Graduation Year


Document Type




Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Marine Science

Major Professor

Pamela Hallock Muller, Ph.D.

Committee Member

David Palandro, Ph.D.

Committee Member

Kathleen Semon Lunz, Ph.D.

Committee Member

Kendra L. Daly, Ph.D.

Committee Member

Chuanmin Hu, Ph.D.


Acropora palmata, A. cervi cornis, restoration, GIS


While populations of nearly all stony coral species along the Florida reef tract have exhibited decline, the most notable decline has occurred in the once-dominant acroporid species (Acropora cervicornis, A. palmata). Both species were listed in 2006 as threatened under the Endangered Species Act. This listing, combined with their continued decline, has resulted in large-scale restoration efforts throughout Florida and the Western Caribbean. Currently, there is little to no information regarding spatial prioritization of sites for these restoration efforts. The primary objective of this dissertation was to utilize species distribution modeling, informed by existing data from the Florida reef tract, to identify sites for restoration of acroporid corals that should have strong likelihood for success.

The initial focus was to use a database of reported field observations, in combination with benthic habitat maps, to model the extent of suitable habitat for Acropora spp. The mapped coral reef and hardbottom classifications throughout Florida, Puerto Rico, and the US Virgin Island reef tracts were used to generate potential-habitat polygons using buffers that incorporated 95% and 99% of reported observations of Acropora spp. Resulting maps demonstrated that A. palmata habitat is relatively well defined, while that of A. cervicornis is more variable and difficult to constrain.

Thus, as the major focus of this dissertation, available monitoring data from the Florida reef tract were used to construct and compare two types of statistical species distribution models, random forest and boosted classification trees, as an approach to inform siting of restoration efforts for A. cervicornis. Boosted classification trees were more accurate than the random forest model at classifying A. cervicornis population trends. Further analyses of the two most important environmental parameters identified by the model, depth and light availability, revealed that reef areas predicted to not have had A. cervicornis present from 1996-2013 were deeper, on average, and had lower light availability and greater variance than areas predicted to have had continuous or transient A. cervicornis presence over this time frame.

This study represents a first step at deriving an ecologically-guided approach to spatial prioritization of restoration efforts. The overarching goal of this project has been to design a strategy for creating models to define and predict where conservation and restoration actions should be most effective, that can be utilized for a variety of coral species. With existing populations mapped, the results can also aid in protecting the limited areas in which these species still occur.