Introduction
The common phrase ‘the devil is in the detail’ implies that sketchy, big-picture thinking is of little use when dealing with real, practical problems. And yet, predictions of climate change and its effects on parasites and disease are inherently uncertain, so that vague warnings of increased risk often seem the default position. In confronting possible changes in parasite epidemiology in an uncertain world, novel strategies for risk-based control wisely remain flexible. However, this flexibility can appear to be either vague or over-complicated to advisors and producers who have become accustomed to well-defined protocols for managing parasite infection.
In fact, both science and application in this area should remain mindful of the diversity of metazoan parasites of veterinary importance, and the complexity of their life cycles and interactions with their hosts, the environment and each other. The behavior of parasite-host systems in changing environments emerges from the fine detail of biological and anthropological characteristics. For example, the spread of ticks and tick-borne diseases is driven not only by climate change, but also by habitat modification and human activities (Leger et al., Reference Leger, Vour'h, Vial, Chevillon and McCoy2013), while the seasonality of tick activity in relation to host migration also appears crucial to timescales of disease emergence (Ogden et al., Reference Ogden, Lindsay and Leighton2013). Predictions that ignore such complexities will usually fail. However, producers and advisors need simple solutions on the ground, and scientists seek generality in their predictions. The tension between general rules of behavior and system-specific properties runs throughout ecology, and disease ecology is no different. Using selected examples, focusing mainly on helminth infections in grazing livestock, this short paper seeks to show how detail, and our knowledge, ignorance and treatment of it, is core to our ability to predict and attenuate the effects of climate change on parasitic disease.
Biological detail
Macroparasites usually have life cycle stages outside the definitive host, whose rates of development and survival are affected by climatic variables such as temperature and moisture. Since parasitic disease often depends on accumulated levels of infection, small changes in these rates could, over time, have large effects on parasite burdens and animal health. Relationships between climate and life history parameters differ between species, and in some cases have apparently been elucidated in some detail. However, this impression is often illusory. Many published experiments test the effects of constant temperatures on parasite stages that are protected from environmental complexity e.g. at fixed temperatures without humidity limitation. More variable conditions in the natural environment could lead to very different behavior, undermining predictions of the effects of climate change on infection pressure. Estimated larval mortality rates of cattle nematodes in the field, for instance, were two to three-fold higher than suggested by laboratory studies (Grenfell et al., Reference Grenfell, Smith and Anderson1986). This could be due to unmeasured or unknown additional environmental factors, such as ultraviolet irradiation (van Dijk et al., Reference Van Dijk, de Louw, Kalis and Morgan2009). The factors and mechanisms behind such discrepancies are often unknown. Predictions of the seasonal abundance of the tick Ixodes ricinus, e.g. found that models developed for the United Kingdom predicted the field data in Spain only if a fitted parameter was added to take account of presumed differences in winter diapause behavior (Dobson et al., Reference Dobson, Finnie and Randolph2011).
Spatio-temporal heterogeneity in microclimate is increasingly acknowledged to drive important epidemiological outcomes. Recent discoveries in this area include the importance of temperature fluctuation on extrinsic incubation periods of malaria parasites in mosquitoes (Paaijmans et al., Reference Paaijmans, Read and Thomas2009) and, in livestock, the importance of rainfall amount and distribution over short timescales on the development and availability of infective trichostrongylid larvae (O'Connor et al., Reference O'Connor, Kahn and Walkden-Brown2008).
Life history parameters can vary between populations of the same parasite species. Predictive transmission models might consequently have to use locally estimated parameters, while projections of future epidemiology under climate change might need to explicitly consider parasite evolution. Thus, e.g. the intestinal nematode of sheep, Nematodirus battus, appears to vary little in core parameters such as temperature thresholds for development, but shows distinct hatching behavior in different parts of the UK, which appears to be adaptive to local climate (van Dijk and Morgan, Reference Van Dijk and Morgan2010). This illustrates that parasites could adapt to climate change quickly and using subtle mechanisms, which are not obvious or easily measured. Crucially, statistically fitting model predictions to data by varying biologically obtuse and experimentally unverifiable scale parameters might provide comfort, but will be of little use when extrapolating models to explore climate change scenarios, because the underlying mechanisms will remain only partially understood and model validation restricted to the range of observed data.
Management detail
Effects of climate change on parasite biology could be exacerbated or counteracted by livestock management, and in some cases even large biological effects can be negated by small management changes. For myiasis in sheep in temperate regions, caused by Lucilia sericata, warmer spring temperatures are predicted to bring forward emergence of adult flies from pupae and dramatically increase risk of blowfly strike in adult sheep at that time of year, but disease would not manifest if either treatment or shearing were also advanced (Wall et al., Reference Wall, Rose, Ellse and Morgan2011). Similarly, a longer season during which temperatures permit development of infective nematode larvae on pasture is likely to lead to higher levels of autumn infection, but only if the grazing period is also extended and if the hosts remain susceptible. Recent modeling studies suggest that host immunity could effectively suppress such end-of-season infections in grazing sheep and render earlier treatment and pasture interventions less relevant (Laurenson et al., Reference Laurenson, Kyriazakis, Forbes and Bishop2012). Laboratory diagnostic data, however, appear to show a net effect of increased clinical parasitic gastroenteritis in sheep in temperate autumn periods (van Dijk et al., Reference Van Dijk, David, Baird and Morgan2008). Other changes in management could dominate effects on seasonal patterns and levels of infection, as for ostertagiasis in cattle, in which housing periods appear to provide the main mechanism of seasonal forcing (Roberts and Grenfell, Reference Roberts and Grenfell1991).
The detail of management must therefore be appreciated and incorporated into climate change predictions in these systems. Ideally, dynamic change in management as a result of climate change would be explicitly incorporated into models of parasite epidemiology, but this would require more detailed understanding of farmer behavior and motivations for change than is currently available.
A simple, direct management change in response to real or perceived increases in risk of parasitic disease is to increase anti-parasitic treatments, but this would be self-defeating through accelerated selection of drug resistance (Kaplan, Reference Kaplan2004). Consequently, control methods must increasingly use targeted approaches at group and individual levels (van Wyk, Reference Van Wyk2001; Besier, Reference Besier2012; Kenyon and Jackson, Reference Kenyon and Jackson2012; Laurenson et al., Reference Laurenson, Bishop, Forbes and Kyriazakis2013). Such integrated parasite management requires risk assessment by the farmer, and animal health scientists have a central role to play in providing them with the necessary tools.
Space and scale
When aiming to turn epidemiological predictions of climate change effects on animal parasites into decision support tools, considerations of space and scale are crucial. Thus, many predictive models exist for parasites of livestock, which focus on general patterns rather than specific on-farm events (Smith and Grenfell, Reference Smith and Grenfell1994; Cornell, Reference Cornell2005). These are useful to advance understanding, but are of little help in guiding decisions on individual farms, especially in real time. At the other end of the spectrum, detailed simulation models have been developed with decision support in mind (e.g. Learmount et al., Reference Learmount, Taylor, Smith and Morgan2006), but the mammoth task of validating these models against sufficient field data, and the tendency to drown in complexity such that key drivers of system dynamics are obscured, mean that these have yet to be widely applied as effective drivers of farmer behavior.
Perhaps the most successful epidemiological decision support systems for nematode control in livestock take the form of decision trees (Ploeger et al., Reference Ploeger, van Doorn, Nijsse and Eysker2008) or schemas to identify appropriate general approaches given prevailing climate (Woodgate and Love, Reference Woodgate and Love2012). For helminths, extending these to specific on-farm predictions of infection risk is difficult mainly because of the inherently spatial nature of transmission (Morgan et al., Reference Morgan, Milner-Gulland, Torgerson and Medley2004). Thus, infection risk is tied to particular pastures, and cannot be predicted from climate alone, without detailed information on prior occupancy by livestock, and infection levels. Gathering such information is too onerous and time consuming to be realistic in most farm situations. In contrast, where parasites are more mobile, pasture history is less crucial, and risk is more directly related to climatic effects on parasite availability. This enables provision of farm-specific advice for blowfly strike, for instance using the flyboss tool (www.flyboss.com.au), compared with more generic advice offered by the wormboss tool (www.wormboss.com.au).
Other spatial considerations in predictive models include the resolution of climatic data, because conditions can vary locally between and even within farms, and between macroclimate measured at meteorological stations and microclimates experienced by parasites. Regional or national forecasts might therefore poorly reflect local risk at farm level, leading to predictions that are too general to be of practical use, even if the underlying science is sound. This charge might be leveled at forecasts for fasciolosis and nematodirosis, which are based, respectively, on summer rainfall and spring soil temperature (Ollerenshaw and Rowlands, Reference Ollerenshaw and Rowlands1959; Smith and Thomas, Reference Smith and Thomas1972), although progress is being made on increasing spatial resolution of the fluke forecast model (Fox et al., Reference Fox, White, McClean, Marion, Evans and Hutchings2011).
A possible solution to issues of poor farm-level resolution and missing information on pasture status is to forecast the climatic component of risk alone, and leave the farmer to interpret this in terms of historical events on the farm, and available management options for evading or attenuating risk. This is equivalent to predicting the hazard component of risk, while delegating assessment (and reduction) of exposure to the end user. Alternative approaches might seek to maximize farm-specific inputs into predictive models, e.g. by using manual or automated data collection through mobile devices. If extended to disease recording, this could provide feedback to improve local prediction accuracy over time, and to provide data for model validation. The best position to take on this continuum of detail will depend on the objectives sought as well as the biological and management characteristics of the system.
Farmer behavior
Significant scientific and technical effort has in the past been devoted to delivering tools for parasite risk prediction and farmer decision support (e.g. Wall et al., Reference Wall, Cruickshank, Smith, French and Holme2002), only for uptake to be low. A significant emerging area of research therefore deals with communication of advice on parasite control, and progressive education of farmers to engage with integrated control strategies and related decision support (Kahn and Woodgate, Reference Kahn and Woodgate2012; Woodgate and Love, Reference Woodgate and Love2012). Better understanding of farmer attitudes to risk and how this influences their decisions (e.g. Marshall, Reference Marshall2010) could be applied to improve uptake of best practice, and would be more constructive than grumbling that farmers never listen. Because farmers using flexible, evidence-based strategies will be best placed to adapt to altered parasite threats under climate change, the details of how prediction tools are configured and disseminated are important. These should take account of impact on production, which has direct meaning for farmers. Finally, it should be recognized that a legitimate component of farmer resilience to changing endemic disease threats might be to not react too quickly to new advice. Farmers slow to adopt the widely promoted ‘dose and move’ approach to nematode control a few decades ago, for instance, are now more likely to be dealing with relatively susceptible and hence tractable parasite populations than those who implemented those recommendations quickly and uncritically.
Applying expertise
Good, farm-realistic predictions of parasite infection risk depend on good science and good scientists, but the reverse is not true. Scientists generally depend for recognition and career advancement on publishing papers in prestigious journals, which will be widely cited by other scientists. Predictions of general patterns of disease and how they might be affected by climate change, based on simplified representations of the general processes involved, are more likely to meet this definition of success than detailed predictions dealing with specific diseases in a more applied context. Sadly, this continues to hold sway, and many contributors of sound models of parasitic disease in livestock have since moved on to more ‘illustrious’ fields, while their general frameworks find limited application in practice. More recently, some thought has been given to aligning research metrics more closely to application (Smith et al., Reference Smith, Ward and House2011). The detail of how science is funded and configured to serve society, although seemingly far removed from the question in hand, is actually its crux.
Concluding remarks
Predictions of climate change effects on parasitic disease on farms, if too general in terms of biology, management factors, and spatial scale, will be of little practical use in guiding control decisions, especially in a new era of emphasis on flexible and integrated approaches to parasite control. The balance between specificity and generality in epidemiological models of parasite transmission, how they fit with farmer attitudes and behaviors relating to risk, and the reward system for scientists developing such models, are all crucial to effective control strategies under climate change. At present, the field lacks balance in this respect, with scarce data and ad hoc consideration of detailed mechanisms often perpetuating over-reliance on simplistic, general outlooks. A change of approach is urgently needed if parasitological predictions are to make any meaningful contribution to farmer adaptation to climate change.
Acknowledgments
I am grateful for funding from EU FP7 project 288975 GLOWORM, and for stimulating discussions with partners in the Parasol and Gloworm projects, including Jan van Dijk, as well as Richard Wall, Jan van Wyk, and our graduate students.
