Monitoring soil conservation techniques via UAV for sustainable production of intercropped forage cactus with reuse water in the Brazilian semiarid region

Context

Sustainable agricultural production in semiarid regions is limited by water scarcity and soil degradation. Forage cactus (Opuntia stricta) has high drought tolerance but requires effective water and soil management to maximize yield.

Objective

To evaluate the effects of mulching and intercropping with gliricidia and moringa, under treated wastewater irrigation, on forage cactus productivity and soil properties, and to assess the utility of UAV-based remote sensing and machine learning for field-scale monitoring and yield prediction.

Materials and methods

A 15-month field experiment was conducted in Northeastern Brazil using a randomized block design (3 × 2 factorial; six treatments; four replicates). Treatments combined mulching (with and without mulch) and intercropping (gliricidia, moringa, or none). Biometric (height, cladode count), biomass (fresh and dry matter yield), and soil (organic carbon, electrical conductivity, moisture) variables were measured. High-resolution UAV multispectral imagery provided vegetation indices. Geostatistical analysis was applied for spatial variability mapping, and a Random Forest model predicted fresh matter yield (R² criterion).

Results

Mulching increased forage cactus height by 21 % and fresh matter yield by 70 % (p < 0.05), raised soil organic carbon by 133 %, and reduced salinity by 61 %. Intercropping with gliricidia further improved fresh matter yield by 59.7 % and soil moisture compared to the non-intercropped control. UAV-derived indices (e.g., NDVI, vNDVI) correlated strongly with measured yields (r > 0.75), and the Random Forest model achieved R² = 0.83 in yield prediction.

Conclusions

Under irrigation with treated wastewater, mulching and intercropping (particularly intercropping with gliricidia) were associated with the highest forage cactus productivity and improved soil indicators in semiarid conditions. UAV monitoring with machine learning supported field-scale diagnostics and fresh matter yield prediction.