1: \begin{abstract}

2: We consider a family of random trees satisfying a Markov branching

3: property. Roughly, this property says that the subtrees above some

4: given height are independent with a law that depends only on their

5: total size, the latter being either the number of leaves or

6: vertices. Such families are parameterized by sequences of

7: distributions on partitions of the integers, that determine how the

8: size of a tree is distributed in its different subtrees. Under some

9: natural assumption on these distributions, stipulating that

10: ``macroscopic'' splitting events are rare, we show that Markov

11: branching trees admit the so-called self-similar fragmentation trees

12: as scaling limits in the Gromov-Hausdorff-Prokhorov topology.

13: 

14: The main application of these results is that the scaling limit of

15: random uniform unordered trees is the Brownian continuum random

16: tree. This extends a result by Marckert-Miermont and fully proves a

17: conjecture by Aldous. We also recover, and occasionally extend,

18: results on scaling limits of consistent Markov branching model, and

19: known convergence results of Galton-Watson trees towards the Brownian

20: and stable continuum random trees.

21: \end{abstract}

22: