Modeling how crop plants evolve under domestication requires estimating among-plant variation in important parameters of the reproductive system, including fecundity -the number of propagules produced -and propagule quality. Measuring these traits poses particular problems in vegetatively propagated crop plants. Unlike seeds, vegetative propagules are not intrinsic biological entities but are prepared by farmers. Propagule number and quality are thus determined by the interaction between plant traits and how farmers prepare propagules. We conducted observations, interviews and experiments to study this interaction in cassava grown by Makushi Amerindians, examining how both sources of variation, in plant traits and in farmers' practices, combine to determine the number and quality of propagules produced. Increased stake mass, determined mostly by stem diameter, leads to increased yield and also to increased asexual 'fecundity' of the resulting plant. Farmers' practices reflect knowledge of this relationship. Diameter is the key criterion in the selection of stems for stakes. Larger diameters are preferred; when thinner stems are used, stakes are cut longer, partially compensating for reduced mass. These results suggest that conscious and unconscious selection to increase 'fecundity' and propagule quality in cassava would act to favor plants with thicker stems. Mean stem diameter is greater, and variation in stem diameter is lower, in little-branched plants. Selection for increased asexual 'fecundity' can thus have led to reduction in the degree of branching, one of the most striking differences between domesticated cassava and its wild ancestors. Measuring variation in asexual fecundity is a key step in analyzing evolution of the mixed clonal/sexual reproductive systems that characterize many vegetatively propagated crop plants.

No abstract