How to choose the statistic metrics to describe the patterns of trees near the tree line? I have tree inventory data: about 700 points of trees with known locations (X;Y;Z) and measured heights and diameters of stems at the breast height (DBH). The trees are located along the elevation gradient from dense forest to treeless tundra, this gradient is also known as treeline transition zone. I have already figured out that trees' disribution is clustered, not random or dispersed. How can I calculate the clustering ability along the gradient? What kind of statistic/spatial statistic metrics can describe the changing heights,DBH or clustering ability along the graidient in the best way?
 A: "Clustering" usually assumes there is more than one cluster.
I'm not convinced you need anything else but a linear model. Tree density will likely be some approx. normal distribution until a certain height, then decrease approximately linearly until trees have disappeared. So $\alpha \mathcal{N}(\mu,\sigma)$ for an elevation dependant piecewise constant/linear $\alpha$.
A: What you likely have here is not clustering, but an inhomogeneous pattern. The intensity of the unmarked point pattern can potentially be modeled as a log-linear function of a continuous covariate: gradient.
$$\lambda(u)=exp(\beta_0 + \beta_1G(u))$$
It's not impossible that the dbh distribution is normal with mean $\mu$ and standard deviation $\sigma$ (hence, $N(\mu,\sigma)$, but it may not be that simple. Dbh data are sometimes multimodal (representing different age classes and/or variation in dbh among tree species) and often right-skewed. 
If the points are independent, then you might be able to model it as multiple realizations of an inhomogeneous Poisson point process (or as one iteration of an inhomogeneous Point process with random labeling of dbh), but if there is dependency, you might be looking at some type of inhomogeneous (and potentially hierarchical) cluster process (e.g., if the location of offspring trees depend on the location of parent trees).
Chapter 14 of Spatial Point Patterns-Methodology and Applications with R (by Baddeley, Rubak, and Turner) is a good introduction to marked point pattern analysis.
