Christian Huyghe, Eduardo D. Bolanos-Aguilar, Christian Ecalle and Bernadette Julier

INRA, Station d’Amélioration des Plantes Fourragères,
F-86600 Lusignan, France

The spatial distribution of the seed yield in an alfalfa canopy was analysed. On one canopy in year 2 of the variety Europe grown at Lusignan, France, with a seed yield of 0.8t/ha, all the number of stems of a sub-plot of 1 m_ were harvested and their individual contributions to seed yield assessed. 260 stems per m_ were present at harvest. Only 68% of these stems contributed to seed yield. The 13% most productive stems produced more than 1.25 g of seeds each and contributed 37% of the total seed yield. The different classes of stems according to their seed production showed little difference in mean seed weight per inflorescence. Their yield mainly varied with the number of inflorescence.

The vertical distribution of the seed yield was analysed on a sample of 30 stems, chosen at harvest as highly productive stems. The different inflorescence-bearing nodes were individually analysed as these nodes also represent a time-scale. The seed production per node declined at the upper nodes both because of a lower frequency of occurrence of the inflorescences and of a lower seed weight per inflorescence. The upper nodes also showed a small decline in the number of seeds per pod and of the mean seed weight. The seed yield per node, the frequency of inflorescence occurrence and the seed weight per inflorescence may be modelled according to their position onto the stems with a mathematical model

y=exp(a x_ + b x + c) with y being the modelled character and x the node position.

When comparing different canopies (locations x sowing years) of the cultivar Europe with contrasted seed yields, it was shown that the variation in the seed production per node was mainly associated to variations in the seed weight per inflorescence and little to the frequency of inflorescence occurrence. The maximum seed weight per inflorescence predicted from the model was highly correlated to the seed weight per inflorescence measured on 30 inflorescences sampled at random in the canopy.

Across 10 environmental conditions and for two cultivars, Europe and Magali, there was a highly significant correlation (r=0.813, P<0.001) between the seed weight per inflorescence and the seed yield.

The physiological mechanisms controlling the variation in the seed weight per inflorescence will have to be investigated.