Solar-induced chlorophyll fluorescence (SIF) is increasingly known as an effective proxy for plant photosynthesis and therefore has great potential in monitoring gross primary production (GPP). However, the relationship between SIF and GPP remains highly uncertain across space and time. Here, we analyzed the SIF (reconstructed, SIFc)-GPP relationships and their spatiotemporal var-iability, using GPP estimates from FLUXNET2015 and two spatiotemporally contiguous SIFc da-tasets (CSIF and GOSIF). Results showed that SIFc had significant positive correlations with GPP at the spatiotemporal scales investigated (p<0.001). The generally linear SIFc-GPP relationships were substantially affected by spatial and temporal scales and SIFc datasets. GPP/SIFc slope of the ev-ergreen needleleaf forest (ENF) biome was significantly higher than those of several other biomes (p<0.05), while the other 11 biomes showed no significant differences in GPP/SIFc slope between each other (p>0.05). We therefor propose a two-slope scheme to differentiate ENF from non-ENF biome and synopsize spatiotemporal variability of GPP/SIFc slope. The relative biases were 7.14% and 11.06% in the estimated cumulative GPP across all EC towers, respectively, for GOSIF and CSIF using two-slope scheme. The significantly higher GPP/SIFc slopes of the ENF biome in the two-slope scheme are intriguing and deserve further study. In addition, there was still consid-erable dispersion in the comparisons of CSIF/GOSIF and GPP at both site and biome levels, calling for discriminatory analysis backed by higher spatial resolution to systematically address issues related to landscape heterogeneity and mismatch between SIFc pixel and the footprints of flux towers and their impacts on the SIF and GPP relationship.