Abstract:Abstract Aim Landscape management and conservation planning require maps of vegetation composition and structure over large regions. Species distribution models ( SDM s) are often used for individual species, but projects mapping multiple species are rarer. We compare maps of plant community composition assembled by stacking results from many SDM s with multivariate maps constructed using nearest‐neighbor imputation. Location Western Cascades ecoregion, Oregon and California, USA . Methods We mapped distributions and abundances of 28 tree species over 4,007,110 ha at 30‐m resolution using three approaches: SDM s using machine learning (random forest) to yield: (1) binary ( RF _Bin); (2) basal area (abundance; RF _Abund) predictions; and (3) multi‐species basal area predictions using a nearest‐neighbor imputation variant based on random forest ( RF _ NN ). We evaluated accuracy of binary predictions for all models, compared area mapped with plot‐based areal estimates, assessed species abundance at two spatial scales and evaluated communities for species richness, problematic compositional errors and overall community composition. Results RF _Bin yielded the strongest binary predictions (median True Skill Statistics; RF _Bin: 0.57, RF _ NN : 0.38, RF _Abund: 0.27). Plot‐scale predictions of abundance were poor for RF _Abund and RF _ NN (median Agreement Coefficient ( AC ): −1.77 and −2.28), but strong when summarized over 50‐km radius tessellated hexagons (median AC for both: 0.79). RF _Abund's strength with abundance and weakness with binary predictions stems from predicting small values instead of zeros. The number of zero value predictions from RF _ NN was closest to counts of zeros in the plot data. Correspondingly, RF _ NN 's map‐based species area estimates closely matched plot‐based area estimates. RF _ NN also performed best for community‐level accuracy metrics. Conclusions RF _ NN was the best technique for building a broad‐scale map of diversity and composition because the modelling framework maintained inter‐species relationships from the input plot data. Re‐assembling communities from single variable maps often yielded unrealistic communities. Although RF _ NN rarely excelled at single species predictions of presence or abundance, it was often adequate to many (but not all) applications in both dimensions. We discuss our results in the context of map utility for applications in the fields of ecology, conservation and natural resource management planning. We highlight how RF _ NN is well‐suited for mapping current but not future vegetation.

Quantitative Evaluation of Species Composition Dissimilarity Within a Community and among Communities

Dissimilarity Measurements and the Size Structure of Ecological Communities

Spatial Heterogeneity in Species Richness and Species Composition

Determining Spatial Heterogeneity in Species Richness of Plant Community

Effectively Inferring Overall Spatial Distribution Pattern of Species in a Map when Exact Coordinate Information is Missing

Spatially Explicit Metrics of Species Diversity, Functional Diversity, and Phylogenetic Diversity: Insights into Plant Community Assembly Processes

Quantifying Effects of Habitat Heterogeneity and Other Clustering Processes on Spatial Distributions of Tree Species

Frequency distribution models for spatial patterns of vegetation abundance

A Comparison And Evaluation Of Four Vegetation Analysis Approaches Based On Remote Sensing Imagery

Quantifying elemental diversity to study landscape ecosystem function

A Probability Distribution Model of Small -Scale Species Richness in Plant Communities.

Evaluation of nine distance-based measures of functional diversity applied to forest communities

Assessing Tree Species Diversity in Forest Ecosystems: A New Approach

A New Cost‐effective Approach to Survey Ecological Communities

Numerical Analysis of Inter-Specific Relationships in the Estuary Salt Marsh Plant Community of Southern Chongming Dongtan, Shanghai

Structural characteristics of different grassland community types

A contribution diversity approach to evaluate species diversity

Multi-scale Object-Based Measurement of Arid Plant Community Structure

A method for measuring the comparability of different sampling methods used in biological surveys: implications for data integration and synthesis.

Species distribution modelling for plant communities: stacked single species or multivariate modelling approaches?

Evaluation of Similarity Measure Methods for Hyperspectral Remote Sensing Data.