The understanding of the changes produced by plant breeding on wheat grain yield and on its determinants, through the study of the behaviour of cultivars released at different times during the breeding process, could be a useful tool to define future selection criteria. The aim of this paper was to examine the genetic improvement effects on bread wheat grain yield potential and on its major physiological determinants in different countries.Generally, it seems that the genetic improvement produced important increases in harvest index, but it did not produce important changes in total biomass. The genetic gain in grain yield ranged from 0.58 g m 2 year 1 (India) to 5.84 g m 2 year 1 (Mexico); reflecting important differences among genetic improvement programmes. However, when the comparisons were made in relative values, the difference was much lower, ranging from 0.16% year -1 to 0.90% year ~, for the previously mentioned countries, respectively. Most of the regions showed relative genetic gains of around 0.45% year 1. Similar patterns of relative genetic gains in harvest index were found.Half of the total gain in grain yield in the past was due to genetic gain in grain yield potential. However, it is expected for the future, that further increases in grain yield will rely more on genetic gain than on technologic gain. Thus, a discussion about possible strategies in future breeding programmes of bread wheat is attempted throughout the paper.The changes in grain yield were mostly associated with changes in number of grains per m 2 rather than with changes in individual grain weight. It appears that the source:sink ratio during the grain-filling period was very high in old cultivars, whereas the source:sink ratio in modern cultivars is more balanced. Further increases in number of grains m 2 could be compensated by a reduction in individual grain weight, if simultaneous improvement in source strength during grain-filling is not achieved.
38The crop attributes which have been changed together with the mentioned traits are discussed and alternative selection criteria are suggested.