In sckott/rphylopic: Get Silhouettes of Organisms from PhyloPic

Authors: William Gearty & Lewis A. Jones

Last updated: r Sys.Date()

knitr::opts_chunk$set(echo = TRUE, include = TRUE, eval = TRUE,
                      out.width = "100%", fig.width = 7.2, fig.height = 7.2,
                      dpi = 200, fig.path = "")

# Introduction Herein we provide three example applications of the **rphylopic** package in combination with the `{ggplot2}` package. However, note that all demonstrated functionality is also available for base R and showcased in a [separate vignette](c-advanced-base.html). # Basic accession and transformation The **rphylopic** package provides robust and flexible tools to access and transform PhyloPic silhouettes. Here we demonstrate this using the example dataset of Antarctic penguins from the `{palmerpenguins}` R package. First, let's load our libraries and the penguin data: wzxhzdk:1 Now, let's pick a silhouette to use for the penguins. Let's pick #2: wzxhzdk:2 wzxhzdk:3 You may have noticed in the preview that the silhouette was a little slanted. Let's rotate it clockwise just a smidgen: wzxhzdk:4 Now, let's draft the plot that we want to make. In this case, let's plot the penguins' bill lengths vs. their flipper lengths: wzxhzdk:5 That's a nice basic plot! But you know what would make it nicer? If we added a penguin silhouette to the plot. Sadly, we don't have a different silhouette for each species (although we could make one...), so let's just go with putting a single silhouette in the top panel. We'll use the `geom_phylopic()` function, which will require us to make a `data.frame`. Note that the `x` and `y` aesthetics specify the center of the silhouette, and the `height` argument specifies how tall the silhouette is in the units of the y-axis. wzxhzdk:6 Isn't that nifty! We can go a step further, though. What if we used little penguins instead of points?! To do that, we can use the `geom_phylopic()` function instead of the `geom_point()` function (in this case, we want to use the same image for each x-y pair): wzxhzdk:7 The default silhouette size for `geom_phylopic()` is as large as will fit within the plot area. That won't work well here. Instead, we can specify a `height` or `width` to use (in y-axis or x-axis units, respectively): wzxhzdk:8 Alternatively, we can vary the height based on some other aspect of data, and the values will be automatically scaled for us. In this case, let's try making the size of the silhouettes relative to the penguins' body masses. Behind the scenes, __rphylopic__ will work its magic and rescale them to values between 1 and 6. Note that, depending on your own data, you may want to customize how the values are scaled by using `scale_height_continuous()` (or `scale_width_continuous()` if you are using the `width` aesthetic) just as you would use `scale_size_continuous()`. We'll just use the defaults here: wzxhzdk:9 Nice! Finally, let's give the female and male penguins different fill colors. Note that the default for `geom_phylopic()` is to not display a legend, so we need to set `show.legend = TRUE`. However, we only want a legend for the fill colors, so we use `guide = "none"` for the size scale. We also want to show the fill color in the legend, so we need to override the shape: wzxhzdk:10 Hmm...the colored dots in the legend are great, but lucky for us, the package also supplies a convenient way to include silhouettes in the legend. Due to technical constraints, you'll need to specify the images/uuids/names again within `phylopic_key_glyph()`. If you supply more than one silhouette to this function, it will cycle through them as it generates legend keys (recycling as needed). Note that `phylopic_key_glyph()` does not currently support the `height`/`width` aesthetics. wzxhzdk:11 Now that's a nice figure! # Geographic distribution In much the same way as generic x-y plotting, the **rphylopic** package can be used in combination with `{ggplot2}` to plot organism silhouettes on a map. That is, to plot data points (e.g., species occurrences) as silhouettes. We provide an example here of how this might be achieved. For this application, we use the example occurrence dataset of early (Carboniferous to Early Triassic) tetrapods from the `{palaeoverse}` R package to visualize the geographic distribution of *Diplocaulus* fossils. First, let's load our libraries and the tetrapod data: wzxhzdk:12 Then we'll subset our occurrences to only those for *Diplocaulus*: wzxhzdk:13 Now, let's plot those occurrences on a world map. `{ggplot2}` and it's built-in function `map_data()` make this a breeze. Note that we use `alpha = 0.75` in case there are multiple occurrences in the same place. That way, the darker the fill color, the more occurrences in that geographic location. wzxhzdk:14 Now, as with the penguin figure above, we can easily replace those points with silhouettes. wzxhzdk:15 Snazzy! Note that while we used the genus name as the `name` aesthetic here, we easily could have done `name = "Diplocaulus"` outside of the `aes()` call instead. However, if we were plotting occurrences of multiple genera, we'd definitely want to plot them as different silhouettes using `name = genus` within the `aes()` call. Also, note that we could change the projection of the map and data using the `crs` and `default_crs` arguments in `coord_sf()`. When projecting data, note that the y-axis limits will change to projected limits. For example, in the Robinson projection, the y-axis limits are roughly -8,600,000 and 8,600,000 in projected coordinates. Therefore, you may need to adjust the `height` argument/aesthetic accordingly when projecting maps and data. wzxhzdk:16 # Phylogenetics Another common use case of PhyloPic silhouettes is to represent taxonomic information. In this example, we demonstrate how to use silhouettes within a phylogenetic framework. In this case, the phylogeny, taken from the `{phytools}` package, includes taxa across all vertebrates. Even many taxonomic experts are unlikely to know the scientific names of these 11 disparate taxa, so we'll replace the names with PhyloPic silhouettes. First, let's load our libraries and data: wzxhzdk:17 We can use a vectorized version of the `get_uuid()` function to retrieve UUID values for all of the species at once. However, just in case we get an error, we wrap the `get_uuid()` call in a `tryCatch()` call. This way, we should get either a UUID or `NA` for each species: wzxhzdk:18 Oh no, we weren't able to find a silhouette for *Myotis lucifugus* (little brown bat)! Good thing we used `tryCatch()`! Given the coarse resolution of this phylogeny, we can just grab a silhouette for the subfamily (Vespertilioninae): wzxhzdk:19 I'm also not a huge fan of the boar picture. Let's choose an alternative with `pick_phylopic()`. wzxhzdk:20 wzxhzdk:21 Now that we've got our phylogeny and UUIDs, we could go ahead and create our figure. However, time for a quick aside. The time required for `geom_phylopic()` and the other **rphylopic** visualization functions scales with the number of *unique* names/UUIDs, not the number of plotted silhouettes. Therefore, if you are plotting a lot of *different* silhouettes, these functions can take quite a long time to poll PhyloPic for each unique name, download the silhouettes, and convert them to be added to the plot. If you plan to use the same silhouettes for multiple figures, we strongly suggest that you poll PhyloPic yourself using `get_phylopic()` or `pick_phylopic()`, save the silhouettes to your R environment, and then these use image objects in the visualization functions (with the `img` argument/aesthetic). Following this advice, let's get image objects for these 11 species before we make our figure. Note that, since we've used `get_uuid()` to get these 11 UUIDs, we know that they are valid, so we don't need to catch any errors this time. wzxhzdk:22 Now let's go ahead and plot our phylogeny with the [ggtree](https://www.bioconductor.org/packages/ggtree) package: wzxhzdk:23 Hmm...that's a bit boring. Let's add a geological timescale to the background using `coord_geo_polar()` from the `{deeptime}` package. Note that we need to use the `revts()` function to reverse the time axis to work with `coord_geo_polar()`. wzxhzdk:24 That's looking a lot prettier! Let's go ahead and add our silhouettes now. Note that we need to attach the `vertebrate_data` object with the `%<+%` operator from `{ggtree}`. wzxhzdk:25 Note that only a single height is specified and aspect ratio is always maintained, hence why the silhouettes all have the same height but different widths. Let's fix some of the silhouettes by rotating them 90 degrees: wzxhzdk:26 And now the finished product: wzxhzdk:27 Beautiful!

{css echo = FALSE} p.caption { display: none; }

sckott/rphylopic documentation built on Oct. 13, 2024, 12:27 p.m.