22/5/06 15:23
NNFE_A_139849
(XML)
Studies on Neotropical Fauna and Environment, August 2006; 41(2): 117 – 122
ORIGINAL ARTICLE
An indirect estimation of the developmental time of Haemagogus janthinomys (Diptera: Culicidae), the main vector of yellow fever in South America
NICOLAS DEGALLIER1, HAMILTON A. DE OLIVEIRA MONTEIRO2, ´ RIO C. SA ´ . FILHO2, & FRANCISCO C. CASTRO2, ORLANDO V. DA SILVA2, GREGO ERIC ELGUERO3 1
Laboratoire d’Oce´anographie et du Climat, expe´rimentation et approches nume´riques (LOCEAN-UMR7159), Paris, France, Evandro Chagas Institute, Arbovirus Laboratory, Bele´m, Para´, Brazil, and 3Institut de Recherche pour le De´veloppement (IRD-UR24), Epide´miologie et Pre´vention, Montpellier, France
2
(Received 30 November 2004; accepted 6 October 2005)
Abstract Yellow fever is a re-emergent disease in many South American countries where its main vector is Haemagogus janthinomys. Epizootics and epidemics have long been associated with the onset of the rainy season, when mosquito densities are higher. Thus, a more precise understanding of the relationship between rainfall and mosquito densities is necessary to evaluate the risk of transmission. Mosquitoes were collected when landing on a volunteer, almost daily for three hours, from 18 June, 1986 to 16 December, 1987, at canopy level, in a rainforest station in Eastern Brazilian Amazonia. A total of 4079 mosquitoes from 25 different species were captured during 547 collecting sessions, and Hg. janthinomys accounted for 18.78% (766 individuals). Cross-correlations were looked for between mosquito densities and rainfall, and auto-correlation calculations were carried out with the mosquito collection series. The first analysis showed a significant correlation index until the 55th day, with two peaks at 11 – 13 and 20 – 24 days. The auto-correlation of the mosquito series showed a regularly decaying correlation index, which remained significant until day lag 34. The study showed indirectly that under a rainforest rainfall regime, the development of female Hg. janthinomys from egg to adult takes 11 – 13 days. Resumo A febre amarela e´ re-emergente nos paises da America do Sul, Haemagogus janthinomys sendo o seu vetor principal. Epizootias e epidemias sa˜o associadas ao inı´cio da estac¸a˜o chuvosa, quando densidades de mosquitos sa˜o mais altas. Portanto, avaliar o risco de transmissa˜o depende em parte de um melhor conhecimento da relac¸a˜o entre chuvas e densidades de vetores. Mosquitos foram coletados, quando atraidos num volunta´rio no nı´vel da copa, quase diariamente e durante tres horas, do 18 de junho de 1986 a 16 de dezembro de 1987, numa floresta de terra firme na Amazoˆnia oriental. Um total de 4079 mosquitos, sendo 25 espe´cies foram coletados durante 547 sesso˜es, Hg. janthinomys contando 18,78% (766 ex.). Correlac¸o˜es cruzadas entre densidades de mosquitos e chuvas diarias foram realizadas, assim como auto-correlac¸a˜o da se´rie do vetor. O ´ındice de correlac¸a˜o cruzada foi significativo ate´ o 558 dia, com dois ma´ximos a 11 – 13 e 20 – 24 dias. A autocorrelac¸a˜o mostrou um ´ındice regularmente decrescente ate´ o 348 dia de prazo. O estudo mostrou que, no regime pluviome´trico da floresta amazoˆnica, o desenvolvimento do ovo ate´ a feˆmea adulta dura 11 – 13 dias.
Keywords: South America, yellow fever vector, larval development, statistic analysis
Introduction The mosquito Haemagogus janthinomys has been known as a vector of sylvatic Yellow fever (YF; Flaviviridae: Flavivirus) for many decades (Trapido &
Galindo, 1956) and has been incriminated as the main vector during recent epizootics and epidemics in Brazil (Degallier et al., 1991, 1992a, 1992b; Vasconcelos et al., 1997). However, due to the impossibility of breeding it in the laboratory, very
Correspondence: N. Degallier, Laboratoire d’Oce´anographie et du Climat, Expe´rimentation et Approches Nume´riques (LOCEAN), T. 45-55, 4e e´t. case 100, 4 Place Jussieu 75252 Paris Cedex 05 France. Fax: +33 01 44 27 38 05. Email:
[email protected] ISSN 0165-0521 print/ISSN 1744-5140 online Ó 2006 Taylor & Francis DOI: 10.1080/01650520500398662
118
N. Degallier et al.
little knowledge is available about its ecology and/or behavior (Pajot al., 1985). Some information is available on its gonotrophic cycle (Herve´ et al., 1985; Alencar et al., 2002), daily biting and dispersal habits (Chadee et al., 1992) but its developmental time and dependence on rain remain mainly unstudied (Galindo et al., 1956). In nature, only the female of this species can be collected because its males are not attracted to the bait, and the breeding places are very difficult to find, preventing any direct estimate of its developmental time. Its eggs, laid inside tree holes, depend on rain water to hatch (Hovanitz, 1946). It was thus hypothesized that a relationship exists between daily rainfalls and adult relative densities. In the present study we will apply various time-series analysis techniques to look for some correlation between rainfall and adult density (Holmes & Birley, 1987; Morettin & Toloi, 1987). The epidemiological implications of the results in the determinism of epidemics and epizootics of Yellow fever will be discussed thereafter. Materials and methods Field work was carried out from 18, June, 1986 to 16, December, 1987 in a forested area 10 km from Bele´m, Para´ State, Eastern Brazilian Amazonia. Mosquitoes were collected daily for a period of three hours, as they were landing on two volunteers, each of whom was sitting on a platform at canopy level. Their protected status against yellow fever has been regularly verified by checking the presence of specific antibodies in their blood. Temperatures at canopy level were measured at the beginning of each three hours of collecting. Daily amounts of rain were obtained from the nearest
meteorological station. There is a marked seasonal variation in both mosquito density and rainfall. Since auto-correlations (Pearson’s coefficient r) may be computed only for stationary time series, the series were detrended by locally weighted regression (Cleveland & Devlin, 1988). Cross-correlation (Pearson’s r) was calculated between the daily relative densities of the mosquitoes and daily amounts of rain, with increasing lag times. Similarly, an auto-correlation function was applied to the mosquito data series, to estimate the decaying of the correlation index. The significance of the r coefficients was then determined by Student’s t test. Hacker et al. (1973) provided a useful summary of these time series analysis techniques for ‘‘less statistically oriented readers’’. Crosscorrelation and auto-correlation functions were plotted against increasing lag times. As a controlling process, the same functions were calculated after randomization of the mosquito data along timescale. All computations were done with the S-plus package (Statistical Sciences. S-plus Guide to Statistical and Mathematical Analysis, version 3.3, Seattle: StatSci, a division of MathSoft, Inc., 1995). Results From a total of 4079 mosquitoes (25 different species) which were captured during 547 collecting sessions, Hg. janthinomys accounted for 766 individuals (18.78%). The values of the cross-correlation index between daily mosquito densities and lagged daily rainfall are significant (2.275t 5 7.93; p 5 0.01) up to the 55th day, with two peaks at 11 – 13 and 20 – 24 days (p 5 0.001) (Figure 1). The autocorrelation analysis of the mosquito series showed a
Figure 1. Cross-correlation index of daily relative densities of Haemagogus janthinomys with respect to daily rainfall at previous lag times; Bele´m, Para´, Brazil. Dashed lines indicate the upper and lower limits of the confidence interval (p 5 0.01).
Development of Haemagogus janthinomys regularly decaying correlation index, which remained significant until day lag 35 (Figure 2). In the two cases, randomization of the timescale resulted in the vanishing of any correlation (Figure 3). The two peaks of the cross-correlation index may represent the time lag for the eggs, once submersed by rain water, to develop into the adults which were collected when landing on the host to pick up their bloodmeal. It is interesting to note that, although the physiological state of the female was not determined (i.e., it was not known which females were searching for their first meal after hatching, or any subsequent meal), a significant correlation occurred until the second generation of hatching. Thus we may confidently infer a 11 – 12-day time lapse between a hatch-inducing rainfall and the first bloodmeal of the female. A period of approximately two months (55th days time lag) is thus tentatively inferred for the development of all eggs available in the tree holes, and the implications of such a result will be discussed in the next section. The auto-correlation index may show a relationship between successive generations of female mosquitoes. However, only a regularly decreasing fate of the index was shown in this case which is an expected behavior of autocorrelation function, as size of population is dependent on size in recent past. Various biological traits may explain this result. Firstly the eggs are drought resistant and do not hatch all at the same submersion (Achmadi, 1989); secondly, the delay between bloodmeal and oviposition plus next bloodmeal (gonotrophic cycle) is much more variable than immature developmental time alone (Achmadi, 1989), introducing a further lack of correlation between egg-laying and adult densities of further generations. However, even within such restrictions, the auto-correlation index
119
was still significant (p 5 0.01) until the 34th day (Figure 2), which allows us to estimate the generation period to be at most 34 days long. Discussion and conclusion Due to the impossibility of breeding Hg. janthinomys in the laboratory and the difficulty of collecting it in its natural breeding pools, only indirect methods were available to determine its development time. One physiological trait which allowed for such an estimation to be done is the dependence of the eggs on rainfall for hatching. Thus cross-correlation statistical techniques gave mean estimates of the time lapse between egg hatching and the first, and probably the second, blood feeding of the females. However, variation of the cross-correlation index showed that the relationship between daily rainfall and the number of collected mosquitoes is not linear. Preliminary analysis showed that seasonally detrended data gave the same estimates as raw data. The auto-correlation analysis of the rainfall data, not reported here, has revealed no special structure, confirming the needlessness of detrending the series. Such a lack of seasonality is certainly a characteristic of the equatorial climate which occurs in this part of the lower Amazonian basin. The decrease in the auto-correlation index until it becomes insignificant may have various explanations. In the Haemagogus case, many eggs in each batch need more than one flooding to hatch (unpublished observations). As non-tree hole breeder mosquitoes also show the same pattern (Hacker et al., 1973), effects of dispersion and mortality may also contribute to the progressive fading out of such autocorrelation. Nonetheless, this study showed indirectly
Figure 2. Auto-correlation index of daily relative densities of Haemagogus janthinomys with respect to densities at previous lag times; Bele´m, Para´, Brazil.
120
N. Degallier et al.
Figure 3. Correlation analysis of daily relative densities of Haemagogus janthinomys with previously randomized time scale. A, Crosscorrelation index with respect to daily rainfall at previous lag times. B, Auto-correlation index with respect to densities at previous lag times; Bele´m, Para´, Brazil. Dashed lines indicate the upper and lower limits of the confidence interval (p 5 0.01).
Development of Haemagogus janthinomys that in primary forest, the development from egg to adult of female Hg. janthinomys took 11 – 13 days. Positive correlations between adult densities and rainfall until the 55th day lag suggest a higher risk of Yellow fever virus transmission during the first two months of the rainy season. This is actually the case for many epizootics of Yellow fever, the first cases of which began to appear 1 – 2 months after the beginning of heavy rains. In Trinidad, the highest density of Hg. janthinomys was observed two months after the beginning of heavy rainfall (Chadee et al., 1992). A higher transmission rate may thus be due mainly to higher densities of the vector during this part of the year. On the contrary, the same study of Chadee et al. (1992) showed that when rainfall decreases just before the drier season, the densities of mosquitoes decreases, but their parity rate increases, which also favors high transmission rates. Despite the standardization of the collecting method, time schedule and regularity of collecting sessions, it was not possible to avoid the occurrence of missing data in the mosquito series. This difficulty is unfortunately very common in chronological series of field data; however, to date no better solution to this problem has been devised. In order to deal with this problem in the present study, the missing points have been filled by linear interpolation between the observed values next to each gap. As all gaps (1 – 3 days) were much shorter than the estimated development time (11 – 13 days), we assumed that the effect of missing data was minimal. Many mosquito-transmitted arboviruses show weather-driven fluctuations because their vectors are dependent on the availability of water in the containers where the immature stages are developing (Reiter, 1988). As has been established for many species, the density of adult females is correlated with the amount of rainfall. The eggs of Hg. janthinomys and most of the Aedini species do not hatch immediately after being laid, and, if embryonated, they can survive one year or longer in a favorable but dry environment, waiting for rain or another mode of flooding (Consoli & Lourenc¸o-de-Oliveira, 1998). Such behavior has favored the quick dispersion of the domestic vectors Aedes aegypti and Ae. albopictus in all tropical countries through commercial exchanges of used tires, on the internal surface of which the mosquitoes lay their dry-resistant eggs (Reiter & Sprenger, 1987). Sylvatic YF vectors also depend on the rain for hatching their eggs, but data about the duration of their development in nature were lacking. The present study showed that it is possible to obtain an estimate of the mean duration of the development from egg to adult female of Hg. janthinomys, despite
121
some practical or methodological difficulties. Such data are important for evaluating the risk of YF epizootics following climatic changes or seasonal fluctuations, and may also help an understanding of the regional distribution of the disease. Acknowledgments To the Evandro Chagas Institute (Ministry of Health), the Conselho Nacional de Pesquisa (CNPq contract n8910042/97-7) and IRD (UR 034) who provided financial and material support. Charly Favier provided us with constructive comments on the manuscript. References Achmadi UF. 1989. Water resource development and its impact on biological vectors in Indonesia. In: Bunnag T, Sornmani S, editors. Proceedings of the Thirtieth SEAMEO-TROPMED Seminar on the Impact of Water Resources Development on the Health of the Communities and Preventive Measures for Adverse Effects. Surat Thani, Thailand, June 13 – 16, 1988. Bangkok, Thailand: SEAMEO-TROPMED Reference Centre, Mahidol University, pp 234 – 240. Alencar J, Degallier N, Guimara˜es AE, Noireau F, Pacheco R, Mello R, Lopes CM. 2002. Ana´lise morfome´trica preliminar entre populac¸o˜es de Haemagogus janthinomys (Diptera: Culicidae). In: Sociedade Entomolo´gica do Brasil – SEB, editor. 19e Congresso Brasileiro de Entomologia. Manaus, AM, Brasil, p 225. Chadee DD, Tikasingh ES, Ganesh R. 1992. Seasonality, biting cycle and parity of the yellow fever vector mosquito Haemagogus janthinomys in Trinidad. Med vet Ent 6:143 – 148. Cleveland WS, Delvin SJ. 1988. Locally weighted regression: an approach to regression analysis by local fitting. J Amer Stat Ass 83:596 – 610. Consoli RAGB, Lourenc¸o-de-Oliveira R. 1998. Principais mosquitos de importaˆncia sanita´ria no Brasil. Rio de Janeiro, RJ: Editora FIOCRUZ. Degallier N, Travassos da Rosa APA, Vasconcelos PF da C, Guerreiro SC, Travassos da Rosa JFS, Herve´ J-P. 1991. Estimation du taux de survie, de la densite´ relative et du taux d’infection d’une population d’Haemagogus janthinomys Dyar (Diptera, Culicidae) ayant fourni des souches de fie`vre jaune en Amazonie bre´silienne. Bull Soc Path ex 84:386 – 397. Degallier N, Travassos da Rosa APA, Herve´ J-P, Travassos da Rosa JFS, Vasconcelos PF da C, Mangabeira da Silva CJ, Barros VLRS de, Dias LB, Travassos da Rosa ES, Rodrigues SG. 1992a. A comparative study of yellow fever in Africa and South America. Cieˆncia e Cultura (Journal of the Brazilian Association for the Advancement of Science) 44:143 – 151. Degallier N, Travassos da Rosa APA, Vasconcelos PF da C, Travassos da Rosa ES, Rodrigues SG, Sa´ GC, Filho, Travassos da Rosa JFS. 1992b. New entomological and virological data on the vectors of sylvatic yellow fever in Brazil. Cieˆncia e Cultura (Journal of the Brazilian Association for the Advancement of Science) 44:136 – 142. Galindo P, Trapido H, Carpenter SJ, Blanton FS. 1956. The abundance cycles of arboreal mosquitoes during six years at a sylvan yellow fever locality in Panama. Ann Entomol Soc Am 49:543 – 547. Hacker CS, Scott DW, Thompson JR. 1973. Time series analysis of mosquito population data. J Med Ent 10:533 – 543.
122
N. Degallier et al.
Herve´ J-P, Sa´ GC, Filho, Travassos da Rosa APA, Degallier N. 1985. Bioe´cologie d’Haemagogus (Haemagogus) janthinomys Dyar au Bre´sil: e´tablissement du cycle gonotrophique au laboratoire et estimation du taux de survie. Cah ORSTOM se´r Ent me´d & Parasitol 23:203 – 208. Holmes PR, Birley MH. 1987. An improved method for survival rate analysis from time series of haematophagous dipteran populations. JanEcol 56:427 – 440. Hovanitz WR. 1946. Comparisons of mating behavior, growth rate, and factors influencing egg-hatching in south american Haemagogus mosquitoes. Physiol Zool 19:35 – 53. Morettin PA, Toloi CM. 1987. Se´ries temporais. Sao Paulo: Atual Editora Ltda. 136 p. Pajot F-X, Geoffroy B, Chippaux JP. 1985. Ecologie d’ Haemagogus janthinomys Dyar (Diptera, Culicidae) en Guyane Franc¸aise. Premie`res donne´es. Cah ORSTOM se´r Ent me´d & Parasitol 23:209 – 216.
Reiter P. 1988. Weather, vector biology, and arboviral recrudescence. In: Monath TP, editor. The arboviruses: epidemiology and ecology. I. Boca Raton, FL: CRC Press Inc. p. 245 – 255. Reiter P, Sprenger D. 1987. The used tire trade: a mechanism for the worldwide dispersal of container breeding mosquitoes. J Amer Mosq Cont Ass 3:494 – 501. Trapido H, Galindo P. 1956. The epidemiology of Yellow Fever in middle America. Exp Parasitol 5:285 – 323. Vasconcelos PF da C, Rodrigues SG, Degallier N, Moraes MAP, Travassos da Rosa JFS, Travassos da Rosa ES, Mondet B, Barros VLRS de, Travassos da Rosa APA. 1997. An epidemic of sylvatic yellow fever in the southeast region of Maranhao State, Brazil, 1993 – 1994: epidemiologic and entomologic findings. Am J Trop Med Hyg 57:132 – 137.
