Mitigating edge effects in paleoclimate reconstructions: a comprehensive review and bias assessment of soil brGDGT-based temperature reconstructions

Classical paleoclimate reconstructions rely on a space-for-time substitution approach, but past environmental conditions may extend beyond modern analogs. This introduces biases in reconstructions, known as the edge effect, stemming from no-analog past conditions, statistical limitations, and eco-physiological effects. Quantification and partitioning of the sources of the edge effect remain limited. Here, we examined temperature reconstructions based on branched glycerol dialkyl glycerol tetraethers (brGDGTs) from global modern soils, and quantified the factors contributing to the edge effect. Our analysis shows that climate variables alone are the dominant contributors to the reconstruction bias, accounting for 46%, while soil properties (23%) and topography (13%) represent additional environmental controls. Biotic factors also contribute, with vegetation accounting for 11% and bacterial community structure for 7% of the total bias. We employed a Sparse Identification of Nonlinear Dynamics approach to dynamically simulate brGDGT-based temperature biases. Applying this simulation to paleosols from the Chinese Loess Plateau, we demonstrate a previous overestimation of Last Glacial Maximum temperatures by 1.2–4.2 °C. These findings advance our understanding of edge effects in both modern and paleo-reconstructions, and highlight the need for caution when using proxies, especially under no-analog conditions.