Where does forest collide?

A forest as an ecological community: a self-organising assemblage of trees and associated organisms that develops, matures, and turns over through time on a site. Frederic Clements's 1916 Plant Succession: An Analysis of the Development of Vegetation framed such vegetation as a community that passes through orderly stages toward a stable end state, treating the forest as a developing whole rather than a loose collection of plants. Alexander Watt's 1947 Pattern and process in the plant community refined this by showing that a forest is a shifting mosaic of patches at different phases of growth and decay. The referent is a living plant community measured by its species, structure, and dynamics, not a jurisdiction created by royal law or an algorithm built from many decision trees.[1][2]

Unit

A stand of trees and associated organisms treated as one community

Recovery means

After disturbance the community regrows through successional stages toward a mature state

Measured by

Quantified by tree density, basal area, species composition, and rates of growth and turnover

In machine learning a random forest is a predictive model built by combining many decision trees, each grown on a random portion of the data and features, whose votes are pooled to classify a case or predict a value. Tin Kam Ho's 1995 Random decision forests introduced growing an ensemble of trees on random subsets of the available features, and Leo Breiman's 2001 Random forests formalised and popularised the method by adding bootstrap resampling of the training cases. The referent is a statistical estimator assembled from many small models, judged by its predictive accuracy, and it takes the word only by metaphor: it has nothing to do with a wooded community or a territory under royal law.[3][4]

Unit

One ensemble of decision trees combined into a single predictor

Recovery means

Not applicable

Measured by

Assessed by predictive accuracy, out-of-bag error, and feature importance on held-out data