Aim Monitoring biodiversity in remote and logistically challenging regions often relies on opportunistically collected, large-scale ecological data that are patchy and lack structured sampling designs. This poses significant methodological challenges for detecting, attributing, and predicting ecological change in species communities. Using the Southern Ocean krill community as a case study, we evaluate the effectiveness of species distribution models (SDMs) in detecting spatio-temporal trends and distinguishing true ecological signals from noise through context-appropriate simulations of simplified species communities. Innovation A wide range of SDMs have been developed over the years. However, their ability to detect change remains uncertain in both space and time. Our comparison of six SDMs explicitly considers how well single-species and joint-species models, as well as statistical and machine learning approaches capture spatial and, more novelly, temporal patterns of change in species distributions. By evaluating both predictive performance and the ability to detect spatio-temporal trends while distinguishing true ecological signals from modelled noise, our study provides practical guidance on the suitability of SDMs for biodiversity monitoring using opportunistic and patchily collected datasets. Main Conclusions Evaluation metrics showed consistent differences among the six SDMs, with statistical models often performing better than machine learning approaches. Models with an underlying structure in their relationship with covariates were effective in capturing species responses and detecting spatio-temporal trends. In contrast, some models consistently misidentified noise as ecological signal. By systematically comparing the strengths and limitations of each approach, our study recommends models with species-specific covariate effects and latent variables for detecting ecological changes in species communities. These findings are particularly relevant for monitoring biodiversity in remote and data-limited ecosystems, where understanding spatio-temporal shifts in species communities is important for informing conservation strategies and protecting vulnerable communities.