Active interventions accelerate native plant recolonization following agricultural abandonment
Data files
Apr 29, 2026 version files 3.66 MB
-
IndicatorSppDatasetJAN2026.csv
3.63 MB
-
README.md
3.20 KB
-
Remnant.analysis.4.2026.R
29.56 KB
Abstract
A key challenge in restoration planning is predicting how passive and active processes may best be used together to promote successful landscape recovery. While passive recolonization of native plant communities can drive recovery, this process is often slow, and success is variable. Active restoration practices, such as reinstating historical disturbance regimes, may accelerate recolonization. However, few studies have empirically tested how specific active interventions can be used to facilitate recovery across heterogenous landscapes.
We conducted a 13-year field experiment monitoring the recolonization of longleaf pine savanna ground-layer communities in landscapes recovering from agricultural use in the southeastern USA. Across 12 sites, we established pairs of 1-ha plots in untilled, fire-suppressed remnant savannas and adjacent pine plantations previously under agriculture (recovering areas). Each pair was randomly assigned a restoration treatment: canopy thinning in the remnant, the recovering area, both, or neither (control), combined with either low or high frequency of prescribed fire. In each recovering area, we measured the spread of remnant indicator plant species along a 100 m transect extending away from the remnant.
We found that remnant savanna indicator species can slowly recolonize recovering areas after 70 years without active interventions. However, high fire frequency and canopy thinning in the remnant or recovering area resulted in 3-4× more remnant indicator species in recovering areas. These patterns were modified by seed dispersal mode: species with vertebrate and short-distance (e.g., by gravity or ants) dispersal modes spread further into recovering areas with combined thinning and burning treatments whereas species dispersed by wind were rare regardless of treatment.
Synthesis and applications: Our results show that combined canopy thinning, and prescribed burning can accelerate natural recolonization in savanna systems. However, additional measures (e.g. seed addition) may be needed for weak-dispersing species. These findings highlight the utility of integrating passive processes with active interventions to conduct efficient landscape-scale restoration.
Dataset DOI: 10.5061/dryad.9cnp5hr03
Description of the data and file structure
We conducted a 13-year field experiment monitoring the recolonization of longleaf pine savanna ground-layer communities in landscapes recovering from agricultural use in the southeastern USA. Across 12 sites, we established pairs of 1-ha plots in untilled, fire-suppressed remnant savannas and adjacent pine plantations previously under agriculture (recovering areas). Each pair was randomly assigned a restoration treatment: canopy thinning in the remnant, the recovering area, both, or neither (control), combined with either low or high frequency of prescribed fire. In each recovering area, we measured the spread of remnant indicator plant species along a 100 m transect extending away from the remnant.
Files and variables
File: IndicatorSppDatasetJAN2026.csv
Variables
- site: Which of the 12 sites this row of data was collected in
- year: Whether the data in this row was collected in 2015 or 2024
- PlotName:Tool to identify which of the nultiple pairs of plots (remnant and recovering areas) was sampled in each site
- treatment:Whether the remnant was thinned ("Harvest") or not ("Control")
- pairedplottreatment:Whether the recovering area was thinned ("Harvest") or not ("Control")
- Configuration: What thinning treatment the plot pair experienced (e.g. thinning in both remnant and recovering areas)
- distance:How far away on the transect (0-100m) the sampled plot was from the remnant
- species: The identity of the species measured in the "presence" and "abundance" columns below. Species are spelled using their 3-letter codes (e.g. SORNUT = Sorgastrum nutans)
- species_cleaned:The identity of the species measured in the "presence" and "abundance" columns below, changed from "species" to reflect the changes in taxonomy between the 2015 and 2024 surveys--species are spelled using their 3-letter codes (e.g. SORNUT = Sorgastrum nutans)
- presence: Whether the species was present in the plot at the given distance or not
- abundance: What abundance that species was in the plot
- WHC: The proportionate amount of water dried soil was capable of holding onto (% of wet weight that is water)
- burns.since.1971:How many prescribed burns had occurred at the site since 1971
- burns.since.2000: How many prescribed burns had occurred at the site since 2000
- tot.burns:Sum of the previous two columns
- burn.frequency: Category for whether a site was burned frequently ("High") or infrequently ("Low")
- DispModeCat: The dispersal mode each species was categorized into (either short distance, animal, or wind-dispersed)
File: Indicator_species_project.R
R script used to analyze data included in IndicatorSppDatasetJAN2026.csv.
Code/software
We performed all statistical tests in RStudio (R version 4.4.3; R Core Team, 2025). The R file in this dataset includes all packages used.
Access information
Other publicly accessible locations of the data:
- NA
Data was derived from the following sources:
- NA
