Public | Automated Build

Last pushed: a day ago
Short Description
Image to reproduce Camac et al BCI mortality analysis
Full Description

Unifying intra- and inter-specific variation in tropical tree mortality

By: James S Camac, Richard Condit, Richard G FitzJohn, Lachlan McCalman,
Daniel Steinberg, Mark Westoby, Joe Wright, Daniel Falster

Maintainers: James Camac and Daniel Falster


Tree death is a fundamental process driving population dynamics, nutrient cycling, and evolution within plant communities. While past research has identified factors influencing tree mortality across a variety of scales, these distinct drivers are yet to be integrated within a unified predictive framework. In this study, we use a cross-validated Bayesian framework coupled with classic survival analysis techniques to derive instantaneous mortality functions for 203 tropical rainforest tree species at Barro Colorado Island (BCI) Panama. Specifically, we develop mortality functions that not only integrate individual, species, and temporal effects, but also partition the contributions of growth-dependent and growth-independent effects on the overall instantaneous mortality rate.

This repository contains the data and code required to reproduce our entire workflow from data cleaning, rerunning the analysis, producing figures and reproducing the manuscript. Details below instructions on how this work can be reproduced.


This paper uses data from Barro Colorado Island (BCI) Panama.
Condit, R., Lao, S., Pérez, R., Dolins, S.B., Foster, R.B. Hubbell, S.P. 2012. Barro Colorado Forest Census Plot Data, 2012 Version. DOI


A preprint of this project has been released on BioRxiv:

Reproducing analysis

We are committed to reproducible science. As such, this repository contains all the data and code necessary to fully reproduce our results. To facilitate the reproducibility of this work, we have created a docker image and set up the entire workflow using remake. Below we outline the two approaches that can be taken to reproduce the analyses, figures and manuscript.

Copy repository

First copy the repository to your a desired directory on you local computer.

This can either be done using the terminal (assuming git is installed)

git clone

Or can be downloaded manually by clicking here.

Download the model fits

Running all model cross validations took approximately 2-months of computing time on a HPC machines. As the model fitting proceedure is not included in the remake workflow. We have however provided the model fits as a release here. The release contains a file called This contains the chain fits from stan's MCMC sampler. The should be unpacked in the parent directory of the mortality_bci. The file contains cached files from using the package remake and should be moved (and unpacked) in the parent directory of mortality_bci. It is not essential to download the .remake folder. If you do, the code will reproduce the paper using cached calculations. If you don't the code will rerun all the preliminary calculations.

Reproducing analysis with remake & docker (Recommended approach)

Each computer is different. Operating systems, software install and the versions of such software are likely to vary substantially between computers. As such it is extremely difficult to develop code that can easily run on all computers. This is where Docker comes in. Docker is the world’s leading software container platform. Here we use Docker because it can readily be used across platforms and is set to install the appropriate software, and software versions used in the original analysis. As such it safeguards code from differences among computers and potential changes in software and cross platform issues.

Setting up Docker

If you haven't installed docker please see here.

We can set up docker two ways. The recommended approach is to download the precompiled docker image by running the following in the terminal/shell:

docker pull traitecoevo/mortality_bci

This image contains all required software (and software versions) to run this analysis.

If however, you would like the recompile the image from scratch the code below can be run. Note this will much slower relative to the docker pull approach.

docker build --rm --no-cache -t traitecoevo/mortality_bci .

The period is important as it tells docker to look for the dockerfile in the current directory

Using docker

Now we are all set to reproduce this project!

Rstudio from within docker

To be able to run the code, we interface with the Rstudio within the docker container by running the following in the terminal/shell:

docker run -v /Users/path/to/repository/:/home/rstudio -p 8787:8787 traitecoevo/mortality_bci

Now just open your web browser and go to the following: localhost:8787/

The username and password is rstudio

Rerunning analysis from within docker

Assuming the model fits have been downloaded (see above), one can now reproduce the outputs by running:


Reproducing analysis without docker (Not recommended)

This option is not recommended as R packages are constantly being updated and backwards compatibility broken. However, if you are adverse to using Docker you can run remake outside docker and willing by using the instructions below. Code was developed under R 3.4.1 (2017-06-30).

Installing remake

First install some dependencies from cran as follows:

install.packages(c("R6", "yaml", "digest", "crayon", "optparse"))

Now we'll install some packages from github. For this, you'll need the package devtools. If you don't have devtools installed you will see an error "there is no package called 'devtools'"; if that happens install devtools with install.packages("devtools").

Then install the following two packages


See the info in the remake readme for further details if needed.

Open a new R session with this project set as working directory. We use a number of packages, these can be easily installed by remake:


Run the following to generate all outputs (analysis, figures, table, manuscript):



If you have any problems getting the workflow to run please create an issue and I will endevour to remedy it ASAP.

Docker Pull Command
Source Repository