Agriculture Reference
In-Depth Information
and diffusion of information technologies and computer programming, above all during the
1980s, have led to the possibility of elaborating a large amount of data and developing very
complex models (Barrett and Nearing, 1998). Nevertheless, the level of complexity and detail
needed for a specific model depends on the objectives and questions being asked of the model
and on the amount of information (data) and time that is available for model building and
testing. In agriculture, models are mainly used for three purposes: i) by researchers to
examine scientific hypothesis or to study and better comprehend biological/
epidemiological/agricultural systems; ii) by governmental agencies as policy analysis tools, to
assist in best management solutions for problem or susceptible areas; iii) by agronomic
technical services or producers as part of computerized decision support systems to evaluate
optimum management practices. The use of a model to predict the outcome of a disease is
desirable to enhance and trigger management opportunities with the aim of reaching high
technological and/or nutritional and/or productivity quality and/or safety of the production
(Boote
et al.
, 1996; Schaafsma and Hooker, 2007; Maiorano, 2008; Maiorano
et al.
, 2009).
Kranz and Royle (1978) classified the epidemiological models according to their main
objective in i) descriptive (models that provide generalise experimental results), ii) predictive
(models that allow prediction of epidemics), and iii) conceptual (models that allow the
identification of the problem by distinguishing cause from effect). The descriptive models are
easy to comprehend, often require fewer inputs, and often are easier to use and apply, but they
have to be calibrated to each new site and year. On the other hand, the conceptual models are
better able to model genotype X environment interaction, but their complexity makes them
more difficult to understand, to use and to apply, and they also require more input information
(Prandini et al., 2008).
The relevant economic impact (section 1) of FHB for the main cereal producing countries
in the world has stimulated many researchers to develop epidemiological models in order to
simulate FHB infection and DON contamination. Most of these models have been included in
computerized decision support systems mainly used to help the operator in the filed to decide
weather or not applying and the correct timing of application of fungicides against FHB.
A review of the models that have been developed for FHB and for DON synthesis is
presented in this section. A similar review has recently been presented by Prandini et al.
(2008).
5.1. Argentina
Moschini and Fortugno (1996) developed nine empirical equations, in the region of
Pergamino, Argentina, associating mean head blight incidence data with temperature and
moisture variables. The models were obtained with linear regression techniques by fitting a
12-year dataset (1978-1990). The two equations showing the best results were validated with
data from Pergamino for the years 1991, 1992 and 1993 and subsequently also in the northern
Argentinean Pampas Region (from 1993 to 1995) by Moschini et al. (2001). The
meteorological variables (temperature, humidity, rainfall) and the time segment of
observation (beginning eight days prior to the heading date and finishing after the
accumulation of 530 degree days) where chosen for their agreement with the fungal
environmental requirements. The authors also incorporated a factor for cultivar susceptibility
Search WWH ::
Custom Search