A rules-based model to predict shrub thicket expansion on Atlantic coast barrier islands

Background/Question/Methods

During the past 40 years, sea level has risen by ~15 cm along the Virginia coast.  Over the same time period eight of the fifteen barrier islands within the Virginia Coast Reserve decreased in area while shrub thicket cover increased by more than 400% .  This is especially notable because the dominant shrub, Morella cerifera, is sensitive to salt.  We developed a rules based cellular automata model to examine thicket development and expansion at several spatial scales: swale, cross-island, island, and island chain, to predict shrub expansion and perhaps contraction in response to continued increases in sea level and to varying rates of erosion.  The model incorporates the habitat polygon for M. cerifera with regard to distributional relationships of distance from the ocean shoreline and elevation above sea level.  Island topography and shoreline position matrices for islands were based on LIDAR imagery collected in 2010.  In addition to seed dispersal, seedling mortality, and age till reproduction, the model incorporates shoreline migration and sea level rise due to climate change. 

Results/Conclusions

When incorporating changes in island area, model predictions of shrub expansion on the northern portion of Hog Island followed patterns of shrub expansion based on a time series of remotely sensed imagery from the past thirty years.  The predictions for shrub area were within 10% of the imagery values.  In general thicket area is coupled to island upland area, except for islands with rapidly changing shorelines.  At a smaller scale, thicket expansion was limited to island regions were shorelines were accreting and the model predicted shrub mortality when shorelines erosion increased exposure to salinity.  Predictions of shrub area and changes in shrub area relative to shoreline change scenarios indicate that shrub thicket expansion and contraction is closely related to shoreline migration.  Further, comparisons across spatial scales demonstrated that this modeling approach may be useful for cross-scale analyses and predictions of the effects of climate change in coastal environments.