Accuracy, precision, and repeatability of loss-on- ignition protocols for estimating soil organic carbon relative to dry combustion

Soil organic matter (SOM), including its primary constituent organic carbon, is fundamental to soil health and a major component of the global carbon cycle. Reliable assessment of SOM stocks requires precise, repeatable measurements and sufficiently large sample sizes to capture spatial heterogeneity and detect slow temporal change. Loss‑on‑ignition (LOI) provides a low‑cost approach for increasing sample throughput, yet the use of multiple LOI protocols has limited comparability across studies and obscured understanding of associated measurement error. This study evaluated three LOI methods for estimating SOM and predicting soil organic carbon (SOC) measured by dry combustion (DC): ignition at 360°C for 2 h (LOI360), 400°C for 16 h (LOI400), and 550°C for 3 h (LOI550). Linear conversion equations were developed to relate LOI‑derived SOM to DC‑SOC, and model performance was assessed using R², RMSE, MAE, MedAE, and Lin’s concordance correlation coefficient. LOI360 and LOI400 produced comparable accuracy and precision in predicting SOC, but LOI400 exhibited the strongest repeatability. LOI400 also provided an effective balance between incomplete SOM combustion at lower temperatures and mineral mass loss at higher temperatures, whereas LOI550 performed poorly due to variable mineral decomposition. Localized conversion equations grouped by landform region did not substantially improve SOC predictions relative to a single generalized equation. Although LOI400 showed greater measurement variability than DC, its low cost enables larger sample sizes, thereby improving characterization of spatial SOC variability. Using the generalized equation SOC = 0.365 * LOI400 − 0.039, LOI400 represents a reliable and economical alternative to DC for SOC estimation in resource‑limited settings.