Disentangling seed availability and establishment filters at alpine treelines through a decade-long field manipulation
Data files
Nov 07, 2025 version files 85 MB
-
brm_models.rds
64.90 MB
-
BRT_model.rds
19.96 MB
-
Code_brm.R
12.58 KB
-
Code_BRT.R
15.51 KB
-
Code_extract_interaction_summaries.R
5.12 KB
-
Code_seed.R
3.42 KB
-
data_sgs.csv
97.47 KB
-
FIGURE_2.R
3.77 KB
-
FIGURE_S2S3.R
7.74 KB
-
README.md
4.56 KB
Abstract
Climate warming is expected to increase upslope shifts of alpine treelines globally. However, the ecological filters controlling tree recruitment in the alpine belt remain poorly quantified, limiting accurate predictions of treeline responses to climate change. To fill this gap, we conducted a decade‑long, factorial field experiment at abrupt (Ordesa) and diffuse (Tessó) Pinus uncinata treelines located in the Spanish Pyrenees. Our experimental treatments manipulated seed addition, herbivory exclusion, shrub competition, and soil scarification. We monitored seedling presence and abundance annually and analyzed their interactions with climate variables, specifically growing-season temperature and snow depth. Seedling recruitment at alpine treelines was strongly filter-limited and varied between sites. Seed addition enhanced emergence at both treeline types, with a steeper, density-dependent response at the diffuse treeline. Herbivore exclusion (1-mm mesh) consistently increased densities across cohorts, underscoring herbivory as a critical biotic filter. Climatic and biotic factors interacted to shape establishment: at the abrupt treeline, warmer growing-season maxima and dense shrub cover suppressed recruitment, while snow depth exerted contrasting effects across cohorts, from protective to limiting. At the diffuse treeline, year-one seedlings peaked at intermediate growing-season maximum temperatures under low shrub cover, whereas older cohorts showed more variable responses, occasionally persisting under dense shrubs when cooler growing season and deeper snow provided facilitative microclimatic conditions. These patterns highlight that both propagule supply and the interplay of climatic stress and biotic interactions jointly determine early recruitment above treeline. Treeline advance under warming occurs only when multiple filters align—adequate seed supply, favourable microclimatic windows, moderated herbivory, and shrub effects that remain facilitative rather than competitive. Because these filters are context- and life-stage dependent, forecasts must move beyond climate envelopes to integrate fine-scale propagule dynamics, episodic heat/snow extremes, and density-dependent biotic interactions. Models coupling these processes with long-term observations will better predict forest–tundra change.
Description of the data and file structure
This dataset contains all the necessary data for reproducing the results of the research, as described in the associated manuscript: ‘Disentangling seed availability and establishment filters at alpine treelines through a decade-long field manipulation’
Files and variables
File: data_sgs.csv
Description: This data records field observations from alpine and treeline ecological experiments. It contains variables describing site conditions, experimental treatments, seedling dynamics, and climatic factors. Each row represents data from one permanent plot in a given observation year.
Variables
- Year: the year when the field observation was conducted
- Site: two treeline types are abrupt (Las Cutas) and diffuse (Tessó) treelines
- Zone: the ecological zone of treeline (Alpine tundra and transition)
- PlotID: a unique identifier for each permanent plot, typically combining site name and plot number (e.g., Las CutasAT1)
- Plot_number: plot number
- Mesh_cage: indicates whether a mesh cage was installed the plot (Caged or Uncaged)
- Treatment: the type of experimental treatment applied to the plot (Control, Seeded, Scarified and Seeded&Scarified)
- 1YS: number of year-one-old seedlings recorded in the plot
- 2YS: number of two-year-old seedlings recorded in the plot
- 3YS: number of three-year-old seedlings recorded in the plot
- prev1YS : the number of seedlings in the previous year corresponding to the current two-year-old cohort (i.e., 1YS from the prior year)
- prev2YS: the number of seedlings in the previous year corresponding to the current three-year-old cohort (i.e., 2YS from the prior year)
- seedlag: seed density in the plot, usually referring to the number of seeds sown per unit area (seeds per m²)
- Shrub: shrub cover percentage (%)
- GSMaxT: average maximum temperature during the growing season (°C)
- GSMinT: average minimum temperature during the growing season (°C)
- GSTP: total precipitation during the growing season (mm)
- SD: Snow depth during the winter season (cm)
- WMMWS: winter mean maximum wind speed (m/s)
File: Code_brm.R
Description: code for Bayesian generalized linear mixed-effects models with zero-inflated negative binomial distribution to examine the effects of treatment measures and environmental drivers on seedling recruitment.
File: Code_seed.R
Description: code for Bayesian generalized linear mixed-effects models with zero-inflated negative binomial distribution to examine the effects of seed density on year-one-old seedlings.
File: Code_extract_interaction_summaries.R
Description: code for extracting the posterior estimates, 89% credible intervals, and directional probabilities from each Bayesian generalized linear mixed-effects model.
File: Code_BRT.R
Description: code for exploring the relative importance of multiple influencing factors and the interaction effects among various variables using a boosted regression tree model.
File: FIGURE_2.R
Description: Code for plotting in FIGURE 2 shows temporal dynamics of seedling density (No. 0.25 m⁻²) from 2014 to 2024 (except 2019) across treeline types, seedling age classes (1YS = year-one seedlings, 2YS, 3YS), and zones (transition vs. alpine tundra).
File: FIGURE_S2S3.R
Description: Code for plotting in FIGURE S2 and S3. Figure S2 shows the interannual variations in the growing season climate at two different treeline study sites in the Pyrenees from 2014 to 2024. Figure S3 shows the zero proportion plots of seedlings at different age classes (1YS = year-one-old seedlings, 2YS, 3YS) under various treatments for the two tree species.
File: brm_models.rds
Description: This file contains the fitted Bayesian generalized linear mixed-effects models (zero-inflated negative binomial distribution) generated using Code_brm.R. These models quantify the effects of experimental treatments and environmental factors on seedling recruitment across sites and years.
File: BRT_model.rds
Description: This file stores the final boosted regression tree (BRT) models produced by Code_BRT.R. The models evaluate the relative importance and interaction effects of multiple ecological and climatic variables influencing seedling recruitment.
Code/software
All statistical analyses were conducted in R (version 4.4.3).
