DEVELOPMENT AND USE OF A GENERAL, USER-FRIENDLY MODEL OF TSETSE CONTROL

MISE AU POINT ET UTILISATION D’UN MODELE DE LUTTE CONTRE LES TSETSE FACILE A UTILISER

G.A. Vale & S.J. Torr

Natural Resources Institute, University of Greenwich
Correspondence: G.A. Vale, 93 The Chase, Mount Pleasant, Harare, Zimbabwe.
E-mail: gvale@healthnet.zw.

Résumé

            Bien qu’il existe des modèles faciles à utiliser pour orienter les stratégies et tactiques de lutte contre de nombreux insectes tels que les moustiques et les mouches dorées des ovins, aucun modèle du genre n’a été mis au point pour lutter contre les tsétsé. Cette omission est inquiétante en raison des nombreuses méthodes alternatives actuelles de lutte contre les tsétsé et du fait que l’éradication à grande échelle est actuellement envisagée par des organisations, par ex. le Programme de Campagne panafricaine d’éradication des tsétsé et de la trypanosomose (PATTEC). Le « Natural Resources Institute », avec l’appui du Programme de production animale du DFID, est en train d’élaborer un modèle pour combler ce vide. Ce modèle, appelé « La muse à tsétsé » permet d’envisager l’utilisation d’appâts artificiels, de bovins traités à l’insecticide, de l’épandage au sol et de la pulvérisation aérienne ainsi que de la technique de l’insecte stérile (TIS) contre des populations circonscrites ou dispersées. L’utilisateur est en mesure d’ajouter au modèle toute une gamme d’intrants selon la dynamique et la répartition des populations de mouches, et selon la performance de base des mesures de contrôle. Il est alors possible de simuler l’impact des mesures utilisées seules ou en combinaisons temporelles ou spatiales, et d’évaluer les coûts et la durée requise pour la lutte. On a montré l’utilisation et les résultats du modèle.

Summary

            Although there are user-friendly models to guide the strategies and tactics for control of many insects such as mosquitoes and sheep blow-flies, there is no such model for tsetse. This omission is worrying in view of the many alternative methods of tsetse control and the fact that large-scale eradication is currently being considered by organisations such as the Pan-African Tsetse and Trypanosomiasis Eradication Programme (PATTEC). The Natural Resources Institute, with support from DFID’s Livestock Production Programme, is developing a model that fills the gap. This model, called Tsetse Muse, allows consideration of artificial baits, insecticide-treated cattle, ground and aerial spraying, and the sterile insect technique used against closed or open populations. The user is able to prime the model with a wide variety of inputs for population dynamics and distribution and for the basic performance of the control measures. It is then possible to simulate the impact of the measures, alone or in various temporal and spatial combinations, and to assess the costs and required duration of control. The use and outputs of the model are demonstrated. 

Problems

            The formation of the Pan African Tsetse and Trypanosomias Eradication Campaign (PATTEC) has renewed the debate about the optimal scale and organisation of tsetse control, and the relative merits of the various types of control techniques (Hargrove, 2003; Torr et al, 2005). Such debate has much of its old intensity, particularly regarding the sterile insect technique (SIT) and the strategy of “area-wide eradication” (Molyneux, 2001; Rogers & Randolph, 2002; Feldmann, 2004). Ideally, the arguments about all of the various strategies and tactics might be resolved by predicting the cost and efficacy of all options, using a standardised predictive procedure that is universally recognised as objective. Could such a procedure ever be found?

            The commonest predictive method is to examine past experiences, but these can be difficult to interpret.  For example, one is often required to determine how the experience with one species at a particular population density in a certain situation applies to another species at a different density elsewhere.  Moreover, the array of experience with the various control options is motley. Some operations were intended to eradicate the flies, others were designed merely to reduce their abundance, and yet others were required to produce barriers to invasion. Some operations were planned and supervised more carefully than others and were costed more completely. Hence, it can be possible to make of history whatever one wants.

            Modelling of various control options offers a complement to the historical approach, especially since we know much about the basic dynamics of tsetse populations (Hargrove, 2004). One joy of modelling is that it relies on fundamental arithmetical realities which, in principle, are beyond argument. However, modelling has its own special problems, as below.

Complexity.

            If the modelling is to be realistic it requires a complex computer-programme that most tsetse control officers have insufficient time to produce. Unfortunately, the simple growth equations reviewed by Barclay (2005) were designed for insects with non-overlapping generations, ie, creatures markedly different from tsetse. In particular, the equations do not consider the age structure of tsetse populations and the long time that it takes to adjust with control measures that reduce breeding instead of killing the flies.

Input assumptions.

            Despite our confidence about the basic biological inputs to use for tsetse, there can be much nervousness about the validity of other inputs that refer to the control measures themselves. For example, what is the death rate imposed by insecticides; what is the cost and sexual competitiveness of a sterilized male? Hence, can the modeller, like the historian, engineer any prediction he fancies?

Limited circumstances.

            Whereas modelling has addressed a variety of control measures in a range of circumstances (Hargrove, 2003), it has not embraced all of the many options that can be considered under a complete range of conditions. For example, what would happen in a particular area if separate parts of it were treated with natural baits, artificial baits, aerial spraying, ground spraying and SIT, using each at a particular specification, and if some parts were left untreated? The few specialist modellers cannot address everybody’s detailed questions.

Solutions

            A general solution to the modelling problems seems to be in making a model that can be employed by all to address their own particular concerns with any type or combination of options in any operational area, using their own input assumptions. Such a model could encourage three processes that reinforce each other. First, it would help to interpret the historical data. Second, it could assist directly the planning of operations in particular situations. Third, and more importantly for resolving the general debates, there would be the opportunity to test everybody’s input assumptions for a wide range of modelled operational circumstances, so leading to the clear and unbiased identification of basic principles that apply whatever inputs are used. The latter process would make the model a novel aid to teaching.

            One user-friendly model, called “Tsetse Plan”, has already been developed and is freely available at httb//www.tsetse.org. However, this model is intended primarily to assist NGOs and other inexperienced agencies to conduct small campaigns with baits. Hence, another, more sophisticated model to deal with the cost and effectiveness of baits, spraying measures and SIT has been drafted for use by specialist planners. This model allows a wide range of input opportunities, covering for example, the even or uneven distribution of tsetse and control measures, used against closed or open populations in favourable or unfavourable habits. The model is called “Tsetse Muse”, and is now available at the tsetse.org website.

            Although Tsetse Muse is presently a draft, it is working and has allowed the production of an article (Vale & Torr, 2005) that exemplifies the value of the programme by using it to compare the costs and effects of two distinct policies of tsetse control: reducing the birth rate and increasing the death rate. The comparisons suggest that reducing the birth rate is the least satisfactory option, and that SIT is a particularly inappropriate form of this, mainly because it takes an inevitably long time to become effective and gives no direct protection against invasion.

            The programme can elucidate many other matters. This is best established by the reader’s own exploration of the programme. To encourage such exploration some further examples of outputs are provided below. Each of these examples considers a population of 2500 males and 5000 females/km2, at an average temperature of 25oC, with a mean daily displacement of 249m for males and 367m for females (Vale & Torr, 2005). Before treatment by baits this population was assumed to be evenly distributed in the operational area and the adjacent invasion source, although in previous modelling (Vale & Torr, 2005) the population was distributed unevenly. An imaginary line, called the invasion front, separated the invasion source and operational area. The population was considered to be eradicated where its density dropped to <0.1 females/km2.

Invasion barriers.

            The efficacy of baits in neutralising invasion of a treated area was one of the first matters considered by professional modellers (Hargrove, 1993). It is pertinent, therefore, to see that the present model can produce outputs of the same type. The specimen outputs now generated by Tsetse Muse (Fig. 1) refer to the stabilised distribution obtained after operating baits evenly within the operational area. Not surprisingly, the greater the daily kill rate imposed by baits the narrower the invaded area. A kill rate of 10% per day, produced by treating 3-4 large cattle/km2 (Vale & Torr, 2005) meant that the invaded area was 9km wide.

            While there is nothing new about these types of indications for barriers, it is novel that anybody can now produce the indications quickly, for whatever control technique and whatever input assumptions are selected.

Residual pockets.

            Let us say that in most of the operational area there are enough cattle to allow a kill rate 10% per day. Let us also say that no cattle occur in a band, 5km wide, that runs parallel to the front at 17-22km inside the operational area. Hence, tsetse are not killed within this band. Despite this, the flies do disappear from the band (Fig. 2) since once the bait campaign begins the number of flies diffusing into the band from the treated area becomes less than the number diffusing out, and the population in the band cannot breed quickly enough to offset this net emigration. The population in the band is eliminated after 4.2 years, as against the 0.5 years it would have taken if the cattle treatments had been evenly applied throughout the whole operational area, ie, inside and outside the band.

Fig. 1

The flies are eliminated from the untreated band in 0.7 years if it is only 1km wide, because the narrower the band the greater the proportion lost per day by outward diffusion. On the other hand, elimination takes 13 years if the band is 7km wide, and tsetse are never eliminated entirely from bands wider than this. All of the above bands were centred at 19-20 km from the front, but the programme allows untreated bands to be located anywhere. The programme also allows exploration of the most economical ways of avoiding or removing the residual infestations. An effective expedient is the deployment of artificial baits down the centre of the bands, it being unnecessary to treat the whole band.

Fig. 2.

Mosaic control.

            Let us imagine that (i) insecticide-treated cattle are used in the first 10km from the front, and also at <30km from the front, and (ii) that the 20km-wide band in between is treated by four aerial applications of non-residual insecticide at 13-day intervals. The cattle treatments and the aerial spraying start at the same time. Each of these control techniques is, in itself, highly effective when applied singly and uniformly against an isolated population. However, when applied in the mosaic now envisaged, the results are disastrous (Fig. 3). Although the flies in the centre of the aerially sprayed block were eliminated after 40 days, when the spraying was complete, flies were still present in the cattle-treated areas then. Some of these flies invaded the aerially sprayed block and started breeding – indeed, some even invaded and deposited pupae during the 40 days. Hence, the population in the aerially-sprayed block increased rapidly to provide a stream of invaders into the cattle-treated areas, so preventing control reaching its full potential there. The upshot was that the flies were eliminated only at distances >36km from the front, not at the distances >9km if the cattle treatment had been used throughout (Fig.1, 10% kill).

            The above types of problem threaten whenever the control measures to be combined operate at contrasting rates or at different times of year. The programme allows such problems and their solutions to be investigated. For example, it shows how the disasters can be avoided by: (i) overlapping the measures, (ii) starting the baits well in advance of the spraying and (iii) choosing where possible the measures that are the most compatible. For example, if each control measure is started at the same time, then insecticide-treated cattle and artificial baits are the most compatible pair; aerial spraying and SIT are the least compatible pair since they work at such different rates.

Refinement

            It is hoped that Tsetse Muse will evolve with contributions by all interested persons, so enhancing the services offered, improving its authority and broadening its ownership. For example, it would be useful to expand the model’s treatments of costs and trypanosomiasis risk, and to add a section on economic analysis. Moreover, it could be particularly instructive to address better one of the great gaps in our knowledge: the dynamics of low density populations. At some point the population must collapse of its own accord, if only because the females cannot find mates (Glasgow, 1963).

Fig. 3

            Since we do not know what the critical density is under various circumstances we do not know at what density we could stop control. Moreover, even if we did know that density our uncertainty about sampling efficiency with sparse populations means that we cannot be sure when that density has been reached. In consequence, do we now continue control operations for much longer than need be, over areas much greater than necessary? Certainly, in Zimbabwe it is usual to maintain baits for at least a year after the last fly has been caught (W. Shereni, personal communication, Harare, 2005) and the same type of thing happened with SIT on Unguja Island (Vreysen, 2000) The problem is further illustrated by Fig. 2: could we stop control outside the pocket after 3 years, or should we wait for the currently suggested 4.2 years, or longer? A confident answer could allow us save many thousands of dollars on each campaign, and many millions Africa-wide.

            The science of low density populations requires stochastic modelling (Hargrove, 2005) to supplement the deterministic approach of Tsetse Muse. More problematically, where can we find a low population of tsetse that maintains itself steadily for the long time required for us to catch enough flies to elucidate matters such as mating success and sampling efficiency, so telling us what inputs to use in models? What tools can we develop for routine monitoring of mating success and sampling efficiency during actual campaigns? One of the most promising possibilities is the use of sterilised male and female tsetse, released in steady known numbers – there would be no chance of their natural increase, and their prior treatment with trypanocides might make then innocuous (Moloo & Kamunya, 1987). Being able to track the individual flies continuously, by harmonic radar (Riley et al., 2005), would be a huge help to the basic studies involved.

Conclusion

            While Tsetse Muse is far from complete as an aid to teaching and creating the plans for tsetse control, it does seem to be a substantial move in the right direction. The extent to which it can assist in settling the various planning debates depends on the degree to which it is refined and accepted by the tsetse control community. The authors have attempted to construct it objectively and to use it with a range of inputs that seem to be the most realistic available, but the resulting outputs may not suit everybody. The authors would be delighted to amend the model in the light of any closely justified criticisms and suggestions. If no such detailed comments are openly offered it might be widely presumed that the model’s outputs are materially correct.

            The immediately important implications of previous outputs have been the warnings (Torr & Vale, 2005) about the proposed use of SIT in integrated management of tsetse populations (Feldmann, 2004). Present outputs show that even with various insecticidal techniques the planning of integrated campaigns can be difficult. On the other hand, Tsetse Muse can help to identify the best integration systems, and to find appropriate methods of avoiding or removing residual pockets of infestation.

            The outputs may well be disappointing to SIT enthusiasts. However, the field of tsetse control is large enough for all to find a niche, provided there is enough lateral thinking. For example, even if SIT is inappropriate as a complete package, the technique does have some individual components that could assist greatly with the economy of other control measures.

Acknowledgements

            The authors are grateful for encouragement from Drs J. Kabayo and S. Haille-Mariam. The Animal Health and Livestock Production Programmes of DFID funded the work but do not necessarily endorse the views expressed.

References

Barclay, H. J. (2005). Mathematical models for the use of sterile insects. Sterile Insect Technique. Principles and Practice in Area-Wide Integrated Pest Management. IAEA, Austria (in press)

Feldman, U. (2004). The sterile insect technique as a component of area-wide integrated pest management of tsetse. The Trypanosomiases. Maudlin et al., (Eds)., CABI, Wallingford, UK, 614pp.

Glasgow, J.P. (1963). The distribution and abundance of tsetse. Pergamon, Oxford, UK

Hargrove, J.W. (1993). Target barriers for tsetse flies (Glossina spp.) (Diptera: Glossinidae). Quick estimates of optimal target densitiesand barrier. Bulletin of Entomological Research  82, 197-200.

Hargrove, J.W. (2003). Tsetse eradication: sufficiency, necessity and desirability. DFID Animal Health Programme, Edinburgh UK, 133pp.

Hargrove, J.W. (2004). Tsetse population dynamics. The Trypanosomiases. Maudlin et al., (Eds). CABI , Wallingford, UK 614pp.

Hargrove, J.W. (2005). Extinction probabilities and times to extinction for populations of tsetse flies Glossina spp. (Diptera: Glossinidae) subjected to various control measures. Bulletin of Entomological Research 95, 1-9.

Moloo, S.K. and Kamunya, G.W. (1987). Suppressive action of Samorin on the cyclical development of pathogenic trypanosomes in Glossina morsitans centralis. Medical and Veterinary Entomology 1, 285-287.

Molyneux, D.H. (2001). Sterile insect release and trypanosomiasis control: a plea for realism. Trends in Parasitology 19, 413-414.

Riley, J.R., Greggers, U., Smith, A.D., Reynolds, D.R. and Menzel, R. (2005). The flight paths of honeybees recruited by the waggle dance. Nature  435, 205-207.

Rogers, D.J. and Randolph, S.E. (2002). A response to the aim of eradicating tsetse from Africa. Trends in Parasitology 18, 534-536.

Torr, S.J., Hargrove, J.W. & Vale, G.A. (2005). Towards a rational policy for dealing with tsetse. Trends in Parasitology (in press).

Vale, G.A. and Torr, S.J. (2005). User-friendly models of the costs and efficacy of tsetse control: application to sterilizing and insecticidal techniques. Medical and Veterinary Entomology 19, 293-305.

Vreysen, M.J.B., Saleh, K.M., Ali, M.Y., Abdulla, A.M., Zhu, Z-R., Juma, K.G., Dyck, V.A., Msangi, A.R., Mkonyi, P.A. and Feldmann, H.U. (2000). Glossina austeni (Diptera: Glossinidae) eradicated on the island of Unguja, Zanzibar, using the sterile insect technique. Journal of Economic Entomology, 93, 123-135.