The global spectrum of tree crown architecture

Abstract

Trees can differ enormously in their crown architectural traits, such as the scaling relationships between tree height, crown width and stem diameter. Yet despite the importance of crown architecture in shaping the structure and function of terrestrial ecosystems, we lack a complete picture of what drives this incredible diversity in crown shapes. Using data from 374,888 globally distributed trees, we explore how climate, disturbance, competition, functional traits, and evolutionary history constrain the height and crown width scaling relationships of 1914 tree species. We find that variation in height-diameter scaling relationships is primarily controlled by water availability and light competition. Conversely, crown width is predominantly shaped by exposure to wind and fire, while also covarying with functional traits related to mechanical stability and photosynthesis. Additionally, we identify several plant lineages with highly distinctive stem and crown forms, such as the exceedingly slender dipterocarps of Southeast Asia, or the extremely wide crowns of legume trees in African savannas. Our study charts the global spectrum of tree crown architecture and pinpoints the processes that shape the 3D structure of woody ecosystems.

Fig. 1. Overview of the global tree allometry database.

a Geographic distribution of the allometric data (n = 374,888 individual trees belonging to 1914 species). Individual tree records were aggregated in 200 × 200 km grid cells (mean number of trees per grid cell = 742). The map was obtained from the Natural Earth database (https://www.naturalearthdata.com) and is displayed using a Robinson projection (EPSG:54030). Relationships between each tree’s stem diameter and its b height (H), c crown diameter (CD) and d crown aspect ratio (CAR) are shown on a logarithmic scale. CAR is defined as the ratio between CD and H, with values lower than 1 indicating a vertical crown profile (H > CD) while values greater than 1 corresponding to a horizontal crown profile (CD > H). Points are coloured according to the aridity index value assigned to each tree based its geographic coordinates, with larger values corresponding to drier conditions (shown in red). Graphical illustration of the approach used to generate size-standardized estimates of e tree height (H_RESID), f crown diameter (CD_RESID) and g crown aspect ratio (CAR_RESID) for each tree species. Regression lines are predicted values obtained by fitting a linear model to the entire dataset (grey points). By comparing predicted and observed value of H, CD and CAR, we quantified how much each species departs, on average, from this general trend and identified ones with greater (blue points) or smaller H, CD and CAR values (red points) than expected given their stem diameters. This approach is conceptually similar to generating species-level predictions of H, CD and CAR at a fixed size (e.g., D = 30 cm), but avoids the need to arbitrarily select a size at which to compare species.

Fig. 2. Global spectrum of tree crown architecture.

Tree crown architectural types and their distribution across biomes for the 1309 tree species for which both height and crown size were measured. a Tree species were grouped into one of nine architectural types based on their size-standardized height (H_RESID) and crown diameter values (CD_RESID). The vertical and horizontal lines mark the 25th and 75th percentile of the data and the size of each circle reflects the crown aspect ratio (CAR_RESID). Examples of tree species that occupy different areas of this crown architectural spectrum are highlighted. b Proportion of species belonging to the nine architectural types for each biome. See Supplementary Table 4 for pairwise comparisons of H_RESID, CD_RESID and CAR_RESID values among biomes, and Supplementary Fig. 5 for a breakdown of the nine architectural types among angiosperms and gymnosperms.

Fig. 3. Variation in tree crown architecture across the tree of life.

Size-standardized estimates of a tree height (H_RESID, n = 1225 species), b crown diameter (CD_RESID, n = 870 species) and c crown aspect ratio (CAR_RESID, n = 868 species) are mapped onto a time-calibrated phylogeny of seed plants. Low values (red) indicate species that are shorter, with narrower crowns and smaller crown aspect ratios than expected given the size of their stem, while high values (blue) indicate the opposite. H_RESID, CD_RESID and CAR_RESID all exhibited phylogenetic signal, with Pagel’s λ values of 0.70, 0.54, and 0.63, respectively (P < 0.001 in all three cases based on a likelihood ratio test). Plant families and genera with mean H_RESID, CD_RESID and CAR_RESID values that are significantly lower (red lines) or higher (blue lines) than zero are highlighted on each phylogenetic tree (see Supplementary Tables 5–6 for full details). Note that only species that were a direct match to those in the phylogeny were used for the phylogenetic analysis.

Fig. 4. Drivers of variation in crown architecture among tree species.

Standardized model coefficients for each predictor variable were obtained by fitting phylogenetic generalised least squares regressions to size-standardized estimates of a tree height (H_RESID, n = 1910 species), b crown diameter (CD_RESID, n = 1313 species), and c crown aspect ratio (CAR_RESID, n = 1309 species). Error bars show both standard errors (thick lines) and 95% confidence intervals (thin lines) of the model coefficients. Significantly positive and negative coefficients are shown in blue and red, respectively, while those for which the 95% confidence intervals overlap with zero are shown in grey.

Fig. 5. Variation in the crown architecture along gradients of tree cover, aridity, and wind gust speed.

Points are species-level estimates of size-standardized a–c tree height (H_RESID, n = 1910 species), d–f crown diameter (CD_RESID, n = 1313 species) and g–i crown aspect ratio (CAR_RESID, n = 1309 species). Fitted lines correspond to phylogenetic generalised least squares model predictions generated by keeping all other predictors fixed at their mean values. Negative values of H_RESID, CD_RESID, and CAR_RESID indicate species that are shorter, with narrower crowns and smaller crown aspect ratios than expected given the size of their stem, while positive values denote the opposite. Values of tree cover, aridity index and wind gust speed represent are means calculated across all individual trees of a given species. Note that the aridity index was log-transformed and that larger values correspond to drier conditions.