The Caatinga biome, South America's largest tropical dry forest, faces critical degradation due to extensive land-use changes, particularly through the conversion of native vegetation into pasture and subsequent pasture degradation. Agroforestry systems (AFS) have been proposed as sustainable alternatives for rehabilitating these degraded landscapes by improving vegetation structure, biomass productivity and soil functionality. This study evaluated the effects of four contrasting land-cover systems: (i) preserved native Caatinga vegetation (reference ecosystem), (ii) long-term pasture dominated by Urochloa brizantha, (iii) low-density agroforestry with widely spaced tree-crop arrangements and (iv) high-density agroforestry with structurally complex and closely spaced woody components-on ecosystem multifunctionality, aboveground biomass production, litter deposition, litter nutrient content and key soil fertility and carbon parameters in Oxisols of the Brazilian tropical seasonal dry forest. We hypothesized that habitat simplification through land-use intensification reduces above- and belowground ecosystem functionality by impairing carbon cycling, nutrient dynamics and soil structure. Our findings support this hypothesis, showing that the natural ecosystem exhibited the highest aboveground biomass (22.47 t ha(-1)), litter deposition (849.98 kg ha(-1)) and litter C, N and P contents (357.40, 10.88 and 3.16 g kg(-1), respectively), significantly outperforming managed systems. Among agroforestry systems, high-density configurations showed significantly higher biomass production (69.76% greater) and improved soil fertility parameters-soil pH (17.65% higher), SOC (18.75% higher), K+ (129.41% higher) and Ca2+ (57.43% higher)-compared to low-density systems (p < 0.05). Soil analyses revealed that the natural ecosystem had superior levels of light fraction organic matter, humic substances and total soil organic carbon, alongside lower bulk density and higher acidity. Conversely, pasture had the highest particulate organic carbon and SOC stock, suggesting contrasting carbon stabilization pathways under different land uses. Principal Component Analysis (PCA) distinguished land covers based on multidimensional litter and soil traits, with key indicators such as pH, base cations (Ca2+, Mg2+), organic carbon fractions and litter nutrient contents explaining much of the variance. These results highlight that complex vegetation systems, including high-density agroforestry, maintain better soil and biomass functionality than simplified land uses. The study contributes valuable insight into sustainable land management practices aligned with the United Nations Sustainable Development Goal 2 (Zero Hunger), emphasizing the need for multifunctional systems to restore ecosystem services and improve productivity in semi-arid regions like the Caatinga.