1: \begin{abstract}

2: Diffusion processes on trees are commonly used in evolutionary biology to model the joint distribution of continuous traits, such as body mass, across species.

3: Estimating the parameters of such processes 

4: from tip values

5: presents challenges because of the intrinsic correlation

6: between the observations produced by the shared evolutionary 

7: history, thus violating the standard independence assumption 

8: of large-sample theory.

9: For instance Ho and An\'e \cite{HoAne13} recently proved 

10: that the mean (also known in this context as selection optimum) 

11: of an Ornstein-Uhlenbeck process on a tree 

12: cannot be estimated consistently from

13: an increasing number of tip observations 

14: if the tree height is bounded. 

15: Here, using a fruitful connection to the so-called

16: reconstruction problem in probability theory,

17: we study the convergence rate of parameter estimation

18: in the unbounded height case. 

19: For the mean of the process, 

20: we provide a necessary and 

21: sufficient condition for the consistency of the maximum 

22: likelihood estimator (MLE) 

23: and establish a phase transition on its convergence rate

24: in terms of the growth of the tree. In particular we show that

25: a loss of $\sqrt{n}$-consistency

26: (i.e., the variance of the MLE becomes $\Omega(n^{-1})$,

27: where $n$ is the number of tips) 

28: occurs when the tree growth is larger than a threshold 

29: related to the phase transition of the reconstruction problem.

30: For the covariance parameters, we give a novel, efficient 

31: estimation method which achieves 

32: $\sqrt{n}$-consistency under natural assumptions on the tree.

33: Our theoretical results provide practical suggestions for the 

34: design of comparative data collection.

35: 

36: \paragraph{Keywords}

37: Ornstein-Uhlenbeck, phase transition, evolution, phylogenetic, consistency, maximum likelihood estimator.

38: % \PACS{PACS code1 \and PACS code2 \and more}

39: % \subclass{MSC code1 \and MSC code2 \and more}

40: \end{abstract}

41: