Accurate mapping of forest carbon stocks and biodiversity is critical for guiding conservation and climate mitigation strategies, particularly in topographically complex mountain regions. We integrated remote sensing data (optical, SAR, LiDAR), topographic variables, and field inventories into a Random Forest Regression using a Nested cross-validation framework to model AGB carbon stock across native forests and pine plantations in the semiarid Andean Patagonia. Specifically, we aimed to: (i) classify vegetation types; (ii) estimate AGB carbon stock and uncertainty; (iii) map the effects of vegetation and topography on AGB carbon stock; and (iv) compare understory biodiversity among vegetation. The model achieved moderate to high performance (R-2 = 0.78, RMSE % = 33.86%). Pine plantations and native forest had significant differences in AGB carbon, but at mid elevations (1400-1500 m a.s.l.) and gentle slopes (0-10 degrees)-, both stored comparable values (100 Mg C ha(-1)). Native forests supported higher biodiversity than pine plantations (species richness: 4.60 +/- 0.34 vs.0.10 +/- 0.07, respectively), and occupied a broader topographic range. Elevation and slope emerged as key drivers shaping the spatial distribution of AGB carbon storage for both vegetation. Management strategies should prioritize the conservation and restoration of native forests in topographic settings where they achieve high AGB carbon stocks while sustaining endemic biodiversity.