Dataset DOI: 10.5061/dryad.cc2fqz6mt

Description of the data and file structure

These datasets contain seed rain and fruiting trees data collected from a long-term tropical forest restoration experiment near Agua Buena, Puntarenas, Costa Rica. See the Islas Project for more information (http://www.holl-lab.com/islas-project.html)

Seed sampling was conducted across seven experimental sites, each site contained three 0.25-ha areas with a tree plantation, natural regeneration, and an applied nucleation restoration plot. Additionally, four mature forests nearby the restoration sites are used as reference forest. Within each plot, 16 seed traps (0.5 m², 1 mm mesh) were installed and monitored over a two-year period (March 2021–February 2023), with seeds collected twice monthly. All seeds ≥1 mm were counted and identified to the lowest possible taxonomic level using a reference collection, resulting in 142,501 seeds from 104 later-successional species across 41 families. For each record, associated information includes trap identity and coordinates, site and treatment, species identity, dispersal mode, growth form, and seed size. Species were classified into successional categories (mid- or late-successional), and seed size was recorded or categorized.

Tree species’ successional stage was determined based on the expert opinion of two taxonomists who have each worked in the region for >20 years, herbarium specimens at Las Cruces, and literature resources (Chazdon et al. 2011 A novel statistical method for classifying habitat generalists and specialists. Ecology 92, 1332–1343. De Souza, 2001. Seed size, seed germination, and seedling survival of Brazilian tropical tree species differing in successional status. Biotropica 33, 447–457. Werden et al. 2020 Effects of dispersal- and niche-based factors on tree recruitment in tropical wet forest restoration. Ecol. Appl. 30, e02139).

Finally, we recorded data on which trees were flowering or fruiting in restoration plots from January 2021 to April 2023. Observations were made during seed collection fieldwork and during bird surveys (4 times per year).

Files and variables

File – Seed_rain_data.csv

Primary dataset containing seed counts at the trap level. Each row represents a sampling event for a given seed trap. It also contains the species-level trait dataset used to assign ecological categories. Key variables include:

Site

Two-letter identifier of the 7 restoration sites used in the study. At four of them (*), a nearby mature forest was selected as a reference forest. These acronyms are mostly based on the original landowner names:

BB = Bambu
EC = El Cenizo
JG = Julio Gonzalez*
LL = Loma Linda*
MM = Melissa's Meadow*
OM = Omar*
SG = San Gabriel

Treatment

Restoration treatment:

NR = natural regeneration
AN = applied nucleation
PL = plantation
RF = reference forest

Trap

Number of the trap where the seeds were collected (1 to 16)

Species

Species name 6-letter code: first three letters for genus, followed by the three first letters of the species (GEN_SPP).

succ_stage

Early- mid- or late-successional species according to expert opinion, herbarium specimens at Las Cruces, and literature resources.

disp_mode

Dispersal syndrome:

ZOO = animal-dispersed
WIND = wind-dispersed
GRAV = barochory
EXPL = explosive dispersal

grow_form

Growth form of the seed species:

UP = understory palm
CP = canopy palm
ET = emergent tree
CT = canopy tree
UT = understory tree
Shrub
Liana
Epiphyte

*Forbs were not taken into account due to the height of our traps (>1 m from the ground)

seed_size

Size classes according to seed width:

1: <5 mm
2: 5 < 10 mm
3: 10 - 15 mm
4: >15 mm

seed_width

Seed size in mm measured or assigned based on the literature

seed_sum

Sum of the number of seeds recorded in each trap along two sampling years.

Missing values are indicated as "NA".

File – Fruiting_tree_data.csv

Records of fruiting individuals within plots are used to infer potential local seed sources. Key variables include:

Date

Date of observation dd-month-yy

Site

Two-letter identifier of the 7 restoration sites used in the study. At four of them (*) a nearby mature forest was selected as a reference forest. These acronyms are mostly based on the original landowner names:

BB = Bambu
EC = El Cenizo
JG = Julio Gonzalez*
LL = Loma Linda*
MM = Melissa's Meadow*
OM = Omar*
SG = San Gabriel

Treatment

Restoration treatment:

NR = natural regeneration
AN = applied nucleation
PL = plantation
RF = reference forest

Species

Species name 6-letter code: first three letters for genus followed by the three first letters of the species (GEN_SPP).

Family

Taxonomical family of the species

Genus

Taxonomical genus of the species

Specific.Epithet

Epithet of the species

Authority

Author(s) of the species name

grow_form

Growth form of the seed species:

UP = understory palm
CP = canopy palm
ET = emergent tree
CT = canopy tree
UT = understory tree
Shrub
Liana
Epiphyte

*Forbs were not taken into account due to the height of our traps (>1 m from the ground)

disp_mode

Dispersal syndrome:

ZOO = animal-dispersed
WIND = wind-dispersed
GRAV = barochory
EXPL = explosive dispersal

succ_stage

Early- mid- or late-successional species according to expert opinion, herbarium specimens at Las Cruces, and literature resources.

seed_size

Size classes according to seed width:

1: <5 mm
2: 5 < 10 mm
3: 10 - 15 mm
4: >15 mm

seed_width_mm

Missing values are indicated as "NA".

File – Trap_location.csv

Site

Two-letter identifier of the 7 restoration sites used in the study. At four of them (*) a nearby mature forest was selected as a reference forest. These acronyms are mostly based on the original landowner names:

BB = Bambu
EC = El Cenizo
JG = Julio Gonzalez*
LL = Loma Linda*
MM = Melissa's Meadow*
OM = Omar*
SG = San Gabriel

Treatment

Restoration treatment:

NR = natural regeneration
AN = applied nucleation
PL = plantation
RF = reference forest

Trap

Number of the trap where the seeds were collected (1 to 16)

Latitude, Longitude: Geographic coordinates of each trap in decimal degrees (WSG86/UTM zone 17 N)

Code/software

Excel (Microsoft Office 2016)

Data processing involved several steps to prepare the dataset for analysis. Early-successional species and selected taxonomically ambiguous groups were excluded to focus on mid- and late-successional species relevant to forest recovery. Seed deposition rate (SDR) was calculated at the seed-trap level as the total number of seeds divided by trap area and sampling duration (seeds m⁻² yr⁻¹), and values were log-transformed to account for high variance in seed counts. Species were grouped by successional stage and seed size (small <5 mm; large ≥5 mm), and dispersal modes and growth forms were assigned based on previously published trait databases. Spatial coordinates were used to calculate distances among traps and account for spatial autocorrelation in subsequent analyses. Species-by-trap matrices were constructed for diversity and composition analyses, and additional datasets were merged (e.g., fruiting tree records) to classify seeds as locally produced or externally dispersed at the plot (treatment within a site) level.