Empirical ecology to support mechanistic modelling: Different objectives, better approaches and unique benefits

Abstract Modern ecological management problems are characterized by large scales, rapid environmental change, multiple stressors and conflicts between local and global conservation objectives. These problems are too complex to address with field studies alone, and statistical models that assume past system behaviours can predict future responses are risky when systems are changing rapidly. Mechanistic simulation models, though, can avoid that assumption while accommodating important natural complexities. Making mechanistic models credible requires empirical studies, but traditional study topics and designs often do not support them well. The models we use for modern problems need empirical studies that provide understanding of life history and autecology of study species, identify general patterns useful for model design and evaluation, collect data of kinds that models show are important and develop submodels and theory for individual‐level mechanisms. Ecologists can better produce such knowledge via research that: (a) is interdisciplinary and across‐level, often designed to understand just enough about individuals to support individual‐based models of populations and communities; (b) is designed to quantify relationships across broad ranges, instead of testing statistical hypotheses; (c) emphasizes relevance and realism over precision; and (d) includes stressful conditions relevant to modern management challenges. Supporting complex management models is rewarding to research ecologists: Modelling identifies crucial yet understudied research topics; models can be used as virtual ecosystems for experiments (including tests of theory) that would be impossible in reality; and supporting models ensures that our work has high impact and contributes to critical issues.