Deriving exploratory temperature-based phenological indicators under data-limited conditions: integrating ERA5 and citizen science

• Understanding climate-biodiversity interaction remains a major gap, particularly in tropical-subtropical ecosystems. • We propose the Cross-Scale Coarse Indicators, which utilizes coarse-scale gridded data and unstructured observations. • The CSCI is guided by 3 ecological premises: Spatial Tolerance, Temporal Smoothing, and Consideration of Interpretability. • CSCIs offer a complementary approach for climate–biodiversity monitoring under data-limited conditions. Phenological shifts are among the most visible biological responses to climate change, with profound implications for ecological interactions, food webs, and predictive modeling. Yet in tropical and subtropical regions, where global biodiversity is concentrated, phenological responses to temperature variability remain poorly quantified due to limited long-term monitoring. This study explores the potential of integrating gridded climate reanalysis datasets with opportunistic citizen science observations to develop exploratory, temperature-based phenological indicators. We propose the Cross-Scale Coarse Indicators (CSCIs), a novel family of indicators designed to address spatial and temporal mismatches in coarse-resolution climate data and unstructured biological records. By incorporating coarse spatial grids and extended temporal windows, CSCIs aim to enhance robustness while preserving biological relevance under uncertain data conditions, as demonstrated in two case studies. Case 1 modeled flowering observations of four tree species ( Crateva religiosa , Cornus controversa , Firmiana colorata , Helicteres isora ) using ERA5-derived temperature indicators. Variable selection was performed using Random Forests, followed by Partial Least Squares regression to assess explanatory power. In Case 2, indicators for Turpinia formosana were derived from finer-resolution historical datasets (0.05°–0.01°) to evaluate indicator stability across scales. Our results demonstrate that CSCIs can capture meaningful temperature–phenology relationships, even with coarse and biased conditions. Despite species-specific difference, consistent indicator convergence across scales supports the utility of the CSCI. These findings highlight CSCIs as a scalable and transferable tool for phenological research in data-scarce regions and a practical foundation for climate-biodiversity assessment where structured monitoring is limited.