Accessions totalling 1425 from the Australian Lupin Collection representing 9 Old World lupin species and Lupinus mutabilis Sweet, a New World species, were sampled for seed coat and pod wall percentage, seed weight, and number of seeds per pod. These traits are of importance to the breeding objective of lowering seed coat and pod wall proportions in crop lupins. Seed coat percentage mean values among the species ranged from 12.7 in L. mutabilis to 33.7 in L. pilosus L. The 4 species that have been subject to selection (L. mutabilis, L. albus L., L. angustifolius L., and L. luteus L.) had lower mean seed coat percentages than the other species with little domestication. The rough-seeded lupin species had higher seed coat percentages relative to the smooth-seeded species except for L. micranthus Guss., which had 31.5% seed coat and small seeds. Within L. angustifolius there was no difference between the mean seed coat percentage values for wild v. domesticated or hard v. soft-seeded entries, although wild accessions of L. angustifolius tended to have lower seed weight, higher pod wall percentage, and more seeds per pod than domesticated accessions. There was no correlation in L. angustifolius germplasm between seed coat percentage and pod wall percentage, indicating that selection for one will not influence the other character. Accessions with the lowest seed coat percentage were from Turkey, Greece, and Cyprus, and those with the lowest average pod wall percentage were from Spain and Cyprus. Mean pod wall percentages ranged from 30.9 in L. albus to 57.1 in L. micranthus; figures that are high compared with other legumes. Large ranges in seed weight were found particularly in L. albus, L. pilosus, and L. angustifolius.

In addition to the germplasm collection, 21 Australian cultivars, released from 1967 to 1999, were evaluated at one site over 2 years for the same traits. There was a negative correlation between seed coat percentage and seed weight for both L. angustifolius historical cultivars and germplasm, indicating that further reductions in seed coat percentage could be achieved by crossing large seeded types with low seed coat types. Pod wall percentage was negatively correlated with both year of release and yield, and positively correlated with days to flowering. These data support other findings that breeding for the reduction of pod wall can lead to yield improvements. The germplasm collection assessed here provides lines with lower seed coat and pod wall compared with what is available in breeding lines or cultivars.