Usage with the paleoTS package

library(StratPal)

This vignette explain how to link the StratPal package with the paleoTS package (Hunt 2006). For details on the underlying data structures, see vignette("StratPal_docs").

Quick summary

To combine the StratPal and the paleoTS packages,

  1. simulate trait evolution using the functions with suffix _sl (specimen level), optionally specifying the additional parameters for intrapopulation variance and number of specimens
  2. build your pipelines as before (e.g., as described in vignette("phenotypic_evolution"))
  3. turn the result into paleoTS format using reduce_to_paleoTS
  4. further analyze or plot the results using the paleoTS package

Motivation

The paleoTS package allows to easily analyze paleontological time series, and StratPal can be connected to it to study how ecological, taphonomic, and stratigraphic effects change our inferences about the tempo and mode of evolution.

The paleoTS package defines its own paleoTS format to store time series. It is a summary format, as it combines trait values measured in multiple specimens into an intrapopulation variance. Because taphonomic and ecological effects act on individual specimens, they can not be modeled directly on the paleoTS format. To circumvent this problem, we introduce a pre_paleoTS format that models trait evolution on the specimen level. Stratigraphic, taphonomic, and ecological effects can act on this format. After all these effects have been applied, we can reduce this data format into the standard paleoTS format, and then analyze it with the standard toolbox of the paloeTS package.

Modeling trait evolution on specimen level, complexity reduction, and plotting

Functions to model trait evolution on specimen level have the suffix _sl, standing for “specimen level”. Internally, they are of S3 class pre_paleoTS. In addition to the options provided to simulate mean trait values, you can simulate strict stasis using strict_stasis_sl. All functions have the same parametrization as the as the other functions for simulating trait evolution, which simulate mean trait evolution. In addition, they take two additional parameters: intrapop_var for the variance of the population around the mean trait value, and n_per_sample for the number of specimens per sampling location.

pre_paleoTS results can be converted to the paleoTS format using the function reduce_to_paleoTS. After this you can plot them with the standard plotting procedure from paleoTS using plot (resp, plot.paleoTS). Note that you can not plot pre_paloeTS objects directly, only after conversion to paleoTS.

library(StratPal)
library(paleoTS)   # needed for plotting

strict_stasis_sl(t = 1:4) |>   # simulate strict stasis on specimen level in `pre_paleoTS` format
  reduce_to_paleoTS() |>       # convert pre_paleoTS to paleoTS
  plot()                       # plot

plot of strict stasis

Modeling of ecological, taphonomic, and stratigraphic effects

Modeling of niches, taphonomic effects, and stratigraphic biases works on pre_paleoTS objects identical to how it is described in the vignettes on event data and modeling phenotypic evolution. As example, we plot a random walk 2 km offshore with 5 specimens per sampling location:

library(admtools)                                  # load admtools for stratigraphic transformation
adm = tp_to_adm(t = scenarioA$t_myr,               # define age-depth model
                h = scenarioA$h_m[,"2km"],
                L_unit = "m",
                T_unit = "Myr")    
set.seed(42)                                      # set seed for reproducibility
seq(min_time(adm), max_time(adm), by = 0.01) |>   # sample every 0.01 Myr
  random_walk_sl(n_per_sample = 10) |>            # simulate random walk on specimen level
  time_to_strat(adm) |>                           # transform into stratigraphic domain
  reduce_to_paleoTS() |>                          # transform into paleoTS format
  plot()                                          # plot

plot of a random walk after stratigraphic transformation

Of course you can also immediately add functions from paleoTS to the pipeline, e.g. to fit models of phenotypic evolution from the simulated data:

set.seed(42)                                      # set seed for reproducibility
seq(min_time(adm), max_time(adm), by = 0.01) |>   # sample every 0.01 Myr
  random_walk_sl(n_per_sample = 10) |>            # simulate random walk on specimen level
  time_to_strat(adm) |>                           # transform into stratigraphic domain
  reduce_to_paleoTS() |>                          # transform into paleoTS format
  fit3models()                                    # fit 3 models to time series
#> Warning in
#> fit3models(reduce_to_paleoTS(time_to_strat(random_walk_sl(seq(min_time(adm), :
#> Sample variances not equal (P = 0 ); consider using argument pool=FALSE
#> 
#> Comparing 3 models [n = 64, method = Joint]
#> 
#>             logL K      AICc      dAICc Akaike.wt
#> GRW    -29.62174 3  65.64349  0.9846877     0.379
#> URW    -30.23104 2  64.65880  0.0000000     0.621
#> Stasis -53.45938 2 111.11548 46.4566839     0.000

References