Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Dev #5

Merged
merged 3 commits into from
Jul 19, 2024
Merged

Dev #5

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
738aafb
Create advanced_functionality.Rmd
NiklasHohmann Jul 19, 2024
70dfd34
expand vignettes
NiklasHohmann Jul 19, 2024
4e2b443
improve formatting
NiklasHohmann Jul 19, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 12 additions & 3 deletions vignettes/StratPal.Rmd
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,19 +23,21 @@ library(admtools)

Welcome to the `StratPal` package. This vignette provides an overview of the structure of the package and preliminaries needed to efficiently use it. We go through _installation_, _dependencies_, provide an overview of the available _example data_, _piping_, and working with _age-depth models_.

Throughout the vignettes there are several **tasks** that you can solve. They are not mandatory for understanding the functionality of the package. Their aim is to explore the available models and data and develop an intuition for stratigraphic paleobiology.

If you want to skip the introduction, go to

```{r, eval=FALSE}
vignette("phenotypic_evolution")
```

for details on how to model stratigraphic paleobiology of phenotypic evolution. Go to
for details on how to model stratigraphic paleobiology of phenotypic evolution, or explore the vignette online on the package webpage here: [mindthegap-erc.github.io/StratPal/articles/phenotypic_evolution](https://mindthegap-erc.github.io/StratPal/articles/phenotypic_evolution.html). Go to

```{r, eval=FALSE}
vignette("event_data")
```

for details on how to model stratigraphic paleobiology of event data such as individual fossils and first/last occurrences of taxa.
for details on how to model stratigraphic paleobiology of event data such as individual fossils and first/last occurrences of taxa, or explore the vignette online on the package website here: [mindthegap-erc.github.io/StratPal/articles/event_data](https://mindthegap-erc.github.io/StratPal/articles/event_data.html).

## Installation

Expand Down Expand Up @@ -77,6 +79,8 @@ or by visiting the package website at https://mindthegap-erc.github.io/admtools/

## Example data

`StratPal` comes with some example data. This data is taken from Hohmann et al. (2024), who simulated a carbonate platform over 2 Myr. More details can be found via `?scenarionA`.

## Piping

In the `StratPal` package and its vignettes, we make heavy use of the base R _pipe_ operator `|>`. While this is not required to run the package, it simplifies the code and makes the underlying logic of a modeling _pipeline_ clearer.
Expand Down Expand Up @@ -111,7 +115,8 @@ You can also use `|>` to pass arguments that are not in the first place. For thi

```{r}
# calculate deciles of normal distribution
seq(0, 1, by = 0.1) |> quantile(x = runif(100), p = _) # pass left hand side to the p argument
seq(0, 1, by = 0.1) |>
quantile(x = runif(100), p = _) # pass left hand side to the p argument
```

## Age-depth models
Expand Down Expand Up @@ -205,3 +210,7 @@ vignette("phenotypic_evolution")
```

for details on how to model stratigraphic paleobiology of trait evolution.

## References

* Niklas Hohmann, Joël R Koelewijn, Peter Burgess, Emilia Jarochowska. Identification of the mode of evolution in incomplete carbonate successions.bioRxiv 2023.12.18.572098; doi: [10.1101/2023.12.18.572098](https://doi.org/10.1101/2023.12.18.572098), Data published under a [CC-BY 4.0](https://creativecommons.org/licenses/by/4.0/) license.
21 changes: 21 additions & 0 deletions vignettes/advanced_functionality.Rmd
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,21 @@
---
title: "Advanced Functionality and Writing Extensions"
output: rmarkdown::html_vignette
vignette: >
%\VignetteIndexEntry{Advanced Functionality and Writing Extensions}
%\VignetteEngine{knitr::rmarkdown}
%\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>"
)
```

```{r setup}
library(StratPal)
```

TBA
138 changes: 100 additions & 38 deletions vignettes/phenotypic_evolution.Rmd
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,77 +14,117 @@ knitr::opts_chunk$set(
)
```

## Introduction

Here we describe the modeling pipeline for the stratigraphic paleobiology of phenotypic evolution. First, let's load all required packages:

```{r setup}
library(StratPal)
library(admtools)
```

## Introduction

Here we describe the modeling pipeline for the stratigraphic paleobiology of trait evolution.
## Modes of evolution in the time domain

## Time domain
### Implemented modes of evolution

The `StratPal` package provides three functions for simulating trait evolution in the time domain:
The `StratPal` package provides three functions for simulating different modes of evolution in the time domain:

- `stasis` simulates evolutionary stasis as independent, normally distributed random variables with mean `mean` and standard deviation `sd`.

- `random_walk` simulates trait evolution following a (potentially biased) random walk with variability `sigma`, directionality `mu`, and initial trait value `y0`. Setting `mu` to a negative (positive) value will make the random walk increase (decrease) over time, `sigma` determines the effect of randomness on the trait values.
- `random_walk` simulates trait evolution following a (potentially biased) random walk with variability `sigma`, directionality `mu`, and initial trait value `y0`. Setting `mu` to a negative (positive) value will make the random walk decrease (increase) over time, `sigma` determines the effect of randomness on the trait values.

- `ornstein_uhlenbeck` corresponds to convergence to a phenotypic optimum, where `mu` is the optimal value (long term mean), `theta` determines how fast `mu` is approached, `sigma` the noise level, and `y0` the initial trait value.

- `ornstein_uhlenbeck` corresponds to convergence to a phenotypic optimum, where `mu` is the optimal value (long term mean), `theta` determines the strength of selection, `sigma` the influence of chance, and `y0` the initial trait value.
### Visualization

You can visualize them using the following pipeline:
You can visualize the different modes of evolution using the following pipeline:

```{r}
seq(0, 1, by = 0.01) |> # times of simulation in myr. simulate over 1 Myr years with 10 kyr resolution
random_walk(sigma = 1, mu = 3) |> # simulate random walk with incresing trait values
plot(type = "l", # plot results
seq(0, 1, by = 0.01) |> # times of simulation in myr. simulate over 1 Myr years with 10 kyr resolution
random_walk(sigma = 1, mu = 3) |> # simulate random walk with increasing trait values
plot(type = "l", # plot results
xlab = "Time [Myr]",
ylab = "Trait value")
```

**Task:** Modify the pipeline to get an intuition for the different modes of evolution. What are the effects of their parameters, and what is their biological meaning?

## Stratigraphic domain

We are interested in the stratigraphic expression of phenotypic evolution, i.e. how is trait evolution within a lineage preserved at a specific location in the stratigraphic record.
We are interested in how phenotypic evolution is preserved in the stratigraphic record, i.e. how is trait evolution within a lineage preserved at a specific location in the stratigraphic record. Here we develop the modeling pipelines to answer this question.

For this, we first define an age-depth model:
### Age-depth models

```{r}
# define ADM 4 km from shore
dist = scenarioA$dist_from_shore[2]
adm = tp_to_adm(t = scenarioA$t_myr,
h = scenarioA$h_m[,dist],
As an example, we compare the the preservation of trait evolution 2 km and 12 km offshore in the carbonate platform in scenarion A (see `?scenarioA`). For this, first define the age-depth models

```{r, figures-side, fig.show="hold", out.width="50%", fig.align='left'}
# define ADM 2 km from shore
adm_2km = tp_to_adm(t = scenarioA$t_myr,
h = scenarioA$h_m[,"2km"],
T_unit = "Myr",
L_unit = "m")

adm_12km = tp_to_adm(t = scenarioA$t_myr,
h = scenarioA$h_m[,"12km"],
T_unit = "Myr",
L_unit = "m")

# plot age-depth model
plot(adm,
plot(adm_2km,
lwd_acc = 2, # plot thicker lines for accumulative intervals (lwd = line width)
lty_destr = 0) # don't plot destructive intervals (lty = line type)
T_axis_lab() # add time axis label
L_axis_lab() # add length axis label
plot(adm_12km,
lwd_acc = 2, # plot thicker lines for accumulative intervals (lwd = line width)
lty_destr = 0) # don't plot destructive intervals (lty = line type)
T_axis_lab() # add time axis label
L_axis_lab() # add length axis label
```

Now we can use the age-depth model to transform the simulated trait values from the time domain into the depth domain using `time_to_strat`:
### Stratigraphic expression of phenotypic evolution

We use the defined age-depth models to transform the simulated trait values from the time domain into the depth domain via `time_to_strat`:

```{r}
# sample every 10 kyr over the interval covered by the adm
# simulate random walk
# transform data from time to strat domain
seq(from = min_time(adm), to = max_time(adm), by = 0.01) |>
random_walk(sigma = 1, mu = 3) |> # simulate random walk
time_to_strat(adm, destructive = FALSE) |>
plot(type = "l",

seq(from = min_time(adm_2km), to = max_time(adm_2km), by = 0.01) |> # sample every 10 kyr over the interval covered by the adm
random_walk(sigma = 1, mu = 3) |> # simulate random walk
time_to_strat(adm_2km, destructive = FALSE) |> # transform data from time to strat domain
plot(type = "l", # plot
orientation = "lr",
xlab = "Stratigraphic height [m]",
ylab = "Trait value")

```

Here we can already see the large jumps in trait values over the gaps, because we are missing a lot of time.
This is what a biased random walk in the time domain would look like if it was observed 2 km offshore in the simulated carbonate platform. The large jumps in traits correspond to long hiatuses caused by prolonged drops in relative sea level. Compare this figure to the random walk in the time domain (without stratigraphic distortions.)

This is already a good start, but there are some small issues: Because we simulate the lineage every 10 kyr, the lineage is sampled irregular in the stratigraphic domain. For example, we get multiple conflicting trait values at height with gaps longer than 10 kyr. To get a more realistic result, we need to prescribe a sampling strategy in the stratigraphic domain. We do this in the following steps:
12 km offshore, the preservation is very different:

1. Determine sampling locations
```{r}
seq(from = min_time(adm_12km), to = max_time(adm_12km), by = 0.01) |> # sample every 10 kyr over the interval covered by the adm
random_walk(sigma = 1, mu = 3) |> # simulate random walk
time_to_strat(adm_12km, destructive = FALSE) |> # transform data from time to strat domain
plot(type = "l", # plot results
orientation = "lr",
xlab = "Stratigraphic height [m]",
ylab = "Trait value")
```

You can also see two jumps, but the first is at a different location compared to the section 2 km offshore. Looking at the age-depth models, we can see why this is: 2 km offshore, the jumps are caused by prolonged hiatuses that are associated with the massive drops in sea level around 0.5 Myr and 1.5 Myr. 12 km offshore, the age-depth model shows fewer gaps, but prolonged intervals with low sedimentation rates at the beginning of the simulation and at around 1.5 Myr. This leads to stratigraphically condensed intervals, at which phenotypic evolution appears to be accelerated. You can see that not all artefactual jumps in traits observable in the stratigraphic record are caused by gaps.

**Task:** How does this effect vary between different modes of evolution (with different parameter choices), and at different locations in the platform (e.g., along an onshore-offshore gradient)? Can you make any general statements about where you see the strongest effects?

### Prescribing a sampling strategy

The above plots already give us a good idea of how stratigraphic effects can change our interpretation of trait evolution.

However, there is a small imperfection. Because we simulate the lineage every 10 kyr, it is sampled irregularly in the stratigraphic domain: If sedimentation rates are high, samples can be multiple meters apart, but if sedimentation rates are low, they are only a few centimeters apart.

To fix this, we want to prescribe where in the stratigraphic domain samples are taken. This allows us to examine the effect that different sampling strategies have on our perception of trait evolution. We do this in the following steps:

1. Determine sampling locations in the stratigraphic domain
2. Calculate what times correspond to said sampling locations via `strat_to_time`
3. Simulate a lineage at said times
4. Transform the simulated trait values back into the stratigraphic domain using `time_to_strat`
Expand All @@ -95,17 +135,39 @@ Let's assume we take a sample every 2 meters. Then our five steps above result i
```{r}
dist_between_samples_m = 2
sampling_loc_m = seq(from = 0.5 * dist_between_samples_m,
to = max_height(adm),
to = max_height(adm_2km),
by = dist_between_samples_m)

sampling_loc_m |> # sampling locations
strat_to_time(adm) |> # determine times where lineage is sampled
random_walk(sigma = 1, mu = 3) |> # simulate trait values at these times
time_to_strat(adm, destructive = FALSE) |> # transform trait values to stratigraphic domain
plot(orientation = "lr",
type = "l", # plot fossil time series
sampling_loc_m |> # sampling locations
strat_to_time(adm_2km) |> # determine times where lineage is sampled
random_walk(sigma = 1, mu = 3) |> # simulate trait values at these times
time_to_strat(adm_2km, destructive = FALSE) |> # transform trait values to stratigraphic domain
plot(orientation = "lr", # plot stratigrahic data
type = "l",
ylab = "Trait Value",
xlab = "Stratigraphic Height [m]")
```

There we have it! This is what the lineage would look like if it would be sampled every 2 meters in a sections 2 km offshore. You nicely see the large jumps over the prolonged hiatuses.
There we have it! This is what the lineage would look like if it would be sampled every 2 meters in a sections 2 km offshore.

**Task:** How does prescribing a sampling strategy change the interpretations of your last task? Can you draw any conclusions about which sampling strategy is best suited for a given environment or mode of evolution?

## Further reading

Go to

```{r, eval=FALSE}
vignette("phenotypic_evolution")
```

for details on how to model stratigraphic paleobiology of phenotypic evolution, or explore the vignette online on the package webpage ([mindthegap-erc.github.io/StratPal/articles/phenotypic_evolution](https://mindthegap-erc.github.io/StratPal/articles/phenotypic_evolution.html)).

See also

```{r, eval=FALSE}
vignette("advanced_functionality")
```

for details on how to expand on the modeling pipelines described here, or explore the vignette online on the package webpage ([mindthegap-erc.github.io/StratPal/articles/advanced_functionality](https://mindthegap-erc.github.io/StratPal/articles/advanced_functionality.html)).


Loading