Carbon isotope discrimination and indirect selection for seed yield in lentil

Carbon isotope discrimination (CID) has been proposed as a secondary trait to indirectly select for improved seed yield and water-use efficiency. To determine the effectiveness of CID to indirectly select for seed yield, 10 diverse lentil (Lens culinaris Medikus) genotypes were grown at four locations in Saskatchewan in 1992 and 1993. Samples were collected for CID determination from leaves at flowering (CIDLF), leaves at maturity (CIDML) and seed at maturity (CIDMS). Variability for CID was present, but no crossover G x E interactions were observed. A significant crossover genotype by sampling date interaction for CID resulted from a change in ranking of genotypes for CIDLF and CIDML. Seed yield was positively correlated with CIDLF and CIDML. The phenotypic correlation between seed yield and CIDLF was 0.82** (df = 8). However, this highly significant relationship was an artifact resulting from the fact that samples for CIDLF were collected for all genotypes on the same day, although PI 244026, an early maturing genotype, flowered about 15 d earlier than the other lentil lines. When PI244026 was removed from the analysis, variability in CID was greatly reduced and the correlation between CIDLF and seed yield approached zero (r = 0.22, df = 7). The correlation between seed yield and CIDML remained positive and significant, but was inconsistent across locations and years. The broad sense heritability for CIDML (0.73 +/- 0.16) was lower than the broad sense heritability for seed yield (0.98 +/- 0.029). These results suggest that under the conditions tested CID should not be used to indirectly select for seed yield in lentil