Study area
The surveys presented in this paper were conducted in the Hurungwe District (Mashonaland West Province, northern Zimbabwe; Fig. 1). In particular, a 110 km long transect was investigated, intersecting the southern escarpment of the Zambezi Valley. The lowlands north of the escarpment are protected areas and are home to a rich variety of wild animals, including elephant, buffalo, impala, warthog and cheetah. In this environment, wildlife provides the main source of blood meals for tsetse flies. The main tree species are Colophospermum mopani, Acacia nigrescens, and Adansonia digitata. The area is also characterized by a dense undergrowth of deciduous bushes [12]. On the highlands south of the escarpment, land use is characterised by mixed crop agriculture, livestock and by higher human population densities. Until recently, a game fence physically marked the boundary between the protected and settled areas. This fence is no longer regularly maintained along its length, but still represents the administrative limit of the protected areas. As such, it continues to mark a sharp discontinuity in land use. The study was conducted over the four seasons (hot-dry, warm-wet, cool-wet, and cool-dry) in Zimbabwe.
Design of tsetse surveys
Transect design
A transect from the floor of the Zambezi Valley (at an altitude of approximately 400 m), up the escarpment and ending on the highlands (at approximately 1200 m) was used for the survey. A random start point was selected for the first sampling location and an interval of 10 km was used for subsequent locations, providing a total of 12 sampling sites along the transect (Fig. 1). A combination of blue/black/blue-screen fly rounds [13, 14] and Epsilon traps were used for sampling.
Screen fly rounds
For each of the 12 sampling sites a 3 km long fly round perpendicular to the transect was carried out. Each fly round was divided into equal sectors of 100 m, which were marked on the ground with white paint. Geographical coordinates were systematically measured with the global positioning system (GPS). Fly rounds involved the use of a blue/black/blue screen attached to an aluminium pole that was baited with tsetse fly odour attractants (3-n propyl phenol, 1-octen-3-ol, 4-methyl phenol). The screen was carried by two men equipped with fly nets for catching all the flies landing on the screen [15]. The surveys were conducted simultaneously on all fly rounds for 5 h (from 08:00 to 10:30 h in the morning and from 14:00 to 16:30 h in the afternoon), taking into consideration the bimodal pattern of tsetse fly activity [16]. Some of our teams were working in National Park areas and limited transport prompted us to end the fly rounds at 16:30 h. This decision was mainly to allow teams to be retrieved from the park before it was dark to avoid encounters with wildlife after sunset. Eleven sessions, with each comprising two fly rounds, one conducted in the morning and the other in the afternoon of each sampling day, were simultaneously conducted at each of the 12 sites, starting on 26 February and ending on 18 November 2014. The study period covered both dry and rainy seasons.
Transect traps
In addition to fly rounds, stationary traps (Epsilon traps) were used to monitor tsetse flies along the transect. Two traps were deployed at each of the 12 sites where fly rounds were conducted. In particular, at each site one trap was located at the exact intersection between the transect line and the fly round line, while the second trap was deployed at a distance of 100 m from the intersection. Efforts were made to optimize trap visibility, by choosing more open sites where traps were not obscured by dense vegetation [17]. The 1 m radius clearings around the traps were intended to prevent damage by veld fires. All Epsilon traps were 90 cm tall and 1 m wide and consisted of phthalogen blue cloth with sections of black cloth inside the trap. The traps were baited with a 1:4:8 mixture of 3-n propyl phenol: 1-octen-3-ol: 4- methyl phenol [18]. Acetone was dispensed from 500 ml bottles with an aperture on the lid of 5 mm in diameter. Traps were maintained and monitored during the same period when fly rounds were conducted.
Additional traps
The results of fly rounds and trapping along the transect were complemented by additional tsetse fly sampling using Epsilon traps carried out in selected sites in the study area. Thirty-nine of these additional traps were located at sites belonging to a network of long term tsetse fly monitoring sites in the Hurungwe District, below and above the Zambezi escarpment. These traps were broadly clustered around five sites, namely Chirundu, Marongora, Makuti, Rukomechi and Mukwichi (Fig. 1). From this long-term (22 years) monitoring dataset, in this paper we present only the catches for a 2-year period (January 2014 to December 2015). In addition, eight additional traps were strategically deployed in suitable habitat in the area south the escarpment, where tsetse flies were expected to be either absent or present at very low densities based on local knowledge. These traps were monitored between December 2014 and March 2015. Apparent density was calculated as the number of flies captured in a trap as a fraction of the number of days the trap was operational.
Tsetse infection
All 92 flies captured along the transect between April and June 2014 (either during fly rounds or in traps) were examined for trypanosome infection with microscopy according to Lloyd and Johnson [19]. Dissection was also performed on 2000 randomly selected flies collected in one of the sites on the valley floor (Rukomechi), where additional traps were deployed (April 2014 and December 2015).
Statistical analysis
Statistical analyses were carried out using Minitab 17, R 3.1.2 and Statistica 8 software [20–22]. Analysis of variance (ANOVA) was conducted for both fly rounds and transect traps to assess whether the difference between means was statistically significant. Graphs for tsetse fly catches using both methods were generated to indicate the level of standard error. T-tests were conducted on morning and afternoon tsetse fly catches on fly rounds as well as on infection rates of tsetse flies captured along the transect compared to those at Rukomechi. Non-parametric tests (Wilcoxon rank sum test) were conducted on tsetse fly species caught in traps and on fly rounds due to the non-normal distribution of the data. Shapiro-Wilk normality test were conducted on data to test the hypothesis of data normality. Where the test indicated data normality, t-test was used and where it did not, the Wilcoxon rank sum test was used.