Using the cornell late blight simulator to test biological questions: a case study on the influence of the source of irrigatio

Using the Cornell late blight simulator to test biological
questions: a case study on the influence of the source of
irrigation water on Phytophthora infestans development
O Frenkel1, D Shtienberg1 and U Yirmiyahu2 Computer simulation models, if properly validated, can be used as research tools. The models can be used for testing biological hypotheses by conducting simulation experiments, reflecting "real-life" situations. To illustrate this approach, a computer simulation model, developed in Cornell University, was used. The model simulates the development of Phytophthora infestans (the causal agent of potato late blight) based on microclimatic data. Below, we will first describe the background and the biological question and hypothesis, and than, the experimental approach used to test the hypothesis. Potatoes are grown in Israel in several distinct regions: the northern Negev (ca. 8,000 ha), the coastal plain (ca. 3,000 ha), the Khula valley in the north (ca. 150 ha) and the Arava valley in the south (ca. 100 ha). All potato fields are irrigated, but the source of the irrigation water varies among the different production areas. Fresh water is used in the coastal plain and the Khula valley, municipal recycled water in the northern Negev and saline water is used in the Arava valley. Late blight is the most destructive foliar pathogen of potatoes in Israel. Observations made in potato fields in the different regions suggested that the potential intensity of the disease differs in the different regions. Whereas severe epidemics occasionally develop in the coastal plain and the Khula valley, epidemics in the northern Negev are usually moderate and late blight seldom develops in potatoes grown in the Arava valley. Several factors (and their interactions) may create these differences, among which are differences in production practices, microclimatic conditions and variation in the aggressiveness of the prevailing P. infestans isolates. However, as the different sources of the irrigation water coincide with the variable intensity of late blight epidemics in the different regions, it is possible that the source of the irrigation water plays a role in determining late blight intensity. To initially test if this possibility is relevant at all, late blight development over time was recorded in autumn 2001 in two potato fields located in the northern Negev. The fields were located at a distance of approximately 1 km apart from each other. One field was irrigated with fresh water whereas the other was irrigated with municipal recycled water. The fields were planted with the same potato cultivar and they were maintained (in respect to irrigation, fertilization, weed, pest and disease management, etc.) similarly. Late blight was observed in the field irrigated with recycled water about two weeks earlier, and the final severity of the disease in that field was higher than in the field irrigated with fresh water (Figure 1A). As the onset of the disease did not coincide in the two fields, a conclusive conclusion on the influence of the source of the irrigation water on the epidemic could not be obtained by comparing the actual disease severity values. Logit transformation of the percent disease severity values enabled to estimate the apparent infection rate (r) of the two epidemics. It was found that late blight developed significantly more rapidly in the field irrigated with fresh water than in the field irrigated with the municipal recycled water (Figure 1B). The apparent infection rate could have been affected by the source of the irrigation water, but also by microclimatic influences. Measurements of temperature and relative humidity in these two fields revealed that the differences between them were negligible (Figure 2). Thus, these data supports the hypothesis that the source of the irrigation water plays a role in late blight development and intensity. 1 Department of Plant Pathology, ARO, the Volcani Center, Bet Dagan, Israel 2 Department of Soil Chemistry and Plant Nutrition, ARO, Gilat Research Center, Israel Figure 1. Effects of the source of the irrigation water on the development of potato late blight in fields irrigated with fresh water or with municipal recycled water. The vertical bars indicate the SE. Figure 2. Changes in the relative humidity (A) and temperature (B) in a representative week as recorded within the canopy of two potato fields in autumn 2001. One field was irrigated with fresh water and the other with municipal recycled water. Even if our conclusion based on these observations is correct and the source of the irrigation water affected late blight intensity in these two fields, it does not necessarily imply that this is one of the factors involved in the differences in late blight intensity between the different potato production areas. As the microclimatic conditions in the northern Negev are markedly different from those prevailing in the coastal plain, and since the microclimate governs late blight intensity, it would not be possible to exclude the influence of the microclimate from that of the source of the irrigation water. One possible way to differentiate between the two factors (and thus to expose the sole influence of the source of the irrigation water) is to take the effects of the microclimate into account in the analysis. This can be done by using the late blight simulator developed in Cornell University. The model was developed in the early 1980's by Bruhn and Fry (Bruhn and Fry, 1981; 1982a; 1982b) and it was revised and validated with independent data set in the early 1990's (Doster et al., 1990). The data used for validation was recorded in upstate New York, the region where the model's parameters were quantified. The validation efforts revealed the model accurately predicted late blight epidemics in that region. However, as the simulator was not yet validated under the local Israeli conditions, before using it for testing the role of the source of the irrigation water on late blight intensity, initial attempts should have been devoted for its validation. The late blight simulator was validated under Israeli conditions using data recorded in four field trails. The trials were conducted in the coastal plain production area in the spring of 2001 and 2002. Each experiment consisted of plots not treated and plots treated with fungicides against P. infestans. In this report, results of the non-treated plots are presented. Disease severity was assessed periodically from the date of disease onset until crop maturity. Temperature and relative humidity recorded within the potato canopy by means of data-loggers were used to run the simulator. Then, predicted epidemics were compared with those actually observed in field. Visual comparison of observed and simulated epidemics revealed that the simulator reasonably predicted the epidemics that actually developed in the field in three out of the four experiments (Figure 3). However, visual comparison is a subjective, non-parametrical validation procedure. A more objective validation of the simulator could be achieved by statistical comparison of predicted vs. observed severities. This analysis suggested that the prediction ability of the simulator was high up to an observed severity of approximately 60%. Above that severity, the model generally under-estimated the actual disease severities and the variation of the differences between observed and predicted values was high (Figure 4A). When observed and predicted epidemics were compared up to disease severity of 60%, the predictions provided by the simulator were adequate. The intercept of the regression equation describing the coincidence between observed and predicted severities (0.4%) did not differ significantly from 0% and the slope of the regression equation (0.93) did not differ significantly (t- test; P<0.05) from 1 (Figure 4B). Based on these findings we concluded that the late blight simulator developed in Cornell University provided satisfactory predictions of P. infestans epidemics in Israel, up to disease severity of 60%. This implies that the model can be used as a research tool for studying biological questions. Figure 3. Using the late blight simulator developed in Cornell University to predict the severity of late blight epidemicsin four field experiments. The experimental fields were located in the coastal plain of Israel and irrigated with freshwater. Squares: observed disease severity in the field; line: simulated disease progress. Figure 4. Comparison of late blight severity observed in four field experiments with the severities predicted by the Cornell late blight simulator. A: all data points recorded in the experiments; B: data points up to observed severity of
60%. The 1:1 line is a theoretical line representing a perfect coincidence between observed and predicted values;
the dashed lines represents + 10% of the 1:1 line; the thick line represents the regression equation between
observed and predicted values.
The next step was to use the model for prediction of late blight intensity in the northern Negev region, where the fields are irrigated with municipal recycled water. Two trials were conducted for this purpose in autumn of 2001. As in the previous trials, disease severity was assessed periodically from the date of disease onset until crop maturity and temperature and relative humidity were recorded within the potato canopy. The predicted epidemics were compared with those actually observed in field. Visual comparison of observed and simulated epidemics revealed that the predictions issued by the simulator overestimated the epidemics that actually developed in the fields (Figure 5). Comparing all observed and predicted epidemics corroborated these conclusions: the slope of the regression equation (1.34) was significantly higher than 1 (t- test; P<0.05) (Figure 6). Figure 5. Using the late blight simulator developed in Cornell University to predict the severity of late blight epidemics in two field experiments conducted in autumn 2001. Both fields were located in the northern Negev and irrigated with municipal recycled water. Squares: observed disease severity in the field; line: simulated disease progress. Figure 6. Comparison of late blight severity observed in four field experiments with the severities predicted by the Cornell late blight simulator. The 1:1 line is a theoretical line representing a perfect coincidence between observed and predicted values; the thick line represents the regression equation between observed and predicted values. Concluding Remarks
Based on the findings described above we concluded that the development of late blight in fields irrigated with municipal recycled water is less severe than could be expected based on the suitability of the environmental conditions to the pathogen. Thus, it turns that some factor(s), which presumably is/are included in the recycled irrigation water, inhibit the rate of P. infestans development. In further studies it was found that Boron, a microelement that exists in high concentrations in recycled water, induces systemic acquired resistance against P. infestans and also against Alternaria solani (the causal agent of early blight). This study demonstrates the possibility of using computer simulation models as a research tool. The late blight simulator developed in Cornell University has been used for addressing biological questions in the last 20 years. However, it was used primarily in New York State. To the best of our knowledge, this is the first attempt to validate the simulator outside upstate New York. Our experience suggests that this particular simulator may be accurate, and thus, could be used as a research tool by researchers in other places and in other environments. Acknowledgments
This research was supported in part by under Grant No. TA-MOU-99-C15-073 US – Israel Cooperative Development Research Program. This project is conducted in collaboration with G A Forbes from CIP, Quito Ecuador and W E Fry from Cornell University, Ithaca, New York, USA. Literature cited
Bruhn J A and Fry W E. 1981. Analysis of potato late blight epidemiology by simulation modeling. Bruhn J A, and Fry W E. 1982a. A statistical model of fungicide deposition on potato foliage. Phytopathology Bruhn J A, and Fry W E. 1982b. A mathematical model of the spatial and temporal dynamics of chlorothallonil residues on potato foliage. Phytopathology 72:1306–1312. Doster M A, Milgroom M G and Fry W E. 1990. Quantification of factors influencing potato late blight suppression and selection for metalaxyl resistance in Phytophthora infestans: A simulation approach. Phytopathology 80:1190–1198.

Source: https://research.cip.cgiar.org/confluence/download/attachments/37192047/CGTE_Frenkel.pdf?version=1&modificationDate=1273383466000