04912nam a2200181 a 450000100080000000500110000800800410001910000220006024500790008226001440016150001040030552042230040965000220463265300180465465300180467270000200469070000200471018698492011-01-19       2010    bl uuuu     u00u1 u    #d1 aLARANJEIRA, F. F.  aModelling Huanglongbing (HLB) spread and eradication procedures in Brazil.  aIn: CONFERENCE INTERNATIONAL ORGANIZATION CITRUS VIROLOGISTS, 18., Campinas, SP, 2010. Proceedings... Campinas: IOCV, 2010. 1 CD-ROM.c2010  a029 PSO Publicado também em: Citrus Research & Technology, Cordeirópolis, v. 31, Suplemento, 2010  aHLB is the most important citrus disease in Brazil. Reported in 2004, so far it has been found in São Paulo, Minas Gerais and Paraná states, where a combined strategy of healthy nursery plants, insecticide spraying and removal of symptomatic plants has been advocated. Moreover, there are mandatory eradication procedures to be followed by growers (IN 53). The IN53 establishes that a given orchard must be eliminated if 28% of its plants are found to be symptomatic, assuming that if visible symptoms reach that level, then all plants are actually already infected. Also, IN53 established quarterly evaluations as the minimum frequency in order to eradicate HLB in an orchard. Despite some success in São Paulo, such measures might not be suitable in other Brazilian citrus regions where the groves are small. In those areas even a limited eradication effort can eliminate an entire orchard and in most cases, growers don?t have enough leverage to keep a continuous spraying program. Because of inherent biological uncertainties concerning HLB (long incubation period, range of primary and secondary infection rates) and also because of the risk for other groves, field experiments can be time-consuming, inconclusive or even impossible to carry out. In this work we used a flexible spatially-explicit model to test hypotheses concerning HLB progress, spread and eradication in citrus groves. One thousand HLB epidemics were simulated, registering the number of susceptible, asymptomatic infectious and symptomatic plants over 120 months at fortnightly steps for combinations of infection rates (high, medium, low) and incubation periods (3, 6, 12, 18, 24 and 30 months). The exponential dispersal kernel with a scale of 18m was used; the average latency period was fixed in one month. The simulated landscape was a target grove of ~9.4ha (40 rows with 96 plants each, 7.0m x 3.5m spacing) surrounded by four neighbouring groves, each of them at least 2ha in area and with 5% of plants infected. The virtual experiment was repeated considering quarterly actions for roguing of symptomatic plants. Data were analysed in terms of proportion of detected plants when 100% are infected, time when 28% are symptomatic, area under disease progress curve, time to unprofitability under an eradication programme and relative risk of IN53 criteria. Simulations showed that the 28% criterion corresponds to 100% of infected plants only in a very limited range of infection rate x incubation period. For incubation periods ranging from 6 to 12 months, maximum infection is reached only when ~45% are symptomatic. Moreover, when the 28% criterion is applied ~40% of the plants are still productive (most of them asymptomatic) no matter the infection rate or the incubation period. When quarterly roguing procedures are applied epidemic stabilization is seldom achieved. Based on a criterion for grove unproductivity, derived from citrus yield distributions in São Paulo, it was found that unprofit- ability is reached earlier under quarterly roguing procedures when compared to allowing HLB incidence to increase up to 28% and then eliminating the entire grove. However, the risk of infection is increased 2.5 fold (measured as the ratio between areas under progress curves) if periodic roguing is not performed. When the same analysis was performed using 45% of detected symptomatic plants as the eradication criterion, the infection risk is increased 7x. The 28% criterion was contrasted with a possible criterion of 45% of detected symptomatic plants to eliminate the entire grove. In this case, the grower would ?gain? between 12 and 60 months of grove life but as a trade-off the infection risk would be 2.5x to 3x higher. The model can be further explored as more accurate data on HLB epidemiology and host-pathogen relationship are obtained. Nevertheless, so far all the results show the brazilian eradication procedures can be improved. Special care should be taken to devise a coherent set of criteria in order to ease the enforcement of eradication procedures. Modelling can play an important role is this task by allowing us to better understand HLB?s epidemiology and helping to screen potential control strategies  aDoença de Planta  aHuanglongbing  aPlant disease1 aDESIMONE, E. R.1 aGILLIGAN, C. A.