Weather affects the severity of many plant diseases, and climate change is likely to alter the patterns of crop disease severity. Evaluating possible future patterns can help focus crop breeding and disease management research. We examined the global effect of climate change on potato late blight, the disease that caused the Irish potato famine and still is a common potato disease around the world. We used a metamodel and considered three global climate models for the A2 greenhouse gas emission scenario for three 20-year time-slices: 2000–2019, 2040–2059 and 2080–2099. In addition to global analyses, five regions were evaluated where potato is an important crop: the Andean Highlands, Indo-Gangetic Plain and Himalayan Highlands, Southeast Asian Highlands, Ethiopian Highlands, and Lake Kivu Highlands in Sub-Saharan Africa. We found that the average global risk of potato late blight increases initially, when compared with historic climate data, and then declines as planting dates shift to cooler seasons. Risk in the agro-ecosystems analized, varied from a large increase in risk in the Lake Kibu Highlands in Rwanda to decreases in the Southeast Asia Higlands of Indonesia.
Original Title
Climate change may have limited effect on global risk of potato late blight
Weather affects the severity of many plant diseases, and climate change is likely to alter the patterns of crop disease severity. Evaluating possible future patterns can help focus crop breeding and disease management research. We examined the global effect of climate change on potato late blight, the disease that caused the Irish potato famine and still is a common potato disease around the world. We used a metamodel and considered three global climate models for the A2 greenhouse gas emission scenario for three 20-year time-slices: 2000–2019, 2040–2059 and 2080–2099. In addition to global analyses, five regions were evaluated where potato is an important crop: the Andean Highlands, Indo-Gangetic Plain and Himalayan Highlands, Southeast Asian Highlands, Ethiopian Highlands, and Lake Kivu Highlands in Sub-Saharan Africa. We found that the average global risk of potato late blight increases initially, when compared with historic climate data, and then declines as planting dates shift to cooler seasons. Risk in the agro-ecosystems analized, varied from a large increase in risk in the Lake Kibu Highlands in Rwanda to decreases in the Southeast Asia Higlands of Indonesia.
Weather affects the severity of many plant diseases, and climate change is likely to alter the patterns of crop disease severity. Evaluating possible future patterns can help focus crop breeding and disease management research. We examined the global effect of climate change on potato late blight, the disease that caused the Irish potato famine and still is a common potato disease around the world. We used a metamodel and considered three global climate models for the A2 greenhouse gas emission scenario for three 20-year time-slices: 2000–2019, 2040–2059 and 2080–2099. In addition to global analyses, five regions were evaluated where potato is an important crop: the Andean Highlands, Indo-Gangetic Plain and Himalayan Highlands, Southeast Asian Highlands, Ethiopian Highlands, and Lake Kivu Highlands in Sub-Saharan Africa. We found that the average global risk of potato late blight increases initially, when compared with historic climate data, and then declines as planting dates shift to cooler seasons. Risk in the agro-ecosystems analized, varied from a large increase in risk in the Lake Kibu Highlands in Rwanda to decreases in the Southeast Asia Higlands of Indonesia.
Climate change may have limited effect on global risk of
potato late blight
ADAM H. S PARKS 1 , 2 , GREGORY A. F ORBES 3 , ROBERT J . HI J MANS 4 and KAREN A. GARRETT 1 1 Department of Plant Pathology, Kansas State University, 4024 Throckmorton Plant Sciences Center, Manhattan, KS 66506, USA, 2 International Rice Research Institute (IRRI), Los Ba~ nos, Laguna 4031, Philippines, 3 International Potato Center, 12 Zhongguancun South Street, Beijing 100081, China, 4 Department of Environmental Science and Policy, University of California, One Shields Avenue, Davis, CA 95616, USA Abstract Weather affects the severity of many plant diseases, and climate change is likely to alter the patterns of crop disease severity. Evaluating possible future patterns can help focus crop breeding and disease management research. We examined the global effect of climate change on potato late blight, the disease that caused the Irish potato famine and still is a common potato disease around the world. We used a metamodel and considered three global climate models for the A2 greenhouse gas emission scenario for three 20-year time-slices: 20002019, 20402059 and 20802099. In addition to global analyses, ve regions were evaluated where potato is an important crop: the Andean Highlands, Indo-Gangetic Plain and Himalayan Highlands, Southeast Asian Highlands, Ethiopian Highlands, and Lake Kivu Highlands in Sub-Saharan Africa. We found that the average global risk of potato late blight increases initially, when compared with historic climate data, and then declines as planting dates shift to cooler seasons. Risk in the agro-eco- systems analyzed, varied from a large increase in risk in the Lake Kivu Highlands in Rwanda to decreases in the Southeast Asian Highlands of Indonesia. Keywords: climate change, food security, Phytophthora infestans, plant disease management, plant pathology, potato late blight Received 11 September 2013 and accepted 12 March 2014 Introduction The risk of production loss from crop diseases, referred to hereafter as disease risk, can be strongly inuenced by weather. Infectious plant disease occurs due to the interaction of three main factors: a favorable environ- ment, a susceptible host, and a competent pathogen (Madden et al., 2007). Therefore, changes in weather due to climate change are likely to affect disease risk (Coakley et al., 1999; Anderson et al., 2004; Garrett et al., 2006). There is growing interest in plant disease risk under future scenarios (Pautasso et al., 2010; Chakr- aborty & Newton, 2011; Juroszek & Von Tiedemann, 2011; Luck et al., 2011; Savary et al., 2011; Sutherst et al., 2011) and how to adapt disease forecasting models to new scales of application for scenario analysis (Seem et al., 2000; Seem, 2004; Garrett et al., 2011; Shaw & Osborne, 2011). Potato late blight is caused by the oomycete Phytoph- thora infestans (Mont.) de Bary. Late blight is well known for its role in the Irish potato famine of the 1840s and today it remains an important disease in glo- bal potato production. Global potato yield loss from diseases, pests, and weeds was estimated to be around 40% of attainable production, with diseases alone accounting for 21% loss (Oerke, 2006), and potato late blight is generally recognized as the most important potato disease. While varieties with partial resistance to late blight are available, most of the popular varieties are susceptible (Forbes, 2012), so disease management is often dependent on the use of fungicides. Models to predict the effect of weather on potato late blight have been evolving for almost a century (Van Everdingen, 1926; Beaumont, 1947), generally drawing on temperature and atmospheric humidity as the most important predictors (Harrison, 1992). Fry et al. (1983) developed the SimCast model, and Gr unwald et al. (2002) further developed the SimCast model and dem- onstrated that it also performed well in a tropical high- land location. SimCast estimates the risk of damaging levels of late blight, expressed as blight units, based on the temperature during the consecutive hours in a day when relative humidity is above 90%. SimCast thus uses hourly weather data as input to calculate the need for a fungicide application to control disease. To rescale the SimCast model for use with larger time steps, from hourly increments to monthly weather observation data, and easier application to wider geographical areas, we previously developed a metamodel of the relationship between weather (temperature and relative humidity) and late blight risk, based on SimCast output Correspondence: AdamH. Sparks, tel. +63 2 580 5600, fax +63 2 580 5699, e-mail: [email protected] 1 2014 John Wiley & Sons Ltd Global Change Biology (2014), doi: 10.1111/gcb.12587 Global Change Biology (Sparks et al., 2011). The metamodel, hereafter referred to as SimCastMeta, uses monthly time-step temperature and relative humidity data to evaluate disease risk expressed as blight units. Hijmans et al. (2000) evaluated contemporary sever- ity of potato late blight indirectly using two tactical decision models, Blitecast (Krause et al., 1975) and Sim- Cast (Fry et al., 1983), to predict the need for a prophy- lactic pesticide application to control late blight in farmers elds. These models were then scaled up to estimate the number of pesticide applications necessary to manage late blight globally. Monthly climate data and a weather generator were used to temporally downscale the hourly weather data necessary for these models. The impacts of climate change on potato late blight have been studied in the Midwestern US (Baker et al., 2005) and Finland (Kaukoranta, 1996; Hannukkala et al., 2007), indicating the potential for increased risk. Our goal was to provide the rst global analysis of climate change effects on potato late blight. We used SimCastMeta to evaluate the effects of climate change emission scenarios dened by the Intergovernmental Panel on Climate Change (IPCC) (2012) and the level of disease resistance on the change in disease risk. Fur- thermore, using the SimCastMeta metamodel to evalu- ate risk illustrates the potential of metamodels for rescaling short-term forecasting models for scenario analysis. Results are discussed in the context of impor- tant potato-producing regions. Materials and methods We used georeferenced, gridded historic climate data (for the reference time period) and future predicted climate data as input in the SimCastMeta model (Sparks et al., 2011) to evalu- ate late blight risk. The EcoCrop crop model, as implemented in the R package dismo (Hijmans et al., 2012; R Core Team, 2013), was used to estimate the three-month potato-growing seasons for each grid cell for reference and future periods. The SimCastMeta model estimates blight units, a measure of the relative risk of damaging levels of potato late blight given an input of monthly temperature and monthly relative humidity (Sparks et al., 2011). Because late blight occurs so widely in potato production systems and the driving factors are gener- ally the weather conditions, temperature and relative humid- ity, we assumed that inoculum is not a limiting factor in areas where potato is grown. We used CRU CL 2.0 grid mean monthly temperature, mean monthly relative humidity and mean monthly precipi- tation data from New et al. (2002) as our reference climate data. These data are at a spatial resolution of 10 arcminutes (344 km 2 ) covering the time period from 1961 to 1990. Future climate emission scenario data were downloaded from the World Climate Research Programmes (WCRP) Coupled Model Intercomparison Project phase 3 (CMIP3) multimodel data (Meehl et al., 2007), that we statistically downscaled and bias-corrected to 10 arcminute resolution using the delta method (absolute difference for temperature, relative differ- ence for precipitation) (Leemans & Solomon, 1993). Only data for the global climate models (GCMs) that provided maxi- mum temperature, minimum temperature, and specic humidity were selected (Table 1), and relative humidity was calculated from these data. Because our goal was to provide a global evaluation of risk, we used an ensemble model approach (Bates & Granger, 1969) to obtain results represen- tative of the current state-of-the-art in climate modeling. Ensemble model averages (Bates & Granger, 1969) of the nonweighted means of the GCM outputs for each climate sce- nario were created for mean monthly temperature and mean monthly relative humidity. The global daily average blight unit accumulation per month was calculated from mean monthly temperature and mean monthly relative humidity using the SimCastMeta meta- model for the 19611990 reference climate normal (referenced from here on by the mid-point year as the 1975 time-slice), and three future 20-year time-slices, referenced hereafter by the mid-point year of the time-slice: 20002019 (2010), 2040 2059 (2050), and 20802099 (2090). The relative risk of damag- ing levels of late blight was then averaged using a three- month moving window to provide the average daily blight unit accumulation for 12 three-month time periods represent- ing three-month potato-growing seasons. Optimal potato planting dates were estimated using the EcoCrop model using mean monthly minimum and maxi- mum temperature, and precipitation data. The rst day of the rst month of a three-month growing season in which plant- ing would produce the highest potato yield for each grid cell was calculated. Optimal planting dates were generated for the reference climate normal and the ensemble GCM outputs for each of the three respective future time-slices. These data were used to estimate late blight risk for what would be a Table 1 Global Climate Models (GCM) selected, which pro- vided maximum and minimum temperature and relative humidity data for the A2 scenario used in SimCastMeta, where model outputs were averaged to create an ensemble model output. Models are freely available for download from the Program for Climate Model Diagnosis and Intercompari- son (PCMDI) and the WCRP Working Group on Coupled Modelling (WGCM) website, http://www-pcmdi.llnl.gov/ ipcc/info_for_analysts.php GCM abbreviation GCM model name BCCR BCM2.0 Bjerknes Centre for Climate Research Bergen Climate Model Version 2.0 CSIRO mk3.0 Commonwealth Scientic and Industrial Research Organization GCM mark 3 INMCM3.0 Institute for Numerical Mathematics Version 3.0 Model 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12587 2 A. H. SPARKS et al. geographic locations most productive potato-growing season in the absence of pests and disease. Total rainfed and irrigated potato production by country (Portmann et al., 2010) was used to remove areas where potato production is currently limited. Several countries were selected for further analysis because they are representative of highland or lowland tropical potato production, areas where potato is an important crop for poverty alleviation and where late blight is difcult to manage because of year-round potato production (Garrett et al., 2009). These countries include Colombia and Ecuador in the Andean Highlands; Ethiopia; Rwanda in the Lake Kivu Highlands region; Nepal in the Himalayan Highlands and Indo-Gangetic Plain; and Indonesia in the South-East Asian highlands. Results Initially global average late blight units increased for the 2010 time-slice relative to the reference climate data (Fig. 1). The predicted global average accumulation in 2050 was similar to the historic baseline and the projec- tions for 2090 were below the historical baseline. The average global temperature for potato-growing areas during growing seasons indicates a sharp increase, fol- lowed by a decline (Fig. 2). Relative humidity exhibited little change (0.9%) throughout the time-slices analyzed (Fig. 3). For a susceptible cultivar, the global average blight units accumulated during 1 month of a three-month growing season was 1.26 for the historical average, and increased to 1.30 for the average of the 2050 time-slice. Many of the areas where potato is grown exhibited relatively little or no change (0.5 to 0.5 difference) in daily mean blight units during the growing season (Figs 4 and 5). However, these changes, while seem- ingly small, are equal to the difference between using a resistant or susceptible variety in the original SimCast Fig. 1 Average global blight unit accumulation as predicted by SimCastMeta during the growing season as predicted by Eco- Crop for four time-slices: 19611990, 1975 time-slice (historic normal); 20002020, 2010 time-slice (A2 scenario); 20402060, 2050 time-slice (A2 scenario); and 20802100, 2090 time-slice (A2 scenario). IPCC A2 emission scenario time-slices are repre- sented here as ensemble model averages of three global climate models. Blight units are a measure of the biological risk of dam- aging levels of late blight of potato developing due to favorable weather conditions and are derived from the SimCast model. Fig. 2 Average global temperature during the growing season appropriate for a given location as predicted by EcoCrop for four time-slices: 19611990, 1975 time-slice (historic normal); 20002020, 2010 time-slice (A2 scenario); 20402060, 2050 time- slice (A2 scenario); and 20802100, 2090 time-slice (A2 scenario). IPCC A2 emission scenario time-slices are represented here as ensemble model averages of three global climate models. Fig. 3 Average relative humidity during the growing season appropriate for a given location as predicted by EcoCrop for four time-slices: 19611990, 1975 time-slice (historic normal); 20002020, 2010 time-slice (A2 scenario); 20402060, 2050 time- slice (A2 scenario); and 20802100, 2090 time-slice (A2 scenario). IPCC A2 emission scenario time-slices are represented here as ensemble model averages of three global climate models. 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12587 CLI MATE CHANGE EFFECTS ON POTATO LATE BLI GHT GLOBALLY 3 model (Fry et al., 1983; Gr unwald et al., 2002). In the original SimCast model, a change in blight unit accu- mulation of 1 or 1 is equivalent to one fungicide application less or more per month respectively for sus- ceptible and moderately susceptible cultivars (Fry et al., 1983; Gr unwald et al., 2002). Blight units increased in parts of East-Central South America, China, Europe, and Canada. The Andes and Himalayan Mountains and Sub-Saharan Africa exhibited a mixture of increas- ing and decreasing blight unit accumulation. Five of the ten countries experiencing the greatest increases in blight units are located in Africa (Table 2). Most of the main potato-producing countries exhib- ited a change in blight units of less than one blight unit per day per growing season (Table 3), with three coun- tries having a decreasing number of blight units. With the minimal changes seen in the main potato producing countries, the effects of host resistance are much less pronounced (Table 3) than among the countries most affected (Table 2). Resistant genotypes are effective in reducing blight risk in all countries, such that no coun- try experiences an accumulation of more than 0.79 aver- age daily blight units during the growing season. In the countries selected to represent specic agro- ecosystems, for a susceptible cultivar late blight risk increased in three of the ve regions: Andean High- lands, Ethiopian Highlands, and Lake Kivu Highlands. In the remaining two regions, Indo-Gangetic Plan and Himalayan Highlands and South East Asian Highlands, there was a slight decrease in the blight unit accumula- tion (Table 4). However, shifts in the blight unit accu- mulations within the growing areas for each country could inuence potato production as some areas expe- rience increased late blight severity while others experi- ence decreased late blight severity. Fig. 4 The change in global potato late blight relative risk as predicted by SimCastMeta model using historical climate normal, 1961 1990 (1975 time-slice) and 20402059 (2050 time-slice) A2 climate scenario for a susceptible potato genotype. IPCC A2 time-slices are ensemble model averages of three global climate models. Blight units are a predictor of biological risk based on weather and potato genotypic resistance. Table 2 Daily average blight unit accumulation and change during the growing season for the ten potato-producing coun- tries experiencing the greatest total increase in blight unit accumulation as predicted by SimCastMeta model using his- toric climate normal, 19611990 (1975 time-slice), and 2040 2059 (2050 time-slice) A2 climate for a susceptible variety. A2 scenario time-slices are ensemble model averages of three glo- bal climate models. Blight units are a predictor of biological risk based on weather and potato genotype resistance Country Susceptible blight units Resistant blight units 1975 2050 Change 1975 2050 Change Rwanda 2.41 3.58 1.17 1.43 2.22 0.79 Burundi 2.53 3.13 0.60 1.54 1.93 0.39 Zimbabwe 1.99 2.43 0.44 1.18 1.44 0.26 Portugal 1.19 1.55 0.36 0.65 0.83 0.18 Mauritius 3.45 3.80 0.35 2.14 2.41 0.27 Uruguay 1.87 2.19 0.32 1.05 1.24 0.19 Estonia 2.39 2.68 0.29 1.38 1.59 0.21 Iraq 0.57 0.80 0.23 0.28 0.41 0.13 Greece 0.86 1.09 0.23 0.45 0.57 0.12 Zambia 2.47 2.70 0.23 1.44 1.55 0.11 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12587 4 A. H. SPARKS et al. An examination of temperature and relative humid- ity in these agro-ecosystems shows that Rwanda has the highest increase in temperature (0.7 C) and relative humidity (2.9%) (Table 5). Nepal shows a decrease in temperature (1.6 C) and relative humidity (2.4%) with a corresponding drop in blight unit accumulation. Discussion In this paper, we focused on the higher emission A2 scenario 2050 time-slice because it is a marker scenario used by the Intergovernmental Panel on Climate Change (2000), it most closely matches current emission levels and because the rst year of the time-slice, 2040, is a realistic goal for long-term research prioritization. In general, under the A2 scenario, global potato late blight risk is likely to increase in the near future and then begin to decrease after the middle of the 21st cen- tury in areas where potato is currently grown. For most potato production in the Northern Hemisphere, this decrease is likely due to a shift in the predominant growing season predicted by the EcoCrop model. The peak season shifts from June to May planting with an increase in other planting dates that occur during cooler parts of the calendar year in the northern hemi- sphere. Because of the higher temperatures by 2050 and 2090 during the historic growing seasons, optimal pro- ductivity in the absence of disease would be achieved by shifting to earlier planting to avoid higher tempera- tures, and this change might also reduce disease risk. However, the predicted effects of climate change are not equal across geographic locations, regions or countries. Not all areas experience an initial increase in late blight risk. The later decrease in late blight risk is because the temperatures move out of the optimal range for P. infestans to infect. In the original SimCast model, 1322 C is the optimal temperature range, requiring the least amount of hours of RH > 90% to accumulate one blight unit. The average temperature observed in the 2090 time-slice is the lowest of all four time-slices in this analysis. In our analysis, the A2 cli- mate data had an increase and then sharp decrease in average temperature across potato-growing regions for the growing seasons, and correspondingly the blight unit accumulations decrease with the increase in tem- perature while relative humidity changes little (0.9%). The SimCast model also treats temperature effects dif- ferently below the 812 C range. At cooler hourly tem- peratures, accumulation of blight units decreases with lower temperatures. Thus, one would anticipate increased late blight risk for the very coolest potato- growing areas (e.g., the highest parts of the Andes). These areas probably represent a small proportion of total potato production in the countries where they occur and may have simply counterbalanced warmer areas, contributing to a small change in risk at the country or regional level over time. The effect of regional changes in temperature and rel- ative humidity is notable in the ecosystems that we compared. Rwanda had a temperature and relative Fig. 5 The change in global potato late blight relative risk as predicted by SimCastMeta model using historic climate normal, 19611990 (1975 time-slice) and 20402059 (2050 time-slice) A2 climate scenario for a resistant potato genotype. IPCC A2 time-slices are ensemble model averages of three global climate models. Blight units are a predictor of biological risk based on weather and potato genotypic resistance. 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12587 CLI MATE CHANGE EFFECTS ON POTATO LATE BLI GHT GLOBALLY 5 humidity prole that was already in the optimal range for late blight development during the historic weather data set. By 2050, the temperature increases by 0.7 C, and relative humidity increases one percent. Contrast this with Nepal or Indonesia, where temperature and relative humidity proles during the predicted growing seasons in both countries are suboptimal for disease development and temperature decreases (Nepal) or increases (Indonesia). Also note, the relatively nearby Ethiopian Highlands are predicted to have a much smaller increase in blight units because the temperature increases and relative humidity remains fairly low for disease development. Where potato late blight risk increases, what are the implications for management? Potato late blight is a challenge to manage, particularly for resource-poor farmers who may have limited access to appropriate fungicides (Kromann et al., 2009; Blandon-Diaz et al., 2011), and limited knowledge of late blight manage- ment. New, effective fungicide compounds have been released in markets in the industrialized countries but these often do not make it to developing countries, or at least not to the more remote areas. Host plant resis- tance, shown in this analysis to be an effective way of adapting to changing risk, would appear to be a better strategy for developing country farmers than fungicide Fig. 6 The change in potato late blight relative risk for the Andean Highlands of Colombia and Ecuador as predicted by SimCastMeta model using historic climate normal, 19611990 (1975 time-slice) and 20402059 (2050 time-slice) A2 climate scenario for a susceptible potato genotype. IPCC A2 time-slices are ensemble model averages of three global climate models. Blight units are a predictor of biological risk based on weather and potato genotypic resistance. 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12587 6 A. H. SPARKS et al. Fig. 7 The change in potato late blight relative risk for the Ethiopian Highlands as predicted by SimCastMeta model using historic cli- mate normal, 19611990 (1975 time-slice) and 20402059 (2050 time-slice) A2 climate scenario for a susceptible potato genotype. IPCC A2 time-slices are ensemble model averages of three global climate models. Blight units are a predictor of biological risk based on weather and potato genotypic resistance. Fig. 8 The change in potato late blight relative risk for the Lake Kivu Highlands region and Rwanda as predicted by SimCastMeta model using historic climate normal, 19611990 (1975 time-slice) and 20402059 (2050 time-slice) A2 climate scenario for a susceptible potato genotype. IPCC A2 time-slices are ensemble model averages of three global climate models. Blight units are a predictor of bio- logical risk based on weather and potato genotypic resistance. 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12587 CLI MATE CHANGE EFFECTS ON POTATO LATE BLI GHT GLOBALLY 7 use, but development and adoption of resistant culti- vars have been slow for several reasons, including the low multiplication rate of potato, lack of functional seed systems in developing countries and the ephemeral nature of resistance in most cultivars that have been released (Forbes, 2012). Current levels of resistance pro- vide some benets, and there is the potential for tech- nological advances to increase the level and/or durability of late blight resistance. There is also the potential for advances in the quality and durability of fungicides used for potato late blight management. Other agronomic practices can also contribute to late blight management, including adjusting planting dates to avoid conditions that favor late blight (Devaux & Haverkort, 1987), use of eld sanitation, and crop geno- type mixtures in the eld (Pilet et al., 2006), though the utility of these practices can also depend on environ- mental conditions (Garrett et al., 2009). Our study of global disease risk includes a number of assumptions that merit discussion. First, our study focused on risk of late blight in the future but did not address the suitability of these areas for potato produc- tion in the future. It is possible that along with a change in risk of late blight, the suitability for potato produc- tion in these areas also changes due to a changing physical or social environment. Second, another aspect of future scenarios that is not addressed by evaluations of average conditions is the potential effect of weather Fig. 9 The change in potato late blight relative risk for the Indo-Gangetic Plain and Himalayan Highlands in Nepal as predicted by SimCastMeta model using historic climate normal, 19611990 (1975 time-slice) and 20402059 (2050 time-slice) A2 climate scenario for a susceptible potato genotype. IPCC A2 time-slices are ensemble model averages of three global climate models. Blight units are a predic- tor of biological risk based on weather and potato genotypic resistance. Fig. 10 The change in potato late blight relative risk for the South-East Asian Highlands in Indonesia as predicted by SimCastMeta model using historic climate normal, 19611990 (1975 time-slice) and 20402059 (2050 time-slice) A2 climate scenario for a susceptible potato genotype. IPCC A2 time-slices are ensemble model averages of three global climate models. Blight units are a predictor of bio- logical risk based on weather and potato genotypic resistance. 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12587 8 A. H. SPARKS et al. extremes and weather variability (Rosenzweig et al., 2001; Garrett et al., 2013). Climate data represent weather means, but not the typical variation in weather. The average effect of late blight across years may be different than the late blight effect in an aver- age year. Late blight risk will likely be affected differ- ently across years as a result of changing weather patterns not represented by climate means. Third, this Table 3 Daily average blight unit accumulation and change during the growing season for the top ten potato-producing countries by number of hectares planted to potato as predicted by SimCastMeta model using historic climate normal, 19611990 (1975 time- slice) and 20402059 (2050 time-slice) A2 climate scenario. A2 time-slices are ensemble model averages of three global climate models. Blight units are a predictor of biological risk based on weather and potato genotype resistance Country Hectares of Potato Susceptible genotype blight units Resistant genotype blight units 1975 2050 Change 1975 2050 Change China 4401727 1.40 1.34 0.06 0.79 0.75 0.04 Russia 3229000 1.12 1.18 0.06 0.61 0.65 0.04 Ukraine 1600000 1.28 1.31 0.03 0.70 0.72 0.02 India 1410000 0.88 0.92 0.04 0.34 0.34 0.00 Poland 811979 2.17 2.21 0.04 1.24 1.27 0.03 Belarus 540000 2.02 1.79 0.23 1.15 1.01 0.14 United States 516590 0.89 0.91 0.02 0.48 0.49 0.01 Germany 284078 1.81 1.96 0.15 1.02 1.11 0.09 Peru 271185 1.45 1.36 0.09 0.71 0.62 0.09 Romania 270000 1.99 2.08 0.09 1.12 1.18 0.06 Table 4 Mean change in blight units from historic climate normal, 19611990 (1975 time-slice) to 20402059 (2050 time-slice) for the A2 climate scenario for select highland or lowland tropical potato production areas where potato is an important crop for pov- erty alleviation and where late blight is difcult to manage because of year-round potato production. A2 time-slices are ensemble model averages of three global climate models. Blight units are a predictor of biological risk based on weather and potato genotypic resistance Agro-ecosystem Country Susceptible genotype blight units Resistant genotype blight units 1975 2050 Change 1975 2050 Change Andean Highlands Colombia and Ecuador 2.25 2.20 0.05 1.03 1.04 0.01 Ethiopian Highlands Ethiopia 0.63 0.71 0.08 0.33 0.37 0.04 Lake Kivu Highlands Rwanda 2.41 3.58 1.17 1.43 2.22 0.79 Indo-Gangetic plain and Himalayan Highlands Nepal 1.49 1.46 0.02 0.85 0.81 0.04 South East Asian Highlands Indonesia 1.85 1.78 0.07 0.75 0.65 0.10 Table 5 Mean change in temperature and relative humidity from historic climate normal, 19611990 (1975 time-slice) to 20402059 (2050 time-slice) for the A2 climate scenario for select highland or lowland tropical potato production areas where potato is an important crop for poverty alleviation and where late blight is difcult to manage because of year-round potato production. A2 time-slices are ensemble model averages of three global climate models Agro-ecosystem Country Temperature (C) Relative humidity (%) 1975 2050 Change 1975 2050 Change Andean Highlands Colombia and Ecuador 22.3 22.8 0.5 83.6 84.7 1.1 Ethiopian Highlands Ethiopia 20.2 21.2 1.0 59.2 62.9 3.8 Lake Kivu Highlands Rwanda 18.4 19.1 0.7 78.7 81.6 2.9 Indo-Gangetic plain and Himalayan Highlands Nepal 15.1 13.5 1.6 73.2 70.8 2.4 South East Asian Highlands Indonesia 25.0 25.4 0.4 82.5 83.6 1.1 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12587 CLI MATE CHANGE EFFECTS ON POTATO LATE BLI GHT GLOBALLY 9 model evaluates average daily risk during the three- month growing season determined to be optimal for potato yield in the absence of disease. The average risk approach used here can be viewed as a conservative estimate of differences in risk, while compound inter- est (Vanderplank, 1963) pathogen reproduction across the season may produce larger differences in disease risk between time-slices and between areas. Fourth, we assumed that inoculum was not limiting, an assump- tion that is generally reasonable for potato late blight. However, for more detailed regional analyses it might be important to evaluate whether inoculum might be limiting in particular locations, and how locations may be linked to sources of inoculum (Sutrave et al., 2012). Effects of regional inoculum load or risk neighbor- hoods could be incorporated in more detailed analyses (Skelsey et al., in review). All else being equal, a loca- tion will experience higher late blight risk if its neigh- bors have higher risk. Spatiotemporal models such as those developed by Skelsey et al. (2009) can model regional interactions of inoculum loads. At larger scales in future scenario analyses, the level of condence in estimates of ne-resolution weather events may not warrant such detailed model evaluation. Finally, our analyses do not include evolution in pathogen responses to the primary driving factors, which in this case are temperature and RH. As noted, the predicted decrease in blight units after 2050 occurs because, at higher temperatures, longer periods of RH are required for infection to occur. Pathogen change at the popula- tion level for temperature and RH responses has been recorded in the past (Mizubuti & Fry, 1998) and even resulted in the reparameterization of a simulation model (Andrade-Piedra et al., 2005). Synthesis of socio-economic models, crop yield mod- els, and data such as generated by SimCastMeta could provide a more integrated view on the effects of climate change on potato yield in the future. Previous studies have incorporated crop growth models with socio-eco- nomic effects, but apparently have not simultaneously included the effects of climate on plant disease (Wei et al., 2009). Other types of more detailed socio-eco- nomic models could incorporate farmers decision- making about use of fungicides and resistant varieties, where the regularity of disease impact may inuence adoption (Lybbert & Bell, 2010). For example, the adop- tion of resistant varieties could be surveyed or modeled and then coupled with the outputs from a plant disease severity model such as SimCastMeta for an estimate of the potential impact of releasing a resistant crop geno- type. Additionally, adoption of plant disease manage- ment that reduces greenhouse gas emissions per unit product can, itself, be considered a form of climate change mitigation (Mahmuti et al., 2009). We have discussed some of the areas for improve- ment in biological and socio-economic models above. Another area for improvement will be the development of coordinated data sets for the global presence of dis- ease (Jeger & Pautasso, 2008; Shaw & Osborne, 2011). Even though potato late blight is one of the most inten- sively studied plant diseases, extensive maps of observed disease severity are not available. This is an even greater problem for less-studied diseases. Without such data sets, ground-truthing of model predictions is limited. Future scenario analyses will also need to be updated, as new information about environmental requirements for pathogens becomes available. A dra- matic example of a change in environmental tolerance is the global spread of more heat tolerant and aggres- sive populations of the wheat stripe rust pathogen (Mi- lus et al., 2006; Hovmoller et al., 2011). There is the potential for P. infestans to develop a different range of temperature optima or greater tolerance for dry condi- tions, a type of evolutionary change, which could mod- ify risk for any of a number of foliar pathogens (Huber & Gillespie, 1992; Caubel et al., 2012). Combining new information about environmental requirements for pathogen lineages, host resistance effects, and climate scenarios will support prioritization in potato research and breeding programs, development agency invest- ment, and extension programs into the future. Acknowledgements We appreciate helpful input from E. De Wolf, D. Hartnett, J.M.S. Hutchinson, H. Juarez, B. Natarajan, A. Nelson, R. Raym- undo, P. Skelsey, J. Stack, and T. Todd. We acknowledge the modeling groups, the Program for Climate Model Diagnosis and Intercomparison (PCMDI) and the WCRP Working Group on Coupled Modelling (WGCM) for their roles in making avail- able the WCRP CMIP3 multimodel dataset. Support of this data- set is provided by the Ofce of Science, US Department of Energy. We also appreciate support by the USAID through the International Potato Center (CIP), by NSF grant EF-0525712 as part of the joint NSF-NIH Ecology of Infectious Disease pro- gram, by NSF Grant DEB-0516046, by the USAID for the SAN- REM CRSP under terms of Cooperative Agreement Award No. EPP-A-00-04-00013-00 to the OIRED at Virginia Tech, by the CGIAR Research Programs Roots, Tubers and Bananas (RTB), and Climate Change, Agriculture and Food Security (CCAFS), and by the Kansas Agricultural Experiment Station (Contribu- tion No. 14-318-J). References Anderson PK, Cunningham AA, Patel NG, Morales FJ, Epstein PR, Daszak P (2004) Emerging infectious diseases of plants: pathogen pollution, climate change and ag- rotechnology drivers. Trends in Ecology & Evolution, 19, 535544. Andrade-Piedra JL, Hijmans RJ, Forbes GA, Fry WE, Nelson RJ (2005) Simulation of potato late blight in the Andes. I: modication and parameterization of the LATE- BLIGHT model. Phytopathology, 95, 11911199. Baker KM, Kirk WW, Stein JM, Andresen JA (2005) Climatic trends and potato late blight risk in the upper Great Lakes Region. HortTechnology, 15, 510518. 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12587 10 A. H. SPARKS et al. Bates JM, Granger CWJ (1969) The combination of forecasts. Operational Research Quarterly, 20, 451468. Beaumont A (1947) The dependence on the weather of the dates of outbreak of potato blight epidemics. Transactions of the British Mycological Society, 31, 4553. Blandon-Diaz JU, Forbes GA, Andrade-Piedra JL, Yuen JE (2011) Assessing the ade- quacy of the simulation model LATEBLIGHT under Nicaraguan conditions. Plant Disease, 95, 839846. Caubel J, Launay M, Lannou C, Brisson N (2012) Generic response functions to simu- late climate-based processes in models for the development of airborne fungal crop pathogens. Ecological Modelling, 242, 92104. Chakraborty S, Newton AC (2011) Climate change, plant diseases and food security: an overview. Plant Pathology, 60, 214. Coakley SM, Scherm H, Chakraborty S (1999) Climate change and plant disease man- agement. Annual Review of Phytopathology, 37, 399426. Devaux A, Haverkort AJ (1987) The effects of shifting planting dates and mulching on late blight (Phytophthora infestans) and drought stress of potato crops grown under tropical highland conditions. Experimental Agriculture, 23, 325333. Forbes GA (2012) Using host resistance to manage potato late blight with particular reference to developing countries. Potato Research, 55, 205216. Fry WE, Apple AE, Bruhn JA (1983) Evaluation of potato late blight forecasts modi- ed to incorporate host resistance and fungicide weathering. Phytopathology, 73, 10541059. Garrett KA, Dendy SP, Frank EE, Rouse MN, Travers SE (2006) Climate change effects on plant disease: genomes to ecosystems. Annual Review of Phytopathology, 44, 489509. Garrett KA, Z u~ niga LN, Roncal E, Forbes GA, Mundt CC, Su Z, Nelson RJ (2009) Intraspecic functional diversity in hosts and its effect on disease risk across a cli- matic gradient. Ecological Applications, 19, 18681883. Garrett KA, Forbes GA, Savary S et al. (2011) Complexity in climate-change impacts: an analytical framework for effects mediated by plant disease. Plant Pathology, 60, 1530. Garrett KA, Dobson ADM, Kroschel J, Natarajan B, Orlandini S, Tonnang HEZ, Valdi- via C (2013) The effects of climate variability and the color of weather time series on agricultural diseases and pests, and decision-making for their management. Agricultural and Forest Meteorology, 170, 216227. Gr unwald NJ, Montes GR, Saldana HL, OaR Covarrubias, Fry WE (2002) Potato late blight management in the Toluca Valley: eld validation of SimCast modied for cultivars with high eld resistance. Plant Disease, 86, 11631168. Hannukkala AO, Kaukoranta T, Lehtinen A, Rahkonen A (2007) Late-blight epidem- ics on potato in Finland, 1933-2002; increased and earlier occurrence of epidemics associated with climate change and lack of rotation. Plant Pathology, 56, 167176. Harrison JG (1992) Effects of the aerial environment on late blight of potato foliage - a review. Plant Pathology, 41, 384416. Hijmans RJ, Forbes GA, Walker TS (2000) Estimating the global severity of potato late blight with GIS-linked disease forecast models. Plant Pathology, 49, 697705. Hijmans RJ, Phillips S, Leathwick J, Elith J (2012) Dismo: species distribution modeling. R package version 0.7-23. Available at: http://CRAN.R-project.org/ package=dismo. Hovmoller MS, Sorensen CK, Walter S, Justesen AF (2011) Diversity of Puccinia strii- formis on cereals and grasses. Annual Review of Phytopathology, 49, 197217. Huber L, Gillespie TJ (1992) Modeling leaf wetness in relation to plant-disease epide- miology. Annual Review of Phytopathology, 30, 553577. Intergovernmental Panel on Climate Change (2000) IPCC special report emissions scenarios: summary for policymakers: A special report of IPCC working group III. Jeger MJ, Pautasso M (2008) Plant disease and global change - the importance of long- term data sets. New Phytologist, 177, 811. Juroszek P, Von Tiedemann A (2011) Potential strategies and future requirements for plant disease management under a changing climate. Plant Pathology, 60, 100112. Kaukoranta T (1996) Impact of global warming on potato late blight: risk, yield loss and control. Agricultural and Food Science in Finland, 5, 311327. Krause RA, Massie LB, Hyre RA (1975) Blitecast - computerized forecast of potato late blight. Plant Disease Reporter, 59, 9598. Kromann P, Taipe A, Perez WG, Forbes GA (2009) Rainfall thresholds as support for timing fungicide applications in the control of potato late blight in Ecuador and Peru. Plant Disease, 93, 142148. Leemans R, Solomon AM (1993) Modeling the potential change in yield and distribu- tion of the Earths crops under a warmed climate. Climate Research, 3, 7996. Luck J, Spackman M, Freeman A, Trebicki P, Grifths W, Finlay K, Chakraborty S (2011) Climate change and diseases of food crops. Plant Pathology, 60, 113121. Lybbert TJ, Bell A (2010) Why drought tolerance is not the new Bt. Nature Biotechnol- ogy, 28, 553554. Madden L, Hughes G, Van Den Bosch F (2007) The Study of Plant Disease Epidemics. The American Phytopathological Society, APS Press, St. Paul, MN. Mahmuti M, West JS, Watts J, Gladders P, Fitt BDL (2009) Controlling crop disease contributes to both food security and climate change mitigation. International Jour- nal of Agricultural Sustainability, 7, 189202. Meehl GA, Covey C, Delworth T et al. (2007) The WCRP CMIP3 multimodel dataset - A new era in climate change research. Bulletin of the American Meteorological Society, 88, 1383-+. Milus EA, Seyran E, Mcnew R (2006) Aggressiveness of Puccinia striiformis f. sp tritici isolates in the South-Central United States. Plant Disease, 90, 847852. Mizubuti ESG, Fry WE (1998) Temperature effects on developmental stages of iso- lates from three clonal lineages of Phytophthora infestans. Phytopathology, 88, 837843. New M, Lister D, Hulme M, Makin I (2002) A high-resolution data set of surface climate over global land areas. Climate Research, 21, 125. Oerke EC (2006) Crop losses to pests. Journal of Agricultural Science, 144, 3143. Pautasso M, Dehnen-Schmutz K, Holdenrieder O et al. (2010) Plant health and global change - some implications for landscape management. Biological Reviews, 85, 729 755. Pilet F, Chacon G, Forbes GA, Andrivon D (2006) Protection of susceptible potato cul- tivars against late blight in mixtures increases with decreasing disease pressure. Phytopathology, 96, 777783. Portmann FT, Siebert S, Doll P (2010) MIRCA2000-Global monthly irrigated and rain- fed crop areas around the year 2000: a new high-resolution data set for agricultural and hydrological modeling. Global Biogeochemical Cycles, 24, GB1011, doi: 10.1029/ 2008GB003435. R Core Team (2013) R: A language and Environment for Statistical Computing, R Founda- tion for Statistical Computing, Vienna, Austria. Available at: URL http://www. R-project.org. Rosenzweig C, Iglesias A, Yang XB, Epstein PR, Chivian E (2001) Climate change and extreme weather events; implications for food production, plant diseases, and pests. Global Change & Human Health, 2, 90104. Savary S, Nelson A, Sparks AH et al. (2011) International agricultural research tack- ling the effects of global and climate changes on plant diseases in the developing world. Plant Disease, 95, 12041216. Seem RC (2004) Forecasting plant disease in a changing climate: a question of scale. Canadian Journal of Plant Pathology, 26, 274283. Seem RC, Magarey RD, Zack JW, Russo JM (2000) Estimating disease risk at the whole plant level with General circulation models. Environmental Pollution, 108, 389395. Shaw MW, Osborne TM (2011) Geographic distribution of plant pathogens in response to climate change. Plant Pathology, 60, 3143. Skelsey P, Kessel GJT, Rossing WaH, van der Werf W (2009) Parameterization and evaluation of a spatiotemporal model of the potato late blight pathosystem. Phyto- pathology, 99, 290300. Sparks AH, Forbes GA, Hijmans RJ, Garrett KA (2011) A metamodeling framework for extending the application domain of process-based ecological models. Ecosphere, 2, art90, doi: 10.1890/ES11-00128.1. Sutherst RW, Constable F, Finlay KJ, Harrington R, Luck J, Zalucki MP (2011) Adapt- ing to crop pest and pathogen risks under a changing climate. Wiley Interdisciplin- ary Reviews: Climate Change, 2, 220237. Sutrave S, Scoglio C, Isard SA, Hutchinson JMS, Garrett KA (2012) Identifying highly connected counties compensates for resource limitations when sampling national spread of an invasive pathogen. PLoS ONE, 7, e37793. Vanderplank J (1963) Plant Diseases: Epidemics and Control. Academic Press, New York. Van Everdingen E (1926) The relation between weather conditions and potato blight (Phytophthora infestans). Tijdschrift over Plantenziekten, 32, 129140. Wei X, Declan C, Erda L et al. (2009) Future cereal production in China: the interac- tion of climate change, water availability and socio-economic scenarios. Global Environmental Change, 19, 3444. 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12587 CLI MATE CHANGE EFFECTS ON POTATO LATE BLI GHT GLOBALLY 11