MORPHOLOGICAL AND CYTOLOGICAL VARIABILITY IN TRIPSACUM DACTYLOIDES (GRAMINEAE)

A total of 235 accessions of Tripsacum dactyloides (L.) L. representing natural populations from 88 locations distributed throughout ten states of the United States were collected in the wild and studied morphologically and cytologically. Multivariate analysis of variance (MANOVA) was applied to 17 morphological characters recorded for 868 specimens from the 88 locations, which were further combined into 25 groups to facilitate analysis. Multivariate F ratios for six comparisons involving different groups were highly significant, indicating that the populations were significantly different on the basis of all 17 characters. Non-significant MANOVA F ratios for two comparisons between sympatric diploid and polyploid accessions showed them to be morphologically similar. Discriminant function (DF) analysis was employed to determine which characters contributed towards group discrimination. No one variable was found to distinguish between groups; rather, a combination of characters was necessary, and the combination changed with each comparison. Accessions were mainly either diploid (2n = 36) or tetraploid (2n = 72), with two triploids (2n = 54) reported from separate locations in Texas. Cytological data revealed that diploids and tetraploids occupy a more sympatric distribution than previously supposed. Morphological analysis suggested that diploids could not always be distinguished from polyploids on the basis of morphology alone, unless the appropriate characters were used. Information from cytological and morphological data showed Tripsacum dactyloides to be composed of many populations covering a wide range of variability, with each population possessing a characteristic combination of morphological attributes. It was concluded that T. dactyloides can be most meaningfully analyzed in terms of particular geographic regions, rather than attempting to elucidate natural relationships by correlating data from the entire United States.