Evolution and improvement of cultivated amaranths

Abstract

This group of amaranths was studied using four domesticated species (A. hypochondriacus, A. cruentus, A. caudatus, A. caudatus var. atropurpureus and A. edulis), two ancestral weedy species (A. hybridus, A. powellii) and eight hybrids, namely A. edulis A. hypochondriacus, A. edulis X A. caudatus, A. edulis X A. caudatus var. atropurpureus, A. caudatus X A. hybridus, A. edulis X A. hybridus, A. caudatus X A. hypochondriacus, A. hybridus X A. hypochondriacus and A. powellii X A. hypochondriacus. The parents have perfectly normal meiosis and pollen and seed fertility. Except for A. powellii and A. cruentus (n = 17), the species have n = 16. However, the hybrids may be divided into three groups. The first group contains A. edulis X A. cruentus, involving parents with n = 16 and 17, which failed totally, although, under the same conditions, crosses between A. powellii (x = 17) and A hypochondriacus (n = 16) and those between species with n = 16 succeeded with ease. The second group is made up of A. edulis X A. hypochondriacus, A. caudatus X A. hypochondriacus, A. caudatus X A. hybridus, A. edulis X A. hybridus and probably also A. powellii X A. hypochondriacus. Of these, the two combinations, A. caudatus X A. hybridus and A. edulis X A. hybridus, did not proceed beyond the two-leaf stage. At pachytene, the other hybrids showed unmistakable evidence of structural hybridity, with deletions, long or short differentiated segments and inversions. Although bivalents were formed, they possessed a chiasma frequency lower than that of either parent. There was total pollen and seed sterility. The third group comprises A. edulis X A. caudatus, A. edulis X A. caudatus var. atropurpureus and A. hybridus X A. hypochondriacus, which did not show serious developmental defects, the F1 being vigorous, with good meiotic pairing associated with a reasonable amount of differentiation in the chromosomes leading to 25- 55% fertile pollen and 49 to 66% threshable seed. In the F2 there were 11-18% unthrifty plants, which disturb the ratios of gene combinations controlling the different characters in the two parents. Plants very near one or both parental phenotypes were recovered, and also those showing different degrees of recombination of characters. Amphidiploids from the F1 hybrids showed the typical autoploid or segmental alloploid type of meiosis indicating that the parental chromosomes are quite homologous. In view of the present experimental evidence and possible parallel mutations in different grains and weed amaranths, it is not certain whether the cases of natural hybridization and, in particular, of introgression can be taken as evidence for or against the two hypotheses proposed by Sauer (1967) on the basis of his brilliant ecogeographical, morphological, ethnobotanical and archaeological studies of this group of amaranths. The only point that can be stated categorically is that A. caudatus has given rise to A. edulis. The dominance of the characters of A. caudatus over those of A. edulis strengthens such a view, but the latter is sufficiently differentiated morphologically and genetically to deserve independent status. A. caudatus var. atropurpureus is a fertile but unstabilized hybrid segregate between A. caudatus and A. edulis. This is borne out by its morphological, cytogenetic and breeding behaviour, and its hybrids with A. edulis, and, above all, by the recovery of plants identical with this variety from the F2 progeny of A. edulis X A. caudatus. Whatever the origin of grain types, at present they exist only in cultivation and appear to have a long history, having been selected for large plant body, huge compound inflorescences, large number of female flowers per glomerule, small and soft bracts and pale coloured seed in a dehiscent utricle. At the same time, there has also been inadvertent selection for higher and correctly balanced amounts of protein, carbohydrate and fat.