (2011) Genetic 501
elimination of dengue vector mosquitoes. Proc Natl Acad Sci USA 108(12): 4772–4775. 502
23
53. Harris AF, Nimmo D, McKemey AR, Kelly N, Scaife S, et al. (2011) Field performance 503
of engineered male mosquitoes. Nat Biotechnol 29(11): 1034–1037. 504
54. Hoffmann AA, Montgomery BL, Popovici J, Iturbe-Ormaetxe I, Johnson PH, et al. 505
(2011) Successful establishment of Wolbachia in Aedes populations to suppress dengue 506
transmission. Nature 476(7361): 454–457. 507
55. Morrison AC, Zielinski-Gutierrez E, Scott TW, Rosenberg R (2008) Defining challenges 508
and proposing solutions for control of the virus vector Aedes aegypti. PLoS Med 5(3): 509
e68. 510
56. Pilger D, Lenhart A, Manrique-Saide P, Siqueira JB, Rocha WT, et al. (2011) Is routine 511
dengue vector surveillance in central Brazil able to accurately monitor the Aedes aegypti 512
population? Results from a pupal productivity survey. Trop Med Int Health 16(9): 1143–513
1150. 514
57. Taliberti H, Zucchi P (2010) Custos diretos do programa de prevenção e controle da 515
dengue no Município de São Paulo em 2005. Rev Panam Salud Publica 27(3): 175–180. 516
58. Breiman L (1992) The little bootstrap and other methods for dimensionality selection in 517
regression: X-fixed prediction error. J Am Stat Assoc 87(479): 738–754. 518
519
520
24
Figure legends 521
522
Figure 1. Study area. Manaus, state of Amazonas, Brazil (A) and Tancredo Neves 523
neighborhood (B) 524
525
Figure 2. Observed dwelling infestation and meteorological variables during the study 526
period. Dwelling infestation (%; left y axis) by Aedes aegypti and Ae. albopictus; total 527
monthly rainfall (mm; right y axis); and monthly averages of daily mean, minimum, and 528
maximum temperatures (°C; left y axis) 529
530
Figure 3. Observed and model-estimated dwelling infestation by Aedes aegypti (A) and 531
Ae. albopictus (B). Monthly model-derived site-occupancy estimates (solid circles, with 95% 532
confidence intervals); monthly observed infestation (empty circles); and Ae. aegypti 533
infestation indices derived from 13 ‘rapid larval surveys’ [28] (red circles in panel A). On the 534
x axis, grey boxes highlight the periods in which city-wide, massive Aedes control campaigns, 535
called Operação Impacto [29,30], took place. Arrows indicate months in which control 536
activities were performed in our study neighborhood (red arrows, interventions included as 537
model covariates) 538
539
Figure 4. Bias in Aedes aegypti (left) and Ae. albopictus (right) observed infestation. 540
Model-derived point estimates (“Model”) correspond to the top-ranking, 38-month dynamic 541
model for each species; observed dwelling infestation recorded during our surveys 542
(“Observed”) and infestation indices for Ae. aegypti reported by the regular vector 543
surveillance system, derived from ‘rapid larval surveys’ [28] (“RLS”). Monthly values (empty 544
circles) and quartiles 50% (horizontal line within box), 25%-75% (box lower-upper limits), 545
10%-90% (short lines), and 0%-100% (bottom-top lines) are shown 546
547
Figure 5. Derived estimates of local extinction probabilities (ε) for Aedes aegypti (A) and 548
Ae. albopictus (B). For each species, ε estimates (bold black lines) and 95% confidence 549
intervals (thin grey lines) were derived from the best-performing (lowest AICc) 38-month 550
model. We also plot variation (z-scores) of average maximum temperatures during sampling 551
days and the previous two weeks (0.5-lag, right y axes in each panel; colored areas); this was 552
the meteorological covariate in the best Aedes aegypti model. On the x axis, grey boxes 553
25
highlight the periods in which city-wide, massive Aedes control campaigns, called Operação 554
Impacto [29,30], took place; note that they coincide with months of very low ε values 555
556
557
558
559
560
561
562
563
564
565
566
567
568
26
Figure 1
27
Figure 2
28
Figure 3
29
Figure 4
30
Figure 5
31
Tables
Table 1. Effects of control interventions, meteorological variables, and dwelling traits on
infestation rates by dengue vectors: dynamic site-occupancy models fitted to a 13-month
dataset
Species/model ∆AICc Covariate β SE CI-lower CI-upper
Aedes aegypti, 13 months
ψ(tmax-0.5-lag),γ(.),p(obs) 0
tmax-0.5-lag –0.65 0.25 –1.14 –0.16
ψ(tmax-0.5-lag,cont),γ(.),p(obs) 0.73
tmax-0.5-lag –0.87 0.36 –1.58 –0.16
cont –0.81 0.62 –2.03 0.41
ψ(tmax-0.5-lag,hou),γ(.),p(obs) 2.41
tmax-0.5-lag –0.66 0.25 –1.16 –0.17
hou 0.22 0.66 –1.08 1.51
Aedes albopictus, 13 months
ψ(tmin-0-lag,hou),γ(.),p(.) 0
tmin-0-lag –0.26 0.12 –0.49 –0.03
hou 0.78 0.40 –0.0005 1.56
ψ(tmin-0-lag,cont1-lag,hou),γ(.),p(.) 0.87
tmin-0-lag –0.27 0.11 –0.49 –0.05
cont1-lag –0.27 0.21 –0.69 0.14
hou 0.79 0.40 –0.002 1.58
ψ(tmin-0-lag,cont,hou),γ(.),p(.) 2.51
tmin-0-lag –0.27 0.12 –0.50 –0.03
cont 0.028 0.19 –0.33 0.39
hou 0.78 0.40 –0.0002 1.56
“(.)” denotes that no covariates entered this part of the model; see text for further details. ∆AICc,
variation of Akaike information criterion (corrected for small sample size) values with respect to the
first-ranking model in each set; β, slope coefficient estimated for each covariate in the corresponding
model; SE, standard error; CI-lower and CI-upper, limits of the 95% confidence interval; tmax-0.5-lag,
standardized mean of maximum daily temperatures over sampling days and the 15 days prior to
sampling; tmin-0-lag, standardized mean of daily minimum temperatures during sampling days and the
previous week; hou, house condition covariate; cont, vector control covariate (same month); cont1-lag,
vector control covariate (previous month); obs, observer covariate; see main text for further details on
covariates
32
Table 2. Meteorological covariate effects on dwelling infestation rates by Aedes aegypti and
Ae. albopictus: dynamic site-occupancy models fitted to a 38-month dataset
Species/model ∆AICc Covariate β SE CI-lower CI-upper
Aedes aegypti, 38 months
ψ(tmax-0.5-lag),γ(.),p(trap,obs) 0
tmax-0.5-lag –0.63 0.14 –0.90 –0.35
ψ(tmax-0-lag),γ(.),p(trap,obs) 0.98
tmax-0-lag –0.57 0.12 –0.81 –0.33
ψ(r0-lag),γ(.),p(trap,obs) 6.29
r0-lag 0.50 0.14 0.23 0.77
Aedes albopictus, 38 months
ψ(tmin-0-lag),γ(.),p(trap) 0
tmin-0-lag –0.59 0.09 –0.77 –0.41
ψ(r1-lag),γ(.),p(trap) 21.4
r1-lag 0.46 0.09 0.28 0.64
“(.)” denotes that no covariates entered this part of the model; see text for further details . ∆AICc,
variation of Akaike information criterion (corrected for small sample size) values with respect to the
first-ranking model in each set; β, slope coefficient estimated for each covariate in the corresponding
model; SE, standard error; CI-lower and CI-upper, limits of the 95% confidence interval; tmax-0.5-lag,
standardized mean of maximum daily temperatures during sampling and the previous 15 days; tmax-0-lag,
standardized mean of maximum daily temperatures during sampling days and the previous week; r0-lag,
standardized mean of daily rainfall during sampling days and the previous week; tmin-0-lag,
standardized mean of daily minimum temperatures during sampling days and the previous week; r1-lag,
standardized mean of daily rainfall over the month before sampling; trap, trap-type covariate; obs,
observer covariate; see main text for further details on covariates
33
CONCLUSÕES
Os programas rotineiros de controle e vigilância entomológica em relação com a
dengue têm um desempenho surpreendentemente precário; na nossa área de estudo, o sistema
de vigilância baseado nos LIRAa subestimou de forma grosseira a proporção de residências
infestadas por Ae. aegypti, e as campanhas de controle tiveram um efeito praticamente nulo
sobre as taxas de infestação predial.
Os dados sugerem fortemente que a implementação de um sistema de vigilância
baseado em armadilhas de oviposição, simples e de baixo custo, poderia ajudar a melhorar a
detecção de Aedes spp. em residências.
O uso de métodos analíticos que incorporam a detecção imperfeita de forma explícita
permitiria avaliar de forma mais rigorosa (e, provavelmente melhorar) os programas de
controle dos vetores da dengue. Além de fornecer estimativas de infestação mais confiáveis,
os modelos evidenciaram que as campanhas de controle de vetores são realizadas durante o
período no qual as probabilidades de extinção local dos vetores são menores (a época
chuvosa). Se realizadas durante os meses mais quentes do ano, as campanhas de destruição de
criadouros potenciais poderiam ter efeitos sinérgicos com o aumento sazonal das
probabilidades de extinção local; esta possibilidade merece ser estudada em projetos de
pesquisa operacional.
Os resultados sugerem que, como ocorre com outros organismos, a detecção
imperfeita dos mosquitos vetores da dengue pode comprometer as conclusões das pesquisas
ecológicas. A incorporação explícita das incertezas do processo amostral é necessária para
fortalecer as nossas inferências sobre ecologia de vetores.
34
REFERÊNCIAS BIBLIOGRÁFICAS
Abad-Franch, F.; Ferraz, G.; Campos, C.; Palomeque, F.S.; Grijalva, M.; Aguilar, H.M.;
Miles, M.A. 2010. Modeling disease vector occurrence when detection is imperfect:
infestation of Amazonian palm trees by triatomine bugs at three spatial scales. PLoS
Neglected Tropical Diseases 4(3): e620.
Ballenger-Browning, K.K.; Elder, J.P. 2009. Multi-modal Aedes aegypti mosquito reduction
interventions and dengue fever prevention. Tropical Medicine and International Health, 14
(12): 1542–1551.
Esu, E.; Lenhart, A.; Smith, L.; Horstick, O. 2010. Effectiveness of peridomestic space
spraying with insecticide on dengue transmission; systematic review. Tropical Medicine and
International Health, 15 (5): 619–631.
Fé, N.F.; Barbosa, M.G.V.; Alecrim, W.D.; Guerra, M.V.F. 2003. Registro da ocorrência de
Aedes albopictus em área urbana do município de Manaus, Amazonas. Revista de Saúde
Pública, 37 (5): 674–675.
Figueiredo, R.M.P.; Naveca, F.G.; de Souza, M.B.; Melo, M.; Viana, S.S.; Gomes, M.P.;
Costa, C.A.; Farias, I.P. 2008. Dengue virus type 4, Manaus, Brazil. Emerging
InfectiousDiseases, 14 (4): 667–669.
Figueiredo, R.M.P.; Thatcher, B.D.; Lima, M.L.; Almeida, T.C.; Alecrim,W.D.; Guerra,
M.V.F. 2004. Doenças exantemáticas e primeira epidemia de dengue ocorrida em Manaus,
Amazonas, no período de 1998-1999. Revista da Sociedade Brasileira de Medicina Tropical,
37 (6): 476–479.
FUNASA. 2001. Dengue: Instruções para pessoal de combate ao vetor. Manual de Normas
Técnicas. 3rd ed. Ministério de Saúde/Fundação Nacional de Saúde. Brasília, Brasil. 84 p.
FUNASA. 2002. Programa Nacional de Controle da Dengue. Ministério da Saúde/Fundação
Nacional de Saúde. Brasília, Brasil. 32 p.
Gratz, N.G. 2004. Critical review of the vector status of Aedes albopictus. Medical and
Veterinary Entomology, 18 (3): 215–227.
Gonçalves da Silva, A.; Cunha, I.C.L.; Santos, W.S.; Luz, S.L.B.; Ribolla, P.E.M.; Abad-
Franch, F. 2012. Gene flow networks among American Aedes aegypti populations.
Evolutionary Applications, no prelo. doi:10.1111/j.1752-4571.2012.00244.x.
35
Gubler, D.J. 1988. Dengue. Em: Monath, T.P., (Ed). Epidemiology of arthropod-borne viral
diseases. CRC Press, Boca Raton, FL, USA. p. 223–260.
Gubler, D.J. 1998. Dengue and dengue hemorrhagic fever. Clinical Microbiology Reviews, 11
(3): 480–496.
Gubler, D.J. 2002a. Epidemic dengue/dengue hemorrhagic fever as a public health, social and
economic problem in the 21st century. Trends in Microbiology, 10 (2): 100–103.
Gubler, D.J. 2002b. The global emergence/resurgence of arboviral diseases as public health
problems. Archives of Medical Research, 33 (4): 330–342.
Gürtler, R.E.; Garelli, F.M.; Coto, H.D. 2009. Effects of a five-year citywide intervention
program to control Aedes aegypti and prevent dengue outbreaks in northern Argentina. PLoS
Neglected Tropical Diseases, 3 (4): e427.
Guzmán, A.; Istúriz, R.E. 2010. Update on the global spread of dengue. International Journal
of Antimicrobial Agents 36 (Suppl. 1): S40–S42.
Guzmán, M.G.; Halstead, S.B.; Artsob, H.; Buchy, P.; Farrar, J.; Gubler, D.J.; Hunsperger, E;
Kroeger, A; Margolis, H.S.; Martínez, E; Nathan, M.B.; Pelegrino, J.L.; Simmons, C;
Yoksan, S; Peeling, R.W. 2010. Dengue: a continuing global threat. Nature Reviews
Microbiology, 8 (12 Suppl.): S7–S16.
Heintze, C.; Garrido, M.V.; Kroeger, A. 2007. What do community-based dengue control
programmes achieve? A systematic review of published evaluations. Transactions of the
Royal Society of Tropical Medicine and Hygiene, 101 (4): 317–325.
Kyle, J.L.; Harris, E. 2008. Global spread and persistence of dengue. Annual Review of
Microbiology, 62: 71–92.
Lambrechts, L.; Scott, T.W.; Gubler, D.J. 2010. Consequences of the expanding global
distribution of Aedes albopictus for dengue virus transmission. PLoS Neglected Tropical
Diseases, 4 (5): e646.
Mackenzie, D.I. 2005. Was it there? Dealing with imperfect detection for species
presence/absence data. Australia & New Zealand Journal of Statistics, 47 (1): 65–74.
Ministério da Saúde. 2005. Diagnóstico rápido nos municípios para vigilância entomológica
do Aedes aegypti no Brasil – LIRAa: Metodología para avaliação dos índices Breteau e
predial. Ministério da Saúde. Brasília, Brasil. 60 p.
36
Reiter, P. 2007. Oviposition, dispersal, and survival in Aedes aegypti: implications for the
efficacy of control strategies. Vector Borne and Zoonotic Diseases, 7 (2): 261–273.
Ríos-Velásquez, C.M.; Codeço, C.T.; Honório, N.A.; Sabroza, P.S.; Moresco, M.; Cunha,
I.C.L.; Levino, A; Toledo, L.M.; Luz, S.L.B. 2007. Distribution of dengue vectors in
neighborhoods with different urbanization types of Manaus, state of Amazonas, Brazil.
Memórias do Instituto Oswaldo Cruz, 102 (5): 617–623.
San Martín, J.L.; Brathwaite, O.; Zambrano, B.; Solórzano, J.O.; Bouckenooghe, A.; Dayan,
G.H.; Guzmán, M.G. 2010. The epidemiology of dengue in the Americas over the last three
decades: a worrisome reality. American Journal of Tropical Medicine and Hygiene, 82 (1):
128–135.
WHO, 2006. Scientific Working Group Report on Dengue.
(http://www.who.int/tdr/publications/documents/swg_dengue_2.pdf) Acesso: 14/02/2012.
WHO, 2012. Global alert and response: Impact of dengue.
(http://www.who.int/csr/disease/dengue/impact/en/) Acesso: 14/02/2012.
37
APÊNDICE
38
APÊNDICE A – Tabela S1, todos os modelos de ocorrência gerados para ambas espécies
Aedes aegypti 38 months
Model Parameters AICc deltaAICc Likelihood AICc
weight ψ(tmax-0.5-lag),γ(.),p(trap,observer) 6 7013.56 0.00 1.00 0.457 ψ(tmax-0-lag),γ(.),p(trap,observer) 6 7014.54 0.98 0.61 0.280 ψ(tmean-0.5-lag),γ(.),p(trap,observer) 6 7016.11 2.55 0.28 0.128 ψ(tmean-0-lag),γ(.),p(trap,observer) 6 7016.62 3.06 0.22 0.099 ψ(rain-0-lag),γ(.),p(trap,observer) 6 7019.85 6.29 0.04 0.020 ψ(tmin-0.5-lag),γ(.),p(trap,observer) 6 7021.15 7.59 0.02 0.010 ψ(tmin-0-lag),γ(.),p(trap,observer) 6 7023.37 9.81 0.01 0.003 ψ(tmin-1-lag),γ(.),p(trap,observer) 6 7024.34 10.78 0.00 0.002 ψ(tmax-1-lag),γ(.),p(trap,observer) 6 7025.22 11.66 0.00 0.001 ψ(tmean-1-lag),γ(.),p(trap,observer) 6 7027.24 13.68 0.00 0.000 ψ(.),γ(.),p(trap,observer) (null) 5 7032.82 19.26 0.00 0.000
Aedes albopictus 38 months
Model Parameters AICc deltaAICc Likelihood AICc
weight ψ(tmin-0-lag),γ(.),p(trap) 5 6418.99 0.00 1.00 1.000 ψ(rain-1-lag),γ(.),p(trap) 5 6440.39 21.40 0.00 0.000 ψ(.),γ(.),p(trap) (null) 4 6466.26 47.27 0.00 0.000
Aedes aegypti 13 months
Model Parameters AICc deltaAICc Likelihood AICc
weight ψ(tmax-0.5-lag),γ(.),p(observer) 5 2593.32 0.00 1.00 0.402 ψ(tmax-0.5-lag,control),γ(.),p(observer) 6 2594.05 0.73 0.70 0.279 ψ(tmax-0.5-lag,house),γ(.),p(observer) 6 2595.73 2.41 0.30 0.121 ψ(tmax-0.5-lag,control,house),γ(.),p(observer) 7 2596.28 2.96 0.23 0.091 ψ(tmean-0-lag,control),γ(.),p(observer) 6 2597.64 4.32 0.12 0.046 ψ(.),γ(.),p(observer) (null) 4 2598.74 5.42 0.07 0.027 ψ(tmean-0-lag,control,house),γ(.),p(observer) 7 2600.24 6.92 0.03 0.013 ψ(yard),γ(.),p(observer) 5 2600.69 7.37 0.03 0.010 ψ(house),γ(.),p(observer) 5 2601.14 7.82 0.02 0.008 ψ(tmean-0-lag),γ(.),p(.) 4 2603.16 9.84 0.01 0.003
Aedes albopictus 13 months
Model Parameters AICc deltaAICc Likelihood AICc
weight ψ(tmin-0-lag,house),γ(.),p(.) 5 2500.23 0.00 1.00 0.339 ψ(tmin-0-lag,house,control-lag),γ(.),p(.) 6 2501.10 0.87 0.65 0.220 ψ(tmin-0-lag,house,control),γ(.),p(.) 6 2502.74 2.51 0.29 0.097 ψ(tmin-0-lag),γ(.),p(.) 4 2502.97 2.74 0.25 0.086 ψ(house),γ(.),p(.) 4 2503.23 3.00 0.22 0.076 ψ(tmin-0-lag,control-lag),γ(.),p(.) 5 2503.74 3.51 0.17 0.059 ψ(house,control-lag),γ(.),p(.) 5 2504.64 4.41 0.11 0.037 ψ(tmin-0-lag,control),γ(.),p(.) 5 2505.38 5.15 0.08 0.026 ψ(house,control),γ(.),p(.) 5 2505.59 5.36 0.07 0.023 ψ(.),γ(.),p(.) (null) 3 2505.87 5.64 0.06 0.020 ψ(control-lag),γ(.),p(.) 4 2507.19 6.96 0.03 0.010 ψ(control),γ(.),p(.) 4 2508.12 7.89 0.02 0.007
39
ANEXOS
40
ANEXO A – Parecer da banca examinadora da aula de qualificação
41
ANEXO B – Parecer da avaliadora do trabalho escrito, Larissa Bailey
42
ANEXO C – Parecer do avaliador do trabalho escrito, Steven Juliano
43
ANEXO D – Parecer do avaliador do trabalho escrito, Ricardo Gürtler
44
ANEXO E – Ata da defesa pública da dissertação
