Optimising targets for tsetse control: Taking a fly’s-eye-view to improve the colour of synthetic fabrics

Abstract

The savannah tsetse flies, Glossina morsitans morsitans and G. pallidipes, are important vectors of Rhodesian human African trypanosomiasis and animal African trypanosomiasis in East and southern Africa. We tested in Zimbabwe whether robust, synthetic fabrics, and innovative fly’s-eye-view approaches to optimise fabric colour, can improve insecticide-treated targets employed for tsetse control. Flies were caught by electrocution at a standard target comprising a 1m x 1m black cotton cloth panel with 1m x 0.5m black polyester net panels on each side. Catches were subdivided by species and sex. Tsetse catches were unaffected by substitution of the black cotton with a blue polyester produced for riverine tsetse targets. Exchanging the net panels for phthalogen blue cotton to simulate the target routinely used in Zimbabwe significantly reduced catches of female G. m. morsitans (mean catch 0.7 times that at standard), with no effect on other tsetse catches. However, significantly greater proportions of the catch were intercepted at the central panel of the Zimbabwe (means 0.47–0.79) versus standard designs (0.11–0.29). We also engineered a new violet polyester cloth using models of tsetse attraction based upon fly photoreceptor responses. With and without odour lure, catches of females of both species at the violet target were significantly greater than those at standard (means 1.5–1.6 times those at standard), and typical blue polyester targets (means 0.9–1.3 times those at standard). Similar effects were observed for males under some combinations of species and odour treatment. The proportions of catch intercepted at the central panel of the violet target (means 0.08–0.18) were intermediate between those at standard and typical blue polyester. Further, the reflectance spectrum of violet polyester was more stable under field conditions than that of black cotton. Our results demonstrate the effectiveness of photoreceptor-based models as a novel means of improving targets to control tsetse and trypanosomiases.

Evaluation of improved coloured targets to control riverine tsetse in East Africa: A Bayesian approach

“…The putatively improved fabrics tested in this work were designed rationally with that intention, and thus prior knowledge clearly formed the basis for our work but was not represented in the above analysis. The design of our violet fabric was based upon fly photoreceptor signals, which can be calculated from fabric reflectance spectra, and provide a discrete set of colour metrics that describe fabric colour properties relevant to flies [8][9][10]. We thus extended our analyses to incorporate these aspects.…”

“…Prior to our work, a large dataset of G. f. fuscipes catches at 37 differently coloured Tiny Targets had been published, comprising 15 separate experiments conducted at the same Kenyan study location as our experiments [4]. By calculating the fly photoreceptor signals that would be elicited by the fabrics used in that study, we previously estimated the ways in which those photoreceptor signals related to the tsetse catches reported, and these models were used to guide the design of our violet fabric [8][9][10]. Thus, we had a clear prior expectation of how fly photoreceptor signals would affect tsetse catch, based upon our earlier work and a large set of pre-existing experimental data.…”

“…The predictors R7p, R7y, and R8y are the three photoreceptor signals found to be most influential upon tsetse attraction in an earlier analysis of these data [8][9][10] (here we have used generic fly photoreceptor sensitivity functions, as recorded in Musca and Calliphora [12]). Since the data available in the prior publication were totals across 10 sampling days per experiment [4], an offset was included to transform predictions to a daily scale and a random intercept for experiment (rather than site and day) was modelled.…”

“…SP23 (P)). We also tested a number of violet fabrics (Fig 1B ): our original violet fabric as tested against savannah tsetse in Zimbabwe [10] (violet; produced using CI Disperse Violet 57 at a dye bath concentration of 7%); and a number of violet fabrics with varying depths of the same dye that were produced as small samples and were unset, resulting in more crinkled fabrics (labelled (P) for prototype to separate them from the fabric previously tested). Based on our prior analysis of fly photoreceptor signals [8,9], we PLOS NEGLECTED TROPICAL DISEASES expected tsetse attraction to increase with dye concentration among these three prototype violet fabrics.…”

“…The above analyses by experiment suggest trends in target performance, but these were obscured by high variability in tsetse catches, and were sometimes inconsistent between experiments. Those analyses used regularising priors, which means that they took no account of prior expectations of target performance, despite the fact that such expectations were overtly drawn from earlier work and informed the design of the targets tested [8][9][10]. We next took advantage of the ability of Bayesian analyses to incorporate these prior expectations and consolidate patterns across experiments.…”