Characterization of depression forests in a karstic island: A geostatistical and multivariate analysis approach

Within a study to determine the effect of introduced feral goats and pigs on the vegetation of Mona Island (Puerto Rico) we considered necessary to define more objectively our study sites, commonly named “bajuras” (depression forests) and regarded as one type of forest.  To characterize and compare three depression forest sites and three plateau forests, we established 250 m transects along the largest axis of each site.  Every 15 m we measured abundance and diversity of plant species (except epiphytes) inside circular plots (3 m radius) for individuals above one meter height, and in two one square meter plots for individuals below one meter.  We also calculated canopy cover and light penetration indexes above the one square meter plots using hemispherical photography.  Soil samples were obtained for physical and chemical analyses.  Geostatistical analyses (semivariance) showed different types of patterning, with depression forests presenting repeated patterns but also strong gradients.  Multivariate analysis (detrended correspondence analysis and detrended canonical correspondence analysis) showed a separation, along the main axis, of plots of depression forests from plots of plateau forests, but also many overlapping plots and effect of the secondary axis on this group.  When included the environmental variables (DCCA), the main axis appears defined by the soil hardness in one direction and P and Na in the opposite direction, meanwhile the secondary axis appears defined by organic matter, N, Mn, and K contents.