Hypothetical vegetation models were made to simulate numerical changes in species populations along a single environmental gradient. A single ordination procedure was evaluated by its ability to detect the ecological information in the hypothetical models. The procedure was reasonably successful when the data were drawn from a short length of the gradient but became progressively less so as longer lengths of the environmental gradient were included in the data. This parallels an increase in the number of stands from which each species is absent in the total data set. Zero values appear to mask ecological information, and an intuitive method of assigning "degree of absence" values to the data is described. After this adjustment, ordination patterns were easier to interpret because ecological information was concentrated in fewer axes.