The study reveals that current methods for analyzing biodiversity change often underestimate uncertainty and misestimate trends due to insufficient consideration of spatial, temporal, and phylogenetic structures in data. Using a new statistical framework, the researchers show that when these structures are accounted for, trends previously identified as increasing or decreasing vanish. This highlights the limited understanding of biodiversity change across large spatial and taxonomic scales. However, incorporating these structures improves local-scale prediction accuracy, offering hope for estimating biodiversity change at policy-relevant scales. The study analyzes ten high-profile biodiversity datasets, which together describe the abundance patterns of over 30,000 populations across 3,100 species and 6,000 locations. The researchers compare their new correlated effect model with traditional models like random intercept and random slope, finding that the correlated effect model significantly improves trend prediction accuracy and reduces uncertainty. This model accounts for spatial, temporal, and phylogenetic non-independence, leading to more reliable trend estimates. The study also shows that despite the large uncertainty in collective trends, the correlated effect model improves local-scale predictions. This is crucial for adaptive conservation strategies. The findings suggest that past estimates of biodiversity change may be unreliable due to the failure to account for non-independence. The study emphasizes the need for improved data collection and more sophisticated statistical methods to accurately assess biodiversity change. The research underscores the importance of considering spatial, temporal, and phylogenetic structures in biodiversity datasets to avoid misestimating trends and uncertainties. The correlated effect model provides a more accurate framework for analyzing biodiversity change, enhancing the reliability of trend estimates and informing conservation policies. The study calls for further advancements in methods to better understand and predict biodiversity change across different scales.The study reveals that current methods for analyzing biodiversity change often underestimate uncertainty and misestimate trends due to insufficient consideration of spatial, temporal, and phylogenetic structures in data. Using a new statistical framework, the researchers show that when these structures are accounted for, trends previously identified as increasing or decreasing vanish. This highlights the limited understanding of biodiversity change across large spatial and taxonomic scales. However, incorporating these structures improves local-scale prediction accuracy, offering hope for estimating biodiversity change at policy-relevant scales. The study analyzes ten high-profile biodiversity datasets, which together describe the abundance patterns of over 30,000 populations across 3,100 species and 6,000 locations. The researchers compare their new correlated effect model with traditional models like random intercept and random slope, finding that the correlated effect model significantly improves trend prediction accuracy and reduces uncertainty. This model accounts for spatial, temporal, and phylogenetic non-independence, leading to more reliable trend estimates. The study also shows that despite the large uncertainty in collective trends, the correlated effect model improves local-scale predictions. This is crucial for adaptive conservation strategies. The findings suggest that past estimates of biodiversity change may be unreliable due to the failure to account for non-independence. The study emphasizes the need for improved data collection and more sophisticated statistical methods to accurately assess biodiversity change. The research underscores the importance of considering spatial, temporal, and phylogenetic structures in biodiversity datasets to avoid misestimating trends and uncertainties. The correlated effect model provides a more accurate framework for analyzing biodiversity change, enhancing the reliability of trend estimates and informing conservation policies. The study calls for further advancements in methods to better understand and predict biodiversity change across different scales.