CONICET | Buscador de Institutos y Recursos Humanos

Questions: Can the species ? environment relationship be understood using cur-rent remote sensing techniques? Can the derived indicators of remotely senseddata serve as a proxy for variables that affect habitat suitability of plant species?Which remote sensing predictors are best associated with woody species occur-rence in a desert environment? How well do models with derived indicators ofremotely sensed data predict the occurrence of these species? What are thepotential distributions of Ramorinoa girolae, Prosopis spp. and Bulnesia retama inthe study area?Location: Ischigualasto Provincial Park, San Juan province, Argentina.Methods: We selected random ﬁeld points from a Landsat 8 OLI to determinepresence/absence of trees species. We calculated Brightness index (BI) using thesame image and used this index to calculate texture measures on a 3 9 3 mov-ing window size. We used the following subset of texture measures: (1) ﬁrst-order: range, (2) second-order: mean, variance, contrast, entropy, secondmoment and correlation. We also calculated Topographic Wetness Index (TWI),slope angle and slope aspect from Global Digital Elevation Model.Results and Conclusion: Second-order mean of BI had an important effect onthe occurrence of target trees species. TWI was an important variable for Prosopisspp. and B. retama, whereas slope angle was important for R. girolae and B. re-tama. In addition, the occurrence of R. girolae was affected by second-order vari-ance of BI and slope aspect; and the presence of B. retama was affected bysecond-order contrast of BI. All the variables that had important effects on theoccurrence of tree species provide environmental information about their differ-ent habitat requirements; therefore, our ﬁndings indicate that the remote sens-ing data are reliable to derive indicators of tree species presence in our studyarea.