Interspecific variation in leaf phenology and its relationship with plant traits in a seasonal tropical forest

Park, John 1 ; Lichstein, Jeremy2; Muller-Landau, Helene3; Wright, S. Joseph3; Bohlman, Stephanie2

Published Dec 06, 2025 on Dryad. https://doi.org/10.5061/dryad.1jwstqk84

Data files

Dec 06, 2025 version files 1.11 MB

README.md

7.23 KB
Table_S1.csv

12.32 KB
Table_S5.csv

1.09 MB

Abstract

Leaf phenology plays an important role in plant life history strategies and in determining ecosystem-level fluxes of carbon and water. In seasonal tropical forests, leaf phenology is highly variable among tree species, but limited quantitative data have hindered our understanding of leaf phenology, and its relationship with other plant traits.

We investigated leaf phenology and its relationship to plant traits in 48 tree species in a seasonally dry tropical forest. Our study combined image time-series acquired by uncrewed aerial vehicles (UAVs) with ground-based forest inventories and plant functional and life-history traits.

Quantitative metrics of phenology revealed wide variation along two dimensions –the degree and timing of deciduousness – with large interspecific variation. We categorized each species to one of three phenological groups: evergreen (25 species, leaf cover always > 60%), early deciduous (six species), and late deciduous (17 species). Early deciduous species were deciduous near the beginning of the dry season and for a relatively short period (mean 22 days); late deciduous species were deciduous later in the dry season and for a longer period (mean 63 days).

In multiple regression models, plant traits explained roughly 50% of the interspecific variation in phenology metrics. The best predictors in these models were xylem vessel area (species with larger vessels tended to have earlier start dates of deciduousness), LMA (negative relationship with the degree of deciduousness), and leaf phosphorus content (positive relationship with the degree of deciduousness).

Synthesis – Co-occurring tree species in a seasonal tropical forest vary widely in the degree and timing of deciduousness, with natural breaks in this variation defining three phenological groups – evergreen, early deciduous, and late deciduous – and continuous variation in phenological traits within the deciduous groups. Interspecific variation in the degree and timing of deciduousness is associated with other plant traits. Combining UAV imagery with ground-based forest inventory and trait data can lead to improved understanding of the complex phenology of tropical forests, which is crucial for accurately predicting carbon and water fluxes under changing climates.

Dataset DOI: 10.5061/dryad.1jwstqk84

Description of the data and file structure

Data from manuscript title "Interspecific variation in leaf phenology and its relationship with plant traits in seasonal tropical forest" from functional ecology. "table_s1.csv" list all the species included in this study, along with their phenology metrics and trait values."table_s5.csv" contain data for model-driven estimated leaf cover time series employed in this study.

Files and variables

Table S1. List of species included in this study, along with their phenology metrics and trait values. The table is provided as a comma delimited file (Table_S1.csv).

Column definitions:

● spcode: species code used in figures.

● genus_species: scientific name of species

● family: taxonomic family of the genus

● group: phenology species group assigned in the paper. We used the species-level phenology metrics to separate species into three groups, species with no or very low peak deciduousness (‘evergreen’: 25 species), species with high peak deciduousness whose main period of deciduousness occurred near the early part of the dry season (‘early-deciduous’: 6 species), and species with high peak deciduousness whose main period of deciduousness occurred after the early dry season (‘late-deciduous’: 17 species). See phenology metrics for the details of metrics used to segregate species into different groups. Evergreen species have PeakDecid < 40%; early-deciduous species have PeakDecid > 40% and MeanDate < 1 February; and late-deciduous species have PeakDecid > 40% and MeanDate ≥ 1 February. Abbreviations: earlyD = early deciduous; lateD = late deciduous.

● group_Meakem: phenology group assigned by Meakem et al. (2018) in the supplementary table S1 ( https://nph.onlinelibrary.wiley.com/action/downloadSupplement?doi=10.1111%2Fnph.14633&file=nph14633-sup-0001-SupInfo.pdf).

● n: number of individuals in our dataset of phenology metrics.

● phenology metrics (multiple columns): We defined multiple metrics to describe different aspects of deciduousness. Each metric was first calculated for individual trees and then calculated for each species. We defined daily deciduousness (%) for a given tree on a given date as 100 minus normalized leaf cover (%).

Annual deciduousness (AnnualDecid) was defined as the unweighted mean of daily deciduousness for the entire year.
Peak deciduousness (PeakDecid) was defined as the maximum value of daily deciduousness during the entire year.
The start, end, and length of the deciduous period (StartDate, EndDate, and LenDecid, respectively) were defined as the first day, last day, and number of days in the time series when deciduousness was above 40%. StartDate and EndDate is defined only for deciduous species.
MeanDate was quantified using circular statistics (Morellato et al., 2010), by representing each daily deciduousness as an angular vector, where the day of the year is represented by an angle (360°/365 days = 0.986° day−1), and the daily magnitudes are equal to the daily deciduousness values. For individual trees, MeanDate is computed as the directional component of the mean vector of daily deciduousness. Species-level MeanDate is computed by first calculating the species-level mean vector from the individual-level mean vectors, and then taking the directional component of that vector.

Please note that mean and standard deviation (SD) are calculated across individuals within species of each metric. For timing metrics (StartDate, EndDate, and MeanDate), SD (days) is the SD of the date, where the date is represented by an integer ranging from 1 (first day of our study: 2 Oct. 2014) to 358 (last day of our study: 24 Sep. 2015).

● nPeak40: number of individuals with PeakDecid > 40%.

● percPeak40: percentage of individuals with PeakDecid > 40%.

● traits (multiple columns): species-level values for each trait included in our analyses of trait-phenology relationships (Table 2). The leaf traits and life-history traits are described in Wright et al. (2010). All traits were collected from BCI or sites within 50 km of BCI. For leaf traits, separate values are reported for sun and shade leaves.

LMA: Leaf mass per area^* (g m⁻²)

LFTHICK: Mean leaf lamina thickness^* (mm)

WSG: Wood specific gravity (g∙cm⁻³)^⁋

DBHmax: Maximum diameter at breast height (mm)

RGR10: Relative diameter growth rate (% year⁻¹) of 10-49 mm diameter saplings

For nitrogen (N) and phosphorus (P), LMA- and normalization-residuals are reported as Nres and Pres (exponentiated values), and the original trait values (percent mass) are reported as Nraw (% mass: 100 × g N / g leaf dry mass) and Praw (mg P / g leaf dry mass), are from Wright (2023).

VesselArea: Mean area of individual xylem vessels (cm²), from Hietz et al. (2017).

Table S5. Model-driven estimated leaf cover time series employed in this study. The table is provided as a comma delimited file (Table_S5.csv).

Column definitions:

● date: date on which each drone flight was conducted.

● tag: unique identifier assigned to individual trees.

● rel_leaf: Relative leaf cover, calculated as the estimated leaf cover at each observation time point divided by the maximum estimated leaf cover of the focal tree.