1. Climate and Tsetse: exploring the effect of climate variability and change on vector biology, population dynamics and distribution in the Zambezi Valley Human African Trypanosomiasis: alleviating the effects of climate change through understanding human-vector-parasite interactions John Hargrove, Glyn Vale, Lisa van Aardenne South African Centre for Epidemiological Modelling and Analysis, Stellenbosch University Climate Systems Analysis Group, University of Cape Town HEALTH & CLIMATE COLOQUIUM 2016

2. What makes the Rekomitjie case study special Studies in the area generally focus on correlations of disease and climate variables which may be quite complex The correlation between climate variables and vector biology, population dynamics and distribution is often more direct Where we have climate data we generally know little about the details of the vector And, often, where we have data on vectors we do not have adequate climate data For tsetse flies (Glossina spp) in the Zambezi Valley of Zimbabwe, however, we have both good climate data and extensive biological data

3. Tsetse and Trypanosomiasis Tsetse (Glossina) infest over 11 million square kilometres of Africa where they transmit trypanosomiasis Males and females feed ONLY on blood. Slow reproductive strategy. Only adults and pupal stages are available

4. Female ovulates Single larva develops inside the mother Larva deposited and buries itself in the ground Adult emerges and matures 20-40 days 8-15 days 8 days Tsetse life cycle

5. Various aspects of the population dynamics of tsetse flies are markedly affected by temperature: A) Pupal duration. B) Inter-larval period. C) Adult mortality. D) Abortion rate.

6. Given that the pupal duration is a closely controlled function of temperature – this duration changes sharply through the year As we shall see this has important implications for modelling population growth

7. The rate at which fat is used during pupal development increases linearly with temperature. Given the non-linear change in pupal duration with temperature this means that pupae developing at very high or very low temperatures use most fat – and will have the shortest time available as tenerals to find their first blood meal.

8. Rekomitjie Climate Rekomitjie Research Station 1988- 2000: monthly mean climatologies Temperatures peak in October/November, just before the onset of the rains. A single summer rainfall season which peak in January/February. An uninterrupted dry season from June to August. NDVI tracks rainfall with a maximum in February after which it decreases to a minimum at the end of the dry season.

9. Rekomitjie Climate Rekomitjie Research Station 1988- 2000: monthly mean climatologies Sprint (September – November) Environment is stressed High temperatures Low availability of water Lack of vegetation

10. Temperature Trends Daily maximum and minimum temperatures have both increased over time. However the magnitude and significance of the trends depend on the variable, season and the historic period. Minimum Temperature Maximum Temperature

11. Tsetse flies have an optimal temperature range (16 < tasmean < 32). The warming trends result in the upper threshold being exceeded more often and for longer periods, especially in the 1990s and after 2010. This has severe implications for the survival of immature stages and newly emerged adults Changes in the timing and duration of optimal temperature conditions

12. Tsetse flies have an optimal temperature range (16 < tasmean < 32). The warming trends result in the upper threshold being exceeded more often and for longer periods, especially in the 1990s and after 2010. This has severe implications for the survival of immature stages and newly emerged adults Changes in the timing and duration of optimal temperature conditions

13. Work in Zimbabwe and elsewhere has led to the identification of a number of odour attractants for tsetse. When used in conjunction with a trap this can produce large catches of tsetse We gauge vector population levels by the numbers of flies caught in odour-baited traps, run for three hours each afternoon

14. The protracted hot spell in October 1992 saw a spectacular decline in tsetse catches. This was followed in 1994 by a further heat-wave, coupled with the lowest annual rainfall on record. The tsetse population has, so far, never recovered to the levels of the late 1980s.

15. Vector borne diseases and climate change Previous work has used a differential equation approach to model trypanosomiasis and its control This work is base on the model developed by Rogers (1988) In the present work we look only to improve the modelling of the vector Rogers, D. J. (1988). A general model for the African trypanosomiases. Parasitology 97, 193-212. Hargrove JW, Ouifki R, Kajunguri D, Vale GA, Torr SJ (2012) Modelling the control of trypanosomiasis using trypanocides or insecticide-treated livestock. PLoS Neglected Tropical Diseases 6(5): e1615. doi:10.1371/ journal.pntd.0001615

16. Cohort modelling of tsetse populations The number of pupae produced on any given day depends on the number of flies that complete pregnancy on that day – and this again cannot be predicted accurately from the temperature on a given day. Rather we need to accumulate the percentage of development completed on each day according to the temperature on that day A pupa is taken to be produced on the day that this percentage exceeds 100% The pupal duration is estimated in an analogous fashion. The number of pupae surviving to emerge depends both on losses due to predation and parasitism (which may be density dependent) and on the temperature dependent losses consequent on the excess use of fat reserves during development

17. What has happened to tsetse populations at Rekomitjie Research Station over the past 55 years? Catches of G. m. morsitans have declined by at least 75% over the past 55 years Using the estimated effects of temperature on mortality derived from Antelope, and adding also density dependent mortality, simulation studies suggest that this population decline may indeed be attributable to increased temperature

18. Projected temperatures for the future at Rekomitjie Research Station If the current rate of temperature increase continues then we can expect that temperature in 2050 will be about 2 0 C higher than it was in 1960 What effect will that have on tsetse in the mid-Zambezi Valley?

19. Simulations suggest that Glossina morsitans morsitans may actually go extinct if temperatures continue to increase at current rates Conversely, with a warmer climate, we might expect tsetse to invade suitable habitat in highveld of countries such as Zimbabwe.

20. What's next? Can the results from this study site be transferable to other locations? What observed climate dataset can / should one use? What climate change projections should one use?

21. Observed climate dataset Watch Forcing Data ERA-Interim Long record (1979-2015) High temporal resolution (3-hour) Multiple variables (precipitation, temperature, wind speed, relative humidity) Low spatial resolution (0.5 degree)

22. GCM Projections Large ensemble Daily resolution Long records Multiple variables Multiple emission scenarios Low spatial resolution Rainfall not skillful Models disagree on amplitude and, for rainfall, the direction of change.

23. Empirical/Statistical downscaling Self-organising map based downscaling methodology Higher spatial resolution Not necessarily higher skill or reduced uncertainty Rainfall Models still disagree on the magnitude and direction of change No significant change relative to model variability is evident in many models Some models do show significant drying in the second half of the century.

24. Mean temperature: time of emergence Trends in downscaled projected mean temperature are all positive and significant. Anthropogenic signal emerges from the model’s variability The absolute anomalies vary primarily in magnitude. Possible bifurcation of models response towards the end of the century

25. Mean temperature: ideal temperature range A consistent message of decreasing frequency of days within the ideal temperature range within the Zambezi River, and increasing frequency over the interior of Zimbabwe.

26. Challenges and the future Modelling is of little value where it is not related to the real world and to real-world data And THE central data requirement for work on climate change is, of course, climate data – including current, historical and future projections data. Because if we do not understand the past we cannot hope to predict what is liable to happen in the future The modelling described here is “work in progress” We would like to extend the approach to model vector and trypanosome population dynamics simultaneously

27. Acknowledgements SACEMA receives core funding from the Department of Science and Technology, Government of South Africa. This work was supported by: UNICEF/UNDP/World Bank/WHO Special Programme for Research and Training in Tropical Diseases (TDR) European Union’s Seventh Framework Program (FP7/2007-2013) under grant agreement 221948 ICONZ (Integrated Control of Neglected Zoonoses) DFID/RCUK programme on Zoonoses and Emerging Livestock Systems (ZELS, BBSRC grant no.BB/L019035/1)