Statistics and Results

We developed a habitat suitability index (HSI) for SAV beds/types through a decision tree approach that utilized parameters of shore vegetation and geographic/topographic information. A decision tree assigns a probability to each of the possible choices (a future event), based on the context of the decision (history). For data analysis we used Classification and Regression Tree (CART) methods in Rusing the rpart function, and plotted using rpart.prp

To interpret these CART trees start at the top and follow the left branches down. The criterion for making a decision about the Habitat Suitability Index (HSI) of a site for that group of SAV is defined by the cutoff value for each parameter (e.g. SAV grpA = seagrass, the first decision says that Distance from Ocean <6.2km is most suitable). The terminal nodes (or leaves) of the tree at the bottom indicate the probability that SAV will be found under that suite of parameters, and the percent of sites sampled that satisfy those conditions (e.g. SAV grpA has a probability = 0.81 of being found <6.2km from the ocean and with a substrate composed of soft mud. 15% of the 106 sites sampled met those criteria).