Data for global pattern of net conspecific negative density dependence

This search resulted in a total of 50 papers (Figure 1). For each selected study, we focused on three different types of density-dependence: (1) Seedling-seedling neighbourhood interactions – for individuals within the same seedling plot, For each seedling, the explanatory variables were the densities of conspecific and heterospecific seedling neighbours within the same 1 x 1 m plot. (2) Tree-seedling neighbourhood interactions within a specified interaction distance, the target individuals are seedlings.

For each seedling, explanatory variables were the density of conspecific and heterospecific tree neighbours (including saplings and adult trees) within a circular neighbourhood of a fixed radius centered on the seedling plot. (3) Tree-tree neighbourhood interactions within certain radius, the target individuals are trees—including saplings and adults, defined in each study (see below). For each individual tree or sapling, explanatory variables were the density metrics of conspecific and heterospecific tree neighbours within a circular neighbourhood of a certain radius centered on the focal tree, the densities are either the total abundance, the total basal area, or the distance weighted basal area.

For papers that contained multiple pairs of conspecifics and heterospecific neighbour densities, for example, a model that included both seedling-seedling neighbour densities and tree-seedling neighbour densities, each pair of densities was considered separately, and for papers that tested multiple life stages, each life stage was considered independently. Data extractionWe identified 50 eligible papers, yielding a total of 350 observations, where each observation represents a paired estimate of conspecific (CDD) and heterospecific (HDD) density dependence extracted from a given paper.

Multiple data points were often extracted from a single paper if it reported density-dependent effects across different forest communities or distinct ontogenetic stages (e.g., seedling vs. adult) or used distinct density predictors (e.g., seedling density vs. adult density) within the same community. For each study, we gathered the mean, standard error (or confidence interval) of each pair of conspecific and heterospeicific neighbor density coefficients.

In cases where models included environmental covariates, we extracted the density-dependent coefficients from the full model. For observations that only reported the mean value with no variation (79 / 350), we only recorded and used the mean values in the analysis. Most coefficients extracted from the literature were originally reported as log odds ratios from generalized linear mixed effect models, for those reported as odds ratios, we applied a natural log transformation to ensure data consistency.

Information from observations where results were presented as graphs were extracted using Plot Digitizer (plotdigitizer.sourceforge.net/). We also collected other key variables, including life-history stage (i.e., seedling, sapling, or tree), as well as the stages of seedlings, according to the description in each paper, we classified early-stage seedling as the new recruit seedlings or seedlings less than 20 cm height, and late-stage seedling as the established seedlings larger than 1 years old, or seedlings more than 20 cm height) and the tree ontogenetic stage (sapling, juveniles and adults).

We also recorded metadata and covariates including author, year, study site location, and, size of seedling plots. We recorded the latitude, mean annual temperature (MAT), and mean annual precipitation (MAP) for each observation. To assess the influence of aridity on the geographic variation in net density dependent effects, we extracted aridity index (AI) values from the Global Aridity Index (version 3; Zomer et al., 2022).