1ESALQ-Departamento de Fitopatologia, Universidade de São Paulo, Piracicaba 13418-900, Brazil; 2Departamento de Fitopatologia, Universidade Federal de Viçosa, Viçosa 36570-000, Brazil

Background and objectives
Sustainability must be defined in relation to systems, rather than to inputs, crops or diseases. To call a system sustainable means that, given stable inputs and a reasonable period of time, internal interactions within the system will keep key performance criteria within a specified range. Possible threats to the sustainability of a system can be classified as (i) internal threats or (ii) external threats. Both threats lead to unsustainability but, according to a systemic view of the process, internal threats are more serious than external threats.

Sustainable agriculture is concerned with the ability of agricultural systems to remain productive in the long run at socially acceptable economic and environmental costs. Sustainable agriculture can be defined in a strict sense or in a pragmatic sense. The first is based on the concept of total social factor productivity (TSFP), and has an absolute meaning but a difficult operational definition. The second allows for agroecosystems with different degrees of sustainability and considers systems more sustainable, given similar inputs and outputs, that (i) use fewer non-renewable inputs and (ii) use fewer external renewable inputs. It is more reasonable, therefore, to reduce the number and impact of unsustainable actions rather than to aim for a theoretical absolute state called sustainability.

The sustainability of coffee rust (Hemileia vastatrix) and Coffea arabica management in Brazil will be analysed in this paper based on epidemiological data collected in six regions and on three cultivars during two consecutive years, and on the actual situation of rust control in Brazil. Some of the conclusions can be extended to other tropical pathosystems.

Results and conclusions
Contrary to what happened historically in other regions of the world, as in Asia and Africa, the introduction of coffee rust in Brazil in 1970 did not result in change of the cultivated varieties. The Brazilian varieties Mundo Novo and Catuai represented, and still represent, more than 90% of the area occupied by coffee in Brazil. From an epidemiological point of view, the two varieties differ from others, such as the Ethiopian Ibaare-Harar, by a shorter leaf-retention period after initiation of sporulation by the pathogen. This characteristic is inherited in a quite dominant way, is likely to be race-non-specific, and is sufficient to eliminate completely the residual inoculum from one season to another in some regions. This effect, due to the usually low infection rate of Hemileia vastatrix, is enough to keep the disease under control in many areas without any need for chemical sprays. In other areas, however, 1-4 sprays may be necessary. The decision to spray or not is based on a fixed calendar schedule modified each year according to (i) date of disease onset; (ii) actual disease incidence; and (iii) current berry yield. This management programme has been tested for many years, can be considered internally robust, and has shown an adequate level of sustainability. However, if treatments are missed because of rain or a damaged sprayer, the infection may increase rapidly enough to cause economic damage, especially when yields are heavy.

A new strategy to manage coffee rust was developed recently and is based on the application once a year of a mixture of systemic fungicide (triadimenol or cyproconazole) + systemic insecticide (disulfoton) to the soil. Control of rust and of the leaf miner Perileucoptera coffeella is adequate, and there is less harm to the environment. This strategy is especially recommended for areas with irregular topography.

From the epidemiological characteristics of coffee rust and other diseases of tropical plantation crops (e.g. South American leaf blight of rubber trees caused by Microcyclus ulei) we consider the reduction initial inoculum to have a more important role than previously accepted.