Species traits mediate environmental responses but not conspecific density dependence in tropical tree saplings
Data files
Oct 29, 2025 version files 78.59 GB
-
Acon15_model_list.RData
9.64 GB
-
Atotal15_model_list.RData
9.64 GB
-
base_data.RData
2.82 MB
-
Bayesian_Model_Fitting.R
7.14 KB
-
closure10_model_list.RData
9.64 GB
-
EnvironmentData.csv
107.04 KB
-
envpc1_model_list.RData
9.64 GB
-
envpc2_model_list.RData
9.64 GB
-
envpc3_model_list.RData
9.64 GB
-
Extract_result.R
8.19 KB
-
Figure_code.R
52.76 KB
-
FigureData.RData
3.31 MB
-
first_level_model.Rdata
1.58 GB
-
PCA_trait_environment_visualization.R
5.96 KB
-
README.md
10.63 KB
-
RGRAnalysisData.csv
2.68 MB
-
Scon15_model_list.RData
9.63 GB
-
stanmodel.zip
77.94 KB
-
Stotal15_model_list.RData
9.49 GB
-
trait_data.csv
16.03 KB
Abstract
Understanding how functional traits mediate species-specific responses to environmental variation and neighborhood interaction is fundamental for linking individual performance to community assembly. We used a hierarchical framework to examine how functional traits along the acquisitive-conservative spectrum mediate growth responses to environmental effects and density dependence. We monitored the growth of 16,717 saplings from 115 tree species over five years in a tropical rainforest and measured 10 functional traits reflecting the acquisitive-conservative spectrum. We employed Bayesian hierarchical models to quantify the relative importance of environmental and density factors on sapling growth and investigate how functional traits mediate species-specific responses to these factors. Sapling growth rates were primarily influenced by soil conditions, light availability (canopy closure), and conspecific adult neighbor density. Acquisitive species exhibited enhanced growth under high light, favorable soil resources, and low aluminum conditions compared to conservative species. However, we found no significant relationship between functional traits and conspecific density dependence. Functional traits mediate environmental responses through divergent resource-use strategies rather than conspecific density dependence. Trait-based mechanisms underlying species coexistence may operate through pathways beyond the acquisitive-conservative spectrum. Our hierarchical modeling provides a framework for disentangling the complex relationships between functional traits, environmental responses, and density dependence in diverse tropical forests.
Data and R code from: Species traits mediate environmental responses but not conspecific density dependence in tropical tree saplings
Author: Xiaohua Ma[1, 2#], Rong Gu[1, 2#], Guoshan Shi[2], Feng Liu[3, 4], Guanghong Cao[3], Yun Deng[2], Shangwen Xia[2], Xiaodong Yang[2], Zhiming Zhang[1*], Luxiang Lin [2, 5*]
[1] School of Ecology and Environment Science, Yunnan University, Kunming 650504, China
[2] Yunnan Key Laboratory of Forest Ecosystem Stability and Global Change, Xishuangbanna Tropical Botanical Garden, Chinese Academy of Sciences, Kunming, Yunnan, 666303, China
[3] Administration Bureau of Naban River Watershed National Nature Reserve, Jinghong, Yunnan 666100, China
[4] Yunnan Academy of Forestry and Grassland, Kunming, Yunnan 650204, China
[5] National Forest Ecosystem Research Station at Xishuangbanna, Mengla, Yunnan, 666303, China
[#] These authors have equal contributions.
[*] Corresponding authors: Zhiming Zhang: E-mail: zzming76@ynu.edu.cn; Luxiang Lin: E-mail: linluxa@xtbg.ac.cn
This README.txt file was updated on 2025-10-24 on "Species traits mediate environmental responses but not conspecific density dependence in tropical tree saplings"
Software Requirements
Data analyses were performed in R version 4.2.2.
The following R packages were used in this analyses:
dplyr (version 1.1.4)
rstan (version 2.32.3)
loo (version 2.6.0)
cowplot (version 1.1.1)
tibble (version 3.2.1)
purrr (version 1.0.2)
ggplot2 (version 3.5.1)
ggh4x (version 0.3.0)
vegan (version 2.7.1)
factoextra (version 1.0.7)
FactoMineR (version 2.12)
ggpubr (version 0.6.1)
Species traits mediate environmental responses but not conspecific density dependence in tropical tree saplings
This data repository consist of 3 data files, 4 code scripts, 101 stan scripts, 11 fitted stan model objects and this README document, with the following data and code filenames and variables
Data files
1.RGRAnalysisData.csv: Data used for constructing Bayesian hierarchical models.
- Family: family of a species.
- species: scientific species name.
- qua: The numbering of 20m × 20m plots.
- dbh_2016 : Diameter (m) at breast height (1.3 m above ground) measured in 2016.
- rgr: Relative growth rate calculated from 2016 to 2021.
- Atotal_15: Total adult neighbor density(cm^2/m), calculated based on the sapling basal area within a 15-meter radius.
- Stotal_15: Total sapling neighbor density(cm^2/m) , calculated based on the sapling basal area within a 15-meter radius.
- Acon_15: Conspecific adult neighbor density(cm^2/m) , calculated based on the sapling basal area within a 15-meter radius.
- Scon_15: Conspecific sapling neighbor density(cm^2/m) , calculated based on the sapling basal area within a 15-meter radius.
- Scon_stem_number: Conspecific sapling neighbor density(ind. / m) , calculated based on the sapling stem count within a 15-meter radius.
- Stotal_stem_number: Total sapling neighbor density(ind. / m) , calculated based on the sapling stem count within a 15-meter radius.
- closure_10 & closure_15 & closure_20: Canopy closure within radius of 10m, 15m, 20m, respectively
- envpc1 & envpc2 & envpc3: The first three principal component axes of physical and chemical properties of soil and topographic variables
2.trait_data.csv: Trait file includes 10 species-specific functional traits and their first two principal components, which were used as group-level predictors in the two-level model. These traits were measured in 2013 (leaf economics traits) and 2014 (wood density) inin adjacent forests of the same type as our study plot (Zhou et al. 2020), following standardized protocols (Cornelissen et al. 2003; Perez-Harguindeguy et al. 2016). Seed mass was measured by collecting seeds within the forest for most focal tree species. For a small number of species, seed mass data were obtained from the "Seeds of Woody Plants in China" (State Forestry Administration, National Forest Farm and Tree Seed and Seedling Workstation, 2001)
- LA: Leaf area (mm^2).
- SLA: Specific leaf area (mm^2 /mg).
- LDMC: Leaf dry matter content (mg/g).
- LT: Leaf thickness (mm).
- LCC: Leaf carbon content (mg/g).
- LNC: Leaf nitrogen content (mg/g).
- LPC: Leaf phosphor content (mg/g).
- LKC: Leaf kalium content (mg/g).
- WD: Wood density (g/cm^3).
- SM: Seed mass (g).
- PC1: the first principal component axis of 10 functional traits.
- PC2: the second principal component axis of 10 functional traits.
3.EnvironmentData.csv: contains 5 topographic factors and 12 soil physical and chemical properties, which were used in the first-level model to characterize the environment.
- plot: plot ID of each 20m × 20m plots.
- AAl: available aluminum content (mg/kg).
- ACa: available calcium content (mg/kg).
- AFe: available iron content (mg/kg).
- AK: available potassium content (mg/kg).
- AMg: available magnesium content (mg/kg).
- AMn: available manganese content (mg/kg).
- AP: available phosphorus content (mg/kg).
- AS: available sulfur content (g/kg).
- TC: total carbon content (g/kg).
- TN: total nitrogen content (g/kg).
- TP: total phosphorus content (g/kg).
- Soil pH: Soil pH.
- Elevation: elevation(m).
- Convexity: convexity.
- Slope: slope.
- AspectSin: Aspect in Sin.
- AspectCos: Aspect in Cos.
- TWI: topographic wetness index.
stan model scripts
- stanmodel.zip: All stan model files for RGR parameter estimation
First-Level Models
Models of canopy closure series:
- model_closure10.stan: The model with canopy closure within radii of 10m
- model_closure15.stan: The model with canopy closure within radii of 15m
- model_closure20.stan: The model with canopy closure within radii of 20m
Neighborhood density model: models with different combinations of density variables including total and conspecific adult and sapling neighbor densities within radius of 15m from the focal saplings.
- nei_model1.stan: The model with total sapling nieghbor density and conspecific sapling neighbor density that were calculated based on basal area.
- nei_model2.stan: The model with total sapling nieghbor density and conspecific sapling neighbor density that were calculated based on stem count.
Second-Level Models
- model_FixFactor_trait.stan: file represents a Bayesian hierarchical model where functional traits are used as group-level predictors for species-specific coefficients of environmental or neighbor density variables. FixFactor represent fixed factor in Neighborhood density model, include "envpc1", "envpc2", "envpc3", "closure", "Acon", "Scon", "Atotal", "Stotal"; trait include "LA", "SLA", "LDMC", "LT", "LCC", "LNC", "LPC", "LKC", "WD", "SM", "PC1", "PC2".
E.g.
- model_closure_LA.stan: leaf area (LA) as the group level predictor for species-specific coefficients of canopy closure (closure).
Note: Since this study involves a large number of Stan models, we have provided detailed annotations only in "model_closure_LA.stan".
R scripts
- Bayesian_Model_Fitting.R: Code for Bayesian parameter estimation of growth models using Stan. This script compiles and executes Stan models via the rstan package in R. Generates Table S4, Table S5. Note that the full execution may take approximately 78 hours to complete on a standard research workstation (varies by hardware; tested on 8-core CPU with 32GB RAM).
- Extract_result.R: Code for extracting posterior coefficients and summary statistics from Bayesian model outputs for visualization and analysis. Generates Table 1, Table S1, Table S6, and Table S7.
- Figure_code.R: R script for the drawing Figure1 to Figure 4 and Figure S5.
- PCA_trait_environment_visualization.R: R script for principal component analysis of trait and environmental variables and visualization, including Figure S1, Figure S3, Table S2 and Table S3.
Intermediate Stan model objects
- base_data.RData: A serialized R data file containing a prepared and standardized dataset for Bayesian hierarchical modeling using Stan, including transformed response and predictor variables, species and plot indices, and standardized functional traits.
- first_level_model.Rdata: This file contains the fist-level model object (newfit1 and newfit2) resulting from a Bayesian hierarchical analysis.
- envpc1_model_list.RData: Contains a list of 12 fitted Bayesian hierarchical models from the second-level analysis, examining how 10 functional traits and their first two principal components influence species-specific responses to Soil PC1.
- envpc2_model_list.RData: Contains a list of 12 fitted Bayesian hierarchical models from the second-level analysis, examining how 10 functional traits and their first two principal components influence species-specific responses to Soil PC2.
- envpc3_model_list.RData: Contains a list of 12 fitted Bayesian hierarchical models from the second-level analysis, examining how 10 functional traits and their first two principal components influence species-specific responses to Soil PC3.
- closure10_model_list.RData: Contains a list of 12 fitted Bayesian hierarchical models from the second-level analysis, examining how 10 functional traits and their first two principal components influence species-specific responses to light availability.
- Atotal15_model_list.RData: Contains a list of 12 fitted Bayesian hierarchical models from the second-level analysis, examining how 10 functional traits and their first two principal components influence species-specific responses to total adult neighbor density.
- Acon15_model_list.RData: Contains a list of 12 fitted Bayesian hierarchical models from the second-level analysis, examining how 10 functional traits and their first two principal components influence species-specific responses to conspecific adult neighbor density.
- Stotal15_model_list.RData: Contains a list of 12 fitted Bayesian hierarchical models from the second-level analysis, examining how 10 functional traits and their first two principal components influence species-specific responses to total sapling neighbor density.
- Scon15_model_list.RData: Contains a list of 12 fitted Bayesian hierarchical models from the second-level analysis, examining how 10 functional traits and their first two principal components influence species-specific responses to conspecific sapling neighbor density.
- FigureData.RData: All outputs derived from Bayesian models for visualization.
Tree census
The tree censuses data were collected from the 20-ha (400 m × 500 m) Tropical Seasonal Rainforest Dynamics Plots in 2017 and 2021. The plot located in Naban River Watershed National Nature Reserve, Yunnan Province, Southwestern China (100°36′E, 22°14′N). All woody stems ≥ 1 cm DBH (diameter at breast height, 1.3 m above ground) in the plot were measured, tagged, mapped and identified to species.
Functional traits measurement
We measured functional traits following the standardized protocol (Cornelissen et al. 2003, Pérez-Harguindeguy et al. 2016). These included key whole-plant traits (wood density, WD, g/cm3; seed mass, SM, g) and leaf traits (specific leaf area, SLA, mm2/mg; leaf area, LA, mm2; leaf thickness, LT, mm; leaf dry mass content, LDMC, mg/g; leaf carbon content, LCC, mg/g; leaf phosphorus content, LPC, mg/g; leaf nitrogen content, LNC, mg/g; leaf potassium content, LKC, mg/g). These traits were measured in in the 20-ha (400 m × 500 m) Xishuangbanna Forest Dynamics Plot in Southwest China, located in Yunnan Province, Southwestern China (101°34′ E, 21°36′ N). For each tree species, 3 - 5 healthy adult individuals were randomly selected in the plot, and 3 intact leaves (including petioles) were collected from each individual during September-October 2013, placed in plastic bags, and transported to the laboratory. Each fresh leaf was scanned and the area was measured using IMAGEJ software (Abràmoff et al. 2004). Leaf thickness (mm) was measured at the center of the leaf lamina, avoiding major veins, using electronic digital micrometer (CANY Co., Shanghai, China) on fresh leaves. The fresh mass of each leaf was measured using electronic balance with a precision of 0.001 g. Leaves were placed in paper envelopes, dried at 70 ℃ for 72 hours to a constant weight, and then weighed. Petioles were removed, leaves were grounded, and carbon and nitrogen contents were determined using a carbon-nitrogen analyzer (Vario MAX CN). Total phosphorus and potassium contents in the leaves were determined according to the LY/T 1270-1999 standard, using ICP-AES and iCAP6300 elemental analyzer, respectively. Specific leaf area was calculated as LA/leaf dry mass, and leaf dry matter content was calculated as leaf dry mass/leaf fresh mass. In 2014, wood density was measured by extracting tree cores from live trees using a sharp increment borer, and their volume was determined using the water displacement method. The cores were oven-dried at 70°C for 72 hours to achieve constant weight and then weighed. Wood density was calculated as the dry weight of the cores divided by their volume (Zhou et al. 2020). Seed mass was measured by collecting seeds within the forest for most focal tree species. For a small number of species, seed mass data were obtained from the "Seeds of Woody Plants in China" (State Forestry Administration, National Forest Farm and Tree Seed and Seedling Workstation, 2001).
Topographic factors were measured by subdividing the plot into 20 m × 20 m quadrats and calculating aspect, slope, elevation, convexity and topographic wetness index (TWI) based on elevations from 10-m stakes placed across the NFDP (Harms et al. 2001; Valencia et al. 2004). TWI is the ratio of upslope contributing area to the local slope (Tarboton 1997; Sørensen et al. 2006). These calculations were performed using the "CTFS" and "RSAGA" packages (Hall 2006; Brenning et al. 2018) in R (R Core Team 2021).
Soil Physical and Chemical Properties
Soil properties were measured using a grid of 40 m × 40 m cells established across NFDP. For each cell, a basic sampling point was established at coordinate (10 m, 10 m) in the cell, and a random sampling point was selected at a random distance of 5, 10, or 20 m along a random x, y or xy direction. Two soil samples were collected from each of the 130 cells (totaling 260 samples) at a depth of 10 cm, excluding litter and humus. For each sample, 12 soil physical and chemical properties were measured: pH, total nitrogen content, total carbon content, total phosphorus content, available iron content, available aluminum content, available potassium content, available calcium content, available manganese content, available magnesium content, available sulfur content, and available phosphorus content . Soil pH was measured using a pH meter in a 1:2.5 (w/v) suspension. Total carbon and nitrogen contents were analyzed using an elemental analyzer (Vario MAX CN). Total Phosphorus (TP) was determined by digestion with nitric acid-hydrochloric acid-hydrofluoric acid and measured using an elemental analyzer (iCAP 7400 ICP-OES). Available iron, aluminum, potassium, calcium, manganese, magnesium, sulfur, and phosphorus were measured using the Mehlich-3 method (Ziadi and Tran 2007).
Reference
Abràmoff, M. D., P. J. Magalhães, and S. J. Ram. 2004. Image processing with ImageJ. Biophotonics International 11:36–42.
Brenning, A., D. Bangs, M. Becker, P. Schratz, and F. Polakowski. 2018. SAGA Geoprocessing and terrain analysis.
Cornelissen, J. H. C., S. Lavorel, E. Garnier, S. Díaz, N. Buchmann, D. E. Gurvich, P. B. Reich, et al. 2003. A handbook of protocols for standardised and easy measurement of plant functional traits worldwide. Australian Journal of Botany 51:335–380.
Hall, P. 2006. The CTFS large plot forest dynamics analyses.
Harms, K. E., R. Condit, S. P. Hubbell, and R. B. Foster. 2001. Habitat associations of trees and shrubs in a 50-ha neotropical forest plot. Journal of Ecology 89:947–959.
Perez-Harguindeguy, N., S. Diaz, E. Garnier, S. Lavorel, H. Poorter, P. Jaureguiberry, M. S. Bret-Harte, et al. 2016. New handbook for standardised measurement of plant functional traits worldwide. Australian Journal of Botany 61:715–716.
R Core Team. 2021. R: A language and environment for statistical computing.
State-owned forest farm and tree seedling work station of state forestry administration. 2001. Seeds of woody plants in China. China Forestry Publishing House, Beijing.
Sørensen, R., U. Zinko, and J. Seibert. 2006. On the calculation of the topographic wetness index: evaluation of different methods based on field observations. Hydrology and Earth System Sciences 10:101–112.
Tarboton, D. G. 1997. A new method for the determination of flow directions and upslope areas in grid digital elevation models. Water Resources Research 33:309–319.
Valencia, R., R. B. Foster, G. Villa, R. Condit, J. C. Svenning, C. Hernandez, K. Romoleroux, et al. 2004. Tree species distributions and local habitat variation in the Amazon: large forest plot in eastern Ecuador. Journal of Ecology 92:214–229.
Zhou C., B. Wang, Y. Deng, J. Wu, M. Cao, and L. Lin. 2020. Canopy structure is an important factor driving local-scale woody plant functional beta diversity. Biodiversity Science 28:1546–1557.
Ziadi, N., and T. S. Tran. 2007. Mehlich 3-extractable elements. Pages 81–88 in M. R. Carter and E. G. Gregorich, eds. Soil sampling and methods of analysis (2nd ed.). CRC Press, Boca Raton, USA.