Integrated demographic strategies are more strongly associated with variation in conspecific density dependence than single traits in tropical tree seedlings

Huanca-Nunez, Nohemi 1 ; Browne, Luke1; Hülsmann, Lisa2; Dent, Daisy3; Rüger, Nadja4; Comita, Liza1

Published Jan 30, 2026 on Dryad. https://doi.org/10.5061/dryad.18931zd85

Data files

Jan 30, 2026 version files 21.51 MB

Data_Script_NHN.zip

21.49 MB
README.md

11.58 KB

Abstract

Conspecific negative density dependence (CNDD) is considered a key mechanism shaping species diversity in plant communities. However, species vary widely in CNDD strength, and the main ecological drivers of this variation remain unclear. We tested whether interspecific variation in CNDD is linked to (i) species’ demographic strategies related to growth–survival and stature–recruitment trade-offs, (ii) functional traits, including wood density, seed mass, maximum height, and four leaf traits, and (iii) species’ relative abundance. Using 18 years of seedling mortality data (145,768 individuals, 260 tree species) from a tropical moist forest in Panama and a robust modeling framework that accounts for potential biases due to non-linearities and variation in baseline mortality, we calculated species-specific estimates of CNDD that quantify the effect of conspecific relative to heterospecific neighbors and tested for relationships between CNDD strength and demographic strategies, functional traits, and relative abundance. CNDD strength varied widely across species and was significantly related to species location along both the growth-survival and stature-recruitment trade-off axes. Fast-growing species and tall, long-lived pioneers exhibited stronger sensitivity to CNDD compared to slow-growing and short-lived species, respectively. With the exception of wood density, single functional traits were not significantly associated with CNDD strength. Rare species experienced stronger CNDD than common species. These findings underscore the importance of life-history strategies over isolated traits in shaping density-dependent effects, suggesting that CNDD variation could reinforce niche differences and influence community composition and species coexistence in tropical forests. Additionally, our results confirm that rare species experience stronger CNDD than common species at the seedling stage, suggesting that CNDD may constrain species abundances and shape tropical tree diversity. Our results show that, with the exception of wood density, single traits were weak predictors of CNDD strength, while integrated demographic strategies captured meaningful variation. CNDD is strongest in rare, fast-growing, and long-lived pioneer species, suggesting that biotic interactions at early life stages align with key life-history trade-offs. Studies that assess and integrate links between CNDD and demographic trade-offs across multiple life stages are needed to understand the combined impact of these mechanisms on tree communities.

Overview

This repository contains the data and R scripts required to reproduce the analyses from the manuscript titled "Integrated demographic strategies are more strongly associated with variation in conspecific density dependence than single traits in tropical tree seedlings."

The study quantifies species-specific CNDD (Conspecific Negative Density Dependence) using long-term seedling survival data and links variation in CNDD to species traits, life-history axes, and relative abundance via meta-regression.

All files are contained in Data_Script_NHN.zip.

Directory Structure and File Descriptions

1. Data Processing

File: Data_processing.R
Purpose: Prepares the raw seedling census data (2001–2018) for modeling mortality.
Variable Definitions (Main Census File)

Raw census variables (e.g., plot location, tag, species code, size, status, coordinates, and related fields) originate from the BCI seedling census dataset (Comita et al. 2023; https://doi.org/10.1002/ecy.4140). Detailed variable definitions, units, and field protocols are provided in Supplementary Metadata S1 of that publication.

Below, we provide a concise variable dictionary for the main census variables used in this study, derived from the original BCI seedling census dataset.

Variable	Description	Units
`id`	Unique seedling identifier (q20.p5.tag)	–
`q20`	20 × 20 m quadrat ID	–
`p5`	5 × 5 m subquadrat ID	–
`tag`	Tag number within plot	–
`spp`	Six-letter species code	–
`fecha`	Census date (MM/DD/YYYY)	–
`censo`	Census number	–
`alt`	Seedling height	mm
`dbh`	Stem diameter at 1.3 m	mm
`tallos`	Number of stems	count
`hojas`	Number of leaves	count
`status`	Seedling status (alive/dead/etc.)	categorical
`x`	Plot x-coordinate	m
`y`	Plot y-coordinate	m
`genus`	Genus name	–
`species`	Species name	–
`family`	Plant family	–
`census.count`	Number of censuses observed	count
`time.since.last.census`	Interval since previous census	years

Additional administrative and field notes columns (e.g., nov, ryan, notes_liza, sjw09.do.not.remeasure) contain observer comments and quality-control flags.

Missing values: Missing or unavailable measurements are coded as NA (e.g., leaf counts not recorded in a given census).

2. Model Fitting

File: 01_fit_model.R
Purpose: Fits species-specific GAMs using conspecific and total seedling density, height, and plot/census as random smooth terms. Calculates:
- AUC for model fit
- Pseudo-R² (compared to reduced model)
- Average Marginal Effects (AME) and relative AMEs (rAME) for CNDD under multiple change scenarios

CNDD Marginal Effects

File: rAME_spp_equilibrium.csv

This file contains species-specific estimates of conspecific negative density dependence (CNDD) derived from generalized additive models fitted in 01_fit_model.R. It summarizes relative average marginal effects (rAME) under defined density-change scenarios.

Variable	Description	Units
`N`	Row index	–
`X`	Species code	–
`term`	Model term (e.g., `con_dens` for conspecific density)	–
`estimate`	Estimated marginal effect	model scale
`std.error`	Standard error of estimate	model scale
`offset`	Offset value used in model	–
`change.value`	Density change value used in AME/rAME calculation	count
`estimate_relative_log`	Relative marginal effect (log-transformed)	log scale
`var_log`	Variance of log-transformed estimate	–
`change`	Density-change scenario (e.g., equilibrium)	categorical
`sp`	Species code	–

Estimates are derived from GAMs with a complementary log–log link and represent effects on mortality risk.

Missing values: Missing values are coded as NA.

Additional Derived and External Data Files

The following external datasets are used in the meta-regression analyses described below. For transparency and reuse, we provide brief descriptions of key variables in each file. Full methodological and variable details are available in the original source publications cited.

1. Functional traits and life-history PCA dataset (Traits_Browne.et.al_and_Westbrook.et.al.csv)

This file contains species-level functional trait data compiled from Browne et al. (2023; https://doi.org/10.5061/dryad.mkkwh713s; https://doi.org/10.1086/659963), Westbrook et al. (2011), and life-history PCA axes from Rüger et al. (2020; https://doi.org/10.1126/science.aaz4797). These traits are used to explain interspecific variation in CNDD.

Column	Description	Units
`SPP`	Six-letter species code	–
`wood_dens`	Mean wood specific gravity	g cm⁻³
`lma`	Leaf mass per area (indirect light; incl. petiole)	g m⁻²
`seed_mass`	Mean seed dry mass (60°C)	g
`leaf_area`	Mean leaf area (indirect light)	cm²
`height_max`	Mean height of up to the six largest individuals in the BCI 50-ha plot	m
`leaf_dmc`	Leaf dry matter content (indirect light)	g g⁻¹
`LamT`	Lamina thickness	mm
`LamFT`	Lamina fracture toughness	J m⁻²
`PC1_ruger`	Life-history PCA axis 1 (Rüger et al. 2020)	unitless
`PC2_ruger`	Life-history PCA axis 2 (Rüger et al. 2020)	unitless
`BA_m2`	Adult species basal area (if included)	m²
`No_trees`	Number of adult trees (if included)	count

Trait measurement protocols and original units are described in Westbrook et al. (2011) and Wright et al. (2010). PCA axes are described in Rüger et al. (2020).

Missing values: Missing or unavailable trait values are coded as NA.

2. Species-level demographic rates (subset_demographic_Ruger.etal.csv)

This file contains species-level demographic rate variables provided by N. Rüger and described in Rüger et al. (2018; https://doi.org/10.1111/ele.12974). Values are in transformed (but not centered or scaled) form and are used as covariates in the meta-regression analyses.

Variable	Description	Units
`sp`	Species code used to match across datasets	–
`G1`–`G4`	Growth rates in four canopy layers	transformed scale (see source)
`S1`–`S4`	Survival rates in four canopy layers	transformed scale (see source)
`Rcapita`	Per-capita sapling recruitment	per capita (transformed)
`seedlingGrowth`	Seedling growth rate	transformed scale (see source)
`seedlingSurvival`	Seedling survival estimate	transformed scale (see source)
`seedN`	Per-capita number of seeds	per capita (transformed)
`seedlingN`	Per-capita number of seedling recruits	per capita (transformed)

For full methodological details and original units, see Rüger et al. (2018) and associated supplementary materials.

Missing values: Missing or unavailable values are coded as NA.

3. Meta-Regression Analyses

Script: Meta-regression.R
Purpose: Tests whether variation in species-specific CNDD is explained by:
- Functional traits
- PCA axes representing major life-history trade-offs (growth-survival and stature-recruitmet).
- Species relative abundance.
- Demographic rates (growth, survival, recruitment from Rüger et al., 2018).
Outputs: Regression summaries, diagnostic plots, and figures.

How to Reproduce the Analysis

Run Data_processing.R to generate the cleaned dataset.
Use 01_fit_model.R to fit GAMs, compute AMEs/rAMEs, and save model summaries.
Use Meta-regression.R to evaluate how CNDD relates to species traits and demographic strategies.

Notes on Access and Data Sharing

Data Source (original):
1. Long-term seedling censuses in Barro Colorado Island, Panama (2001–2018) find it at https://doi.org/10.1002/ecy.4140 (Comita et al. 2023)
2. Life-history PCA scores find it at https://10.1126/science.aaz4797 (Rüger et al. 2020) and species demographic rates find it at https://doi.org/10.1111/ele.12974 (Rüger et al. 2018)
3: Functional traits: find it at https://doi.org/10.5061/dryad.mkkwh713s and https://doi.org/10.1086/659963 (Browne et al., 2023) and (Westbrook et al., 2011)