A method for the diagnosis of yield limitation in maize was tested. The tes
t was based on a survey carried out in west Burkina-Faso, with a total of 4
37 farmers, over a period of 3 years. Data on weather, soil, farm, maize (c
ultural practices, and, on two plots per field, components of yield) were r
ecorded. The diagnosis is based on analysis of yield components, compared w
ith their potential values. 'Realisation indices' for different phases of g
rowth are defined to quantify local stresses, and to locate them in relatio
n to time. A treatment is proposed which allows us to distinguish long-last
ing stresses (such as fertility) from transient or random stresses. In our
conditions, long-lasting stresses appeared to be the most frequent. They re
sult more from cultural practices than from environmental factors. The inte
nsity of the random stresses also appeared to depend on cropping techniques
. The method provided a satisfactory analysis of the different types of sit
uations encountered in the survey. The diagnosis is limited by data availab
le on the abiotic factors which are sources of stress. (C) Inra/Elsevier, P
aris.