Abstract
Insects of Triatominae subfamily are vectors of the parasite Trypanosoma cruzi, the etiological agent of Chagas disease affecting millions of people in Latin America. Some of these vector species, like Triatoma infestans, live in the human neighborhood, aggregating in walls or roof cracks during the day and going out to feed on animal or human blood at night. Except for their feeding specialization, these insects share this cycle of activities with many gregarious arthropod species. The understanding of how sex and T. cruzi infection affect their aggregation and geotaxis behavior is essential for understanding the spatial organization of the insects and the parasite dispersion. Experiments with non-infected and infected adults of T. infestans show that the insects presented a high negative geotaxis and aggregative behavior. Males had a higher negative geotaxis and a higher aggregation level than females. The aggregation level and the negative geotaxis were stronger in infected insects than in non-infected ones, the difference between sexes being maintained. The importance of these results is discussed in term of parasitic manipulation, dispersion of the vector and strategy of its monitoring.
Introduction
Chagas disease is one of the most important neglected tropical diseases with 6-7 million people who are estimated to be infected, and 20% of the population who are at risk1–3. This vector-borne disease is caused by the parasite Trypanosoma cruzi (Kinetoplastida: Trypanosomatidae) and is mainly transmitted by contact with infected feces/ urine of hematophagous insects of the Triatominae subfamily (Hemiptera: Reduviidae). Currently, 149 extant species have been described worldwide, and all of them are considered as potential vectors4. Most of them live in sylvatic habitats, and only a dozen of species are regarded as vectors of major epidemiological importance due to their capacity to live in the surrounding of the human dwellings where they find stable shelters and food abundance5. Triatoma infestans is the main vector in the Southern Cone of South America. Except for their feeding specialization, the domiciliary species share similar lifestyle and cycle of activities with many gregarious arthropods including other synanthropic species like cockroaches6–8. During the daytime, they assembled in dark and sheltered places such as cracks in the walls or roof, or behind objects hanging on walls. At night, they leave their shelter to actively seek a host upon which to feed and then, they come back to a resting place to digest. The digestive phase can last from some days to several weeks according to the blood meal size, the individual and the environmental conditions9.
The control strategy for Chagas disease relies mainly on the control of the domestic vectors through chemical control1. Faced with the increased of the insecticide resistance exhibited by these insects, and with the reinvasion of the dwellings by residual or sylvatic population of triatomines10–12, it is necessary to study the behaviors leading to a better understanding of their distribution and their dispersion. In this perspective, aggregation and geotaxis are key behaviors. Knowing them better and understanding how the parasite dispersion may influence them is fundamental. Indeed, aggregation is a widespread behavior that results from a response of individuals to environmental heterogeneity, and from social interactions involving attractions between individuals13–15. The social interactions maintain the group cohesion and the associated adaptive values of group living. In triatomines, protection against predation is usually evoked as the main benefice of clustering, but surviving might also be enhanced thanks to protection against hydric loss, and to a higher probability of coprophagy, symbiont exchange, and of sex encounters, as it was shown for other insects16–20. Aggregation in triatomines was investigated with a focus towards the substances that mediate it, and on the factors that modulate the aggregative response21–26. All these works analyzed nymphal instars behavior response; in adults very few is known except that they can aggregate around feces25. Geotaxis, also called gravitaxis, is a crucial behavior involved in insect orientation27. Animals can exhibit locomotion that is gravitationally directed vertically down or up (positive or negative geotaxis, respectively). Geotaxis in triatomine has been poorly described, T. infestans was just reported as being more concentrated in the upper half of the walls in houses or chicken houses17,28. Moreover, to our knowledge, no studies were conducted to analyze the synergy or conflict between gregariousness and geotaxis in triatomines.
It is well-known that parasites can modify physiological, behavioral, and/or morphological traits of their hosts to increase their fitness, even if it is at the cost of the host fitness29. The latter usually means that infected hosts will behave in ways that facilitate the transmission of the parasite30,31. Literature about the effects and possible manipulation of triatomines behavior by T. cruzi is relatively sparse, covering only seven species: Mepraia spinolai, Panstrongylus megistus, Rhodnius pallescens, Rhodnius prolixus, Triatoma brasiliensis, Triatoma dimidiata and T. infestans. Authors have been especially interested in the parasite’s effects on four groups of the host’s behavior: life-history trait, feeding, defecation, and dispersion/ locomotion. It seems that T. cruzi increases the development time and biting rate, and decreases the longevity and defecation time in M. spinolai32,33, but no change was observed in P. megistus34, R. prolixus35, T. dimidiata36, T. infestans37, and almost no change in T. brasiliensis38. The reproduction was decreased by T. cruzi in T. brasiliensis38. The dispersion was higher in infected females of T. dimidiata than in non-infected females; no effect was found in males39. Moreover, T. dimidiata individuals infected with T. cruzi were found to have larger wings than non-infected ones40. In R. pallescens, T. cruzi infection did not significantly impact flight initiation, but it was observed that infected females flew significantly faster than males from 30 s to 2 min after flight initiation41. The locomotory activity of R. prolixus was decreased by infection: the total number of movements was 20% less than that observed in non-infected insects42. The time to find a host for an infected M. spinolai was almost twice as fast as for a non-infected insect33. In conclusion, modification of the triatomine traits seems to be species-dependent, age-dependent, sex-dependent, and even environment/ physiology-dependent.
In this work, video-recorded experiments were conducted to study aggregation and geotaxis in adults of T. infestans and to analyze the effect of the infection with T. cruzi. Our hypotheses, based on the literature, were that these two behaviors – gregariousness and geotaxis - are strongly intertwined and are increased in infected individuals. In each experiment, ten insects (non-infected females, non-infected males, infected females, or infected males) were dropped at the base of a vertical wall covered with a paper sheet allowing the bugs to climb. Spatial positions of each insect were extracted from the video every five minutes until 150 minutes, permitting the following of the dynamics and the calculation of the size and spatial stability of the clusters. We demonstrate that both sexes exhibit a high clustering and a high negative geotaxis, males revealing a higher response than females. Interestingly, the T. cruzi infection significantly strengthens both behaviors in both sexes.
Results
Negative geotaxis: spatial distribution and total population
The bugs quickly climbed on the wall and stayed there, demonstrating a high negative geotaxis; and after an exploratory phase, the insects began to cluster and rest (see Supplementary Fig. S1 online). After 10 min, more than 80% of the individuals were on the wall (90% after 20 min) for the four conditions; and this proportion remained constant until the end of the experiment where no statistical difference was detected between the four conditions (Fig. 1). The bugs were mostly located in the upper half of the setup; the median vertical position reached a plateau value (stationary state) after 15 min with a value greater than 35 cm for the four conditions (Fig. 2). Their vertical distributions at 150 min (end of the experiment) revealed a statistical difference between sexes, e.g., males were located higher than females, and also between non-infected and infected individuals, e.g., infected males were higher than non-infected males (Fig. 3). These trends were also discovered inside the top strip of 4 cm (40-44 cm), a zone corresponding to 10% of the total area of the setup and where 56% (80%) of the non-infected (infected) males and 33% (44%) of the non-infected (infected) females were located (see Supplementary Fig. S2 online).
At the end of the experiment, more than 80% of the individuals had a vertical orientation from which around 70-80% had the head turned towards the top of the setup (± 30°). For the four conditions, individuals were not uniformly distributed (Rao’s test < 0.01), but rather centered on 0 (V-tests < 0.001, Fig. 4). When the distributions of individual orientations were compared, no difference appeared between sexes. Interestingly, the infection affected the orientation of the males which demonstrated a higher proportion of insects with the head towards the bottom when infected (Fig. 4).
To summarize, without being infected, both sexes exhibited a high negative geotaxis that was higher for males than for females (Fig. 3). Most of the insects of both sexes were oriented the head toward the top (Fig. 4). The T. cruzi infection strengthened this bug’s geotaxis, especially for males.
Global Clustering
The median aggregated bug fraction increased up to reach a plateau around 35 min, gathering around 70% and 90% of non-infected and infected males respectively, and 40% and 60% of non-infected and infected females respectively (Fig. 5). A statistical difference was detected at 150 min between sexes (males showed a higher aggregated fraction than females), and between infected conditions (infected bugs with a higher aggregated level than non-infected ones) (Fig. 5, see also Supplementary Fig. S3 online). Insects in all conditions tended to gather in one or two clusters. The biggest cluster assembled 40% (70%) of the aggregated non-infected (infected) population in males, and 30% (40%) of the aggregated non-infected (infected) population in females (Fig. 5). No difference was observed between sexes, neither between infection condition (Fig. 5). When the structure of the clusters was compared between infected sexes, clusters of infected males looked more compact, with a significantly smaller distance between aggregated individuals; and they also looked denser, with a higher K-density (Fig. 6, see also Supplementary Fig. S4 online).
At the end of the experiments, individuals were very stable in space: the median fraction of individuals that moved less than 1 cm was greater than 60% for the four conditions, and no difference between sexes and infection condition was detected (Fig. 7). The biggest cluster also showed a strong spatial stability, with no statistical difference between the four conditions (Fig. 5).
In order to verify that the fraction of aggregated individuals was not directly due to the method of calculating this fraction and the increase of the bugs density at the top of the setup, 20,000 repetitions of groups of N simulations were performed (N = 16 for NegFem, N = 15 for NegMal, N = 9 for PosFem, and N = 12 for MalPos). For each simulation, 10 points were vertically distributed following the experimental vertical distribution of the bugs at 150 min (see Supplementary Figure S5 online), and homogeneously horizontally distributed. For each group of simulations, the mean fraction of aggregated individuals was calculated for each repetition. The mean aggregated fractions obtained in the simulations were 0.26, 0.42, 0.36, and 0.72 for NegFem, NegMal, PosFem, and PosMal respectively, revealing that an increase of the geotaxis leads to a rise in the observed aggregation level. However, the probability of observing a mean aggregated fraction higher or equal to the corresponding experimental one was P < 0.0001 for all the conditions, demonstrating that the observed phenomenon involved an active aggregation due to the inter-attraction between individuals.
The fraction of aggregated individuals in a strip of 0.5cm was proportional to the fraction of the population settled in this strip (Fig. 8). The slope of the regression line was the lowest for the non-infected female condition, and the highest for the infected male condition, being intermediate and similar for the two other conditions. The slopes of the linear regression were compared computing a model including the interaction between the total number of bugs and the conditions: a significant interaction was found (F3, 348 = 30.04, P < 0.001), giving a slope equal to 0.61 for non-infected females, 0.85 for non-infected males, 0.84 for infected females, and 0.92 for infected males. The comparison of the slopes between the four conditions showed that the non-infected females’ slope was lower than the three other conditions’ slopes (P < 0.001), the infected males’ slope was higher than the three other conditions’ slopes (P < 0.02), and the non-infected males’ slope was not different from the infected females’ slope (P = 0.98). These results demonstrated that, for the same density, the aggregation was higher for males than for females and for infected insects than for non-infected ones.
Discussion
This work represents the first detailed analysis of aggregation and geotaxis in adult males and females of T. infestans, and how both sexes are affected by T. cruzi infection. As shown before in nymphal instars43,44, adults exhibited an active aggregation due to the inter-attraction between individuals, illustrating the social character of these insects. A stable aggregation emerged for both sexes, but the fraction of aggregated individuals and the density of the clusters were higher for males than for females. This difference between genders was maintained under T. cruzi infection, but the latter reinforced the gregariousness in both sexes.
Our results are in agreement with those of previous studies. Indeed, a multi-factorial analysis (using species, development stages and, feces source altogether) of the aggregative response of individuals to feces shows that the aggregation level was lower (but not statistically different) for females than for males25. It is well established that clustering or reduction of the inter-individual distances of social and subsocial/ presocial arthropods reduces various stresses and therefore energy consumption18,45,46. Different studies show that clustering reduces water loss18–20,46. We hypothesize that the clustering of T. infestans individuals provides a similar benefit. As their weight is lower than females, males could be under higher hydric stress, leading them to a stronger aggregation. Moreover, it could be more adaptive for females to aggregate less to disperse their eggs and increase their probability of survival. In our experiments, males and females were supposed to be in similar physiological status due to their comparable period of starvation (8-10 days), but in the case of infection, T. cruzi and T. infestans compete for nutrients, and bug individuals show reduced resistance to starvation when they are infected47. It might be speculated that infected bugs were more starved and therefore exhibited a stronger aggregation to reduce the cost of the different stresses.
We know very little about the distribution of triatomines inside a dwelling. Even if all the stages are found in the upper part of the walls17,28, our results suggest that males and females rest in a stratified manner on the walls, males being above of females. Domestic cockroaches have a similar way of life than triatomines, except the feeding habits. Periplaneta americana, for instance, shows a preference for vertical areas, and males were vertically positioned above females48. Due to the vertical air current, they can detect the female pheromone easily and orient themselves towards them. In triatomines, the sex pheromone is emitted by the female metasternal glands, inducing males moving towards the females (positive anemotaxis)9. On a wall, due to the difference in temperature between the bottom and the top (up to 6°C 17), an air current move towards the top of the wall could allow the males to feel at some distance the sexual pheromone released by the females. Most individuals in this study showed a vertical orientation with the head towards the top, allowing them to escape quickly from predators generally located below when positioned vertically on a wall. T. cruzi infection enhanced negative geotaxis, especially in males, where a higher proportion of infected individuals faced their heads toward the bottom of the experimental wall.
Detection of T. cruzi by direct microscopic observation is known as being less efficient than by polymerase chain reaction (PCR), especially in case of low parasitemia49,50. It is consequently possible that a small proportion of non-infected insects would have been detected as infected by PCR. Despite this handicap, the difference observed between groups was statistically significant. In the same way, a part of the experiments with non-infected insects was composed of a mix of non-infected and infected insects. The latter implies that a small proportion of infected individuals inside a group of non-infected bugs is not enough to observe a change at the group level. Another interesting question is whether all the discrete typing units (DTUs) of T. cruzi and even strains of these DTUs, will influence the behavior of the bugs in a similar manner. This point was recently asked by Peterson et al., demonstrating that different strains of T. cruzi had different consequences in life history outcomes of R. prolixus51. Thus, more experiments are necessary to understand how T. cruzi affects the mechanisms underlying the geotaxis and the clustering, from a physiological and a behavioral point of view.
Behavioral alterations upon infection are called parasitic manipulation when they are adaptive for the parasite, altering phenotypic traits of its host in a way that enhances its probability of transmission. Some examples where the parasitism affects the geotaxis and the gregarious behavior of the hosts were described52-57. Is there any advantage to T. cruzi to enhance the negative geotaxis and the aggregation behavior in males of T. infestans? Two hypotheses can be put forward: 1) expand its spread, by increasing the longevity of the vector, away from ground predators, and/ or by allowing a higher rate of coprophagy and cleptohaematophagy through a stronger aggregation; and 2) facilitate the host-finding; it could be easier for them to find a food source by being higher on a vertical surface. Actually, these insects are known to fall near a host from the ceilings when they wander in a host search behavior58. Ramirez-Sierra et al. (2010) have already reported an increase of the dispersion on the field of infected females of T. dimidiata39. Infected nymphs of R. prolixus exhibited, on the contrary, a reduction of their locomotory activity42.
A low height device like the setup used in these experiments allowed us to highlight differences, between sexes, and between infected and non-infected insects. The questions that the results generate showed that little is known about the spatial distribution of the insects in their natural conditions and how they behave. Our results predict that in a natural/ anthropic environment the percentage of infected insects should increase with the height of the settlement. More experiments have to be carried out to understand the dispersion and aggregation behaviors of T. infestans, both in the laboratory and in the field. For example, one of our hypotheses concerns the influence of the height of the setup: the higher the setup, the greater would be the spatial segregation between the four categories. The response of the bugs should be modulated according to factors like the development stages and the physiological condition of the insects, the bug density, the numbers of available and suitable shelters, and the infection of the bugs. Another interesting question is about how mixed groups would distribute (both sexes, and/ or both infected and non-infected bugs). Indeed, in addition to those at work in monospecific aggregation new effects come into play among which segregation, whereby the different populations select different patches, plays a prominent role59.
Our results also open a new vision for controlling/ monitoring triatomines on the field, suggesting that there is a higher risk of T. cruzi infection in bugs located the upper part of walls/ rooms. More precise studies regarding bug distribution within microhabitats under field conditions should help to improve control and monitoring by trapping60. Finally, our results lead us to propose simple tests easily feasible in the field, based on geotaxis and the aggregative behavior of the bugs, to detect infected insects.
Methods
T. infestans specimens were collected in dwellings from Yacuiba Municipality (Gran Chaco region), Department of Tarija, Bolivia, in the area Tierras Nuevas (S21.748334, W63.561866, 621 m asl) - San Francisco de Inti (S21.818193, W63.588042, 600 m asl). The infection rate of the captured insects, determined by analyzing drops of feces under a light microscope, was 47.5 ± 21.7%. Bugs were reared at 26±1°C, 60±15% RH, 12:12 night:dark cycle, in plastic pots containing a folded piece of kraft paper commonly used in the insectarium. They were fed on hens once every two weeks. Four conditions were then studied: non-infected males, non-infected females, infected males and infected females (abbreviated as NegMal, NegFem, PosMal, and PosFem in Figures). Fifteen and sixteen experiments were carried out with non-infected males and females respectively, and twelve and nine experiments with infected males and females respectively. Due to a problem in the insectarium, nine experiments using non-infected insects (4 and 5 experiments in males and females respectively) included some infected insects actually. A test for T. cruzi infection of the insects from these experiments was realized again determining a proportion of the infected individuals being less or equal to 20%. At 150 min, the aggregated fraction from the weakly infected experiments was closer to the fraction observed in non-infected group than to the fraction observed in infected groups (Anderson-Darling k-sample test: TkN = 6.08, P < 0.001; number of observations: 11 for non-infected males (NM) and for non-infected females (NF), 9 for infected females (IF), 12 for infected males (IM), 4 for weakly infected males (WIM) and 5 for weakly infected females (WIF); Anderson-Darling all-pairs comparison test: females: NF vs WIF: P = 0.45, WIF vs IF: P = 0.03; males: NM vs WIM: P = 0.91, WIM vs IM: P = 0.08). Therefore, these experiments were included in the non-infected group.
Setup and Methods
A glass aquarium was used (50 × 20 × 50 cm) to avoid escaping of T. infestans which is unable to climb on glass walls. Insects were allowed to climb on one of the vertical surfaces of this aquarium (50 × 50 cm) offered by a paper sheet (kraft paper, 43 × 44 cm). The glass setup was washed, and the paper changed at the end of each experiment. It was illuminated by a centered 60W incandescent light bulb, placed at 50 cm behind the wall covered by the paper sheet. The paper guaranteed a homogeneous illumination of the setup. A video camera (Sony DCR-SR68) placed in front of the setup recorded the bug activity for 150 min. A 1 m high polystyrene wall surrounded the setup to isolate it. Experiments were conducted in a quiet and dark room to avoid any disturbance, at the beginning of the photophase. Ten bugs (8-10 days of starvation) were dropped on the bottom of the setup. They explored their environment rapidly and climbed on the wall. From the recordings, a snapshot was extracted at 1 min, 5 min and then every 5 min up to 150 min (31 snapshots in total). A processing program allowed us to record the spatial position of the thorax of each bug on each snapshot. With these spatial coordinates, the inter-individual distances were computed. As the length of an adult bug is on average 2.5 cm, and due to a tactile (legs or antennae) or visual perception, two individuals were considered as aggregated when they were at a distance less or equal to 4 cm.
Indexes and statistics
Several indexes of position and aggregation were calculated using processing programs: 1) the number of individuals on the paper sheet; 2) the number of aggregated individuals; 3) the number and the size of the clusters; 4) the spatial stability of the individual (% of individuals that were found at time t+1 in a circle of 10mm in radius centered on the coordinate of the insect at time t); 5) the spatial stability of the biggest cluster (study of the distance between the centroid of the biggest cluster at time t+1 and the centroid of the biggest cluster at time t). Finally, the individual position of the insects in the setup at the end of each experiment was analyzed, recording the vertical orientation of the bugs (position 0: head towards the top), to put forward a privileged position. A vertical orientation was defined as inside an angle of ±30° to the vertical, head oriented towards the top or the bottom. Outside this range, the insect is not considered in a vertical position anymore. The structure of the clusters was also compared between infected males and infected females, conditions where bigger clusters emerged. Each cluster was considered as an undirected network where each node was an individual. Links between nodes were established when the distance between them was less or equal to 4 cm (the threshold for considering aggregation). The cluster K-density (ratio of the number of edges divided by the number of possible edges) was compared for clusters with a size greater than three individuals.
The comparisons between conditions were made using the Anderson-Darling k-sample test61. In case of obtaining a P < 0.05, an Anderson-Darling all-pairs comparison test was performed. These statistics were calculated using the functions adKSampleTest and adAllPairsTest of the PMCMRplus package of R62,63. Circular statistics were carried out with Oriana 4.02 (Kovach Computing services). Uniformity of data was tested using Rao’s test, mean comparisons using V-test, and distribution comparisons using Mardia-Watson-Wheeler pairwise test. The structure of the clusters was analyzed using the igraph package in R64. The linear regression was done using the lm function, and the comparison of the slopes of regression with the lsmeans package of R, using Least-squared means (predicted marginal means)65.
Author contributions statement
SD and JLD designed the experiments. SD collected the data. SD, GMRA and JLD analyzed the results. SD and JLD wrote the main manuscript text. SD and GMRA prepared the figures. All authors reviewed the manuscript.
Competing interests
The authors declare no competing interests.
Data availability statement
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.
Acknowledgments
S. Depickère is particularly grateful to T. Chavez, F. Lardeux, Don Hugo and E. Siñani for logistic support. S. Depickère thanks the FYSSEN Foundation.