Tuesday, August 4, 2009 - 3:20 PM

COS 48-6: Self-organized spatial pattern formation in Hamamelis virginiana

David N. Allen, University of Michigan

Background/Question/Methods

In a 500m by 400m Michigan forest in which all the stems above 10cm gbh have been mapped we find a striking spatial distribution of Hamamelis virginiana individuals.  The understory tree grows in non-random clearly defined patches that do not correlate with any edaphic or environment characters.  We hypothesize this pattern is an example of self-organized spatial pattern formation.  Mathematician Alan Turing in 1952 described how pattern could form in a homogeneous environment in which ‘particles’ (chemical reactants, individuals) facilitated their own growth at a small scale and inhibited their own growth at a larger spatial scale.  It is thought that this process of pattern formation is thought to be responsible for the striking banded vegetation observed in some semi-arid systems.
Results/Conclusions

We hypothesize this Turing-mechanism of pattern formation is responsible for the non-random spatial of distribution of Hamamelis virginiana individuals.  In this system the putative Turing-mechanism results from very local dispersal (small scale facilitation) and Janzen-Connell recruitment limitation (larger scale inhibition). Hamamelis virginiana disperse through mechanical ballistic projection, so local dispersal is well established.  A species-specific seed parasitistiic weevil Pseudanthonomus hamaelidis is most likely responsible for such a Jazen-Connell effect.  Individuals in large patches have a higher rate of seed parasitism and produce fewer seeds.  In addition, a higher density of Pseudanthonomus hamaelidis was found over wintering in Hamamelis virginiana large patches compared to small patches and around isolated trees.  Thus we have at least the beginnings of the processes necessary for Turing-mechanism spatial pattern formation.  These results have been incorporated into a spatially explicit model that produces qualitatively similar patterns to the observed distribution.