Problems of population structure exemplified by the invasive tropical shrub Parkinsonia aculeata l. in northern Australia

Age-class frequency distributions are valuable means of describing plant populations because they can be used to infer population history. Variables other than age are also often used to describe plant populations, either because they more accurately reflect an attribute of interest, or because it is difficult to determine age. However, interpretation of frequency distributions based on variables other than age can be problematic. We discuss these problems and illustrate them using data from six populations of the invasive rangeland shrub Parkinsonia aculeata L. We used three different measures of plant size: height, canopy diameter and stem cross-sectional area. Structures based on these measures were compared with structures based on three different estimates of above-ground biomass derived from them. For each variable, structures differed greatly between populations, and for each population, they were strongly dependent on the variable used to describe it. Population structures based on three-dimensional variables (above-ground biomass) tend to be more strongly positively skewed than those based on two-dimensional (area) measures of plant size. These in turn are more strongly positively skewed than those based on one-dimensional (height, diameter) measures. The statistical basis of this general phenomenon is discussed. The results highlight the difficulties of deriving histories and projecting futures of populations from size-class frequency distributions without accompanying knowledge of the temporal patterns of change in size variables as plants grow.