Quantifying the Impact of Forestry Practices on Biodiversity Using Causal Inference Methods

There is cumulative evidence that human impact and land-modifications negatively affect global biodiversity. However, understanding in more detail to what extent specific human interventions—particularly forestry practices—affect biodiversity on a local basis is a complex and currently under-explored challenge. Investigating the causal links between these practices and biodiversity changes is essential for evidence-based conservation and restoration strategies.

This study applies causal inference methods to quantify the impact of different forestry management strategies on biodiversity across thousands of observational plots sampled over several decades across the entirety of Sweden which is part of the Swedish National Forest Inventory (Riksskogstaxeringen) (One of the world's oldest continuous forest inventory which started from 1923). These long-term and large-scale observational studies provide a rich dataset that contains information on several types of human impacts in the forest, lists of detected species, as well as other biotic and abiotic features.

Our modeling approach explicitly distinguishes association from causation, thereby providing robust insights into how forestry interventions affect biodiversity. We estimate conditional causal effects, allowing us to examine how the impact of forestry practices varies across ecological contexts such as forest age, biogeographic region. This conditional perspective is critical for identifying heterogeneous responses and tailoring management strategies to local conditions.

To capture nonlinear dynamics in how biodiversity responds after a forestry management strategy, we employ piecewise linear spline regression within the causal framework. This enables us to model how biodiversity indicators changes through time. We also conditioned the analysis with ecological context similar to before. By integrating spline-based regression with causal inference, we achieve a flexible yet interpretable framework that can capture more ecological complexity while maintaining statistical rigor.

The findings address critical gaps in forest ecology by moving beyond descriptive assessments toward causal understanding. Ultimately, this work can provide policymakers and forest managers with actionable evidence to design sustainable forestry practices that balance economic demands with biodiversity conservation goals.