CLIMATE CHANGE AND PLANT DISEASE MANAGEMENT

Transcription

1 CLIMATE CHANGE AND PLANT DISEASE MANAGEMENT Stella Melugin Coakley Department of Botany and Plant Pathology, Oregon State University, Corvallis, Oregon 97331; Harald Scherm Department of Plant Pathology, University of Georgia, Athens, Georgia 30602; Sukumar Chakraborty CSIRO Tropical Agriculture, CRC for Tropical Plant Pathology, University of Queensland, Queensland 4072 Australia; Key Words host-pathogen interactions, risk and impact assessment?annu. Rev. Phytopathol : Copyright c 1999 by Annual Reviews. All rights reserved climate variability, epidemiological models, global warming, Research on impacts of climate change on plant diseases has been lim- Abstract ited, with most work concentrating on the effects of a single atmospheric constituent or meteorological variable on the host, pathogen, or the interaction of the two under controlled conditions. Results indicate that climate change could alter stages and rates of development of the pathogen, modify host resistance, and result in changes in the physiology of host-pathogen interactions. The most likely consequences are shifts in the geographical distribution of host and pathogen and altered crop losses, caused in part by changes in the efficacy of control strategies. Recent developments in experimental and modeling techniques offer considerable promise for developing an improved capability for climate change impact assessment and mitigation. Compared with major technological, environmental, and socioeconomic changes affecting agricultural production during the next century, climate change may be less important; it will, however, add another layer of complexity and uncertainty onto a system that is already exceedingly difficult to manage on a sustainable basis. Intensified research on climate change related issues could result in improved understanding and management of plant diseases in the face of current and future climate extremes. INTRODUCTION Global climate change is a major topic of discussion within both scientific and political forums. By mid-1998, evidence was mounting that 1998 would set a record high for global mean temperature, with 1997 holding the current record. Combined /99/ $

2 400 COAKLEY SCHERM CHAKRABORTY ocean and land temperatures were 0.25 C warmer from January to May 1998 than previously recorded (154). September 1998 was the hottest on record (104 years) in the United States, with a mean of 20.6? C; the earlier record of 20.2 C was set in 1931 (11). Increasingly, scientists are convinced that human activity, primarily in the form of increased emissions of carbon dioxide (CO 2 ) and other greenhouse gases (predominantly methane and nitrous oxide), is a major contributor to the warming trend. Indeed, the Second Assessment Report of the Intergovernmental Panel on Climate Change (IPCC), which was edited in 1995 (59), concluded that (i) greenhouse gas concentrations have increased and will continue to do so given their long life in the atmosphere; (ii) climate has changed over the past century in response to increasing atmospheric CO 2, with the global mean surface air temperature increasing between 0.3 and 0.6 C; (iii) anthropogenic aerosols such as sulfur dioxide tend to produce negative radiative forcing (i.e. a cooling trend) but are relatively short-lived; and (iv) climate model simulations using future emission scenarios of greenhouse gases and aerosols suggest an increase in global mean temperature between 1 and 3.5 C by the year 2100 (67). Although the IPCC report represents the broad consensus among many scientists from a wide range of disciplines, alternative interpretations of the evidence presented in the report do exist (61, 138). The IPCC report was the basis for discussion at the Kyoto Conference in December 1997 that led to a climate treaty ratified by more than 150 countries. The United States signed the treaty in November 1998 but ratification by Congress is not expected within the next two to three years. The Kyoto Protocol could result in reductions in six greenhouse gases including CO 2, methane, and nitrous oxide. It calls for an average 5.2% reduction (from 1990 levels) of CO 2 emissions by 38 industrialized nations by 2012, although increasing emissions from developing countries could offset these reductions (85). Curbing of CO 2 emissions is a very charged political issue in the United States (73), in part because of the position that any reduction in fossil fuel consumption would be damaging to the economy. Nevertheless, scientists optimistically provide a road map and recommendations of the technology necessary to achieve a reduction to below 1990 CO 2 levels (102, 115). Is global warming an absolute? As recently as 1994, Time magazine carried a feature article The Ice Age Cometh? (74), describing the January blizzard that resulted in record low temperatures in dozens of cities in the United States. This extreme expression of the weather is not inconsistent with the notion that climate change would lead to increased climate variability and more frequent or more severe climatic extremes (39, 55, 92, 113, 162). Underlying all discussions is the expectation that as climate changes, so does the human response to it. With this in mind, there have been numerous attempts to predict the potential impacts of climate change on human enterprises, including agriculture (1, 38, 101, 117, 118). Despite the important role of plant pathogens in limiting agricultural production and food supply (100), there has been limited research to assess the potential impacts of climate change on plant diseases (22, 24, 26, 28). For example, a recent

3 ?CLIMATE CHANGE AND DISEASE MANAGEMENT 401 monograph entitled Climate Change and the Global Harvest: Potential Impacts of the Greenhouse Effect on Agriculture (117) devoted only two pages (out of 324 total) to plant diseases. In contrast, potential impacts on infectious diseases of humans have received considerably more attention, including a monograph published by the World Health Organization (91) and a stand-alone chapter in the 1995 IPCC report (90). Suggestions in these publications that a warmer climate will bring increases of human disease have been challenged as premature and with little basis in fact, based on examination of historical patterns and the general absence of research in this area (147). It is with this example that we enter into a review of what appears probable, what may be possible, and what is unlikely in terms of climate change and its impacts on plant diseases. Mahlman (84) commented on and updated the summary of IPCC s 1995 report. He classified the relative certainty of change into several categories. Virtually certain facts include that atmospheric concentrations of greenhouse gases would continue to rise, largely caused by human activities such as burning of fossil fuel and changes in land use. These gases, which may remain in the atmosphere from a decade to centuries, act to heat the planet because of absorption and re-radiation of infrared radiation. Changes in other radiatively active substances (e.g. sulfur aerosols) and increased cloudiness caused by greater evaporation in a warmer climate may offset some of the greenhouse effect. Another virtually certain fact is that earth s surface has warmed about 0.5 ± 0.2 C during the past century. The expected rate of increase is now at 0.1 C per decade (69). Under virtually certain projections (99% likely to happen), Mahlman (84) included the forecast that the stratosphere would continue to cool as CO 2 levels rise and that global mean concentrations of water vapor in the lower troposphere would increase (approximately 6% per 1 C warming). A very probable projection would have a greater than 90% chance of being correct; under this category, Mahlman (84) predicted that a doubling of atmospheric CO 2 (from a current concentration of 360 ppm) would result in a warming between 1.5 and 4.5 C. The rate of evaporation would increase in a warmer climate, which would lead to an increase in global precipitation of 2 ± 0.5% per 1 C warming. Higher latitudes in the Northern Hemisphere are expected to experience above-average increases in both temperature and precipitation. Mahlman s final category ( probable projections, which have a greater than two thirds chance of occurring) included the forecast that there would be decreases in soil moisture because of increased temperatures, although this could be offset by simultaneously increased precipitation. Further, changes in mean climate would probably be accompanied by changes in the frequency and magnitude of climate extremes; globally, this would include an increased probability of warm events and decreased probability of cold events. COMPONENTS OF CLIMATE CHANGE

4 402 COAKLEY SCHERM CHAKRABORTY? IMPACTS ON CROPS Hansen et al (49) proposed an easily understood climate index based on heating degree days and frequency of precipitation to monitor climate change. For Asia and western North America, the index indicates that climate change should be evident already. Indeed, an increasing number of recent studies involving analyses ranging from satellite temperature data (50) to borehole temperatures (108) is lending support to the notion that climate change is occurring now. Mahlman (84) concluded that it is virtually certain that human-caused greenhouse warming is going to continue to unfold, slowly but inexorably, for a long time into the future. What is much less certain is the magnitude of climate change and its impacts on biological and ecological processes and on human enterprises. Climate change will influence the geographical distribution and growth of plant species around the world. The magnitude of these impacts would vary depending on the species involved and their growth patterns (e.g. annuals vs perennials; agricultural crops vs natural vegetation). Natural vegetation would be affected by factors such as competition, migration, and recovery from disturbances, and therefore new combinations of species are likely to occur (41). In managed systems (agriculture and forestry), any effects of climate change have important socioeconomic implications for the countries involved, hence considerable efforts have been directed toward agricultural impact assessment (1, 38, 101, 117, 118). How a particular agroecological region will fare under climate change will depend largely on society s ability to use available knowledge, technology, and financial resources. Nicholls (98) analyzed historical trends in Australian wheat yields and estimated that recent climate changes were responsible for as much as 30 to 50% of the observed yield increase, with most of the variation explained by an increase in minimum temperature. Goudriaan & Zadoks (44) summarized a large body of evidence suggesting that elevated CO 2 would result in increased photosynthesis and water use efficiency. This would lead to yield increases in most crops in most production conditions. Idso & Idso (60) reviewed hundreds of CO 2 enrichment studies done over ten years and reported that in most cases, there was evidence of growth-enhancing effects. IPCC s 1995 report (112) predicted that a doubling in CO 2 could increase yields of several major crops by an average of 30%. In contrast, recent data taken in field-scale experiments with elevated CO 2 suggested only an 8 to 12% wheat yield increase under optimal growing conditions (45). Further, evidence is now accumulating that crop production would be affected differentially depending on latitude. Although increases in yields are expected at mid and high latitudes, there may be decreases at lower latitudes where food needs are greatest (112, 117). Any direct yield gains caused by increased CO 2 or climate change could be offset partly or entirely by losses caused by phytophagous insects, plant pathogens,

5 ?CLIMATE CHANGE AND DISEASE MANAGEMENT 403 and weeds. It is therefore important to consider these biotic constraints in studies on crop yields under climate change. Unfortunately, nearly all previous climate change yield studies were done under exclusion of insects, pathogens, or weeds. For example, yield simulations using growth models have been completed for various crops in different agroecological zones (43, 113), but only the models for rice have incorporated disease aspects in a mechanistic form (77 80). The focus of this review is on how climate change manifested by elevated CO 2, increased temperature, and changes in precipitation may affect plant diseases and their management. Considerable literature is available on the impacts of other aspects of global change, particularly air pollutants (26, 31, 82). For example, the effects of ozone on plant metabolism, crop yield and productivity (54, 72), plant health (87, 124, 156), host-pathogen interactions (53, 125), host defense mechanisms, and interactions with pathogenic and saprophytic organisms have been extensively studied. Similarly, studies on the effects of increased UV-B on plants (119, 153), pathogens (4, 71, 111), phyllosphere yeasts (46), and disease development (2, 37, 47, 87, 104) are well represented in the literature. Most research on how climate change influences plant diseases has concentrated on the effects of a single atmospheric constituent or meteorological variable on the host, pathogen, or the interaction of the two under controlled conditions (21, 24 26, 28, 83, 87). Interactions are clearly more complex in the real world, however, where multiple climatological and biological factors vary simultaneously in a dynamic environment. Climate change has the potential to modify host physiology and resistance and to alter stages and rates of development of the pathogen. The most likely impacts would be shifts in the geographical distribution of host and pathogen, changes in the physiology of host-pathogen interactions, and changes in crop loss. Another important impact may be through changes in the efficacy of control strategies. New disease complexes may arise and some diseases may cease to be economically important if warming causes a poleward shift of agroclimatic zones and host plants migrate into new regions. Pathogens would follow the migrating hosts and may infect remnant vegetation of natural plant communities not previously exposed to the often more aggressive strains from agricultural crops. Facultative parasites with broad host ranges would mostly fall in this category, although obligate parasites may also expand their host range to infect plants in their proximity (35, 127). The mechanism of pathogen dispersal, suitability of the environment for dispersal, survival between seasons, and any change in host physiology and IMPACTS ON PLANT PATHOSYSTEMS Geographical Distribution of Host and Pathogen

6 404 COAKLEY SCHERM CHAKRABORTY? Physiology of Host-Pathogen Interactions ecology in the new environment will largely determine how quickly pathogens become established in a new region. Changes may occur in the type, amount, and relative importance of pathogens and affect the spectrum of diseases affecting a particular crop. This would be more pronounced for pathogens with alternate hosts. For perennial species in plantation and natural forests and orchards, plants may continue to be grown in existing regions under marginal climatic conditions. The need for large tracts of suitable land, infrastructure such as fruit- or timber-processing facilities, and the high cost of establishing a new plantation could limit the mobility of perennial crops. Plants growing in marginal climate could experience chronic stress that would predispose them to insect and disease outbreaks. Warming and other changes could also make plants more vulnerable to damage from pathogens that are currently not important because of unfavorable climate. Infection of eucalypts by Phytophthora cinnamomi is favored in wet soils at temperatures of 12 to 30 C (107), hence the pathogen does not pose a serious threat to the susceptible Eucalyptus spp. grown in southeastern Australia. This situation may change with an increase in temperature. Similarly, poplar clones in northern Europe may experience increased damage from the thermophilic rust, Melampsora alli-populina (76). If the frost-line moves north in the Northern Hemisphere, higher winter temperatures could be accompanied by increased survival of insects (109, 145). For virus-vector aphids, this could lead to higher incidence of virus diseases, especially in those regions where the timing of virus arrival is linked to winter survival and spring flight of aphids (51). Barley yellow dwarf potyvirus (BYDV) is an example of a virus that causes more severe disease following mild winters. Since BYDV is exclusively vectored by aphids, increased survival of pathogen reservoirs could greatly increase the economic losses caused by infection. Similar increases in viruses of potato and sugar beet have also been observed following warmer winters (19, 51, 81, 150). Elevated CO 2 Increases in leaf area and duration, leaf thickness, branching, tillering, stem and root length, and dry weight are well-known effects of increased CO 2 on many plants (13, 167). Based on a review of literature, Manning & von Tiedemann (87) suggested that elevated CO 2 would increase canopy size and density, resulting in a greater biomass of high nutritional quality. When combined with increased canopy humidity, this is likely to promote foliar diseases such as rusts, powdery mildews, leaf spots, and blights. The decomposition of plant litter is an important factor in nutrient cycling and in the saprophytic survival of many pathogens. Increased C:N ratio of litter is a consequence of plant growth under elevated CO 2 (5). Evidence from pot and field studies indicates that decomposition of high-co 2 litter occurs at a slower rate. Increased plant biomass, slower decomposition of litter, and higher winter

7 ?CLIMATE CHANGE AND DISEASE MANAGEMENT 405 temperature could increase pathogen survival on overwintering crop residues and increase the amount of initial inoculum available to infect subsequent crops. In recent studies on host-pathogen interactions in selected fungal pathosystems, two important trends have emerged on the effects of elevated CO 2. First, the initial establishment of the pathogen may be delayed because of modifications in pathogen aggressiveness and/or host susceptibility. Colletotrichum gloeosporioides showed delayed or reduced conidial germination, germtube growth, and appressorium production when inoculated onto susceptible Stylosanthes scabra plants under increased CO 2 (23). Similar effects in other pathosystems include a reduction in the rate of primary penetration in Erysiphe graminis (barley powdery mildew) (57) and a lengthening of the latent period in Maravalia cryptostegiae (rubbervine rust) (S Chakraborty, MP Weinert & JR Brown, unpublished data). In these examples, host resistance may have increased because of changes in host morphology, physiology, nutrients, and water balance. A decrease in stomatal density (7, 13, 167) increases resistance to pathogens that penetrate through stomates. In barley, although the thickness of epicuticular wax did not play a role in resistance to E. graminis, plants in elevated CO 2 were able to mobilize assimilates into defense structures including the formation of papillae and accumulation of silicon at sites of appressorial penetration (57, 58). High-CO 2 wheat had an average 14% reduction in nitrogen concentration in its shoot tissue (30) that was associated with decreased susceptibility to powdery mildew (151). The second important finding has been an increase in the fecundity of pathogens under elevated CO 2. Following penetration, established colonies of E. graminis (57) and C. gloeosporioides grew faster under 2 CO 2. Sporulation per unit area of infected tissue was increased several-fold under elevated CO 2 for C. gloeosporioides (23) and for M. cryptostegiae (S Chakraborty, MP Weinert & JR Brown, unpublished data). There are few studies on how elevated CO 2 may influence diseases other than those caused by fungi. Malmstrom & Field (86) examined the effects of 2 CO 2 on oats infected with BYDV. With virus infection, the oats showed a greater biomass response to CO 2 enrichment than did the healthy plants. Root biomass increased more in the infected plants than in those not infected. It is possible, therefore, that elevated CO 2 could alter the epidemiology of BYDV by increasing the virus reservoir due to improved winter survival of infected plants as the result of increased root biomass and water use efficiency (86). Elevated Temperature Increases in temperature can modify host physiology and resistance. Considerable information is available on heat-induced susceptibility and temperature-sensitive genes (33, 42, 123). For example, a rise in temperature above 20 C can inactivate temperature-sensitive resistance to stem rust in oat cultivars with Pg3 and Pg4 genes (89). In contrast, lignification of cell walls increased in forage species at higher temperatures (165) to enhance resistance to fungal pathogens (140).

8 406 COAKLEY SCHERM CHAKRABORTY Impacts would, therefore, depend on the nature of the host-pathogen interactions and the mechanism of resistance. Agricultural crops and plants in natural communities may harbor pathogens? as symptomless carriers (32, 66), and disease may develop if plants are stressed in a warmer climate. Host stress is an especially important factor in decline of various forest species. Climate extremes such as drought may increase invasion by Armillaria spp. that are not normally very pathogenic (76, 114). High temperatures may increase the damage caused by diseases such as Scleroderris canker on lodgepole pine (65, 76). Such projections, however, do not consider other factors that can enhance the resilience of forest ecosystems to climate change, which led Loehle (75) to conclude that there is a systematic bias toward alarmist predictions in projections of tree health response to climate change. Crop Loss At elevated CO 2, increased partitioning of assimilates to roots occurs consistently in crops such as carrot, sugar beet, and radish. If more carbon is stored in roots, losses from soilborne diseases of root crops may be reduced under climate change. In contrast, for foliar diseases favored by high temperature and humidity, increases in temperature and precipitation under climate change may result in increased crop loss. The effects of enlarged plant canopies from elevated CO 2 could further increase crop losses from foliar pathogens (23, 87). Unfortunately, canopy characteristics have not featured prominently in plant pathology research, despite their influence on microclimate and pathogen dispersal (3). Recent developments in three-dimensional modeling of plant architecture ( virtual plants ) offer new opportunities to integrate canopy architecture with microclimate effects and pathogen dispersal (166). Chakraborty et al (23) studied dispersal of and infection by C. gloeosporioides under ambient weather conditions in the field on S. scabra plants that had been raised under 1 or 2 CO 2 in controlled environment chambers. Plants from the two CO 2 environments were exposed to naturally occurring inoculum in the field on different dates, and conidial dispersal and infection were monitored. The enlarged canopy of plants grown under elevated CO 2 trapped more conidia that, together with increased humidity in the denser canopy, led to more severe anthracnose than on plants grown under 1 CO 2 (Figure 1). In a separate experiment, disease was reduced when plants were grown under elevated CO 2 and inoculated with C. gloeosporioides inside the controlled environment (23). The contrasting results help to highlight dangers in extrapolating effects observed under controlled environments to field situations. It is difficult to arrive at realistic predictions of yield losses that may result from a slow and gradual climate change by extrapolating findings from controlled environment studies which generally considered only two contrasting CO 2 levels. Indeed, host and pathogen populations are apt to adapt to gradual changes in CO 2 concentration. Long-term field studies utilizing free-air CO 2 enrichment (FACE technology) (136) or similar open-air facilities (99) offer the best approach to

9 ?CLIMATE CHANGE AND DISEASE MANAGEMENT 407 Figure 1 Cumulative number of lesions (a) and disease severity expressed as percent leaf area affected (b) caused by Colletotrichum gloeosporioides on susceptible Stylosanthes scabra plants that had been raised under 1 or 2 CO 2 in controlled environment chambers before being exposed to naturally occurring inoculum in three field plots for 48 h on five separate occasions. Data are means and standard errors from five exposures with three plants from each field plot. Source: S Chakraborty, IB Pangga, PM Room &DYates (unpublished data). studying disease development and crop losses caused by polycyclic pathogens. Runion et al (120) presented preliminary data on the effects of 2 CO 2 on the phyllosphere and rhizosphere microflora of cotton at the FACE-site in Arizona. They found that soil infestation by Rhizoctonia solani tended to be greater under elevated CO 2 at one of their sampling dates; a laboratory bioassay of the soil, however, showed no increased damping-off potential. Regional Impact Assessment There have been some efforts to estimate impacts of climate change on plant diseases at regional or country levels. An assessment in New Zealand (110) concluded that disease problems in the kiwifruit and pome fruit industries would probably be amplified by increases in temperature and precipitation. Areas on the northern island such as Hawkes Bay and Nelson are likely to experience increased levels of Botrytis, Sclerotinia, blossom blight, and fire blight. In contrast, the impacts on the vegetable industry should be minimal because this industry is mostly annual and

10 408 COAKLEY SCHERM CHAKRABORTY? IMPACTS ON DISEASE MANAGEMENT intensive in nature and management changes required to mitigate climate change impacts may be made more easily. In a review of potential impacts of climate change on Australia, plant pathologists developed qualitative impact assessments for important diseases of major agricultural crops (22). For wheat, the analysis indicated that existing cultivars would experience increased severity of stripe rust, Septoria tritici blotch, and BYDV, but take-all would be reduced in warm and wet soils. It was further suggested that increases in soil temperature may reduce Verticillium wilt in potato and tomato but increase sugarcane root rot caused by Pachymetra chaunorhiza. Increases in summer temperature and the frequency of precipitation would favor apple powdery mildew but reduce peach leaf curl. Carter et al (20) summarized results of five years work on the impacts of climate change on agriculture in Finland. The assessment included pathogens such as Phytophthora infestans and the root knot nematode, both of which were predicted to increase because of additional pathogen generations in a warmer climate. Another preliminary, country-wide assessment of impacts on important plant diseases was presented by von Tiedemann (155) for Germany. Because of the rudimentary knowledge about impacts of climate change on plant pathosystems, it is impossible to predict implications for disease management with any certainty. It is prudent to assume, however, that effects would occur chiefly through influences on host resistance or chemical and biological control agents. Particular attention is needed to identify cases where the efficacy of disease management may be reduced under climate change. Host Resistance Cultivar resistance to pathogens may become more effective because of increased static and dynamic defenses from changes in physiology, nutritional status, and water availability (see Physiology of Host-Pathogen Interactions). Durability of resistance may be threatened, however, if the number of infection cycles within a growing season increases because of one or more of the following factors: increased fecundity, more pathogen generations per season, or a more suitable microclimate for disease development. This may lead to more rapid evolution of aggressive pathogen races. In a pilot study, Chakraborty et al (unpublished data) monitored evolution of C. gloeosporioides on S. scabra under elevated CO 2. A susceptible cultivar was grown in a controlled environment under 1 or 2 CO 2 and inoculated with three isolates of the pathogen. For each isolate, conidia collected from infected host tissue were used to inoculate a second group of plants of the same cultivar. Successive groups of plants were inoculated with conidia arising from the previous infection cycle to simulate polycyclic disease development and

11 ?CLIMATE CHANGE AND DISEASE MANAGEMENT 409 Figure 2 Disease severity (on a scale from 0 to 9) caused by Colletotrichum gloeosporioides on susceptible Stylosanthes scabra plants under 2 CO 2 over eight cycles of infection. For each of the three isolates (indicated by different symbols and lines), successive groups of plants were inoculated with conidia arising from the previous infection cycle to simulate polycyclic disease development and pathogen evolution over time. The regression lines for the three isolates were significantly different from zero (P < 0.05). Source: S Chakraborty, PA Wilson & IB Pangga (unpublished data). pathogen evolution over time. After each cycle, measurements were made on components of pathogen aggressiveness, such as fecundity, lesion size, lesion number, and disease severity. Preliminary results suggested a significant trend toward increased disease severity (Figure 2); further, two of the three isolates showed a gradual increase in fecundity under elevated CO 2 after eight infection cycles. Climate change could affect the efficacy of crop protection chemicals in one of two ways. First, changes in temperature and precipitation may alter the dynamics of fungicide residues on the crop foliage. Globally, climate change models project an increase in the frequency of intense rainfall events (39), which could result in increased fungicide wash-off and reduced control. The interactions of precipitation frequency, intensity, and fungicide dynamics are complex, and for certain fungicides precipitation following application may result in enhanced disease control because of a redistribution of the active ingredient on the foliage (128). Neuhaus et al (96) applied simulated rain to potato foliage at two intensities (6 and Chemical Control

12 410 COAKLEY SCHERM CHAKRABORTY? Microbial Interactions 30 mm h 1 ) and found that the higher rate significantly reduced the fungicide residue that could be measured with a chemical assay, but that there was no difference in disease between the two treatments when the leaves were challenged in a bioassay with Phytophthora infestans. Second, morphological or physiological changes in crop plants resulting from growth under elevated CO 2 could affect uptake, translocation, and metabolism of systemic fungicides. For example, increased thickness of the epicuticular wax layer on leaves (13, 167) could result in slower and/or reduced uptake by the host, whereas increased canopy size could negatively affect spray coverage and lead to a dilution of the active ingredient in the host tissue. Both factors would suggest lowered control efficacy at higher concentrations of CO 2. Conversely, increased metabolic rates because of higher temperatures could result in faster uptake by and greater toxicity to the target organism. In a pilot study with the herbicide chlorotoluron, Edis et al (34) showed that a resistant biotype of the weed blackgrass (Alopecurus myosuroides) became more sensitive to herbicide application when grown under elevated CO 2. The authors hypothesized that this was due to changes in herbicide uptake and translocation because of altered stomatal physiology. Despite the potential for important interactions, no similar studies evaluating the impacts of climate change variables on physiological aspects have been published for fungicides. Climate change may alter the composition and dynamics of microbial communities in aerial and soil environments sufficiently to influence the health of plant organs (4, 46, 122). Changed microbial population in the phyllosphere and rhizosphere may influence plant disease through natural and augmented biological control agents. A direct effect of elevated CO 2 is unlikely in the soil environment as the microflora there is regularly exposed to levels 10 to 15 times higher than atmospheric CO 2. Trees grown in soils of poor nutrient status, especially nitrogen, favor colonization of roots by arbuscular mycorrhizal fungi (70). The relationship between elevated CO 2 and mycorrhizae is not well understood (139), and there are conflicting reports on how it may be influenced by the nutrient status of the plant and soil. If a lower nitrogen status of plant tissue under increased CO 2 results in more mycorrhizal colonization, this could improve plant health through improved nutrient uptake. Similar confusion exists on the potential role of vesicular-arbuscular mycorrhizae and ectomycorrhizae in the suppression and biological control of plant pathogens. Mycorrhizae can have positive, negative, or neutral effects on plant disease, and their role is not well understood despite numerous studies on the subject (105). Clearly, the influence of mycorrhizae on plant health under climate change requires further research. Changes in temperature may have highly nonlinear effects on tri-trophic interactions of host, pathogen, and biocontrol agent. In wheat (152), a rise in temperature

13 ?CLIMATE CHANGE AND DISEASE MANAGEMENT 411 from 17 to 22 C resulted in an increase in aphid (Sitobion avenae) reproduction by 10%; at the same time, however, predatory activity by lady beetle (Coccinella septempunctata) adults increased by 250%. Aphid damage was reduced further because of earlier maturity of the crop. Similar data are not available for tri-trophic interactions involving plant pathogens. Management of climate change will put additional pressure on agencies responsible for exclusion as a plant disease control strategy (64). In some regions, certain diseases of economic concern do not currently occur because the climate has precluded the causal agents from becoming established. Use of Geographical Information Systems and climate matching tools may assist quarantine agencies in determining the threat posed by a given pathogen under current and future climates. This approach was used by Sansford & Baker (126) to assess the risk of establishment of Karnal bunt in the cereal-growing regions of the European Union. Most of what has been said about plant disease in relation to climate change is based on qualitative, rule-based reasoning. This approach seems attractive because of the substantial body of knowledge already available regarding the environmental requirements of plants and their pathogens (25, 28, 36). For example, it seems plausible that increased air temperature would result in a poleward expansion of the geographical range of pathogens and in more generations per year (109); that elevated winter temperatures would increase survival and hence the amount of initial inoculum in many pathosystems (28); and that greater continental dryness during summer (6) would reduce risk of infection by pathogens that require leaf wetness or saturated soils for infection. But will this really be what happens? The answer will depend on complex interactions of atmospheric, climatic, and biological factors with technological and socioeconomic changes that are exceedingly difficult to predict (24). It would appear, therefore, that in all but the simplest cases these interactions are not amenable to qualitative analyses. Hence, quantitative (modeling) approaches, which allow one to investigate multiple scenarios and interactions simultaneously, will become more important for impact assessment (28). Guidelines for such model-based assessments are needed, and Sutherst et al (146) and Teng & Yang (149) have given a framework. With few exceptions (28, 146, 148), not enough attention has been given to modeling approaches and the analytical tools needed for quantitative impact assessment in plant pathology. Quarantine and Exclusion IMPACT MODELS Climate Matching Climate matching involves the calculation of a match index to quantify the similarity in climate between two or more locations. The match index is based on

14 412 COAKLEY SCHERM CHAKRABORTY? Empirical Models variables such as monthly minimum and maximum temperatures, precipitation, and evaporation. Software packages for climate matching include BIOCLIM (18), CLIMEX (143), HABITAT (158), and WORLD (9). These packages often come with additional useful features such as internal algorithms for generating climate surfaces through interpolation between stations. Climate matching may be used for climate change impact assessment by identifying those locations on the globe with a current climate that is most similar to the predicted future climate at the location of interest. An analysis of the plant disease problems at the matching locations, for example based on disease distribution maps (161), would allow predictions to be made about future disease risk at the location of interest. Booth et al (10) used climate matching to identify regions suitable for Cylindrocladium leaf blight on Eucalyptus spp. in Southeast Asia and around the world. They first established a simple rule for presence or absence of the disease based on long-term means of temperature and precipitation. This rule was then implemented in a climate matching program to identify high-risk regions in Africa, Australia, Latin America, and Southeast Asia under current climate. Further, two climate change scenarios were run for locations in Southeast Asia. The results suggested an increase in disease risk in northern Vietnam, southern Laos, and eastern Thailand. These predictions are consistent with limited field observations indicating that severe disease can occur in these regions during years with extreme weather. Possible effects of climate change on Phytophthora cinnamomi, a soilborne oomycete with an extremely wide host range, were considered by Brasier (15) and Brasier & Scott (16). This pathogen requires warm, wet soils and is hence limited primarily to tropical and subtropical regions (76, 88, 107). More recently, P. cinnamomi has been associated with oak declines in southern and Mediterranean Europe. It was hypothesized (14) that this may be an early indication of climate warming as the pathogen may have become more active because of higher soil temperatures and/or increased host susceptibility caused by stress (e.g. more frequent winter droughts in the region). For a more formal impact assessment, Brasier & Scott (16) used the CLIMEX climate matching program (143) to map regions in Europe favorable or unfavorable for this pathogen under present and future climate scenarios. The climate change simulations suggested that the pathogen could extend its range further north, although it appeared unlikely that it could become established in those regions where winter temperatures are low such as central and eastern Europe (15). It was further hypothesized that the pathogen s host range could increase if spread occurs into regions where it is currently absent. Four diseases of two major crops in China, wheat and rice, were examined by regression analysis to determine how they have varied through time and whether this may relate to recent increases in mean and minimum temperatures (169). Rice blast and wheat scab have increased sharply since the 1970s. The wheat acreage

15 ?CLIMATE CHANGE AND DISEASE MANAGEMENT 413 infected with powdery mildew has become more extensive, whereas stripe rust has decreased steadily (169). This may be related to the increased spring and early summer temperatures and would be consistent with the changes in stripe rust observed in the Pacific Northwest associated with climate variability (27). Boag et al (8) used data from soil samples collected during the European Plant- Parasitic Nematode Survey to assess the possible impacts of climate warming on the geographical range of virus-vector nematodes. Initial analyses of nematode presence-absence data suggested a close association between mean July soil temperature and nematode distribution. Based on this result, the authors predicted that climate change could result in increased nematode and virus problems in northern Europe; they estimated that a 1 C warming would allow the species in study to migrate northward by 160 to 200 km (95). Although nematodes migrate very slowly, humans are credited with efficiently disseminating them. Hence, nematode spread into new regions could put a wide range of crops at risk; additionally, introduction of new crops into a region could also expose them to infestation by nematode species already present. Changes in precipitation, which were not considered in these analyses, could influence nematode distribution on a large scale, although previous findings had suggested that soil moisture would not affect nematode distribution in most agricultural soils in northern Europe (8, 95). In a similar study, Jahn et al (62) utilized long-term plant disease monitoring records collected by the State Plant Protection Service in the former German Democratic Republic (GDR) to develop empirical climate-disease models for 15 individual host-pathogen combinations. These models were then used with various climate change scenarios to predict possible changes in infestation levels in a future climate. Calculations with the most realistic scenario (a temperature increase of 1 C combined with a decrease in precipitation of 30%) indicated that leaf rusts of wheat and barley and powdery mildew of sugar beet could increase substantially, reaching levels between two and five times as high as under the current climate. Infestation levels on small grains by powdery mildews would remain virtually unchanged, whereas those caused by foot rots and leaf blotch diseases would decrease. Most notable was a decrease in potato late blight to a mere 16% of its current level. The authors cautioned against over-interpreting their results, which were based on calculations with data from only 1 of 14 regions in the former GDR. Population Models A very different conclusion regarding the importance of potato late blight under climate change was reached by Kaukoranta (68). This author developed degreeday models for the emergence of potatoes and the date of late blight outbreaks in Finland. The two models were coupled and extended by including leaf area expansion of the crop as a function of thermal time; calculating radiation interception as a function of leaf area; transforming the intercepted radiation to tuber dry matter; and simulating the effects of late blight on tuber dry matter through a

16 414 COAKLEY SCHERM CHAKRABORTY? Simulation Models reduction in green leaf area, assuming that disease reduced leaf area to zero within 14 days after the predicted outbreak. Model parameters were obtained and model validation was done using data from a three-year field and greenhouse study. The combined model was then used with various temperature change scenarios to predict possible changes in potato yield and yield losses caused by late blight in a warmer climate. The results suggested that tuber yield could increase by 2tha 1 per 1 C warming in the absence of late blight. This potential yield gain was almost completely offset when late blight was considered, chiefly because late blight outbreaks occurred 4 to 7 days earlier and the period during which the crop was susceptible was lengthened by 10 to 20 days per 1 C warming. This study did not consider possible yield-enhancing effects of elevated CO 2, nor did it incorporate the effects of changes in precipitation on late blight. Simulation models have been used extensively to predict yields of various crops in different agroecological zones under climate change (43, 113). Biotic yieldreducing factors such as insects, pathogens, and weeds have, however, been largely ignored in these simulations (148). Because of this shortcoming, the development of linked disease-crop models is an important objective within the overall goal of developing a predictive capability for agricultural impact assessment and mitigation (130, 146, 148). For at least one key crop, rice, preliminary analyses considering the combined effects on yield of increased temperature, elevated UV-B radiation, and rice blast disease (Pyricularia grisea) have been done using a coupled simulation model (77 80). The model consisted of a physiological rice growth model and a leaf blast epidemic simulator, linked via the quantitative effects of leaf blast on photosynthesis and biomass production (78). Climate change was imposed by increasing mean temperature in fixed increments and by either including or omitting effects of UV-B on the host and pathogen (77). The results suggested that elevated UV-B could result in direct yield losses of 10%. Impacts of increased temperature varied by agroecological zone, with an increase in blast and associated yield losses in cool, subtropical rice production regions (e.g. Japan) and a decrease in humid tropics and subtropical regions (e.g. the Philippines). The authors cautioned that the results must be considered preliminary as the simulations did not include neck and panicle blast, two other important symptom types caused by P. grisea. Further, increased CO 2 was not considered, nor were changes in precipitation as preliminary analyses had indicated that the combined model was insensitive to changes in rainfall (77). Modeling for Impact Assessment: Prospects and Limitations What does the future hold for the use of models for climate change impact assessment in plant pathology? On the one hand, several new and exciting developments offer considerable prospect for modeling; examples include the emergence of object-oriented programming and increasing interest in the development

17 ?CLIMATE CHANGE AND DISEASE MANAGEMENT 415 of integrated impact assessment models. Object-oriented programming is an approach to model-building that is highly modular and generic, hence allowing model development by non-modelers and avoiding the substantial costs associated with developing and maintaining computer code for conventional simulation models (141, 144). Integrated assessment models include multiple drivers of global change (e.g. land-use changes in addition to climate changes) and multiple effects (e.g. biological and socioeconomic outcomes), with the goal of providing comprehensive advice on impacts and mitigation to policymakers and stakeholders (146). A proposed integrated case study of impacts of global change on vector-borne diseases of crops, livestock, and humans was outlined by Sutherst et al (142). On the other hand, numerous challenges remain regarding the use of models for climate change impact assessment. For example, no studies using state-of-the-art climate change scenarios [e.g. those predicted by various General Circulation Models (GCMs)] have been published in plant pathology. Instead, authors have relied on simplified sensitivity analyses that assumed fixed changes in mean temperature or precipitation, regardless of time of year or geographical location. There is now increasing evidence that the most important effects of climate change will be felt not through changes in mean climate, but through increased climate variability and more frequent or more severe climatic extremes (39, 55, 92, 113, 162). These effects must be considered in future impact assessments by plant pathologists (131). This could be accomplished through a better understanding of the effects of natural, medium-range climate variability (such as that caused by El Niño and similar climate cycles) on plant pathosystems, as discussed below. The preoccupation with temperature in many climate change impact assessments is unfortunate as it overlooks the fact that changes in precipitation have more pronounced effects on the development of many plant diseases, particularly those caused by fungi. Unfortunately, precipitation, which is temporally and spatially discontinuous, is still difficult to simulate with current GCMs (6, 137). Most GCMs predict increases in precipitation on a global scale (6) that could result in a more favorable environment for plant pathogenic fungi. This could be offset by other factors, e.g. greater dryness during summer because of increased evaporation (6) and possible reductions in dew deposition because of increased cloudiness and higher night-time temperatures. Recent developments in statistical downscaling algorithms offer progress to the goal of making precipitation predictions with sufficient temporal and spatial resolution to be useful in epidemiological models (56, 121, 163, 164). The magnitude of climate change in a high-co 2 world is still uncertain. Depending on the GCMs used and their underlying assumptions, these uncertainties result in a wide range of possible scenarios for the future. Kacholia & Reck (63), who compared 108 climate projections published between 1980 and 1995, reported that predicted global mean air temperature changes under 2 CO 2 ranged from to +8.7 C. As stated above, uncertainties in precipitation scenarios are even greater. Uncertainty in model inputs can compromise the reliability of the outputs of impact assessment models because of error propagation. Hence, the