Shyamolina Ghosh, University of Kansas

Lawrence W. Sheppard, University of Kansas

Daniel C. Reuman, University of Kansas

This repository records the complete workflow that produced the paper. All analyses can be reproduced and the paper and supporting information recompiled (see below).

Knit makefile.Rmd using R markdown. If all dependencies are in place (see next section) this should re-compute all analyses of the paper, resulting in three pdfs: MainText.pdf (the main text of the paper), SuppMat.pdf (the supporting information file for the paper), and makefile.pdf (notes on the compilation process - can be useful for error mitigation in the event of failure).

The knit may several hours, depending on your computer speed and other factors. Subsequent knits, if any, can be faster because packages will be installed (see below) and because intermediate results are cached.

If you try to knit MainText.Rmd or SuppMat.Rmd directly, you may have some success, but cross-document references and other features will fail so this is not recommended.

To compile the documents from the command line, use the following: Rscript -e "library(knitr); knit('makefile.Rmd')".

• R
• R markdown
• R studio
• latex
• bibtex

Code uses the R checkpoint package. This is set up in the master file makefile.Rmd in the R chunk checkpoint_chunk, which contains the following line of code specifying a date :

checkpoint("2019-01-13",checkpointLocation = "./")

The checkpoint package then automatically scans through other files looking for other required R packages. It then downloads and installs the newest versions of those packages available on the given date. This helps ensure that re-compiling the document uses exactly the same code that was originally used, in spite of package updates and other changes. This can take some time on first run (you are warned) but it is faster on subsequent runs because the packages are already installed. This also means that R package dependencies should only be the checkpoint package, since that package should scan for other packages and install them locally. Quite a few MB disk space are used (approx. 289).

The open source program pandoc converts documents from one format to another. Here, the knitr package uses it to convert the markdown files into latex format so that they can then be turned into PDF files. Installers for multiple operating systems are available here: https://pandoc.org/installing.html.

The makefile makes a system call to pdflatex, so software supporting that needs to be installed:

• On Windows, you can use Miktex (https://miktex.org/howto/install-miktex),
• On Linux, install latex (e.g., sudo apt-get install texlive), and
• On Mac, use the MacTeX installer (http:https://www.tug.org/mactex/)

If you find additional dependencies were needed on your system, please let us know: [email protected]. The compilation process was tested on Ubuntu 16.04 using R version 3.4.4 and R studio version 1.1.463. We have endeavored to list all dependencies we can think of above, but we have only compiled on our own machines, so we cannot guarantee that additional dependencies will not also be needed on other machines. This repository is intended to record a workflow, and is not designed or tested for distribution and wide use on multiple platforms. It is not guaranteed to work on the first try without any hand-holding on arbitrary computing setups.

Knitting the makefile automatically produces a lot of 'intermediate' files. Files ending in .tex are the converted documents from .Rmd including all the R code output and the rest (files ending .log, .aux, .lof, .lot, .toc and .out ) are intermediate files that pdflatex uses to keep track of various parts of the document. Some of these can be useful for diagnosing problems, if any.

The authors were partly funded by US National Science Foundation grant 1714195 and the James S McDonnell Foundation, and the California Department of Fish and Wildlife Delta Science Program. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation or the McDonnell Foundation or the California Department of Fish and Wildlife Delta Science Program.