1: \begin{abstract}

2: Bayesian hierarchical models with latent Gaussian layers  have proven very flexible in capturing complex

3: stochastic behavior and hierarchical structures in high-dimensional spatial and spatio-temporal

4: data. Whereas simulation-based Bayesian inference through Markov Chain

5: Monte Carlo may be hampered by

6: slow convergence and numerical instabilities, the inferential framework of Integrated

7: Nested Laplace Approximation (INLA) is capable to provide accurate and relatively fast analytical

8: approximations to posterior quantities of interest.  It heavily relies on the use of

9: Gauss--Markov dependence structures to avoid the numerical bottleneck

10: of high-dimensional nonsparse matrix computations. With a view

11: towards space-time applications, we here review the principal

12: theoretical concepts, 

13: model classes  and inference tools within the

14: INLA framework. Important elements to construct space-time models are certain spatial Mat\'ern-like

15: Gauss--Markov random fields, obtained as approximate solutions to a

16: stochastic partial differential equation. Efficient implementation of statistical

17: inference tools for a large variety of models is available  through

18: the \texttt{INLA} package of the \texttt{R} software. 

19: To showcase the  practical use of \texttt{R-INLA} and to illustrate its principal commands and syntax,  a comprehensive simulation experiment is  presented using  simulated

20: non Gaussian space-time count data with a first-order autoregressive dependence

21: structure in time.

22: \end{abstract}

23: