Data and code from: Alternative vegetation states in old-field grasslands: does dead biomass from native and non-native species matter?
Data files
Sep 28, 2026 version files 461.71 KB
-
Exp1_Biomass.xlsx
10.05 KB
-
Exp1_Vegetation.xlsx
24.87 KB
-
Exp2_SeedBank.xlsx
14.58 KB
-
Figure1_ExperimentalDesign.png
400.79 KB
-
README.md
11.43 KB
Abstract
Aims: Vegetation in old fields can exist as alternative states dominated by native or non-native species. Dead biomass accumulation may play a key role in community dynamics. We aimed to disentangle the role of living and dead biomass in maintaining alternative states, represented by native, regenerating tussock grasses and non-native, invasive tussock grasses.
Location: Inland Pampa, Buenos Aires Province, Argentina.
Methods: We conducted a factorial field experiment manipulating the removal of living biomass, dead biomass (standing dead and surface litter), or both in two patch states: one native dominated by Paspalum quadrifarium and one invasive state dominated by Festuca arundinacea. Species richness and cover were monitored for one year. We also assessed propagule availability using a greenhouse seedling emergence assay with soil and litter samples from both patch states.
Results: Only simultaneous removal of living and dead biomass increased species richness, independently of patch state, but establishment was strongly context dependent. Dominant species persistence was unaffected by biomass removals. Other species, mostly non-native, established almost exclusively in invaded patches, mainly following combined dead and living biomass removal, whereas native patches remained resistant to establishment. Seed bank richness and seedling emergence were higher in soil than in litter, with native seedlings more abundant than non-native ones. However, native species showed extremely limited establishment in the field.
Conclusions: Persistence of alternative native and non-native states was driven more by rapid recovery of dominant grasses and patch-specific establishment filters than by dead biomass accumulation alone. Despite the native-dominated seed bank, post-disturbance establishment was largely restricted to non-native species, suggesting a decoupling between propagule availability and community reassembly. Overall, a single aboveground disturbance event seems insufficient to destabilize these alternative states. Understanding interactions among disturbance, community assembly, and regeneration may contribute to ecological theory on alternative stable states and evidence-based restoration of old-field grasslands.
Dataset DOI: 10.5061/dryad.ghx3ffc3t
Description of the data and file structure
Field data were collected from a factorial field experiment designed to evaluate how the removal of living and dead aboveground biomass affects plant community structure in old-field grasslands invaded by Festuca arundinacea, which coexist with patches dominated by the native grass Paspalum quadrifarium. The experiment was established in December 2023 within an enclosed old-field grassland containing scattered patches dominated by Festuca arundinacea and Paspalum quadrifarium. In each patch state, five plots were established and subdivided into four (4) 1 m² subplots assigned to biomass removal treatments (control, dead biomass removal, living biomass removal, and combined removal). Species richness and percent cover were recorded through repeated censuses following treatment application to assess community responses over time.
In addition, soil and litter samples were collected in each block to assess the seed bank associated with each patch state. Samples from Festuca- and Paspalum-dominated patches were processed in a greenhouse germination assay, where soil and litter substrates from each patch type were spread in germination trays. All emerging seedlings were identified and counted to characterize the pool of species available in each patch state.
Files and variables
File: Exp1_Biomass.xlsx
Description:
This file contains aboveground biomass measurements collected in a factorial field experiment manipulating the removal of living and dead biomass in grassland patches dominated by Festuca arundinacea (FA) and Paspalum quadrifarium (PQ). Biomass was harvested, separated by component, dried, and weighed.
Variables
- block: Integer. Experimental block identifier (1–5). Each block contains one subplot per treatment and patch state.
- patch_state : Categorical. Dominant species of the patch where the subplot is located.
Levels: FA: Festuca arundinacea–dominated patch; PQ: Paspalum quadrifarium–dominated patch - treatment: Categorical. Biomass removal treatment applied to the subplot.
Levels: CONTROL: no biomass removal; DR: dead biomass removal; LR: living biomass removal; DLR: combined dead and living biomass removal
- litter_g: Numeric. Dry mass of litter (detached dead biomass on the soil surface), in grams per square meter (g m⁻²).
- standing_dead_g: Numeric. Dry mass of standing dead biomass, in grams per square meter (g m⁻²).
- living_removal_g: Numeric. Dry mass of living biomass removed by the treatment, in grams per square meter (g m⁻²).
- dead_removal_g: Numeric. Dry mass of dead biomass removed by the treatment (litter + standing dead), in grams per square meter (g m⁻²).
- total_removal_g: Numeric. Total dry mass removed by the treatment (living + dead biomass), in grams per square meter (g m⁻²).
File: Exp1_Vegetation.xlsx
Description:
This file contains vegetation data from repeated censuses of a factorial field experiment manipulating living and dead biomass removal in grassland patches dominated by Festuca arundinacea and Paspalum quadrifarium. The dataset includes complete information for all censuses (1–4), with species-level percent cover and accumulated species richness used to describe community dynamics over time.
Variables
-
year: Integer. Sampling year.
-
month: Categorical. Sampling month.
-
censo: Integer. Census number.
-
patch_state: Categorical. Dominant species of the patch where the subplot is located.
Levels: FA: Festuca arundinacea–dominated patch; PQ: Paspalum quadrifarium–dominated patch -
treatment: Categorical. Biomass removal treatment applied to the subplot.
Levels: CONTROL: no biomass removal; DR: dead biomass removal; LR: living biomass removal; DLR: combined dead and living biomass removal -
block: Integer. Experimental block identifier.
-
IDG: Categorical. Plot identifier combining patch state and block.
-
Species-level cover variables: from pasqua to ver_sp columns. Numeric. Percent cover (%) of each species per 1 m² subplot, recorded in each census.
Aggregated variables
- cover_other: Numeric. Total percent cover (%) of all non-dominant species, excluding Festuca arundinacea and Paspalum quadrifarium.
- cover_other_native: Numeric. Percent cover (%) of non-dominant native species.
- cover_other_nonnative: Numeric. Percent cover (%) of non-dominant non-native species.
- accumulated_richness_total: Integer. Species richness accumulated across censuses.
- accumulated_richness_native: Integer. Accumulated richness of native species.
- accumulated_richness_nonnative: Integer. Accumulated richness of non-native species.
File: Exp2_SeedBank.xlsx
Description: Seed bank availability was assessed by collecting soil and litter samples from each block in patches dominated by Festuca arundinacea and Paspalum quadrifarium. Samples were processed in a greenhouse germination experiment, where soil and litter from each patch state were spread in germination trays. All emerging seedlings were identified to species level and counted, allowing the characterization of species richness and seedling abundance available in each patch state and substrate.
Variables (Sp names)
- spcode: Character. Species code used as column name in Sheet1.
- sp: Character. Scientific name (species or genus level).
- origin: Categorical. Origin of the species in the study region. Levels: Native; Non native.
Variables (Sheet1)
- block: Integer. Experimental block identifier (1–5).
- patch_state: Categorical. Dominant species of the patch where the subplot is located.
Levels: FA: Festuca arundinacea–dominated patch; PQ: Paspalum quadrifarium–dominated patch - substrate: Categorical. Germination substrate. Levels: soil: mineral soil; litter: surface litter layer
- ID: Character. Unique identifier for each experimental unit, combining patch state and block
- fresh_weight_g: Numeric. Fresh biomass weight of collected material.
Units: grams (g). - richness_total: Integer. Total species richness.
- richness_other: Integer. Species richness excluding the dominant species (Festuca arundinacea or Paspalum quadrifarium).
- richness_other_natives: Integer. Richness of non-dominant native species.
- richness_other_nonnative: Integer. Richness of non-dominant non-native species.
- seedlings_total: Integer. Total number of seedlings.
- seedlings_other: Integer. Seedlings excluding the dominant species.
- seedlings_other_natives: Integer. Seedlings of non-dominant native species.
- seedlings_other_nonnatives: Integer. Seedlings of non-dominant non-native species.
- Species-level seed bank variables:: From fesaru to phyvis columns. Numeric. Number of emerged seedlings per species in each germination tray (counts). Each column represents one species recorded in the seed bank assay, for both soil and litter substrates.
File: Figure1_ExperimentalDesign.png
Description: Schematic of the experimental design of the biomass removal experiment (Experiment 1). Top: area closed to agriculture dominated by Festuca arundinacea (Fa, red) with patches dominated by Paspalum quadrifarium (Pq, blue); five main plots were established in each patch state (n = 10). Bottom: each main plot was divided into four 1 m² subplots to which biomass removal treatments were applied (C: control, DR: dead removal, LR: living removal, DLR: dead and living removal). Scissors indicate the type of tissue removed and grass leaf color indicates the tissue remaining in the field (green = living, brown = dead).
Code/software
All analyses were conducted in R (free, open-source statistical software; R Core Team).
Scripts are fully reproducible using base R plus openly available CRAN packages. Analyses were run in R ≥ 4.2.x (the workflow does not rely on version-specific features; any recent 4.x version is sufficient).
Packages used: dplyr, tidyr, glmmTMB, car, emmeans, DHARMa. All packages are free and available from CRAN.
Workflow overview
- Data preparation
- Filtering by experiment, census, patch state, and treatment.
- Aggregation when needed (e.g., mean cover across censuses 3–4).
- Conversion of grouping variables to factors to ensure correct model interpretation.
- Model fitting
- Species richness: GLMMs with Poisson error.
- Seedling abundance: GLMMs with negative binomial (nbinom2) error.
- Percent cover data: GLMMs with Tweedie error distribution.
- Random effects account for spatial or experimental structure (e.g.,
IDG,block,ID).
- Inference
- Fixed effects tested using Type III Wald χ² tests.
- Post-hoc comparisons performed with emmeans, conditional on interaction significance.
- Diagnostics
- Model assumptions and dispersion evaluated using DHARMa simulated residuals.
- Overdispersion explicitly checked for Poisson models.
- Outputs
- Model summaries, ANOVA tables, pairwise comparisons, diagnostic tests, and summary tables of mean cover and richness.
File: R_MODELS.R (also hosted in Zenodo)
Description: R script containing all statistical analyses reported in the associated article. The script loads the three data files (Exp1_Vegetation.xlsx, Exp1_Biomass.xlsx, Exp2_SeedBank.xlsx) using relative paths, so the working directory must be set to the folder containing them. It is organized in the following sections:
- EXP1 RICHNESS: accumulated species richness (total, native, non-native) at census 4; GLMMs with Poisson error, treatment × patch state as fixed effects and plot (IDG) as random effect.
- EXP1 DOMINANT COVER: cover of Festuca arundinacea (fesaru) and Paspalum quadrifarium (pasqua), averaged over censuses 3 and 4; GLMMs with Tweedie error.
- EXP1 COVER OF OTHER SPECIES: cover of non-dominant species (total, native, non-native) in F. arundinacea patches, averaged over censuses 3 and 4; GLMMs with Tweedie error. Also includes summary tables of mean cover per species and treatment.
- EXP2 RICHNESS: seed bank species richness (total, native, non-native non-dominant species); GLMMs with Poisson error, patch state × substrate as fixed effects and block as random effect.
- EXP2 SEEDLING DENSITY: number of emerged seedlings (total, F. arundinacea, P. quadrifarium, and non-dominant species); GLMMs with negative binomial error (nbinom2) and sample (ID) as random effect.
For each model, the script reports Type III Wald χ² tests (car::Anova), pairwise comparisons (emmeans) and residual diagnostics (DHARMa).
Experiment 1: Dead and living biomass removal
In December 2023, we established a factorial experiment manipulating dead and living biomass in patches dominated by Festuca arundinacea (Fa) and Paspalum quadrifarium (Pq) (Figure 1). Five main plots were established per patch state (five in Fa patches and five in Pq patches), and within each plot, four 1 m² subplots were randomly assigned to one of the following biomass removal treatments: control (C, no removal), dead removal (DR), living removal (LR), and dead and living removal (DLR). Dead biomass removal involved the removal of standing dead material and surface litter; living biomass removal consisted of cutting green leaves and stems while minimizing disturbance to dead material; combined dead and living biomass removal involved removing all aboveground biomass and surface litter. The biomass collected was separated by tissue type (living; dead = standing dead + litter), dried at 60 °C, and weighed to quantify the amount removed in each treatment (Table 1). The timing of the removals coincided with a period in which both species were in a vegetative state but had not yet reached their maximum growth, thereby avoiding advantages associated with phenological differences.
Figure 1 (file: Figure1_ExperimentalDesign.png). Biomass removal experimental design. Top: area closed to agriculture dominated by Festuca arundinacea (Fa, red) with patches dominated by Paspalum quadrifarium (Pq, blue). Five main plots (red and blue squares) were established in each patch state (n = 10). Bottom: main plots were divided into four 1 m² subplots to which biomass removal treatments were applied. C: control, DR: dead removal, LR: living removal, DLR: dead and living removal. Scissors indicate the type of tissue removed and grass leaf color indicates the tissue remaining in the field (green = living, brown = dead).
Vegetation surveys were conducted in April, October, and December 2024, and March 2025, allowing us to capture short-term (c.a., 15 months) recovery dynamics and the recruitment of species with different growing seasons. In each permanent plot, a 1 m² frame subdivided into 25 sub-quadrats (20 × 20 cm) was placed to visually estimate the relative cover of species present and dead tissues (standing dead and litter) in each sub-quadrat. Cover values from the 25 sub-quadrats were averaged to obtain a single value per plot for each variable. Species richness was calculated as the total number of other species (distinct from the dominant F. arundinacea and P. quadrifarium) accumulated per treatment across the four surveys. This approach allowed us to quantify the pool of species available at the site that could potentially establish, regardless of whether individuals were present in all surveys. For vegetation cover analyses, we averaged the values from the last two surveys (December and March) to capture the growing seasons of the dominant species. This approach allowed us to assess the recovery of the dominant vegetation (F. arundinacea and P. quadrifarium) and the establishment of other species one year after treatment application.
Experiment 2: Seed bank
To assess propagule availability in the seed bank, we conducted a seedling emergence assay (Roberts & Neilson 1981). Propagules can be retained both in surface litter and in the soil (Ruprecht & Szabo 2012). Accordingly, two substrate types were included to distinguish the seed-trapping effect of dead material deposited on the soil surface (litter) from the soil seed bank sensu stricto (upper 5 cm of soil).
In April 2024, samples of both substrates (litter and soil) were collected at the experimental site. Four 25 × 25 cm frames were placed within each of the five blocks per patch state of the living and dead biomass removal experiment, outside the subplots where treatments were applied and on sites completely covered by litter. Thus, each sample represented an equivalent area of litter and soil (4 × 0.0625 m² = 0.25 m²). Within each frame, surface litter was manually collected, taking care not to disturb the material to avoid seed loss. Once the surface was cleared, the upper 5 cm of soil were collected from the same frame area.
All samples were bagged by substrate type and patch state and weighed fresh. Each sample was then mixed with sterile commercial substrate and perlite and placed in trays in a greenhouse. Trays were watered regularly until November 2024. All emerging seedlings were identified, counted, and removed. The assay ended when seedling emergence ceased, under the assumption that all viable seeds had germinated. For each tray, abundance was recorded as species richness and number of emerged individuals.