%0 Journal Article %A Bart Haegeman %A Jean-François Arnoldi %A Shaopeng Wang %A Claire de Mazancourt %A José M. Montoya %A Michel Loreau %T Resilience, Invariability, and Ecological Stability across Levels of Organization %D 2016 %R 10.1101/085852 %J bioRxiv %P 085852 %X Ecological stability is a bewildering broad concept. The most common stability measures are asymptotic resilience, widely used in theoretical studies, and measures based on temporal variability, commonly used in empirical studies. We construct measures of invariability, defined as the inverse of variability, that can be directly compared with asymptotic resilience. We show that asymptotic resilience behaves like the invariability of the most variable species, which is often a rare species close to its extinction boundary. Therefore, asymptotic resilience displays complete loss of stability with changes in community composition. In contrast, mean population invariability and ecosystem invariability are insensitive to rare species and quantify stability consistently whether details of species composition are considered or not. Invariability provides a consistent framework to predict diversity-stability relationships that agree with empirical data at population and ecosystem levels. Our findings can enhance the dialogue between theoretical and empirical stability studies. %U https://www.biorxiv.org/content/biorxiv/early/2016/11/11/085852.full.pdf