1: \begin{abstract}

2: Mixed finite element methods with flux errors in $H(\ddiv)$-norms and div-least-squares 

3: finite element methods require a separate marking strategy in obligatory adaptive mesh-refining.

4: The refinement indicator $\sigma^2(\T,K)=\eta^2(\T,K)+\mu^2(K)$ of a finite 

5: element domain $K$ in an admissible triangulation $\T$ consists of some 

6: residual-based error estimator $\eta(\T,K)$ with some reduction property under local 

7: mesh-refining and some data approximation error $\mu(K)$. Separate marking means either D\"orfler

8: marking if $\mu^2(\T) \leq \kappa \eta^2(\T)$ or otherwise an optimal data approximation algorithm runs with

9: controlled accuracy as established in \cite{CR09, safem2015}. 

10: 

11: The axioms are abstract and sufficient conditions on the estimators $\eta(\T,K)$ and data

12: approximation errors $\mu(K)$ for optimal asymptotic convergence rates. The enfolded set of axioms simplifies \cite{CFP14} for collective marking,

13: treats separate marking established for the first time in an abstract framework, generalizes \cite{CCP-lsfem}

14: for least-squares schemes, and extends \cite{CR09} to the mixed FEM with flux error control in $H(\ddiv)$.

15: \end{abstract}

16: