Models of Clonal Growth in Solidago Altissima

Publisher Website
Google Scholar
Abstract
Patterns of clonal growth are described for the old-field perennial Solidago altissima in three fields located in the Finger Lakes region of New York [USA]. S. altissima shows considerable variation within fields for branching angles, rhizome lengths and numbers of daughter rhizomes. There also was variation between fields for rhizome lengths. Statistical analyses for autocorrelation over time suggest that S. altissima rhizome lengths and branching angles are independent of each other, and of previous branching angles and rhizome lengths. The modal direction for S. altissima clonal growth was 0.degree.. Because clonal growth is highly variable, deterministic models should not be used to describe the spread of S. altissima clones. Instead, the process of clonal growth is consistent with stochastic simulation models and random-walk models: the critical assumption of these models, that branching angles and rhizome lengths are not correlated over time, appears to be satisfied by this clonal plant. This study represents the first application of random-walk models to the vegetative spread of a clonal plant species. There was no statistical difference between observed ramet displacements from a point of origin and displacements expected in a correlated random walk. This result allowed an exact formula to be used to make quantitative predictions about the expansion of S. altissima clones. Random-walk models can be approximated by diffusion models when time is large. Therefore, an estimate was developed of the effective rate of diffusion, i.e. the diffusivity that would be observed if long-term data were gathered on the expansion of S. altissima clones. This approach yielded an effective diffusion constant estimate of 322 cm2 year-1 for S. altissima clones. Field data and computer simulations were used to compare the accuracy of the (effective) diffusion constant estimate developed here to the estimate obtained by standard regression techniques. If only short-term data are available, the correlated random-walk estimate is more accurate than the standard regression estimate. If long-term data are available, the two approaches yield similar results.