Assessing forest degradation using multivariate and machine-learning methods in the Patagonian temperate rain forest

Area of Science:

• Ecology
• Forestry
• Environmental Science

Background:

• Forest degradation and deforestation are major contributors to global greenhouse gas emissions.
• Quantifying the threshold between degraded and non-degraded forests remains a significant challenge.

Purpose of the Study:

• To determine the critical threshold of forest degradation in southern Chile's temperate evergreen rain forests.
• To identify key forest stand factors indicative of degradation status.

Main Methods:

• Established 160 plots (500 m²) across a gradient of forest alteration.
• Measured variables including species richness (native and exotic), soil nutrients, and tree density.
• Applied multivariate and machine-learning analyses to classify degradation levels.

Main Results:

• Exotic species richness (diameter at breast height < 10 cm) and tree density (diameter at breast height > 10 cm) were key indicators.
• Forest stands with ≥5 exotic species or <200 trees/ha were classified as degraded.
• Pristine forests were characterized by >1,000 trees/ha.

Conclusions:

• A machine-learning based methodology was developed to successfully identify forest degradation status.
• The findings provide valuable data for managers and policymakers in classifying and mapping degraded forests.