Spatial Gradients of Ecosystem Health Indicators across a Human-Impacted Semiarid Savanna

Area of Science:

• Environmental Science
• Soil Science
• Remote Sensing

Background:

• Soil organic carbon (SOC) dynamics are influenced by complex interactions between soil properties, human activities, and climate.
• High-resolution ecosystem health maps are crucial for understanding these drivers and informing land management strategies.
• Savanna ecosystems face challenges related to soil health and require targeted management interventions.

Purpose of the Study:

• To develop accurate, high-resolution maps of key ecosystem health indicators in South African savanna landscapes.
• To identify the primary drivers influencing soil organic carbon distribution and dynamics.
• To provide data for informed strategic land management decisions aimed at enhancing ecosystem conditions.

Main Methods:

• Systematic field surveys and collection of 320 topsoil samples across two 100-km² landscapes.
• Utilized mid-infrared spectroscopy for predicting soil properties (SOC, pH, sand, clay) with validated models.
• Employed RapidEye satellite imagery and random forest modeling for mapping soil erosion prevalence and herbaceous cover at 5-m resolution.

Main Results:

• Mid-infrared spectroscopy models showed good predictive performance for SOC (RMSEP 1.3), pH (0.2), sand (5), and clay (3.6).
• Mapping models for soil erosion prevalence (AUC 0.80) and herbaceous cover (0.86) demonstrated strong validation results.
• Increasing SOC and pH correlated with decreasing sand content, with soil texture, erosion, and climate identified as key SOC drivers.

Conclusions:

• Despite overall low SOC, soil texture, erosion prevalence, and climate significantly influence SOC dynamics in the studied savanna ecosystems.
• Accurate, high-resolution mapping of ecosystem health indicators is feasible and valuable for land management.
• The generated data will support strategic land management decisions to improve ecosystem health in the region.