Access these datasets on Dryad https://doi.org/10.5061/dryad.2547d7wzv

Empirically derived stage-based population models were collected from the COMPADRE Plant Matrix Database v6.22.5.0 (created 2022-05-11; Salguero-Gomez et al. 2015) that (1) were ergodic and irreducible, (2) were modelled on an annual time step, and (3) did not explicitly parse clonal growth into a separate matrix. This subset resulted in 1,606 matrices representing multiple years and/or populations of 317 plant species.

Life history traits were estimated from the matrix population models using the R package Rage (Jones et al. 2022).

Plant matrix population models were used to simulate asymptotic growth, demographic and environmental stochasticity and test the impact of initial population size, population structure, stochasticity, and life history on the odds of extinction. The impacts on the odds of extinction were tested within 10 years to represent the population dynamics within a testable period to the odds of extinction within 100 years to represent expected long-term population persistence.

Datasets included:

R code:

“Exist_” refers to populations projected from stable stage distribution while “Novel_” refers to populations projected from either the seed or seedling stage. “_Asymptotic”, “_demostoch”, “_environ”, and “_environ-demo” refer to the treatment of each simulation.

The asymptotic and demographic simulations use a single population matrix for each 100-year simulation. Simulations were initiated with one of five abundances (1, 10, 50, 100, 1000). For existing populations, the initial population size was distributed among stages proportional to the stable stage distribution (or the stage with the largest abundance when the number of stages was greater than the initial abundance). For novel populations, the initial population size was assigned to either the smallest non-dormant stage (to represent seedlings) or to the seed stage with the largest abundance.

Species with three or more matrix population models which have equal number of stages and life history category were grouped for the environmental stochasticity and environmental and demographic stochasticity treatments.

The resulting datasets each contain:

·       Time2Ext: the simulated year when the population dropped to zero indiviudals. NA if the population persisted beyond 100 simulated years.

·       start: the treatment including “SSD_Asymptotic” for simulations started at stable stage distribution with asymptotic conditions, “SSD_demographic” for demographic stochasticity, “SSD_environ” for environmental stochasticity, and “SSD_environdemo” for environmental and demographic stochasticity.

·       StPopSz: the initial population size for the simulation

·       LH: for the categorical life history of fast or slow pace and iteroparous or semelparous mode of reproduction. ‘Fast’ pace have an age of reproductive maturity less than three years, ‘slow’ have an age of reproductive maturity of three or more years. ‘Iteroparous’ populations have three or more years of mature life expectancy while ‘semelparous’ have fewer than three years of mature life expectancy. The four categories are FI, FS, SI, and SS for fast iteroparous, fast semelparous, slow iteroparous, and slow semelparous.

·       DegreeItero: Demetrius’ entropy calculated from vectors of age-specific survivorship and fecundity (for a single matrix in asymptotic and demographic stochasticity simulations or averaged over matrices for environmental stochasticity and environmental and demographic stochasticity treatments).

·       ShapeItero: the ‘shape’ value from the R package Rage shape_rep function. Either the value for a single matrix or averaged over matrices.

·       AgeMature: the age of reproductive maturity measured using https://compadre-db.org/Education/article/other-key-life-history-traits-from-mpms

·       MatrixID: the matrix population model(s) used in the simulation.

"2023_PhylogenyCovarianceMatrix.Rmd" explores the phylogentic relationships of the species within the dataset.

"NovelExist_ParityShape" has the JAGS code for the log-odds of extinction.

"_treatments.Rmd" are code to run the 100 year projections for both population histories (existing, novel) and treatments (control = asymptotic, demographic stochasticity, environmental stochasticity, demographic and environmental stochasticity). These pull matrices from "plants_data.Rdata" which is available on the Git repository.

Sharing/Access information

Datasets, code for dataset construction, and data analyses can be found on our GitHub respository:

• https://github.com/DenverBotanicGardens/TraitsExtinctionRisk

Data includes:

"COMPADRE_matrixdata.Rmd" pulls and subsets plant matrix population models from https://compadre-db.org/.

Matrix population models are saved as an .Rdata file, "plants.data". Tips for data structure and manipulation can be found on the COMPADRE database website (https://compadre-db.org/)

“plants_data” contains plant matrix population models and associated information gathered from the COMPADRE Plant Matrix Database v6.22.5.0.

Data was derived from the following sources:

• COMPADRE Plant Matrix Database v6.22.5.0

Code/Software

Scripts are written in R4.2.2 and require JAGS (https://sourceforge.net/projects/mcmc-jags/).