In toddellis/outbreakR: Dendrochronological outbreak reconstructions in R

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

outbreakR

outbreakR provides simple tools to extract corrected indices and insect outbreak reconstruction data from dendrochronological time series data.

Installation

Install the development version from GitHub with:

# install.packages("devtools")
devtools::install_github("toddellis/outbreakR")

Using outbreakR

outbreakR comes with two demo datasets:

ob_host: A host dataset comprised of tree-level ring-width indices from Douglas-fir trees sampled across four study sites in 2014.
ob_nonhost: A nonhost master chronology -- the first principal component -- built from ponderosa pine trees sampled across study sites from the same region.

library(outbreakR)
head(ob_host)
head(ob_nonhost)

Corrected indices

To extract insect outbreak data, outbreakR uses two simple functions. The first produces corrected indices by removing climatic noise from the host tree-level data, leaving an assumed biological signal.

The default settings assume wide-format data, as expected from the output of dplR's RWI functions.

Just be aware of the additional settings for cor_index().

foo <- cor_index(ob_host, ob_nonhost)

head(foo)

As an alternative, it also takes long-format data:

ob_host %>%
  tidyr::gather('id', 'rwi', -year) %>%
  cor_index(., 
            ob_nonhost,
            format = 'long') %>%
  head()

Many dendrochronological datasets come in wide format rather than long, and treat years as the row names. This can also be handled here, but it's really recommended that you avoid deprecated data formats like this:

## Reformat both host and nonhost to mimic older formatting
host_old <- ob_host %>%
  as.data.frame()
row.names(host_old) <- host_old$year
host_old <- host_old %>%
  dplyr::select(-year)

nonhost_old <- ob_nonhost %>%
  as.data.frame()
row.names(nonhost_old) <- nonhost_old$year
nonhost_old <- nonhost_old %>%
  dplyr::select(-year)

head(host_old)

## Specify years_as_rownames = TRUE
## Note: This does not work with long-format data
cor_index(host_old, nonhost_old, years_as_rownames = T) %>%
  head()

foo %>%
  ggplot(aes(x = year, y = ci)) +
  theme_classic() +
  geom_point(alpha = 0.05) +
  geom_smooth(method = 'gam', 
              formula = y ~ s(x, k = 12),
              color = 'darkred',
              fill = 'hotpink')

Outbreak reconstructions

The outbreak function extracts either binary outbreak/non-outbreak time series data at the tree level, or produces proportional outbreaks showing the percentage of trees with an outbreak in a given year.

outbreak(foo) %>%
  tail(10)

outbreak(foo, prop = FALSE) %>%
  tail(10)

outbreak(foo, prop = FALSE) %>%
  mutate(site = substr(tree_id, 1, 3)) %>%
  group_by(site, year) %>%
  summarise(infected_trees = sum(outbreak)) %>%
  ggplot(aes(x = year, y = infected_trees, color = site)) +
  theme_classic() + 
  geom_point(alpha = 0.1) + 
  geom_line(size = 1.2, alpha = 0.4)

We can extract the sites here and run it individually per site. I may try to implement this in the future, but for now, this would provide a workaround:

## add site variable
foo <- foo %>%
  dplyr::mutate(site = substr(tree_id, 1, 3)) 
## pull out unique sites
sites <- foo %>%
  .$site %>%
  unique()

## create empty list to fill
site_outbreaks <- list()

## loop through site list
for (i in 1:length(sites)) {
  site_outbreaks[[i]] <- foo %>%
    dplyr::filter(site == sites[i]) %>%
    outbreak() %>%
    dplyr::mutate(site = sites[i])
}

site_outbreaks <- dplyr::bind_rows(site_outbreaks)

site_outbreaks %>%
  ggplot(aes(x = year, y = outbreakProp, color = site)) +
  geom_line() +
  theme_classic()