Arid and semi-arid forest ecosystems represent the largest biomes on Earth. However, research on identifying their species using remote sensing techniques is still limited. Understanding the spatial distribution of vegetation is crucial for precision management. This can be achieved through methods that allow for the individual identification and classification of species, which are essential for accurately estimating forest inventory. The objective of this study was to identify and classify forest species present in a xeric shrubland (arid and semi-arid region) based on multispectral images, red-green-blue (RGB) images, and Light Detection and Ranging (LiDAR) data. All images and data were drone-captured. Machine learning algorithms such as Adaptive boosting (AB), Gradient boosting machine (GBM), Xtreme gradient boosting (XGB), Classification and regression trees (CART), Random Forest (RF), and Support vector machines (SVM) were employed. RF yielded better results for species and shrub class classification, with an accuracy of 0.64 and a Kappa coefficient of 0.56. Classification accuracy values per species were 0.73 (E. antisyphilitica), 0.70 (opuntias), 0.67 (palms), 0.65 (L. tridentata), 0.59 (trees and shrubs), and 0.55 (A. lechuguilla), all of which were obtained by combining the three types of data used. Spectral variables contributed the most metrics, followed by Li-DAR and RGB. The results support the adoption of remote drone-mounted sensing systems for characterizing the complex forest vegetation in arid and semi-arid regions, thereby providing a decision-support tool for its management.