Data and code from: Succession in a tropical dry forest: A test of the chronosequence and inference of community assembly dynamics
Data files
Aug 03, 2026 version files 41.09 KB
-
Carrington_et_al._data_files.zip
24.41 KB
-
README.md
16.68 KB
Abstract
To better understand successional changes and drivers of changes in forests, we must 1) test assumptions of successional studies and 2) quantify patterns in species and functional traits in stands varying in successional age. We examined patterns in species composition and functional traits along a chronosequence of tropical dry forest (TDF) sites in North Key Largo, Florida in 1993. To test patterns identified along the chronosequence, we quantified longitudinal change in species and functional traits by remeasuring the sites in 2013. We tested the following predictions: 1) dynamic changes within sites in species composition and functional trait composition will corroborate a pattern indicated by the chronosequence; 2) a growth-survival demographic tradeoff will be consistent with increasing dominance of evergreen species over the course of succession; and 3) forest succession will demonstrate convergence in both composition and function. Data include tree species basal areas, community-weighted trait data, and environmental vectors (site age, site soil depth, site elevation). Also included are data analysis files and results of analyses for principal component analysis (PCA) of species by functional traits, delineation of functional groups (deciduous, evergreen, and leaf exchanger species) and basal areas of functional groups over the course of succession, NMDS analyses of species basal area by site age and community-weighted traits by site age, and analyses of convergence of species composition and functional traits by site age class using the DispRity package in R. These data may be reused, and there are no legal or ethical considerations associated with reuse.
Dataset DOI: 10.5061/dryad.c2fqz61rt
Description of the data and file structure
Files and variables
File: Carrington_et_al._data_files.zip
Description: Compressed file containing all data and data analysis files.
File: Sites.txt
Description: Names and site numbers of sites sampled only in 1993, in 1993 and 2013, and only in 2013.
- Name: Name of site
- Age: Age of site in years; age indicates number of years since site disturbance (i.e., start of secondary succession).
- Site_No: Site assigned number
File: Soil_dpth.csv
- Name: Name of site
- Soildepth: Soil depth (cm). Soil depth is an average of 25 measurements along a transect at each site.
File: Site_elevation.csv
- Name: Name of site
- Site_No: Site assigned number
- Elevation: Site elevation (m). Site elevation is an average of 4 to 10 measurements along each transect (determined for 10-meter lengths for transects 40 – 100 m in length), from LiDAR-derived United States Geological Survey 3D Elevation Program (3DEP) Datasets (U.S. Geological Survey. 2021. USGS National Map 3DEP Downloadable Data Collection: U.S. Geological Survey). Elevations were not measured for LNDFLLS (site #6), N9 (site #16), or CRDSNDE (site #19).
File: LL_SEAS.csv
Description: Leaf longevities and seasonality by species.
- Species: Species code. AMY = Amyris elemifera, ARD = Ardisia escallonioides, ATE = Ateramnus lucida, BOU = Bourreria ovata, BUR = Bursera simaruba, COC = Coccoloba diversifolia, EUGA = Eugenia axillaris, EUGF = Eugenia foetida, GUA = Guapira discolor, GUEE = Guettarda elliptica, GUES = Guettarda scabra, KRU = Krugiodendron ferreum, LYS = Lysiloma bahamensis, MET = Metopium toxiferum, NEC = Nectandra coriacea, PIS = Piscidia piscipula, PIT = Pithecellobium guadeloupense, REY = Reynosia septentrionalis, SID = Sideroxylon salicifolium, SWI = Swietenia mahagoni
- ll: Leaf longevity (days). Leaf longevity was estimated for each species by plotting the total number of leaves surviving from the 1990 cohort (first counted in January 1990, and not counting new leaves added during subsequent months) by month and estimating the half-life of leaves in days from a curve (either straight line or quadratic function) fit to the plotted data. Estimated leaf longevity for each species was twice the leaf half-life.
- Seasonality: Leaf seasonality. Leaf seasonality was calculated as the coefficient of variation of the total number of leaves counted by month (including new leaves added), from January 1991 to January 1992.
File: Traits.csv
Description: Species functional trait values. This file was used as the input file for PCA of species by functional traits.
- SLA: Specific leaf area (cm^2/g)
- WD: wood density (g/cm^3)
- MULTI: frequency of multiple stems (%)
- d13C: Leaf stable carbon isotope ratio (o/oo)
- TP: leaf total phosphorus (µg/g)
- TN: leaf total nitrogen (%)
- N:P: leaf N:P (ratio)
- HT:DBH: tree height:diameter at breast height (ratio)
- CrownA/DBH: tree crown area/diameter at breast height (cm)
- Htmax: maximum tree height (m)
- ll: leaf longevity (days)
- Seasonality: leaf seasonality (days)
File: PCA_speciesbytraits.txt
Description: Output file for ordination of species by functional traits. Conducted a principal component analysis (PCA) using PC-ORD 6.08.
File: Species_traits.csv
Description: Input file for one-way MANOVA analyzing differences in functional traits among three functional groups.
- Fgroup: Functional group. D = deciduous, E = evergreen, LE = leaf exchanger
- Species: Species code. In addition to the 21 species codes defined earlier, this file contains FIC = Ficus citrifolia.
- SLA: Specific leaf area (cm^2/g)
- WD: standardized value for wood density (number of standard deviations above or below mean for all species)
- freq multiple stems: standardized value for frequency of multiple stems (number of standard deviations above or below mean for all species)
- d13C: standardized value for leaf stable carbon isotope ratio (number of standard deviations above or below mean for all species)
- TP: standardized value for leaf total phosphorus (number of standard deviations above or below mean for all species)
- TN: standardized value for leaf total nitrogen (number of standard deviations above or below mean for all species)
- N:P: standardized value for leaf N:P (number of standard deviations above or below mean for all species)
- HT/DBH: standardized value for tree height:diameter at breast height (number of standard deviations above or below mean for all species)
- CrownA/DBH: standardized value for tree crown area/diameter at breast height (number of standard deviations above or below mean for all species)
- Htmax: standardized value for maximum tree height (number of standard deviations above or below mean for all species)
- ll: standardized value for leaf longevity (number of standard deviations above or below mean for all species)
- Seasonality: standardized value for leaf seasonality (number of standard deviations above or below mean for all species)
File: MANOVA_fgroup.txt
Description: One-way MANOVA analyzing differences in functional traits among three functional groups was followed by univariate ANOVAs conducted individually for each functional trait. Following statistically significant (p < 0.05) ANOVAs, we used Tukey’s post-hoc tests to detect differences in functional traits among functional groups. We conducted the analysis using R.
File: 1994_ba.csv
Description: Input matrix for NMDS analysis of 1993 data examining relationships between species basal area and environmental variables. Basal areas are relativized within each site to the maximum species value.
- Siteno: Each value indicates the assigned site number and the year sampled; e.g., 1_93 = site #1 sampled in 1993.
- AMY: basal area (m^2/ha) for Amyris elemifera
- ARD: basal area (m^2/ha) for Ardisia escallonioides
- ATE: basal area (m^2/ha) for Ateramnus lucida
- BOU: basal area (m^2/ha) for Bourreria ovata
- BUR: basal area (m^2/ha) for Bursera simaruba
- COC: basal area (m^2/ha) for Coccoloba diversifolia
- EUGA: basal area (m^2/ha) for Eugenia axillaris
- EUGF: basal area (m^2/ha) for Eugenia foetida
- FIC: basal area (m^2/ha) for Ficus citrifolia
- GUA: basal area (m^/ha) for Guapira discolor
- GUEE: basal area (m^2/ha) for Guettarda elliptica
- GUES: basal area (m^2/ha) for Guettarda scabra
- KRU: basal area (m^2/ha) for Krugiodendron ferreum
- LYS: basal area (m^2/ha) for Lysiloma bahamensis
- MET: basal area (m^2/ha) for Metopium toxiferum
- NEC: basal area (m^2/ha) for Nectandra coriacea
- PIS: basal area (m^2/ha) for Piscidia piscipula
- PIT: basal area (m^2/ha) for Pithecellobium guadeloupense
- REY: basal area (m^2/ha) for Reynosia septentrionalis
- SID: basal area (m^2/ha) for Sideroxylon salicifolium
- SWI: basal area (m^2/ha) for Swietenia mahagoni
File: Evectors1994.csv
Description: Environmental vectors (site age, site soil depth, and site elevation) as input file for NMDS analysis of 1993 data examining relationships between species basal area and environmental variables, and relationships between community-weighted traits and environmental variables.
- Age: Age of site (years) in 1993; age indicates number of years since site disturbance (i.e., start of secondary succession).
- Elevation: Site elevation (m). Elevations were not available for sites 6, 16, or 19.
- Soildepth: Soil depth (cm)
File: 1993_ba_nmds.txt
Description: R code and results for NMDS analysis of 1993 species basal area data. We used NMDS (“metaMDS” function in “Vegan” package in R) to examine relationships among communities and combined it with a vector-fitting technique to understand the relationships between forest species composition and important environmental variables.
File: 2013_ba.csv
Description: Input matrix for NMDS analysis of 2013 data examining relationships between species basal area and environmental variables. As in 1994_ba.csv, basal area values (m^2/ha) are relativized within each site to the maximum species value. Variable names follow the same naming conventions as in 1994_ba.csv.
File: Evectors2013.csv
Description: Environmental vectors (site age, site soil depth, and site elevation) as input file for NMDS analysis of 2013 data examining relationships between species basal area and environmental variables, and relationships between community-weighted traits and environmental variables.
- Age: Age of site (years) in 2013; age indicates number of years since site disturbance (i.e., start of secondary succession).
- Elevation: Site elevation (m). Elevation was not available for site 19.
- Soildepth: Soil depth (cm)
File: 2013_ba_nmds.txt
Description: R code and results for NMDS analysis of 2013 species basal area data. We used NMDS (“metaMDS” function in “Vegan” package in R) to examine relationships among communities and combined it with a vector-fitting technique to understand the relationships between forest species composition and important environmental variables.
File: 1994_Cwtdtraits_nolabels.csv
Description: Input matrix for NMDS analysis of 1993 data examining relationships between community-weighted traits and environmental variables. Each trait value is a community-weighted mean (CWM); i.e.,
CWMip = Σμifi ,
where i = species, p = site, and μi and fi are the mean trait value and relative abundance (proportion of stand basal area) of the species i in site p. Although this matrix has no variable labels, each line represents an individual site sampled in 1993 (sites are identical to those in 1994_ba.csv). Each column represents the following functional traits:
- SLA: Specific leaf area (cm^2/g)
- WD: wood density (g/cm^3)
- MULTI: frequency of multiple stems (%)
- d13C: absolute value of leaf stable carbon isotope ratio (o/oo)
- TP: leaf total phosphorus (µg/g)
- TN: leaf total nitrogen (%)
- N:P: leaf N:P (ratio)
- HT:DBH: tree height:diameter at breast height (ratio)
- CrownA/DBH: tree crown area/diameter at breast height (cm)
- Htmax: maximum tree height (m)
- ll: leaf longevity (days)
- Seasonality: leaf seasonality (days)
File: 1993_trait_nmds.txt
Description: R code and results for NMDS analysis of 1993 community-weighted trait data. We used NMDS (“metaMDS” function in “Vegan” package in R) to examine relationships among communities and combined it with a vector-fitting technique to understand the relationships between forest species composition and important environmental variables.
File: 2013_Cwtdtraits_nolabels.csv
Description: Input matrix for NMDS analysis of 2013 data examining relationships between community-weighted traits and environmental variables. Each trait value was a community-weighted mean (CWM); i.e.,
CWMip = Σμifi ,
where i = species, p = site, and μi and fi are the mean trait value and relative abundance (proportion of stand basal area) of the species i in site p. Although this matrix has no variable labels, each row represents an individual site sampled in 2013 (sites are identical to those in 2013_ba.csv). Each column represents the following functional traits:
- SLA: Specific leaf area (cm^2/g)
- WD: wood density (g/cm^3)
- MULTI: frequency of multiple stems (%)
- d13C: absolute value of leaf stable carbon isotope ratio (o/oo)
- TP: leaf total phosphorus (µg/g)
- TN: leaf total nitrogen (%)
- N:P: leaf N:P (ratio)
- HT:DBH: tree height:diameter at breast height (ratio)
- CrownA/DBH: tree crown area/diameter at breast height (cm)
- Htmax: maximum tree height (m)
- ll: leaf longevity (days)
Seasonality: leaf seasonality (days)
File: 2013_trait_nmds.txt
Description: R code and results for NMDS analysis of 2013 community-weighted trait data. We used NMDS (“metaMDS” function in “Vegan” package in R) to examine relationships among communities and combined it with a vector-fitting technique to understand the relationships between forest species composition and important environmental variables.
File: ba2013mds.csv
Description: Input file for test of convergence in species composition across the chronosequence. Each row in this file represents an individual site sampled in 2013 (sites are identical to those in 2013_ba.csv). Each column represents an ordination axis from nonmetric multidimensional scaling (NMDS) ordination of species basal areas in 2013 sites. Data in this file comprise the output from the R code (“metaMDS” function in “Vegan” package) included in 2013_ba_nmds.txt.
- NMDS1: First NMDS axis. Values represent loadings on this axis for individual sites.
- NMDS2: Second NMDS axis. Values represent loadings on this axis for individual sites.
- NMDS3: Third NMDS axis. Values represent loadings on this axis for individual sites.
File: trait2013mds_nolabels.csv
Description: Input file for test of convergence in functional traits across the chronosequence. Although this matrix has no variable labels, each row represents an individual site sampled in 2013 (sites are identical to those in 2013_ba.csv). Each column represents an ordination axis from nonmetric multidimensional scaling (NMDS) ordination of community-weighted trait values in 2013 sites. Data in this file comprise the output from the R code (“metaMDS” function in “Vegan” package) included in 2013_trait_nmds.txt.
File: DispRity_analyses.txt
Description: R code and results for Tests for convergence in species composition (basal areas) and function (community-weighted trait values) across the chronosequence: We used the “dispRity” package in R to compare inter-site dissimilarities (disparities) within three age classes of sites, as sampled in 2013: < 50, 51 – 90, and > 90 years since abandonment.
File: TraitCVsbyAgeClass.csv
Description: Coefficients of variation (CV) of traits across site age classes: To identify functional traits driving patterns of inter-site dissimilarities across the chronosequence, we calculated coefficients of variation (CV) of all community-weighted traits for each of the three site age classes and compared trends in coefficients of variation among all functional traits across the age classes. CV values are unitless.
- AgeClass: Age categories of sites (years)
- CV_SLA: Coefficient of variation for specific leaf area values
- CV_WD: Coefficient of variation for wood density values
- CV_MULTST: Coefficient of variation for frequency of multiple stems values
- CV_d13C: Coefficient of variation for leaf stable carbon isotope ratio values
- CV_TP: Coefficient of variation for leaf total phosphorus values
- CV_TN: Coefficient of variation for leaf total nitrogen values
- CV_N:P: Coefficient of variation for leaf N:P values
- CV_HT:DBH: Coefficient of variation for tree height:diameter at breast height values
- CV_CA/DBH: Coefficient of variation for tree crown area/diameter at breast height values
- CV_Htmax: Coefficient of variation for maximum tree height values
- CV_LL: Coefficient of variation for leaf longevity values
- CV_Seas: Coefficient of variation for leaf seasonality values
Code/software
PCA of species by functional traits: PC-ORD 6.08
MANOVA, univariate tests, Tukey tests conducted in R. R code is included in MANOVA_fgroup.txt.
NMDS examining relationships between species basal area and environmental variables for 1993 and 2013 data; and examining relationships between community-weighted traits and environmental variables for 1993 and 2013 data: We used the “metaMDS” function in “Vegan” package in R. R code for the four NMDS analyses is included in 1993_ba_nmds.txt, 2013_ba_nmds.txt, 1993_trait_nmds.txt, and 2013_trait_nmds.txt.
Tests for convergence in species composition and function across the chronosequence: We used the “dispRity” package in R. R code is included in DispRity_analyses.txt.
In 1993, we sampled tree composition in 23 stands that varied in age from 14 years to more than a century since abandonment following agriculture or other land-clearing activities in the North Key Largo, Florida tropical dry forest. In 2013, following 20 years without a significant hurricane, 19 of the stands were resampled, and 7 additional stands were added. In both surveys, sampling in each stand was concentrated on a nested belt transect, 40-100 m in length. A center line was established, and stem diameter at 1.4 m was measured for each tree 1-5 cm diameter at breast height (DBH) that was rooted within 1 m of the line. Trees 5-25 cm DBH were measured within a plot that extended 2 m from the center line, and all trees >25 cm DBH were measured if they fell within 5 m of the line. Soil depth was measured in 2015-17 by probing to bedrock with a 1-cm diameter rod at 25 evenly spaced locations along the axis of each transect. Surface elevation was determined for 10-meter lengths along each transect from LiDAR-derived United States Geological Survey 3D Elevation Program (3DEP) Datasets with a vertical accuracy of 10 cm root mean square error (RMSE) and a horizontal resolution of 1 m.
We selected thirteen functional traits for analysis. Data for tree height (HT), diameter at breast height, i.e., 1.4 m (DBH), crown area (CA), and the presence of multi-stemmed structure (MULTI) were collected from all trees ≥1 cm DBH rooted within 20 x 20 m plots near ten of the transects. Means for HT:DBH and CA:DBH (crown area:dbh) were determined within each stand and averaged across all plots where the species was present. The percentage of multi-stemmed trees (MULTI) was also determined based on species averages across plots. MAXHT was the height of the tallest individual among all measured stems in the ten plots. Wood density (WD) was determined from three 1-2 cm DBH branch samples collected from 3-5 individuals per stand. It was calculated as the ratio of the oven-dry mass of the wood sample divided by the mass of water displaced by its fresh (green) volume. Specific leaf area (SLA), percent nitrogen and phosphorus (N and P), and δ13C were determined from three leaves collected from 3-5 individuals of each species present in each plot. Recently expanded sun leaves were sampled, or in cases of understory species, the most illuminated leaves on the plant were sampled. Estimates of leaf longevity and seasonality were based on a 1991-92 study of leaf demography. Two individuals of 21 common Key Largo trees were selected along a trail near one of the transects. Single meristems from two separate branches were selected for monitoring of leaf production/survival. Beginning in January 1991 and continuing at monthly intervals through January 1992, we counted leaf numbers within two cohorts: marked leaves present at the initiation of the study, and leaves added during the next 12 months. To estimate leaf longevity, for each species we plotted the total number of leaves surviving from the 1990 cohort (first counted in January 1990, and not counting new leaves added during subsequent months) by month and estimated the half-life of leaves in days from a curve (either straight line or quadratic function) fit to the plotted data. Estimated leaf longevity for each species was twice the leaf half-life. Leaf seasonality was calculated as the coefficient of variation of the total number of leaves counted by month (including new leaves added), from January 1991 to January 1992.
We used a principal component analysis (PCA) to produce a detailed ordination of tree species by functional traits. We used a plot of leaf longevity vs. leaf seasonality to group species into three functional groups. We analyzed differences in functional traits by functional group using a one-way MANOVA followed by univariate ANOVAs conducted individually for each functional trait. Following statistically significant (p < 0.05) ANOVAs, we used Tukey’s post-hoc tests to detect differences in functional traits among functional groups. We then quantified changes in mean basal area of the functional groups over the chronosequence of sites. We used non-metric multidimensional scaling (NMDS; “metaMDS” function in “Vegan” package in R) to examine relationships among communities and combined it with a vector-fitting technique to understand the relationships between forest composition and important environmental variables. The ordination was applied to tree species abundances from all 27 sites sampled in 1993 and/or 2013. We used basal area as the measure of abundance, relativized to the maximum value observed for each species. We used the Bray-Curtis metric as a measure of dissimilarity among sites. The environmental variables of interest – stand age, soil depth, and elevation – were standardized to mean = 0 and standard deviation = 1 prior to being fit to the ordination. We used the “envfit” function in the “Vegan” package in R to fit the stand age, soil depth and elevation environmental vectors to each NMDS ordination. For each environmental vector, statistical significance of fit to the ordination was calculated through a permutation test, and goodness of fit was the squared correlation coefficient (r2). A similar ordination was performed on a site-by-functional trait data set. The functional trait values were community-weighted means (CWM) (Violle et al. 2007; Hulshof et al. 2013), i.e.,
CWMip = ∑µifi
where i = species, p = site and μi and fi are the mean trait value and relative abundance (proportion of stand basal area) of the species i in site p.
To test for convergence in composition and function across the chronosequence, we used the “dispRity” package in R to compare inter-site dissimilarities (disparities) within three age classes of sites, as sampled in 2013: < 50 (n = 6 sites), 51 – 90 (n = 14 sites), and > 90 years since abandonment (n = 5 sites). The “dispRity” package measures and compares the volume in multivariate space occupied by sites in each of the age classes. We conducted this analysis twice—once using the 3-dimensional NMDS ordination of species basal area data for sites and once using a similar ordination of community-weighted trait data. To identify functional traits driving patterns of inter-site dissimilarities across the chronosequence, we calculated coefficients of variation (CV) of all community-weighted traits for each of the three site age classes and compared trends in coefficients of variation among all functional traits across the age classes.