The impact of land-use change (LUC) on soil organic carbon (SOC) has been a wide concern of land management policymakers because CO₂ emissions induced by LUC have been the second largest carbon source worldwide. However, due to insufficient data quality and limited biome coverage, a global big picture of the impact of LUC on SOC is still not clear. This study conducted a meta-analysis on 288 independent observations sourced from 62 peer-reviewed papers to provide a global summary of the change in SOC after the conversion of primary forests into other land-use types. The conversion of primary forest to cropland resulted in the most severe SOC loss (−33.2%), followed by conversion into plantation forests (−22.3%) and secondary forests (−19.1%). Nonetheless, SOC increased by 9.1% after a conversion from primary forests into pasture. More SOC loss was found at sites with lower precipitation for primary forests converted to cropland and plantation forests. The SOC loss decreased consistently with increasing mean annual temperature (MAT) for all four types of LUC. Moreover, the loss of SOC tended to worsen over time when primary forests are converted to cropland or plantation forests. In contrast, SOC loss recovered over time following conversion to secondary forests. The gain of SOC gradually increased over time after conversion to pastures. To conclude, the changes in SOC are related not only to the land-use type but also to precipitation, temperature and turn years after LUC. Due to limited data, this study focuses on soil profiles within 30 cm depth, and future research should explore SOC dynamics induced by LUC at greater depths. Overall, cases of SOC loss of approximately 30% following deforestation were very common (except for conversion to pasture), and the results of this study show that the loss of SOC following LUC should be carefully considered and monitored in land management.

Seasonal and spatial variations in Leaf Area Index (LAI, or plant area index, PAI) are challenging to delineate in tropical forests because the changes are often too subtle to be reliably detected. Many existing DHP software tools offer one‐click processing of all images using auto‐threshold segmentation (e.g. HemispheR, HemiPy and Hemisfer), but they produce results with large discrepancies. Some software (e.g. CAN‐EYE) requires manual tuning of each image, making large‐scale analysis impractical. We analysed 19,044 images from four tropical forest subtypes and found that using coding‐free AI software to process hemispherical images can significantly improve the consistency of canopy‐sky segmentation, thereby improving the detection of temporal and spatial variation of PAI. The results show that replacing the auto‐threshold segmentation with AI substantially reduced inter‐software disagreement, with the mean inter‐software range for PAI decreasing from 4.31 to 1.19, and delineating correct seasonal and spatial patterns. CAN‐EYE could identify the correct seasonal patterns but produced less accurate results than the CAN‐EYE‐AI integrated approach due to subjective bias. The high consistency achieved after AI integration enables reliable cross‐site and temporal comparisons. As users can combine the AI segmentation with other DHP software, our integrated, affordable and coding‐free method offers wide applicability and high consistency of PAI measurements, facilitating the advancement of tropical forest monitoring and research.