The behaviour of tsetse flies in an odour plume

The tsetse flies Glossina pallidipes Austen and G. m. morsitans Westw. (Diptera: Glossinidae) are obligatory blood feeding insects that do not live in close association with their hosts (mainly mammals). Tsetse flies are relatively long lived insects and have to take a blood meal regularly. Tsetse flies use smell and vision to find their hosts. In the last decade, many aspects of tsetse foraging and host-location behaviour have been elucidated. A range of kairomones has been identified. These kairomones can be used to which increase the number of tsetse caught with a trap or a target. However, some aspects of tsetse behaviour remained unclear. Not much known about the effects of the identified kairomones on the host-location behaviour of tsetse at long distance for example. Also, the effect of the physiological state of a tsetse on its foraging behaviour, was unclear.This thesis describes a field study, conducted in the Zambezi valley, Zimbabwe, that aimed at clarifying these questions. Chapter 2 and 3 describe studies of the foraging behaviour of tsetse. Studies of tsetse host location behaviour are presented in the following five chapters.Foraging behaviourIn the field, G. pallidipes and G.m. morsitans are mostly caught in the early morning and the late afternoon. Spontaneous activity of G. morsitans in the laboratory also showed peaks in the morning and evening. How the activity patterns in the field are and what the effects of sampling methods on the apparent rhythm are, was not clear, however. In chapter 2 , 1 describe studies in the diurnal rhythm in catches with different devices.From before sunrise until after sunset, hourly catches of G. m. morsitans and G.pallidipes were made from a stationary unbaited electric net; an ox flyround; an electrified target; an epsilon trap and a biconical trap. The latter three were baited with artificial host odour, consisting of acetone, 1-octen-3-ol (henceforth termed octenol), 3-n-propylphenol and 4-methylphenol.Catches of tsetse were low from dawn to early afternoon, peaking just before sunset. Despite the broad similarity in diurnal patterns, there were some consistent differences between the sampling methods. The fraction of the daily catch caught in the afternoon with the target or the traps was larger for males than for females. The catch at the unbaited electric net probably gave the best estimate of the diurnal rhythm of flight activity of tsetse. Compared to the net catch, trap catches of tsetse were relatively high during the middle of the day and low in the early morning and late afternoon. The differences were attributed to sampling biases of traps. The pattern of the target catches was not significantly different from the pattern of unbaited net catches. The catch pattern at the mobile bait was significantly different from that of the net. This is attributed to the response of tsetse sitting on vegetation, which were not sampled by the unbaited net, to the ox fly-round. It is concluded that target catches can be used to monitor diurnal rhythms in tsetse activity. The pattern of Stomoxynae -catches on targets resembled that of tsetse. However, no peak was evident in the electric net catches. Catches of the tabanid Philoliche (Stenophara) zonata Walker showed a sharp peak in the early afternoon.The daily patterns in catches at the unbaited net and the baited target were similar. This suggested that active search was the most important strategy for tsetse to find a host. In chapter 3 , the catches of targets and unbaited nets and the tsetse sitting on vegetation were studied in more detail to obtain evidence for this suggestion.The catches of G.pallidipes and G.m. morsitans at a target baited with odour (acetone, octenol and two phenols) were positively correlated with catches of the same species at an unbaited net. No correlation existed between target catches and hand net catches of tsetse flies sitting on the vegetation. G.pallidipes females caught at a target and at an unbaited net were older than those caught from vegetation. Of the female G.pallidipes caught at the target, 46 % were in the first 3 days of pregnancy. Of those caught at the unbaited net, significantly fewer, 21 %, were in this stage. G.pallidipes males caught from vegetation contained more fat (3.07 ± 0.333 mg) than those caught at the unbaited net (2.06 ± 0.339 mg) or at the target (2.19 ± 0.218 mg). The target catches consisted mostly of tsetse that were already in flight when they sensed the stimuli from the target. Target catches were biased towards female G.pallidipes in the first 3 days of pregnancy.Host-location behaviourTo investigate the effects of odour composition and odour release rate on hostlocation behaviour of tsetse, mark-release-recapture studies were used. Before these experiments were started, the behaviour of tsetse just after their release was studied at 10 m downwind of an odour source with a video camera and with electric nets. This work is described in chapter 4 . Tsetse were collected with traps, marked and released. Only tsetse that were recaptured on the same day were analyzed. The study focused on G.pallidipes, because it is more readily caught in traps than G.m. morsitans.Video studies showed that in the absence of odour, 46 % of the released G.pallidipes turn downwind and 32 % turned upwind. Tsetse left the release box at a constant rate and appeared to avoid each other. When an artificial odour mixture containing carbon dioxide, acetone, octenol and phenols was used, 35% turn downwind, significantly less than in the absence of odour, and 37 % turned upwind. Tsetse left the release box later than in the absence of odour and not at a constant rate. Tsetse did not avoid each other.When the release box was placed in a complete ring of electrified nets, only natural ox odour changed the distribution of tsetse over the electric nets compared to the no odour treatment. The artificial odour mixture, with and without carbon dioxide, had no effect on the distribution of tsetse over the electric nets. Most tsetse were caught while flying in a downwind direction.The electric nets have an efficiency of about 50 %. In this experiment, with a complete ring of nets, the tsetse that survive their first contact with the net, could thus be killed on another net. The difference between the video study and the electric net study is attributed to this effect. The small differences between the odour treatment and the control was probably due to frequent changes in wind direction. Odour did not always reach the box because of these changes. However, if odour is present at the box, tsetse released or departing from the box react to the presence of odour immediately.Marked G.pallidipes were released downwind of an odour source and the percentage recaptured at the source on the same day was measured. The influence of release distance, odour composition ( chapter 5 ) and odour release rate ( chapter 6 ) on the recapture percentages were studied.In the absence of odour, 1.3 % of the marked tsetse released from a box or refuge were recaptured, independent of the distance between release point and odour source. When natural ox odour or a blend of carbon dioxide, acetone, octenol and phenols was dispensed, untransformed recapture percentages of box-released tsetse decreased from 18 % for tsetse released at 10 m to 2 % for tsetse released at 100 m. Recapture percentages were significantly higher than in the absence of odour at all release distances for ox odour and for release distances up to 75 m downwind for the artificial odour. When a combination of acetone, octenol and phenols or carbon dioxide on its own was dispensed, recapture percentages decreased from 6 % for tsetse released at 10 m to 0 % for tsetse released at 100 m. With these odours, recapture percentages were higher than in the absence of odour when tsetse were released at 20 m from the source, but were lower than recaptures in the presence of ox odour or the artificial mixture with carbon dioxide. Recapture percentages of flies leaving a refuge were higher than those of box-released tsetse. Proximity of source had no effect on the recapture percentage of refugeleaving tsetse.The odour mixture consisting of carbon dioxide, acetone, 1-octen-3-ol and phenols was released at three different rates. The medium odour release rate, was identical to the rate used in the comparison of odour composition and the results were similar. The low odour release rate, a quarter of the medium rate, only attracted tsetse from 20 m and less; the recapture percentage declined from 5% at 10 m to 2 % at 30 m. The decrease in odour release rate thus caused a decline in plume length and a decline in host location efficiency or recruitment of tsetse to the plume. When the medium release rate was increased fourfold, the resulting high odour release rate did not increase the recapture percentages from the same distances, nor did the distance from which tsetse were recaptured increase, compared to the medium rate.From the experiment described above, we can conclude that: (1) tsetse have an efficient host-location mechanism, 80 to 100 % of the tsetse that detect the odour at less than 30 m from the source found the source; (2) tsetse can detect an ox at 100 m distance downwind.An odour plume consists of 'pockets' of odour that are moved by the wind. The flux of kairomones in an 'odour pocket' might be the only cue used by tsetse. The flux is determined by the dose and the number of kairomone compounds present in the mixture. The number of kairomone compounds in the odour pocket seems to be more important than the dose of the components.To elucidate the differences in recapture percentages between different odour treatments, I studied the effects of odour composition on the flight behaviour of released tsetse. This work is described in chapter 7 .Marked G. pallidipes were released from a box at 10, 20 and 30 m downwind of an odour source. Tsetse were recaptured the same day on a 9 m wide wall of electric nets at 4 m upwind of the release box, or at the inside of an incomplete ring of eight nets at 4 m around the box. The incomplete ring was only used when tsetse were released at 10 m downwind of the source. Four odours were tested: natural ox odour; an artificial blend of acetone, octenol and phenols; the same blend with carbon dioxide and carbon dioxide alone.When tsetse were released at 10 m downwind of a source of natural ox odour, they avoided flying upwind. Dense vegetation did not influence their flight direction. Artificial odour did not influence the flight direction of tsetse released at 10 to 20 m downwind of the source in open vegetation, but tsetse avoided entering dense vegetation and preferred flying into crosswind oriented 'game trails'. About 35 % of the recaptured tsetse released at 30 m from the source of the artificial blend with or without carbon dioxide, flew within 20° of the prevailing wind direction. Tsetse released in the absence of odour or in the presence of the other odours flew away at random.The results suggest that tsetse change from straight fast upwind flight to more sinuous and slower flight when the flux of kairomones increases.In chapter 8 I studied the effect of the physiological state of tsetse on their host-location behaviour. G.pallidipes were marked and released from refuges and boxes placed at 10, 20 and 30 m downwind of an odour baited visual target. The physiological state of all tsetse recaptured on the same day, samples of the populations from which the marked tsetse originated and a sample of unmarked tsetse caught together with the marked tsetse, was assessed by analysis of their fat and haematin content. Before fat and haematin analysis, the females were dissected to determine their ovarian age and pregnancy stage.Unmarked tsetse caught with a target had on average fed 2.5 days previously. Among the females from the refuge, the variance in haematin decreased with progressing pregnancy. Females in the last stage of their pregnancy all had a high haematin content, indicating that most had fed less than 24 h ago before being captured. Tsetse recaptured from the refuge had a higher haematin content than unmarked tsetse. Recaptured refuge- leaving females were younger than unmarked females. The fat and haematin content were not related to the distance from which females were recaptured with the target. The females caught at a target baited with acetone, 1-octen-3-ol and phenols contained less fat than females caught at a target baited with the same odour plus C0 2 . The fat content of recaptured males leaving from a refuge, increased with distance of release. There was no such correlation with boxreleased males.It appears that haematin or the time since the last blood meal controls the search strategy of tsetse. Resting tsetse with high energy reserves will feed from a host that comes close by, but do not search actively for a host. The fat content, and in females also the pregnancy stage, influence whether tsetse will take a blood meal from a host that has been found.The research described in this thesis has resulted in an improved understanding of the foraging and host location behaviour of tsetse. It is now clear that tsetse can detect a host at 100 m distance and that long range host location behaviour of tsetse is influenced by odour flux rather than individual kairomones in the odour mixture. Furthermore, the effect of the physiological state of the tsetse on its behaviour is now more clearly understood. Targets do not catch all tsetse that are searching for a host. To ensure success, target control operations must be sustained, even when stationary baits no longer catch tsetse.