For regeneration, accessions are germinated in small pots in a greenhouse during October and transplanted to irrigated field plots in winter. Normally, 50 plants per accession are established in a 6 m row. In the following spring/summer, dropped pods are vacuumed, threshed, cleaned, and shipped to Pullman,WA for storage and distribution. During regeneration, plants were evaluated for:
Winter vigor : visual scale (1=most, 9= least), determined on 2,466 accessions.
Growth habit : visual scale (1=most, 9= least), determined on 2,222 accessions.
Days to flowering: days from planting to first open flower, determined on 2,487 accessions
Above ground biomass : visual scale (1=most, 9= least), determined on 2,487 accessions
Seed production potential : visual scale (1=most, 9= least), determined on 2,481 accessions.
Weight per 100 seeds (g) has also been determined on 1,776 accessions. For the thirty-eight species examined during the regeneration process, a wide range of agronomic types were found within most species. Eleven of the most important species (M. italica, littoralis, lupulina, minima, murex, obicularis, polymorpha, rigidula, rugosa, scutellata, truncatula) were compared in more detail. M. polymorpha had the highest winter vigor and the fewest average days to flowering at 112. M. lupulina had the lowest winter vigor and the most days to flowering at 145. For growth habit, M. scutellata was the most erect and M. littoralis the most prostrate of the eleven species. Biomass production was highest for M. murex and lowest for M. rugosa. M. rugosa also had the lowest potential seed production with M. obicularis the highest. As expected, M. scutellata had the highest mean 100 seed weight (1.79 g) and M. minima had the lowest (0.12 g).
Canonical discriminate analysis was used to determine if distinct groupings would form using the evaluation data of the eleven species. (Seed weight data was excluded because of the significantly fewer observations available then for the other evaluation factors.) Some separation among species was observed, especially between M. lupulina, M. polymorpha, and a grouping made up of M. scutellata and M. rugosa. For the remaining seven species, the means of canonical functions were quite closely grouped and their standard deviations overlapped. The results suggested that the variability observed between many of the species was too small to allow distinct groupings. But since variation for the evaluation factors within species was high, selection within species appears possible.