Portal de Periódicos da CAPES

Patterns of house infestation dynamics by non-domiciliated Triatoma dimidiata reveal a spatial gradient of infestation in rural villages and potential insect manipulation by Trypanosoma cruzi

2009; Wiley; Volume: 15; Issue: 1 Linguagem: Inglês

10.1111/j.1365-3156.2009.02422.x

1365-3156

María Jesús Ramírez-Sierra, Melba Herrera-Aguilar, Sébastien Gourbière, Eric Dumonteil,

Insect symbiosis and bacterial influences

Objective Chagas disease is a major vector-borne parasitic disease in Latin America, primarily transmitted to humans by triatomine vectors. Non-domiciliated triatomine species such as Triatoma dimidiata in the Yucatan peninsula, Mexico, can transiently invade houses and are emerging as a major challenge to control Trypanosoma cruzi transmission to humans. We analyzed the spatio-temporal spreading of house infestation by T. dimidiata in four rural villages. Methods Triatomines were collected in four rural villages over a 2 years period, and the spatio-temporal patterns of infestation were analyzed. Results Triatomines were consistently more abundant at the periphery of villages than in centers, indicating a much higher risk of T. cruzi transmission at the periphery. Male T. dimidiata were found further in the center of the village, while females remained closer to the periphery, suggesting differential dispersal capabilities between sexes, although the timing of dispersal appeared identical. Surprisingly, infected females were consistently collected in houses much further from the surrounding bushes than non-infected females, while the distribution of males was unaffected by their T. cruzi infection status, suggesting an increased dispersal capability in infected females. Conclusion The spatial structure of infestation should be taken into account for the prioritization of vector control activities within villages, and spatially targeted interventions may be explored. A potential vector manipulation by T. cruzi, observed for the first time in triatomines, may favor parasite transmission to new hosts. Objectif: la maladie de Chagas est une maladie parasitaire majeure à transmission vectorielle en Amérique latine, principalement transmise aux humains par les vecteurs triatomes. Les espèces de triatomes non domiciliés telles que Triatoma dimidiataà Mexico, dans la péninsule du Yucatan, peuvent transitoirement envahir les maisons et émerger en tant qu'un défi majeur pour la lutte contre la transmission de Trypanosoma cruzi aux humains. Nous avons analysé la propagation spatio-temporelle de l'infestation des maisons par T. dimidiata dans quatre villages. Méthodes: Les triatomes ont été collectés dans quatre villages sur une période de deux ans et les profils spatio-temporels des infestations ont été analysés. Résultats: Les triatomes étaient toujours plus abondants à la périphérie des villages que dans le centre, indiquant un risque beaucoup plus élevé de transmission de T. cruzià la périphérie. Les mâles de T. dimidiata ont été plus retrouvés vers le centre des villages, tandis que les femelles restaient plus près de la périphérie, suggérant des différences de capacités de dispersion selon le sexe, même si les moments de la dispersion semblaient identiques. De façon surprenante, les femelles infectées étaient systématiquement collectées dans des maisons situées beaucoup plus loin dans les buissons environnants que les femelles non infectées, alors que la répartition des mâles n'était pas affectée par leur statut d'infection par T. cruzi, suggérant une capacité de dispersion plus élevée chez les femelles infectées. Conclusion: La structure spatiale de l'infestation devrait être prise en compte pour la hiérarchisation des activités de lutte antivectorielle dans les villages et des interventions spatialement ciblées pourraient être explorées. Une manipulation possible des vecteurs par T. cruzi, observée pour la première fois chez les triatomes, pourrait favoriser la transmission du parasite à de nouveaux hôtes. Objetivo: La enfermedad de Chagas es una enfermedad vectorial parasitaria importante en Latinoamérica, principalmente transmitida a los humanos por vectores triatominos. Las especies triatominas no domiciliadas, tales como Triatoma dimidiata en la Península del Yucatan, Méjico, pueden invadir las casas transitoriamente y están convirtiéndose en un reto importante para el control de la transmisión de Trypanosoma cruzi a humanos. Hemos analizado la dispersión espacio-temporal de la infestación de hogares por T. dimidiata en cuatro poblados rurales. Métodos: Se recolectaron triatominos en cuatro poblados rurales durante dos años, y se analizaron los patrones espacio-temporales de infestación. Resultados: Los triatominos fueron consistentemente más abundantes en la periferia de los poblados que en el centro de los mismos, indicando un mayor riesgo de transmisión de T. cruzi en la periferia. Se hallaron machos de T. dimidiata más cerca del centro del poblado, mientras que las hembras se mantenían en la periferia, sugiriendo una dispersión diferencial de capacidades entre sexos, aunque el tiempo de dispersión parecía ser idéntico. Sorprendentemente, las hembras infectadas fueron siempre recogidas dentro de casas mucho más alejadas de los arbustos que las hembras no infectadas, mientras que la distribución de los machos no se veía afectada por su estatus de infección con T. cruzi, sugiriendo una mayor capacidad de dispersión entre las hembras infectadas. Conclusión: La estructura espacial de infestación debería tenerse en cuenta para la priorización de las actividades de control vectorial dentro de los poblados, y deberían explorarse las intervenciones con énfasis espacial. Una posible manipulación vectorial de T. cruzi, observada por primera vez en triatominos, podría favorecer la transmisión del parásito a nuevos hospederos. Chagas disease is a major vector-borne parasitic disease in Latin America, with 9.8–11 million infected people and 60 million at risk of infection (Moncayo & Ortiz Yanine 2006; Schofield et al. 2006). It is caused by the protozoan parasite Trypanosoma cruzi, which is transmitted to humans primarily by triatomine vectors. After decades of intensive vector control programs throughout most of Latin America, house infestation by domiciliated triatomines has been dramatically reduced, but infestation by non-domiciliated triatomine species is emerging as a major challenge to further reduce T. cruzi transmission to humans. Indeed, conventional spraying control strategies are of limited efficacy against these triatomines (Dumonteil et al. 2004; Sanchez-Martin et al. 2006), even when insecticide formulation and application are optimized, because of the recurring infestation by new immigrating insects (Barbu et al. 2009). Alternative strategies, aimed at reducing triatomine entry into houses such as insect screens or impregnated curtains may be more effective (Herber & Kroeger 2003; Barbu et al. 2009). However, further optimization of vector control strategies requires detailed knowledge of house infestation dynamics and triatomine dispersal, which remain poorly understood. Extensive field collections of Triatoma dimidiata in both rural and urban areas of the Yucatan peninsula have shown a very clear and reproducible seasonal pattern of transient house infestation by predominantly adult triatomines during the months of April through July, associated with a very limited colonization of the domiciles (Dumonteil et al. 2002, 2009; Dumonteil & Gourbière 2004; Guzman-Tapia et al. 2005, 2007). Analysis of the population stage-structure (Dumonteil et al. 2002), population genetics studies (Dumonteil et al. 2007), and mathematical modeling (Gourbière et al. 2008) indicated that house infestation relies on a seasonal dispersal of adult triatomines from nearby peridomestic and/or sylvatic sites, although the relative contribution of these habitats is still unclear. Increased foraging for better feeding sources may contribute to the seasonal dispersal of T. dimidiata, while suboptimal feeding status and fecundity in the houses may contribute to the ineffective colonization (Payet et al. 2009). House infestation and re-infestation is often spatially clustered around highly infested or colonized sites, as observed for Triatoma infestans and Triatoma guayasana (Cecere et al. 2004, 2006; Vazquez-Prokopec et al. 2005). Clustering of houses infested by T. cruzi-positive T. infestans has also been observed in Peru (Levy et al. 2006). In the case of non-domiciliated T. dimidiata in urban Merida, no spatial clusters were observed and proximity of houses to the periphery of the city rather than housing quality was a major risk factor for infestation (Guzman-Tapia et al. 2007). Similarly, an initial theoretical exploration of house infestation dynamics by T. dimidiata using a spatially explicit mathematical model predicted that house infestation would start at the periphery of a village, before spreading in the entire locality as infestation progresses from the forest to the village center (Slimi et al. 2009). This spatio-temporal pattern suggested a higher risk for T. cruzi transmission in houses at the periphery of the villages, which should be confirmed in the field because of its potential implications for the prioritization of vector control interventions. Dispersal of infesting triatomines is also likely to influence spatial distribution and spreading of infestation. Direct information on triatomine dispersal has been difficult to obtain (Lehane & Schofield 1981; Schweigmann et al. 1988), but dispersal of several triatomine species is thought to be influenced by sex, reproductive and nutritional status of the bugs (Wisnivesky-Colli et al. 1993; Ceballos et al. 2005; Gurevitz et al. 2007). The dispersal of gravid females is considered particularly important for the colonization of new habitats such as a new house (Schofield et al. 1999; Carbajal de la Fuente et al. 2007). In the present study, we examine in detail the spatio-temporal dynamics of house infestation by non-domiciliated T. dimidiata in several rural villages of the Yucatan peninsula, to obtain microgeographic risk maps of T. cruzi transmission, and to test modeling predictions of a specific spatial structure of infestation. We also investigated potential differences in infestation patterns by male and female bugs, and the effect of T. cruzi infection on triatomine distribution, to obtain some indirect insights on the underlying dispersal mechanisms. Field work took place in four rural villages from the Yucatan state, Mexico. Villages are surrounded by a mixture of secondary bush vegetation and agricultural land called sylvatic or bush area. Bokoba has 1955 inhabitants in 494 houses, Sudzal has 1141 inhabitants in 297 houses, Sanahcat has 1509 inhabitants in 381 houses and Teya has 1966 inhabitants in 518 houses (INEGI 2005 population census). Houses are of various types, ranging from wooden sticks and adobe walls with thatched roof (about 25%), to cement blocks and concrete houses with cement or tiled floors (about 75%), and all are surrounded by large yards/peridomiciles consisting of a mixture of open space, bushes and trees. All houses were identified and georeferenced with a hand held global positioning system (GPS). Insects were collected by community participation, which is highly reliable for entomologic surveys (Dumonteil et al. 2002, 2009). Workshops and individual visits to households were organized in these villages, to provide information on Chagas disease and the vector. Households were then instructed to collect triatomines found inside their house in plastic vials/bags labeled with their name, address and date of capture, and deposit them at the local Health Center of their village, where all collected bugs were stored under the supervision of health personnel (Dumonteil et al. 2009). As noted before (Guzman-Tapia et al. 2005), this collection strategy relies on household's interest, which may vary, but there are no reasons for these variations and particularly underreporting of bugs to follow any spatio-temporal patterns. Bugs were gathered from the local Health Centers during regular visits to the villages, every 2-3 weeks, from August 2006 to September 2008. Triatomines were given unique ID codes, and further identified as male, female or the corresponding larval stages. T. cruzi infection was determined by PCR on a subset of bugs (n = 339) (Dumonteil et al. 2002). Feeding status was estimated using fresh weight (0.1 mg precision) of a subset of bugs (n = 55) brought to the laboratory within 2 days of their collection in the field (Ceballos et al. 2005; Payet et al. 2009). A subset of female bugs was also dissected (n = 156) and the number of eggs in their ovaries was counted as an indicator of reproductive status and/or of potential fecundity (Lopez et al. 1999; Payet et al. 2009). All laboratory data were imported into a georeferenced database for spatial analysis. A Google Earth satellite image was imported into a geographic information system (GIS) database in ArcView 3.2 (Environmental Systems Research Institute, Redlands, CA, USA), and georeferenced using GPS coordinates from field landmarks, to provide background maps of the villages. Digital sketch maps of the major structures of the villages (roads and constructions) were overlaid on the satellite image. Coordinates of all inhabited houses from the villages were imported in the database and matched to the corresponding structures on the satellite image. Maps of triatomine collections in the villages were produced for different time periods ranging from 1 month to 2 years to investigate house infestation dynamics. Theoretical village boundaries with the surrounding bush/sylvatic area were drawn based on the satellite images and field observations. Distances of all houses (infested and non-infested) and of triatomine collection sites to the village boundary and the surrounding bush area were calculated. The average distances between infested houses and village boundaries were compared with the expected distribution of the distances obtained from 100 000 Monte Carlo resampling of an identical number of houses within the villages, which simulated random distributions. The empirical P value of this permutation test was determined by the percentile corresponding to the observed average distance on the expected sampling distribution established by permutations (Diggle 2001; Perry et al. 2006). The same procedure was used to analyze the average distance between triatomine collection sites and the boundaries of the villages and bugs were randomly reassigned to houses. Comparison of male/female distance distribution as well as that of infected/non-infected bugs was performed by 100 000 random labeling permutations of bug positions (also referred to as marked point pattern analysis), and the determination of an empirical P value as above, with bug sex or infection status being randomly reassigned (Diggle 2001; Perry et al. 2006; Atkinson et al. 2007). Bonferroni correction was applied to account for the post-hoc analysis of subgroups (Bland 2000). In addition, we classified all houses into three distance categories (to ensure a comparable number of houses in each category): 0–80 m, 81–200 m and >200 m from the bush area outside the villages, to compare the proportion of infested houses and the abundance of triatomines per house in different areas of the villages. Comparisons of quantitative data (triatomine density, egg number, body weight) among groups were performed by Wilcoxon or Kruskal–Wallis tests, depending on the number of groups. Bias in the sex-ratio was analyzed by exact binomial test. Timing of temporal variations was assessed by Spearman rank correlation tests with different time lags. A total of 795 T. dimidiata were collected in the four villages over 2 years (Table 1). The majority of insects collected were adults (93.3%), with a clear female-biased sex-ratio (63.6% of females, P < 0.0001). To evaluate triatomine spatial distribution in the studied villages, we first mapped the cumulative distribution of all bugs collected in each village over 2 years. As observed in the village of Teya, infested houses and triatomines appeared more abundant at the periphery, while the village center was much less infested (Figure 1a). The spatial distribution of triatomines was similar in the other three studied villages as evidenced by the analysis of the distance of infested houses and triatomines from the surrounding bushes. Indeed, both infested houses and triatomines were not randomly distributed in the villages, but were significantly closer to the surrounding bush area in all the villages except for infested houses in Sanahcat (Figure 1b–e). Pooling our data from all the villages, houses were located on average at 140 m from the bushes, whereas infested houses were found at about 125 m (Permutation test, P < 0.001 with Bonferroni correction), and triatomines at about 110 m from it (Permutation test, P < 0.001 with Bonferroni correction). Accordingly, triatomine density in the peripheric zone corresponding to a maximum distance of 80 m from the bushes was more than double than that observed in the two zones further away from the bushes (0.77 vs. 0.31 and 0.33 bug/house, respectively). Similarly, the proportion of infested houses was much higher in the village zone closer to the bush area, reaching 20.7% of houses, compared to about 13.3% of houses in the other village zones (χ2 = 14.5, d.f. = 2, P < 0.001). Taken together, these data clearly pointed out the existence of an important infestation gradient in all four villages, with the houses located at the periphery of the villages at much greater risk of been infested, and with a much higher density of triatomines. Triatomine distribution patterns in Yucatan villages. (a) Map of the village of Teya. Circles indicate the cumulative triatomine collections over a 2 year period, with the size proportional to the number of bugs. Shaded gray areas indicate village zones defined by the distance from the surrounding bushes. Houses (small squares) and streets are also shown. (b–e) Distance from the surrounding bush area for all the village houses, infested houses and triatomines, for the villages of Bokoba, Sanahcat, Sudzal and Teya. Data are presented as means ± SEM. * and ** indicate a significant difference with the distribution of all the houses (Permutation tests, P < 0.05 and P < 0.001, respectively). We then examined the temporal variations of this particular spatial distribution, to assess whether this higher risk at the periphery would be observed only transiently or during most of the year. Monthly bug collections indicated important seasonal variations in bug presence in the houses. Importantly, the proportion of infested houses was consistently higher in the peripheric zone than in the more central zones of the villages during the 2 years of monitoring (Figure 2a). Similarly, monthly triatomine densities were higher in the zone closer to the surrounding bushes (<80 m) for most months (19/24 months, Figure 2b). These data indicate that the higher level of infestation at the periphery of the villages was maintained during the entire infestation period. Temporal variations in house infestation and triatomine abundance in different village zones. (a) Monthly variations in the proportion of infested houses and (b) Monthly variations in triatomine density, according to the indicated village zones defined from the surrounding bushes. Finally, further analysis indicated that while the spatial patterns of infestation in the villages is conserved from year to year, the majority of houses infested during the first infestation season were not infested during the following season [72.7% (117/161)]. Rather, infestation during the second season occurred mostly in previously uninfested houses, suggesting that the infestation pattern did not depend on the infestation of the previous year. Because dispersal capabilities are thought to be influenced by the sex of the bugs (Schofield et al. 1992; Gurevitz et al. 2006, 2007), we further investigated the infestation pattern of male and female T. dimidiata. Mapping of males and females in Teya indicated a greater female-bias at the periphery compared to the village center areas (χ2 = 6.78, d.f. = 2, P = 0.033; Figure 3a). Comparison of the average distance from the village boundary of males and females bugs from all four villages confirmed a distinct distribution according to sex as female T. dimidiata were found significantly closer to the surrounding bushes than males (permutation test, P = 0.040 with Bonferroni correction; Figure 3b). Interestingly larval stages of all stages were spread in several houses and presented a spatial distribution identical to that of female bugs, as evidenced by a similar average distance from the surrounding bushes (Figure 3b), indicating a higher risk of colonization for houses located closer to the periphery of the villages. On the other hand, temporal analysis did not reveal any differences in the timing of house infestation by male or female bugs (Figure 3c). Both sexes initiated house infestation simultaneously during the month of March, and ended it during the months of July–August (rs = 0.93 between male and female collections without any time lag, d.f. = 24, P < 0.001). Distribution pattern of male and female Triatoma dimidiata. (a) Distribution map of males and females in the village of Teya. Pie charts of the sex-ratio are shown at each location. Sex-ratio in each village zone (0–80 m, 81–200 m and >200 m) were significantly different (χ2 = 6.78, d.f. = 2, P = 0.033). (b) Average distance from the surrounding bushes of males, females and larval stages from the four villages. * indicates a significant difference with female bugs (Permutation test, P = 0.040) (c) Temporal variations in male and female T. dimidiata abundance in the four villages. Males and females followed a similar time course with no time lag (rs = 0.93, d.f. = 24, P < 0.001). Finally, because T. cruzi transmission is directly associated with infected bugs, we investigated the relative distribution of infected and non-infected triatomines. A total of 339 T. dimidiata were diagnosed for T. cruzi infection, giving an identical infection rate in males and females [36/141 (25.5%) vs. 63/187 (25.2%), respectively], but a lower infection rate in larval stages [1/11 (9.1%)]. Infected bugs were distributed throughout most of the different villages rather than clustered in a few hot spots, but detailed analysis indicated that in all villages T. cruzi-infected females were found much further from the surrounding bushes than non-infected females (122 ± 13 vs. 89 ± 6 m, permutation test, P = 0.036 with Bonferroni correction), while the distribution of males was unaffected by their T. cruzi infection status (permutation test, P = 0.56; Figure 4a). Conversely, infection rates in males were not significantly different between village zones (χ2 = 2.15, d.f. = 2, P = 0.34), while infection rates in females were over two fold higher in the village center compared to the two peripheric zones (44.7%vs. 18.5% and 21.6%, respectively, χ2 = 9.21, d.f. = 2, P = 0.010). Importantly, T. cruzi infected females also presented a significantly lower body weight compared to their non-infected counterparts (P = 0.028; Figure 4b). However, T. cruzi infection did not seem to affect their reproductive capacity as no differences were observed in egg content in non-infected and infected females (4.8 ± 0.8 vs. 5.7 ± 1.5 eggs per female, respectively, P = 0.55). Finally, temporal analysis indicated a similar kinetics of house infestation by infected and non-infected triatomines of both sexes (rs = 0.83 between infected and non-infected female collections without any time lag, d.f. = 12, P < 0.001; and rs = 0.57 between infected and non-infected male collections without any time lag, d.f. = 12, P = 0.017; Figure 4c,d). Distribution patterns of infected and non-infected triatomines. (a) Average distance from the surrounding bushes of male and female bugs from the four villages according to their infection status. * indicates a significant difference between infected and non-infected bugs (P = 0.036) (b) Average fresh body weight of male and female bugs according to their infection status. * indicates a significant difference between infected and non-infected bugs (P = 0.028). (c) Time course of infected and non-infected female bug collections. Infected and non-infected females followed a similar time course with no time lag (rs = 0.83, d.f. = 12, P < 0.001). (d) Time course of infected and non-infected male bug collections. Infected and non-infected males followed a similar time course with no time lag (rs = 0.57, d.f. = 12, P = 0.017). A detailed understanding of house infestation dynamics by non-domiciliated triatomines is required for the design and optimization of effective vector control programs. We presented here a first extensive field observation of the spatio-temporal dynamics of house infestation by T. dimidiata in several rural villages in the Yucatan peninsula, Mexico. The consistent observation of a spatial gradient of house infestation in four villages clearly indicated that houses located at the periphery and closer to the surrounding bush area are at a much higher risk of infestation by non-domiciliated T. dimidiata (at least two-fold higher) than houses in the village center. Confounding factors that may explain this spatial structure, such as variations in household participation and sampling bias, can be ruled out as such variations would be expected to be randomly distributed in space and time. Differences in housing quality can also be ruled out as construction type (stone/block/concrete vs. adobe/sticks/thatched roof) does not affect house infestation by non-domiciliated T. dimidiata in either rural (Dumonteil et al. 2002) or urban areas (Guzman-Tapia et al. 2007). The limited relevance of housing quality is further supported by our observations of infestation occuring in different houses at different times rather than being systematically clustered in a few poorly build houses. On the other hand, this spatial structure is in good agreement with a seasonal infestation by bugs dispersing from surrounding sylvatic areas and entering houses on their paths, as predicted by a previous mathematical model (Slimi et al. 2009). Although the relative contribution of bugs from sylvatic and peridomestic habitats is not clear, it is very likely that sylvatic bugs would infest peridomestic sites and houses at the periphery of the village more easily and frequently than those located in the center. Thus, our data do suggest that the spatial gradient of infestation we observed is due to the dispersal of sylvatic bugs surrounding the villages. The observation that this spatial gradient is maintained during the entire infestation season suggests a continuous influx of sylvatic bugs into the village during this period, which should thus be taken into account for vector control. Assuming that most infesting bugs collected originated from peripheric sites, the distances between collection sites and the village boundary measured in our study provide indirect information on the maximum range of dispersal distance of T. dimidiata, which would be at most 95–120 m. Although little is known on dispersal distances of triatomines, this seems in reasonable agreement with field tracking studies suggesting that T. infestans and T. sordida are able to fly over distances of 100–200 m (Schofield et al. 1991, 1992). However, some individuals may fly up to 1 km, although the difficulties of following triatomines over larger distances make such estimates uncertains (Schofield et al. 1991, 1992). The observation of a differential spatial distribution of male and female T. dimidiata in the villages suggests differential dispersal capabilities of these non-domiciliated bugs. Indeed, as mentioned above, collection bias by households or housing structure favoring one or the other sex can reasonably be ruled out. On the other hand, sex has often been associated with dispersal capabilities, although this relationship is not clear. For example, sex-bias in populations has often been interpreted as a sign of preferential dispersal of a specific sex (Monroy et al. 2003a,b; Payet et al. 2009). Direct measurement of flight initiation indicated that it is more frequent in females for T. infestans and T. melanosoma (Schofield et al. 1992; Galvao et al. 2001; Gurevitz et al. 2006), possibly because more females have fully developed flight muscles (Gurevitz et al. 2007). Taken together, our findings suggest that although more T. dimidiata females than males may disperse, they would do so over shorter distances. In addition, the dispersal of gravid females is thought to be particularly important for the colonization of new habitats (Schofield et al. 1999; Carbajal de la Fuente et al. 2007). In this respect, it is striking that larval stages followed an identical spatial distribution as females, i.e. mostly restricted to houses localized at the periphery of the villages. The colocalization of female and larval stages is consistent with the suggestion that female T. dimidiata may intend to colonize the houses, but that this colonization later fails (Dumonteil et al. 2002; Payet et al. 2009). Analysis of the spatial distribution of infected and non-infected triatomines further confirmed that T. cruzi transmission risk is greater in houses located at the periphery of the village, as expected from the general spatial gradient observed. In addition, it also revealed a striking effect of T. cruzi infection on T. dimidiata. Indeed, infected females were found to be distributed much further inside the village than non-infected ones, while distribution of males was unaffected by their infection status. As above, sampling bias can be ruled out as it would be distributed randomly between sexes, whether infected or not. Similarly, the distribution of infected bugs in many different houses in all four villages, the transient infestation of houses and the low infection rate in larval stages allow to rule out a bias due to T. cruzi transmission hot spots in the village due to infected families or domestic reservoirs of the parasite such as dogs. Thus, the most parsimonious explanation is that T. cruzi infection induced an increased dispersal of female T. dimidiata. Further field and laboratory studies on the effects of T. cruzi on female bug behavior should help understand this phenomenon. Parasite manipulation of vector hosts to facilitate their transmission is indeed a major paradigm in host-parasite relationship (Hurd 2003), and it has been observed for several insect vectors including sandflies and Leishmania (Rogers & Bates 2007), mosquitos and Plasmodium (Schwartz & Koella 2001), among others. On the other hand, T. cruzi is generally believed to have little effect, if any, on triatomines (Schaub 1989; Takano-Lee & Edman 2002). Earlier studies suggested that T. cruzi infection reduced Rhodnius prolixus feeding (D'Alessandro & Mandel 1969), and it was recently found that T. cruzi interfered with development time and longevity of Mepraia spinolai (Botto-Mahan 2009). In the case of T. dimidiata, infection has been found to have no effect on its development time, longevity and fecundity in laboratory colonies (Zeledón 1981), and our observations suggest potential manipulation of this species by T. cruzi for the first time. We may speculate that an increased dispersal of infected females would favor the dissemination of the parasite to new hosts, while it appears rather detrimental to the insect as infected females also presented a lower nutritional status, possibly due to the cost of the additional dispersal. Females may be more susceptible than males for being manipulated by T. cruzi, in agreement with observations on Mepraia spinolai (Botto-Mahan 2009). This may be due to sex-differences in resource allocation trade-off between parasite defense and other life-history traits such as dispersal and reproduction. In that respect, it is also interesting to note that infected and non-infected females had an identical reproductive status as determined by their egg content, in agreement with studies on laboratory colonies (Zeledón 1981). In conclusion, house infestation dynamics in rural villages of the Yucatan peninsula, Mexico, results in an important spatial gradient of infestation during most of the year, with houses located at the periphery and closer to the surrounding bush areas being at a much higher risk of infestation by non-domiciliated T. dimidiata than houses located in the village center. This spatial structure is likely due to a seasonal influx of sylvatic/peridomestic bugs from the periphery of the villages and it should be taken into account for the prioritization of vector control activities within villages. Also, it would be of interest to explore the possibility of spatially targeted interventions to reduce triatomine spreading in these villages. Finally, our results reveal a potential vector manipulation by T. cruzi, which significantly increases female T. dimidiata dispersal, possibly to favor parasite transmission to new hosts. This observation has major implications for the understanding of vector-parasite interactions in Chagas disease and warrants further studies. This study was funded by grant # 2005-1-13 790 from CONACYT-SALUD to ED. We thank the villagers for their interests and participation, as well as personnel from the local Health Centers for their assistance.