Title: Seasonality and the Factors that Lead to Multi-annual Cycles through Resonance in a Model of Malaria

The potential of seasonal forcing in determining multiannual epidemic outbreaks is well known in epidemiological models. In particular the epidemic cycles of childhood diseases can be understood using a forced SIR model. In this talk I will introduce this phenomenon and explain a method that makes it easy to examine the likelihood of different resonance outcomes in these models. I will then discuss a baseline Malaria model that uses these approaches to ask the question of what characteristics may lead to multiannual cycles through resonance in this system. I will then turn to an example where seasonal forcing can explain the evolution of a parasite strategy of latent infection. In both examples, the models emphasise the importance of seasonality to the ecology (epidemiology) and evolution of disease.

Title: Identifying and Isolating Influences on Regional Climate: Implications for Disease Prediction

Recent studies arising from both statistical analysis and dynamical disease models demonstrate a link between the incidence of cholera, a paradigmatic water-borne bacterial illness endemic to Bangladesh, and the El Niņo ? Southern Oscillation (ENSO). We use a regionally coupled, or pacemaker, climate model to investigate physical links between ENSO and the regional climate of Bangladesh. We find that, following winter El Nino events, summer monsoon rainfall typically increases over Bangladesh, providing a physically plausible link between this dominant large-scale climate forcing and the regional scales that are more relevant to disease dynamics. We also demonstrate that the pacemaker methodology can be used to identify additional large-scale forcings associated with regional climate variations through the application of nonparametric statistical techniques. We find strong associations between variations in Bangladesh rainfall and summer Indian Ocean temperature anomalies, suggesting an additional source of potential predictability for cholera risk.

Title: Seasonality, Climate Change and the Dynamics of Infectious Diseases of Wildlife

The dynamics of infectious diseases in the wildlife may exhibit a strong climatic footprint. In many cases, outbreaks are clearly synchronized with seasonal fluctuations in temperature, humidity and rainfall patterns. Climatic fluctuations can affect the infective agent either directly by modifying the life-expectancy of free living stages or, indirectly, through changes in behaviour, demography (timing of reproduction, mortality, etc.), abundance (birth pulses, resources availability) and immune response of the host. This may result in turn in a change of probability of transmission between susceptible and infected animals or between susceptible hosts and infective stages/propagules. In the present work we illustrate two examples of how seasonality in meteo-climatic variables can affect the dynamics of infectious diseases caused by micro and macroparasites. In the first case, we investigate how seasonal fluctuations in demography of the host affect the dynamics of rabies epidemics and we show how short-living, fast-reproducing host species may respond to seasonality differently than long-living, slowly reproducing ones. The second example is about the effect of seasonality in the development of hypobiosis (arrested stage of development of parasite larvae in the gut mucosa of the definitive host), a strategy carried out by a number of nematodes species to overcome harsh environmental conditions - such as extremely drought summers or very cold winters - during which survival of free-livings stages (and, thus, the probability of infection) is low or negligible. In both cases, modifications in the seasonal forcing due to anthropogenic climate change have the potential to alter the parasite burden, the prevalence of infective individuals and the abundance of the host. In the temperate areas of Europe, for instance, winter seasons have been progressively milder in the last 30 years and possible effects both on host demography and on the transmission of infectious diseases have begun to be observed.

Title: Predictability of Climate Indices Related to Infectious Diseases

This talk reviews the extent to which certain climate indices related to infectious diseases can be predicted. Much of this review is made possible by a new statistical technique, called predictable component analysis, that optimizes the dectection of predictability and greatly clarifies the spatial structures that can be predicted. This new technique is discussed, and reasons why this technique is superior to previous techniques will be explained. Special attention is focused on quantifying the degree to which state-of-the-art prediction models can predict precipitation and temperature anomalies in Africa, a region in which a connection between climate and infectious diseases is believed to exist.

Title: Using Scenarios to Plan for Climate Uncertainty

Scenarios are important tools for long-term planning, and there is great interest in using integrated models in scenario studies to plan for climate uncertainty. However, scenario definition and assessment are creative, as well as scientific, efforts. Using facilitated creative processes, an interdisciplinary group at the University of Arizona Science and Technology Center for Sustainability of Semi-Arid Hydrology and Riparian Areas (SAHRA) has worked with stakeholders to define regionally significant scenarios that encompass a broad range of hydroclimatic, socioeconomic, and institutional dimensions. The regional scenarios subsequently inform the definition of local scenarios that work with context-specific integrated models that, individually, can address only a subset of overall regional complexity. Challenges in defining and constructing scenarios include stakeholder skepticism of complex integrated models, communication with multidisciplinary teams, tradeoffs between considering many scenarios rather than a baseline and few variants, and downscaling global and regional possibilities to meaningful local corollaries. Our experience highlights the need for a structured approach to scenario studies, including clear identification of model assumptions, strengths and limitations of models and historical observations, and uncertainties associated with scenario results, as well as comparison of scenario definitions and results across multiple studies. When considered as one of many important issues faced by decision makers, climate variability and change can be effectively integrated into risk management strategies.

Title: Climate and Health: Prediction from Seasonal to Climate Change Time-scales

Under the European Commission's 6th Framework Programme, a 5-year project, known as Ensemble-based Predictions of Climate Changes and their Impacts (ENSEMBLES), was initiated in 2004, and is a major undertaking by more than 60 institutes from 20 countries, mainly in Europe. Within the project an ensemble climate forecast system is being developed for use across timescales ranging from seasonal to centennial, and for spatial scales ranging from global to local. The model system will be used to construct probabilistic climate forecasts and scenarios, and will be used to drive a wide range of application models including those for public health. In the past, assessments of the impacts of climate variability and change have generally considered only changes in climate means, but in numerous applications, particularly health, it is the climatic extremes that are more important. The focus on extreme events in the ENSEMBLES project will be particularly useful in this regard. In this presentation an overview of the ENSEMBLES project, with specific focus on the prediction of health impacts will be provided.

Title: Risk Maps for Lyme Disease Emergence in Canada: Vector Biology, Climate Change and Public Health

Climate change in anticipated to cause global changes in the risk from vector-borne diseases. Here I discuss the ways that climate change can impact on vector-borne disease occurrence and risk, and focus on Lyme disease as a pertinent example. The USA has suffered an epidemic of Lyme disease, which began in the late 1970s and peaked in 2002 when over 21000 cases were reported. In Canada, Lyme disease is an emerging infection due to recent expansion of the range of the tick vector Ixodes scapularis, which may in part be due to a warming climate. A comprehensive understanding of the ecology of I. scapularis and its hosts has allowed us to predict the scope and direction of potential range expansion of I. scapularis under current climate conditions. With field information on tick dispersion by migratory birds we have developed risk maps for the occurrence of I. scapularis populations that are validated in the field. We are now using the risk maps to project how climate change may affect the dispersion of I. scapularis into Canada. The public health objectives of the risk maps are clear: rational targeting of surveillance and intervention effort to provide early warning and control respectively. However, the extent to which ecological information can integrate with the different spatial scales and quality of input information (including projected temperature data, knowledge on I. scapularis-endemic areas), in a way useful for public health purposes, requires continued study.

Title: Here and here; now and then. How we incorporate Space and Time into epidemiological understanding

Space, time and space-time are important ingredients of disease risk that have historically often been ignored in describing and explaining epidemic outcomes. The debate about whether or not highland malaria in Africa increased due to global warming shows clearly that statements made at, or about, one spatial scale may be inappropriate at different spatial scales. The spread of bluetongue in Europe is probably the best example of a vector-borne disease changing its distribution because of climate change; the disease's distribution changed only in certain places, and at times and in ways that we would expect from our current knowledge of its environmental dependence. Bluetongue also gave us a surprising example of the adaptation of a disease to the changing situation, by transferring from its traditional Tropical vectors to new, Palaearctic vectors in the European region, thus potentially greatly increasing its range.

In trying to understand malaria, blue-tongue and a whole range of other diseases we can build simple mathematical models that attempt to incorporate key features of space and/or time that are felt to be important. Compartmental R0 models were used in the UK foot and mouth outbreak modelling, and are being developed for a number of other livestock diseases. We can challenge the simplistic assumptions of such models by examining real environmental variability at scales appropriate for each disease and/or vector in turn; an example is given of how spatial wavelet analysis can show us the environment as 'seen' by a disease or its vector. Temporal disease models can switch from one sort of dynamical behaviour to another with a slight change in a key variable or parameter. Whilst we have useful ways of capturing environmental seasonality (i.e. temporal variability) as detected by satellites, the same statistical approaches are not very revealing when applied to disease data. A method derived from life-table analysis of animal populations is proposed that examines the changing relative contributions of each transmission parameter through time.

All statistical and mathematical methods of understanding disease outbreaks involve simplification of the real-world story. With simplification often comes corruption. Analysis usually treads a fine line between simplification and corruption.

Title: Hydrologic Monitoring and Modeling of Mosquito-Borne Disease Transmission

In recent years the effect of climate change on human health, including infectious disease, has become an issue of increasing interest. I argue that a true understanding of the connection between climate and infectious disease requires a better understanding of the processes through which local environmental variability affects disease system ecology. I use the example of West Nile virus transmission to illustrate how the identification of local disease system response to the environment can be used to develop larger-scale and longer-term prediction of disease outcomes. Multiple methods are used in this study, including physical, ecological and statistical models, and the operational use and policy implications of the findings are discussed.

Title: Forecasting Parasitic Disease in Domestic Animals: an Historical Perspective

This talk will focus on Fasciola hepatica, (the common liver fluke) a digenean parasite of sheep and cattle (and , not infrequently, people) throughout the temperate world. F. hepatica causes serious and sometimes fatal disease in the host it infects. It has been the subject of numerous modeling studies over the last 50 years and because all the extra-mammalian stages of the parasite life cycle are exquisitely sensitive to the vagaries of the microclimate they inhabit, F. hepatica provides an ideal context for discussing the modest successes and extravagant failures that have attended attempts to link climate with disease.

The earliest, and simplest, forecasting models for fascioliasis appeared in the late 1950s. they were straightforward attempts to measure soil surface moisture during those months when the temperature permitted the development of the extra-mammalian stages. Long developmental time lags and the existence of an intermediate host which provided a refugium from adverse microclimates meant that it was possible to assess the likelihood of serious disease in grazing hosts some two or three months before it could actually be recognized. These systems, calibrated using 200 years of accumulated climatic and disease data, provided reasonably reliable warnings on a regional basis. Multiple variations on the original model were devised, and some of them are still in regular use fifty years later.

With the advent of greater computing power, a surge of papers on the affect of climate on the development and mortality of the free living stages, and the entrance of ecologists into what had previously been a largely veterinary enterprise, there was a succession of attempts to refine these forecasting systems using detail models of the population biology of the parasite. The intent was to create models that could be used on a much more local basis (at individual farm level, if possible) to predict the likely level of disease and at the same time offer plausible defensive strategies using a combination of grazing management and drug therapy. All of these attempts failed.

Occasionally, we still see claims that such dynamic models could manage this, but the prospect (at least in this context) seems bleak. On the other hand, the effort was not a complete waste. Models detailing the population biology of F. hepatica, still find use in evaluating the relative effectiveness of chemotherapeutic and chemoprophylactic strategies. They also have explanatory value with respect to the development of anthelmintic resistance.

Title: Inferring Local Climate from General Circulation Model's Projections. An Overview of Statistical Downscaling Approaches.

I will present a number of statistical downscaling techniques that are used as alternatives to dynamical downscaling approaches through high resolution regional models. There are simple techniques that amount to more or less sophisticated interpolation of the relative coarse GCM output. There are approaches that "train" a statistical relation between large scales and small scales by observed records and apply the estimated functional form to model output. There are resamplingappraches and weather generator solutions. I will use two examples from recent work with colleagues at NCAR and at University of Newcastle, UK, to anchor my presentation but I will try to provide a good representation of the wide range of approaches out there as well.

Title: Spatial and Temporal Patterns of Malaria in Africa at Landscape and Continental Scales

I first explore observed and projected future spatial and seasonal patterns of malaria occurrence in Africa using a simple, continental grid model (0.5 degrees lat-long) of climatic suitability that is driven by temperature and precipitation only. A complex pattern of changed climatic suitability emerges, which may be indicative of the direction of regional changes in malaria transmission. Changes in rainfall are at least as important as increased temperature. How might such changes be manifest at the landscape scale? Using hydrological models, we have shown that surface flow accumulation is an important predictor of malaria transmission in an upland area of Tanzania, in addition to altitude. We have also used optical and microwave remote sensing to map fine grain spatial (<50m) and temporal (monthly) patterns in the distribution of mosquito breeding habitat in lowland areas of The Gambia and Tanzania, and subsequently to estimate malaria transmission. Breeding habitat is patchy, leading to highly variable transmission across relatively short distances in these landscapes. Demonstrating that more mosquitoes are found near breeding habitat is hardly new. What is new is the possibility of mapping these features across very large areas and incorporating this information into transmission models. In addition to temperature dependencies, this is a key challenge to understand the impact of climate change upon local vector dynamics and malaria transmission in Africa.

Title: Influence of Climate on the Ecology and Evolution of Host-pathogen and Vector-parasite Interactions

Most studies of host-pathogen and vector-parasite systems assume key epidemiological parameters such as virulence to be constant across time and space. However, virulence is not a constant but is a dynamic outcome of the interaction between pathogen growth and host defence. For systems involving ectotherm hosts and/or vectors, virulence and resistance are strongly environmentally context dependent. Moreover, the responses are not necessarily linear or immediately predictable, because they derive from a complex 'genotype-by-genotype-by environment' interaction. Even subtle non-linearities and asymmetries between temperature responses of host and parasite mean that very small changes in temperature can alter the net outcome of an interaction considerably. Such effects can have profound consequences for disease dynamics and host-parasite coevolution. To better understand the ecology and evolution of host-pathogen interactions and predict the consequences of climate change we need to consider not just basic measures of ambient temperature, but the fine-scale thermal environment in which the host?parasite interaction is actually played out.

Title: Climate Information for Public Health Decision Making

African countries striving for socio-economic development are increasingly aware of the challenge imposed by climate variability and change. It has been argued at the highest level that Africa is one of the most vulnerable regions in the world to the impacts of a variable and changing climate. However, todate , most national development plans, poverty reduction strategy papers and sectoral strategies in climate-sensitive sectors such as health have paid minimal attention to climate variability and even less to climate change. The failure to address the importance of managing climate variability may impact severely on the achievement of national and international development targets such as the Millennium Development Goals; many of which are directly or indirectly related to climate and, or, health.

Success in achieving such goals requires not only financial investments but also improved information and analysis tools and the capacity to bridge the gaps between information suppliers' and information users' needs. In the case of public health it is recognized that many diseases are climate sensitive: sensitive to seasonality, year to year variations and trends. A developing body of work is also showing the importance of using climate information in impact assessments for health interventions where the outcome or the intervention is climate sensitive.

In 2005 a 'Gap Analysis for the Implementation of the Global Climate Observing System Programme in Africa' was commissioned by the UK's Department for Internatioanl Development to better understand how improvements in climate data and climate services could impact on development outcomes. The analysis identified four major areas for investment. These are:

   1) Integrating climate into policy
   2) Integrating climate into practice at scale
   3) Climate services
   4) Climate observations

In this paper we present demonstrations of an integrated approach to improving climate services and climate risk management uptake in the health sector in Africa using examples taken from malaria, meningitis and onchocerciasis control initiatives.