Plant disease and global change – the importance of long-term data sets

Authors


(*Author for correspondence: tel +44 020 759 42428; fax +44 020 759 42601; email m. jeger@imperial.ac.uk)

Research on the interactions between climate change and plant diseases is rapidly catching up with the importance of the topic (Garrett et al., 2006). On the one hand, modelling studies are providing increasingly realistic scenarios for the influence on plant diseases of changes in the magnitude and variability of temperature, precipitation and other climatic variables. Recent examples include models predicting an increase under projected climate change in (i) severity of Plasmopara viticola epidemics on grapes in an important wine-producing Italian region near Turin in 2030, 2050 and 2080 (Salinari et al., 2006), (ii) the range and severity of epidemics of Leptosphaeria maculans on oil seed rape (Brassica napus) in the UK for the 2020s and 2050s (Evans et al., in press), and (iii) the distribution and local impact of a range of forest pathogens (Biscogniauxia mediterranea, Cryphonectria parasitica, Melampsora spp., Phytophthora cinnamomi and Sphaeropsis sapinea) in France at the end of the 21st century (Desprez-Loustau et al., 2007a). On the other hand, short-term, local experiments have demonstrated the impacts of predicted global change on plant health. Recent examples include: (i) a study showing that elevated atmospheric CO2 concentration increases the risk of infection with rice blast (Magnaporthe oryzae) and the percentage of rice (Oryza sativa) plants affected by sheath blight (Kobayashi et al., 2006); (ii) an experiment demonstrating species-specific responses to increased ozone concentrations of the susceptibility of young beech (Fagus sylvatica) and spruce (Picea abies) trees to Phytophthora citricola (Lüdemann et al., 2005); and (iii) a 12-yr warming experiment with heaters suspended over plots in a mountain meadow in Colorado, USA in which there was a change in the prevalence of different species of plant pathogens (Roy et al., 2004). These approaches are complementary, as they each have limitations. Modelling can only work if the long-term predictions are not overturned, for example by new insights into the co-evolutionary dynamics of plant–pathogen interactions; and short- to medium-term local experiments may not take into account time lags and signal accumulations (for nitrogen (N) fertilization: Strengbom et al., 2001; for CO2 concentration: Körner, 2006; for N deposition and raised temperature: Wiedermann et al., 2007). A study by Shaw et al. in this issue of New Phytologist (pp. 229–238) provides an example of how the limitations of models and experiments can be overcome by making use of a long-term (1844–2003) UK data set on the occurrence of two key world-wide pathogens of wheat (Triticum aestivum) (Stukenbrock et al., 2006).

‘The use of herbarium specimens in this way and their analysis using PCR techniques provide a unique way of characterizing changes in pathogen prevalence over historical time.’

In the Rothamsted Broadbalk experiments, a series of wheat plots each receiving a different amount of fertilizer have been grown and monitored almost continuously since 1843. Using PCR data from those plots, Shaw et al. combine the rigour of an experimental setting with the advantages of over a century of data. They show that in the long run the annual incidences of Phaeosphaeria nodorum and Mycosphaerella graminicola in the grain of wheat are, respectively, positively and negatively related to the emission of sulphur dioxide (SO2) across the whole of the UK and that these associations largely account for the variations in the relative prevalences of the two pathogens over this period of time. The use of herbarium specimens in this way and their analysis using PCR techniques provide a unique way of characterizing changes in pathogen prevalence over historical time. Emissions and atmospheric concentrations of SO2 have markedly declined over the last two decades in Europe and North America, thus roughly going back to the concentrations observed at the start of the Rothamsted experiment (c. 1–2 megatonnes of sulphur per year; Bearchell et al., 2005). The decline in SO2 emissions in Europe has also been related to improvements in forest health (e.g. Zirlewagen et al., 2007), and it would be interesting to know whether forest tree endophytes (which can behave as mutualists or not, depending on host conditions; Sieber, 2007) have responded in a similar way to wheat pathogens. SO2 emissions have recently increased in China, where forest health is generally declining (e.g. Wang et al., 2007). The findings of Shaw et al. suggest that the relative prevalences of P. nodorum and M. graminicola in China's wheat fields should have gone the opposite way to those at Rothamsted. However, Shaw et al. also provide evidence that the short-term variability in the presence of these two pathogens in the leaves of wheat is associated with weather conditions such as summer temperature and spring rainfall. Together with sea levels (DeSantis et al., 2007), temperature and rainfall are the two climatic factors that are most likely to be widely affected by future global change, and alterations in these factors are expected to have a wide range of impacts both on plants and on their pathogens (Ingram, 1999).

There is increased use in plant ecology of data sets extending over periods longer than one century, not only from fossil series, pollen records and dendrochronology (see also Woodward, 2007). For example, Tait et al. (2005) analysed changes in species richness and composition of the flora in the Adelaide Metropolitan Area for the period 1836–2002. For Turin, Italy, Isocrono et al. (2007) showed that lichen species richness has increased from 1792 to the present day, an indication of the improving air quality of that conurbation. For the flora of the Coliseum in Rome, Caneva et al. (2005) found a marked decrease in the presence of species typical of a cool and wet climate from 1643 onwards. For the living collections of the botanical gardens of the world, Pautasso & Parmentier (2007) showed that age (up to more than 400 years for the oldest gardens) explains one-fifth of the variance in the current species richness. Similar long-term data sets, however, are still rare, particularly in relation to plant and tree mortality (Fig. 1). In a recent example, an analysis of the genomes of barley yellow dwarf viruses in herbarium specimens (1894–1958) showed that this disease may have facilitated the invasion of introduced grasses in California (Malmstrom et al., 2007). For the UK, Jones & Baker (2007) showed that, at least for the period 1970 to 2004, the number of recorded introduced plant pathogens has not increased with time. However, this might well change in a markedly different climate. The issue of exotic pathogens is of utmost importance in relation to climate change. Global warming will not only act on pathosystems already present in a certain region, but will favour the emergence of new diseases, both because the distributional range, temporal activity and community structure of pathogens will be modified (e.g. Desprez-Loustau et al., 2007b; Shaw et al. 2007), and because the phenology and conditions of the hosts will be altered (e.g. Lonsdale & Gibbs, 1996). Add to this the long-distance introductions of pathogens as a result of the increasing globalization of trade (e.g. Jeger et al., 2007) and the challenges for end-of-the-century plant pathology are likely to become as complex as they will be unprecedented (Fig. 2).

Figure 1.

Frequency distribution of 67 studies modelling tree mortality published between 1997 and 2006, according to the length of the period upon which models were based. Sixty-seven of the 141 papers retrieved contained data about the study plots used to validate models. About 60% of these 67 models were based on surveys of less than 10 yr.

Figure 2.

A scale-dependent view of the effects of climate change on the disease triangle (this is formed by the interactions amongst host, pathogen and environment; see e.g. Scholthof, 2007). Climate change implies transformations of the environment of pathosystems over local, regional and continental scales, and these are in turn expected to increase the distribution and virulence of many plant pathogens, as well as cause a mismatch between host distributions and the range of conditions to which they are best adapted.

Not all changes in pathosystems are necessarily related to climate change (e.g. Rogers & Randolph, 2006), but the evidence that climate change can profoundly influence host–pathogen dynamics is growing, not only for plant diseases but also for animal and human diseases (e.g. Purse et al., 2005; Haines et al., 2006). There is a need not only for interdisciplinary collaboration between epidemiologists and climate scientists (Huntingford et al., 2007), but also for more awareness of investigations relating to climate change and diseases of plants, animals and humans in the three scientific communities, as the science involved is similar and analytical techniques are transferable and do not need to be reinvented (see e.g. Cazelles et al., 2007). One issue that is still rarely addressed and is not resolved by the availability of long-term data sets is the potential spatial scale dependence of responses of plant pathosystems to climate change, particularly if local studies focus on habitat patches where disease is disproportionately present (Holdenrieder et al., 2004; Strengbom et al., 2006). Climate change will affect plant pathosystems at a variety of levels of integration (from genes to populations and from ecosystems to distributional ranges) and in most aspects of epidemic development (from environmental conditions to host vigour and susceptibility, and from pathogen virulence to infection rates). Climate change is likely to have a profound impact on plant–pathogen interactions, and will thus represent a world-wide interdisciplinary challenge not only for the long-term sustainability of crop production but also for the understanding of biodiversity dynamics in a changing world and for the success of conservation biology activities.

Acknowledgements

Many thanks to Ottmar Holdenrieder and Sally Power for kindly commenting on a previous version of the draft and suggesting some key references.

Ancillary

Advertisement