Secondary salinization and its relationship to irrigation are strong incent
ives to improve the tolerance of crops to salinity and to drought. Achievin
g this through the pyramiding of physiological traits (phenotypic selection
without knowledge of genotype) is feasible. However, wide application of t
his approach is limited by the practicalities of assessing not only the par
ents, but also large numbers of individuals and families in segregating gen
erations. Genotypic information is required in the form of markers for any
quantitative trait loci involved (marker-assisted selection) or of direct k
nowledge of the genes. In the absence of adequate candidate genes for salt
tolerance, a quantitative trait locus/marker-assisted selection approach ha
s been used here. Putative markers for ion transport and selectivity, ident
ified from analysis of amplified fragment length polymorphism, had been dis
covered within a custom-made mapping population of rice. Here it is reporte
d that none of these markers showed any association with similar traits in
a closely related population of recombinant inbred lines or in selections o
f a cultivar. Whilst markers will be of value in using elite lines from the
mapping population in backcrossing, this has to be considered alongside th
e effort required to develop and map any given population. This result caut
ions against any expectation of a general applicability of markers for phys
iological traits, It is concluded that direct knowledge of the genes involv
ed is needed. This cannot be achieved at present by positional cloning. The
elucidation of candidate genes is required. Here the problem lies not in t
he analysis of gene expression but in devising protocols in which only thos
e genes of interest are differentially affected by the experimental treatme
nts.