Water vapour and heat combine to elicit biting and biting persistence in tsetse
© Chappuis et al.; licensee BioMed Central Ltd. 2013
Received: 14 May 2013
Accepted: 9 August 2013
Published: 19 August 2013
Skip to main content
© Chappuis et al.; licensee BioMed Central Ltd. 2013
Received: 14 May 2013
Accepted: 9 August 2013
Published: 19 August 2013
Tsetse flies are obligatory blood feeders, accessing capillaries by piercing the skin of their hosts with the haustellum to suck blood. However, this behaviour presents a considerable risk as landing flies are exposed to predators as well as the host’s own defense reactions such as tail flicking. Achieving a successful blood meal within the shortest time span is therefore at a premium in tsetse, so feeding until replete normally lasts less than a minute. Biting in blood sucking insects is a multi-sensory response involving a range of physical and chemical stimuli. Here we investigated the role of heat and humidity emitted from host skin on the biting responses of Glossina pallidipes, which to our knowledge has not been fully studied in tsetse before.
The onset and duration of the biting response of G. pallidipes was recorded by filming movements of its haustellum in response to rapid increases in temperature and/or relative humidity (RH) following exposure of the fly to two airflows. The electrophysiological responses of hygroreceptor cells in wall-pore sensilla on the palps of G. pallidipes to drops in RH were recorded using tungsten electrodes and the ultra-structure of these sensory cells was studied by scanning and transmission electron microscopy.
Both latency and proportion of tsetse biting are closely correlated to RH when accompanied by an increase of 13.1°C above ambient temperature but not for an increase of just 0.2°C. Biting persistence, as measured by the number of bites and the time spent biting, also increases with increasing RH accompanied by a 13.1°C increase in air temperature. Neurones in wall-pore sensilla on the palps respond to shifts in RH.
Our results show that temperature acts synergistically with humidity to increase the rapidity and frequency of the biting response in tsetse above the levels induced by increasing temperature or humidity separately. Palp sensilla housing hygroreceptor cells, described here for the first time in tsetse, are involved in the perception of differences in RH.
Tsetse flies are obligate haematophagous insects that pierce the skin of the host with the haustellum to lacerate capillaries and suck blood. Tsetse are diurnal and their host-seeking behaviour is closely correlated to the activity of their hosts and predators . Biting behaviour is the last step in host-seeking behaviour and represents a critical risk for tsetse. Indeed, upon landing on the host flies are exposed to predators and to the host’s defence such as grooming and tail flicking . Tsetse are k-strategists and are among the dipterans with the longest life span , that can extend up to several months in the field . Tsetse such as Glossina pallidipes feed every three to four days [5, 6]. The trade-off between feeding and risk avoidance could explain the flies’ strategy of infrequent host visits but feeding to repletion when they do . As blood is not directly available to tsetse on landing, cues emanating from the skin stimulate tsetse to bite where they can readily find blood, thereby minimising the time spent on the host and so reduce risk. The heat of mammalian skin elicits biting in tsetse  and in other ectoparasite arthropods [8–13]. But heat is not the unique factor in eliciting biting as it is a multi-modal sensory response, i.e. diverse stimuli can affect this behaviour. Van Naters, et al. showed that the combination of a chemical cue, such as uric acid, and heat increased the time spent probing by G. fuscipes fuscipes. Gatehouse  also found that the probing response of the stable fly Stomoxys calcitrans was multi-modal as an increase in humidity with the addition of ammonia induced more flies to probe than increases in humidity alone. Later, Gatehouse [16, 17] illustrated the importance of hygroreception in the biting responses of this species by showing that increases in humidity were as efficient at inducing the biting response as heat alone or the combination of humidity plus heat. The importance of humidity in host-seeking behaviour in tsetse was underlined by experiments conducted in a wind tunnel by Evans and Gooding . They showed that an increase in both moisture and air temperature is better at eliciting upwind flight to a source of CO2 in G. morsitans morsitans than heated air alone. Aedes aegypti is also more attracted to combined heat and humidity increases than to heat alone in an olfactometer . Heat and moisture emanating from the host [20, 21] are clearly important host cues for haematophageous insects.
To our knowledge, the interaction of heat and humidity has not been investigated in the biting behaviour of tsetse. To study this we built an experimental setup that allows us to quantify the biting response of G. pallidipes to an increase in temperature accompanied by increasing humidity or to increasing humidity accompanied by a temperature increase of a fraction of a degree. We demonstrate how combined temperature and humidity increases serve to influence the biting response, response latency, biting persistence and the dynamics of this fundamental behaviour in tsetse. In addition, we report on neurones with hygroreceptive properties present in basiconic sensilla on the maxillary palps of G. pallidipes.
G. pallidipes (Austen) pupae were supplied by the International Atomic Energy Agency (IAEA) Siebersdorf Laboratories, Austria. The G. pallidipes colony originated from flies collected near Tororo, Uganda in 1975 that were reared successively at the University of Amsterdam, NL, the University of Bristol, UK, and at the IAEA-Siebersdorf Laboratories since 1986. Imagos were kept in 2 climate chambers offset by 2 hours with 10 h light at 26°C, 85% relative humidity (RH) and 14 h dark at 22°C, 85% RH. Sexes were separated at emergence. The behaviour of 3-day post-emergence unfed flies was observed during their two daily activity peaks in the first 1 h 30 min and last 2 hours of the photophase. The response of each fly was only tested once.
Additional file 1: Example of a video record used to quantify the biting responses of G. pallidipes. The holding cage was slid from an acclimatisation airflow at 24.5°C, 7.3% RH to a test airflow (red light on) where the female was exposed to an increase of 13.1°C and 72.4% RH. The fly responded by regularly passing its haustellum through the cage netting. Latency is the time between the moment the red light turned on and the onset of the first bite, i.e. the first time the haustellum crosses the netting. To limit the size of the file only 22 s of the high resolution video record is shown. (MP4 14 MB)
Temperature, RH and corresponding partial pressure of water vapour differences in the air streams
Temperature increase [°C]
RH increase [%]
Partial pressure of water vapor increase [hPa]
Number of flies tested
0.2 ± 0.2
1.1 ± 0.5
0.4 ± 0.2
10.8 ± 1.1
3.4 ± 0.3
60.6 ± 0.4
18.0 ± 0.2
75.4 ± 0.7
23.5 ± 0.5
3.0 ± 0.2
1.4 ± 0.04
1.0 ± 0.02
11.1 ± 0.3
4.5 ± 0.1
60.0 ± 0.4
22.4 ± 0.2
13.1 ± 0.1
0.2 ± 0.2
2.7 ± 0.2
10.5 ± 0.4
9.4 ± 0.3
34.5 ± 0.6
24.7 ± 0.3
60.0 ± 0.6
41.8 ± 0.5
72.4 ± 1.3
49.8 ± 0.8
To quantify biting, a bite was defined as the act of passing the haustellum through the netting (Additional file 1). No substrate was supplied beneath the mosquito netting, and flies were only exposed to the temperature and humidity stimuli provided in the air streams. The time the red LED switched on and each time the haustellum crossed the netting and was withdrawn, was recorded with the video analysis programme Kinovea (version: 0.8.14) . For statistical analysis, the program R (version: 2.15.0)  was used and the level of significance was set at 0.05.
Proportions of responding flies were analyzed using a generalized linear model (GLM) with a logit link function. Latency was measured as the time between the moment the cage arrived in the test flow (red light on) and the onset of the first bite. To determine if a treatment increased or reduced latency, Cox proportional hazards regression was used with the R package survival to allow accounting for flies that did not respond as censored data.
As flies bit successively, the number of bites was log transformed and analysed by a factorial ANOVA, following a test for homogeneity of variance with a Bartlett test. Multi comparisons between treatments were made with Tukey’s Honest Significant Difference test (Tukey HSD). Data on time spent biting, average time per bite and average time between successive bites were analysed with a GLM with a reciprocal link function (Gamma family). Where necessary, multi comparison tests were done with a GLHT function depending on the model used (below) with a Tukey matrix of contrast (R-package: multcomp). As the number of responding flies exposed to an increase of 0.2% RH and 13.1°C was low (5 over 34 individuals tested), these data were pooled with data obtained for an increment of 10.5% RH at the same temperature increase. No significant difference in deviance was observed by pooling these two categories (ANOVA). Each step in reducing the Cox model and GLM was controlled by a deviance analysis following a Χ2 distribution (ANOVA). Temperature, RH, partial pressure and saturation pressure of water for experiments were analysed with Expdata (version: 1.2.1, Sable Systems) and are shown in Table 1.
A modified version of the set up used by Taneja and Guerin  was used to record action potentials from receptor cells within basiconic sensilla on the palps of G. pallidipes. After anesthesia with CO2, the fly was immobilized in a disposable micropipette tip (Kartell Spa, Noviglio, Lombardia, Italy) cut at the tip and held horizontally with the head protruding at the narrow end. The disengaged haustellum was attached to the tip with sticky tape. A V-shaped silver reference electrode was placed between the protruding head and the end of the micropipette tip. Mouthparts, except the maxillary palps, and eyes were embedded in dental cement (Ketac™ Cem radiopaque, 3 M ESPE, Seefeld, Bayern, Germany) to prevent movement of the head. A 0.1 mm diam. tungsten recording electrode sharpened by electro polishing in saturated KNO2 solution was placed randomly in the proximal 2 mm of the lateral external palp groove (Figure 2) and connected, in succession, to a high impedance preamplifier (×10; Syntech, Kirchzarten, Baden-Württemberg, Germany), an AC/DC amplifier (UN-03, Syntech) and the intelligent data acquisition controller (IDAC) analog-digital converter (USB-IDAC box, Syntech) to a computer. Spikes were recorded with Autospike (Syntech). Spikes were considered when the amplitude was greater than 1.9 - 2 times the standard deviation of the entire signal amplitude. The number of spikes counted for the first 2 s of stimulation was divided by 2 to estimate the spike frequency in Hz. Statistical analyses of spike frequencies were performed by ANOVA following a test for homogeneity of variance with a Bartlett test and multi comparisons between treatments were made with Tukey’s Honest Significant Difference test.
To test the responses of neurones to changes in RH, palps were exposed to five RH drops from 100% (−0.02% ± 0.3; -26.4% ± 1.7; -44.1% ± 4.0; -69.4% ± 4.8; -91.1% ± 3.4), obtained by mixing moist and dry air at different proportions with a stimulator. For this, two airflows were mixed by manually adjusting two flow controllers (C05K510418, Norgren, Lichfield, Staffordshire, United Kingdom) for a desired moist to dry air ratio. The flow controllers mounted on solenoid valves (RPE 3105 NC 230 V/AC, RPE S.r.l., 22070 Carbonate, Lombardia – CO – Como, Italy) were arranged in six pairs providing six channels triggered by an external electronic control that allowed the manual selection of the desired RH shift. Moist and dry airflows were produced by splitting charcoal-filtered air at 0.4 bar into two channels isolated from each other with non-return valves (T51P0008, Norgren). The moist channel was first humidified by bubbling it through a 2 L gas-wash bottle filled with distilled water immersed in a water-bath at 35°C. This humidified air was cooled to room temperature (25°C) by passing it through a copper coil before bubbling it through a second glass bottle (1 L) filled with distilled water immersed in a second water-bath at room temperature. The dry channel air passed through copper tubes immersed in the same water-baths as the moist airflow. Stimulation with RH drops was made by simultaneously closing the first water-saturated channel and opening one of the five other channels for 8 s. At the output, RH of the airflow was measured with a hygrometer (HMP50-60, above). The air speed at the output of the stimulator was measured with an anemometer (Thermo-air anemometer, above) and set at 1 m/s. Temperature of the output airflow was measured daily with a thermistor (PT-6 Physitemp, Science products GmbH, Hofheim, Hessen, Germany) at 25.4°C ± 0.2 for the water-saturated flow and 25.0°C ± 0.4 for the dry flow (8.9% RH).
All parameters used above to describe the biting behaviour of G. pallidipes do not take into account the fact that the biting response occurs over time. It appears that flies did not bite with the regular frequency of a standing wave (Figure 5). In the first 5 seconds the frequency was at its maximum: 5.8, 5.9, 8.3, 8.4 bites per 5 s, respectively, at RH increments of 0.2-10.5%, 34.5%, 60% and 72.4% and then biting frequency decreased. This drop in biting frequency was related to RH and was highest for the 34.5% RH increase and lowest for the 72.3% RH increase. This resulted in biting termination at 30s for the 22 responding flies at the 34.5% RH increase, at 85 s for 35 responding flies at the 60% RH increase, and one fly was still biting at 120 s of the 38 responding flies at the 72.4% RH increase (Figure 5).
When the temperature shift was minimized to 0.2°C, RH increments of any size did not significantly (GLM, P = 0.092) increase the proportion of flies biting: of the tested flies (Table 1), 0, 0, 10 and 2.5% responded to the RH increments of 1.1, 10.8, 60.6, 75.4%, respectively. Furthermore, when females were exposed to an increase of 3°C accompanying RH increases of 1.4, 11 and 60% (Table 1) only one of 10 females responded to the 11% RH increase and the other RH increases failed to elicit biting.
When RH increases were minimized to less than 1.5%, only a temperature increase of 13.1°C succeeded in eliciting biting: 14.7% of the 34 flies tested responded (Figure 3A). However, this comparison needs to be qualified: RH increases at different temperatures do not produce the same increase of water vapour pressure in air (Table 1). An increase of 13.1°C and 0.2% RH results in water vapour pressure of 2.7 hPa, roughly the same order as produced by an increase of 0.2°C with a 10.8% RH increase (3.4 hPa) or by a 3°C increase with a 11.1% RH increase (4.5 hPa). Nevertheless, with similar minimal water vapour increases of between 2.7 to 4.5 hPa none of the 39 flies responded to an increase of 0.2°C and only one female out of 10 responded to a 3°C increase.
Whereas RH increases accompanying minimal temperature increases or temperature increases accompanying minimal RH increases succeeded in inducing biting in no more than 14.7% of flies, the proportion of responding flies reached nearly 95% when both temperature and humidity were increased by 13.1°C and 72.4% RH (Figure 3A). This indicates a strong synergism between RH and temperature.
On the lateral side of each palp at approximately 800 μm from the tip, a ventro-lateral field of wall-pore sensilla extends over a distance of 1200 μm and is 30 μm in width (Figure 2A & B). These hairs are wall-pore single-walled (wpsw) basiconic sensilla (Figure 2C). They are short (10–15 μm in length) with a diameter of 1.6 μm in the middle and 2.6 μm at the base (Figure 2C). The cuticular apparatus is a thin single wall (100 nm thick) pierced by numerous pores (30–40 nm in diameter) with pore-tubules extending into the lumen shaft. The lymphatic space contains few or more dendritic branches (up to 72) and can be filled with filaments which make it electron dense (Figure 2C).
At the base of the sensillum three sensory cells are associated with three enveloping cells (thecogen, tricogen and tormogen). From the thecogen cell a short sheath emerges surrounding the outer dendritic segments. To describe the fine structure of the outer dendritic segments and their branching pattern, all of the 56 sensilla examined (18 flies) presented two swollen dendritic segments at the base of the sensilla (Figure 2C & D). The swelling of the outer dendrites is not due to an uptake of water driven by a hypotonic fixing solution, as the same swelling was also found using a hypertonic fixative.
Our results show that combined RH and temperature increases act synergistically to induce the biting response in tsetse. Moreover, at a temperature increment of 13.1°C, as RH increased, the flies responded more rapidly and in higher numbers. The persistence of biting behaviour measured by the number of bites and the time spent biting also increased with increasing RH. The dynamics of the biting response correlated over time with the amount of water in warm air, with a biting frequency that was higher at the beginning of the 2 min exposure and then fell off rapidly, but flies adapted more slowly at higher RH increments.
Latency and the proportion of flies responding are, respectively, negatively and positively correlated to RH increases at the 13.1°C air temperature increment. This indicates a very close link between the integration of thermoreception and hygroreception to initiate biting behaviour in tsetse. G. pallidipes and other savannah species exploit mainly mammals to obtain a blood meal [28, 29]. These hosts lose heat and water through their skin , creating a humidity gradient between the skin and the environment [20, 30, 31], equipping these animals with a means of regulating body temperature through evaporation. Indeed, water droplets are formed by aprocrine glands during sweating  indicating that the humidity is close to saturation in the air near the skin (>90% at 37°C, 57 hPa). Taking into account the environmental conditions prevailing during the dry season in Africa (<40% at 30°C, 17 hPa) , differences in water partial pressure of more than 40 hPa could occur between a vertebrate host and its environment during the dry season and in the range of 38 hPa during the wet season (60% at 25°C, 19 hPa). This corresponds to the range of water vapour partial pressure differences tested here. The combination of heat and water vapour assist tsetse to find hosts as demonstrated by Evans and Gooding  and we show here how temperature and humidity combined serve to elicit the biting response at the feeding site. Our findings corroborate those of Khan and Maibach  who found that heat must be combined with humidity to elicit biting in another dipteran, Ae. aegypti. Previous studies have demonstrated that heat elicits a biting response in tsetse flies as described by Dethier  and later by Brady  and this has been amply exploited in the development of artificial feeding systems using membranes over heated blood for mass production of these vectors of disease . Our results suggest that it is the change in RH accompanied by an increase in temperature that induces the biting response in flies coming from cooler and dryer conditions. We did not investigate the biting response of tsetse to RH increases at a constant temperature of 37°C for example or to temperature increases for flies maintained at high RH conditions. An increase in one factor, heat for example, may be sufficient to induce biting when a fly is already in a very humid environment or vice versa. Increasing temperature serves to increase the metabolic rate in tsetse  but, in the absence of an increase in humidity, a temperature increment alone is insufficient to induce the biting response. The water balance state of a fly could influence its eagerness to bite. Our flies were maintained under constant and high RH conditions before experiments. The role of water vapour can also differ according to the degree of haematophagy in a species: tsetse hardly responded to RH increments alone in this study whereas an increase only in humidity is sufficient to elicit probing in the stable fly that is both haematophagous and a nectar feeder [16, 17] and in the nectar feeding hawkmoth .
As well as affecting response latency and the number of flies responding, increasing water vapour accompanied by an increase in temperature also increased the persistence of the biting response in G. pallidipes. This indicates that heat and humidity combined constitute an adequate stimulus to induce flies to persist in biting, a behaviour that represents a critical risk for tsetse . Van Naters, et al. found that two sensory modalities presented together, heat and uric acid, increased the biting persistence of G. fuscipes fuscipes compared to flies stimulated by heat alone. This interaction between sensory systems increased the time spent probing but not the numbers of bites. In our conditions the interaction between thermoreception and hygroreception increased the number of bites and correspondingly the time spent biting. Moreover, for a temperature increment of 13.1°C the average time per bite decreased with increasing RH above 34.4%. This suggests that the more adequate the stimulus, the quicker the fly withdraws its haustellum to try again when haustellum extension fails to reach the skin under optimum stimulus conditions, i.e. the haustellum is quickly brought back to initiate a new biting attempt. Since the average time between two consecutive bites remains unchanged, this suggests that flies repeat a stereotypic positioning behaviour between biting attempts that assists them to reach the skin through the host’s fur. The frequency of biting events was correlated with the amount of water in warm air but was not stable in time, being higher in the first 10s of the response, before falling rapidly thereafter. However, flies adapted more slowly at the higher dose of water vapour in warm air.
Neurones responded with a significant increase in spike frequency in all wpsw palp basiconic sensilla exposed to a decrease in RH. When these sensilla were also exposed to an increase of 9°C in dry air, no significant response in spike frequency was observed (data not shown). The overall responses represent complex spike patterns and our recordings did not permit to differentiate between the responses of the three neurones within the basiconic wpsw sensilla on the palps. However, we can conclude from our recordings that these sensilla contain one or more dry cells. Wet cells also occur in basiconic sensilla on the palps, but were very rare as only four were found in preliminary recordings from more than 60 sensilla. Tsetse are thus able to perceive RH changes with dry and wet cells increasing their spike firing frequency as humidity decreases or increases. To our knowledge, this is the first report of hygroreceptor cells in tsetse. Bursell  previously demonstrated that several tsetse species including G. pallidipes are more active in dry air than in moist air and that antennal removal did not modify the response, indicating that hygroreceptor cells responsible for the observed behaviour occur elsewhere on the fly. Evans and Gooding  showed that in the presence of CO2, moist and heated air is more efficient at eliciting upwind flight in tsetse than heated air alone. Temperature and humidity conditions affect water loss in tsetse flies , so hygroreceptor cells could evidently serve to guide flies in the selection of resting sites where they would be subject to lower rates of water loss, as proposed by Chown, et al. and demonstrated in mosquitoes .
The ultra-structure of the basiconic wpsw sensilla housing the dry cells on the palps of G. pallidipes is very similar to typical olfactory wpsw sensilla of insects . We recorded responses to the host-related volatiles 1-octen-3-ol, n-valeric acid and dimethyl-trisulphide from cells in these sensilla (data not shown), confirming earlier findings by Lewis  of responses by these cells to carboxylic acids. The fact that basiconic sensilla of insects can contain both hygro- and chemoreceptor cells and has already been documented (cf. ), but to our knowledge this has never been reported before for testse. All basiconic sensilla examined present 2 neurones with swollen dendrites that occupy all the lymphatic space at the base of each sensillum, separating the sensillar lymph into proximal and distal selections. The swelling remained unaffected by a hypertonic fixative solution. The transduction mechanism for the coding of differences in RH is unknown, but it appears clear that pressure exerted on the membrane of hygroreceptor cells could play a key role . We propose that as RH levels drop, water loss from the wall-pore sensillum could cause the inflated dendrite to be deformed as a result of a pressure difference exerted on either side of the dendritic swelling.
Our results show that temperature acts synergistically with water vapour increments in air in inducing the biting response in tsetse and that palp sensilla housing hygroreceptor cells are involved in the perception of differences in RH. As such, the tsetse biting response exploits the homoeothermic requirements of hosts where evaporative water loss is a crucial means of lowering body temperature. The findings also underline the perception of a substance as essential to life as water in the biting behaviour of obligate haematophagous insects like tsetse where host animals are the only source of both nutrients and imbibed water.
The experimental design described here that permits quantification of the biting response of tsetse to heat and humidity constitutes a means to study inhibition of this crucial response in trypanosome transmission. Disruption of either thermoreception or hygroreception is likely to be sufficient in considerably reducing the biting response. The experimental set up also provides a means to investigate the effect of trypanosome infection on tsetse biting responses.
We thank Debora Zoia, Kevin Thiévent and François Walter for their contribution to the building of the experimental set-up to record tsetse biting responses and for data collection. We thank Claude Hêche and Daniel Varidel of the Technical Service of the Faculty of Science, University of Neuchâtel for precious technical support, and the IAEA Seibersdorf Laboartory, Vienna, Austria for supplying tsetse fly pupae. We also wish to thank anonymous reviewers and colleagues of the Animal Physiology Laboratory, University of Neuchâtel, especially Andrew McMullin, Alexandre Gurba, Sébastien Kessler and Jérôme Frei, for their pertinent comments on the manuscript.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.