Matthew Strimas-Mackey

One of the first things I took on when I started at the Cornell Lab of
Ornithology was creating the [auk R
package](https://cornelllabofornithology.github.io/auk/) for accessing
eBird data. The entire eBird dataset can be downloaded as a massive text
file, called the eBird Basic Dataset (EBD), and auk pulls out manageable
chunks of the dataset based on various spatial, temporal, or taxonomic
filters. I'm often asked "how do I extract data from within a polygon?"
(usually "polygon" is replaced by "shapefile", but I try to avoid that
word since there's [good reasons to stop using
shapefiles](http://switchfromshapefile.org/)). Rather than answer these
questions individually, I thought I'd do a quick post about how to do
this with auk. Note that, at the time of posting, this requires auk
version 0.4.1 or higher, which can be installed with:

::: {.highlight}
``` {.chroma tabindex="0"}
install.packages("auk")
```
:::

For more details on auk and eBird data in general, including how to get
access to the EBD, it's worth reading the first two chapters of the
[eBird Best Practices
book](https://cornelllabofornithology.github.io/ebird-best-practices/).
For the sake of speed and smaller file size, I'll be working on a subset
of the EBD containing all Northern Bobwhite records from 2019, which I
obtained using the [EBD custom download
form](https://cornelllabofornithology.github.io/ebird-best-practices/ebird.html#ebird-size-custom),
and you can [access
here](https://github.com/mstrimas/strimasdotcom/raw/master/content/post/2020-04-02-extracting-ebird-data-polygon/ebd_norbob_201901_201912_relFeb-2020.zip).
However, everything I'll show in this post works equally as well (just a
lot slower!) on the full EBD. For this example, let's say we want to
extract all records from within a polygon defining [Bird Conservation
Region](https://nabci-us.org/resources/bird-conservation-regions/) 27
([Southeastern Coastal
Plains](https://nabci-us.org/resources/bird-conservation-regions-map/#bcr27)).
A GeoPackage of this region is available on the GitHub repository for
the eBird Best Practices book, [download
it](https://github.com/CornellLabofOrnithology/ebird-best-practices/raw/master/data/gis-data.gpkg),
place it in the `data/` subdirectory of your RStudio project, then load
it into R with:

::: {.highlight}
``` {.chroma tabindex="0"}
library(sf)library(auk)library(dplyr)poly <- read_sf("data/gis-data.gpkg", layer = "bcr")
```
:::

If you have a shapefile, replace `"data/gis-data.gpkg"` with the path to
your shapefile and omit `layer = "bcr"`. Now that we have a polygon,
extracting eBird data is a two step process:

1.  Extract data from the EBD that's within a bounding box containing
    the polygons using the function `auk_bbox()`. This is necessary
    because due to the way auk works under the hood, it can only filter
    to ranges of latitudes and longitudes.
2.  Import the resulting data into R and further subset it to just the
    observations that fall within the polygon.

Fortunately, step 1 is made easier by `auk_bbox()` accepting spatial
`sf` or `raster` objects and automatically calculating the bounding box
for you. For example,

::: {.highlight}
``` {.chroma tabindex="0"}
auk_ebd("data/ebd_norbob_201901_201912_relFeb-2020.txt") %>%   auk_bbox(poly)
```
:::

``` {tabindex="0"}
Input   EBD: /home/maelle/Documents/ropensci/roweb2/content/technotes/2020-04-16-extracting-ebird-data-from-a-polygon/data/ebd_norbob_201901_201912_relFeb-2020.txt Output   Filters not executedFilters   Species: all  Countries: all  States: all  BCRs: all  Bounding box: Lon -91.6 - -75.5; Lat 29.3 - 37.3  Date: all  Start time: all  Last edited date: all  Protocol: all  Project code: all  Duration: all  Distance travelled: all  Records with breeding codes only: no  Complete checklists only: no
```

Notice that the output of the above command says
`Bounding box: Lon -91.6 - -75.5; Lat 29.3 - 37.3`, which are the bounds
of the smallest square that contains the polygon. Let's follow the
method [outlined in the Best Practices
book](https://cornelllabofornithology.github.io/ebird-best-practices/ebird.html#ebird-extract)
to extract some data! We'll get all observations on complete checklists
from May to August inside the bounding box of the polygon:

::: {.highlight}
``` {.chroma tabindex="0"}
f_out <- "data/ebd_norbob_poly.txt"auk_ebd("data/ebd_norbob_201901_201912_relFeb-2020.txt") %>%   # define filters  auk_bbox(poly) %>%   auk_date(c("*-05-01", "*-08-31")) %>%   auk_complete() %>%   # compile and run filters  auk_filter(f_out)
```
:::

The results were output to a file, which you can read in with
`read_ebd()`.

::: {.highlight}
``` {.chroma tabindex="0"}
ebd <- read_ebd("data/ebd_norbob_poly.txt")
```
:::

The data are now in a data frame and it's time to proceed to step 2:
further subset the data to only keep points within the polygon. First
we'll convert this data frame to a spatial `sf` object using the
`latitude` and `longitude` columns, then well use `st_within()` to
identify the points within the polygon, and use this to subset the data
frame. Note that we have to be careful with our coordinate reference
system here: `crs = 4326` specifies that the EBD data are in
unprojected, lat-long coordinates and we use `st_transform()` to ensure
the polygons and points are in the coordinate reference system.

::: {.highlight}
``` {.chroma tabindex="0"}
# convert to sf objectebd_sf <- ebd %>%   select(longitude, latitude) %>%   st_as_sf( coords = c("longitude", "latitude"), crs = 4326)# put polygons in same crspoly_ll <- st_transform(poly, crs = st_crs(ebd_sf))# identify points in polygonin_poly <- st_within(ebd_sf, poly_ll, sparse = FALSE)
```
:::

``` {tabindex="0"}
although coordinates are longitude/latitude, st_within assumes that they are planar
```

::: {.highlight}
``` {.chroma tabindex="0"}
# subset data frameebd_in_poly <- ebd[in_poly[, 1], ]
```
:::

Finally, let's create a simple map showing the EBD observations before
(in black) and after (in green) subsetting the data to be within the
polygon.

::: {.highlight}
``` {.chroma tabindex="0"}
par(mar = c(0, 0, 0, 0))plot(poly %>% st_geometry(), col = "grey40", border = NA)plot(ebd_sf, col = "black", pch = 19, cex = 0.5, add = TRUE)plot(ebd_sf[in_poly[, 1], ],      col = "forestgreen", pch = 19, cex = 0.5,      add = TRUE)legend("top",        legend = c("All observations", "After spatial subsetting"),        col = c("grey40", "forestgreen"),        pch = 19,       bty = "n",       ncol = 2)
```
:::

<figure>
<img src="plot-1.png"
alt="map of all occurrences, with the polygon in grey, and the points kept after spatial subsetting in green" />
</figure>

Looks like it worked! We got just the points within the polygon as
intended. Two final notes:

1.  If you're working with the full EBD (a 200+ GB file), you'll need to
    follow step 1 and subset the data using auk prior to importing
    into R. However, if you've used the custom download form to get an
    EBD subset, your file is likely small enough that you can read the
    data directly into R with `read_ebd()` and skip straight to step 2.
2.  If your intention is to eventually [zero-fill the
    EBD](https://cornelllabofornithology.github.io/ebird-best-practices/ebird.html#ebird-zf)
    to produce presence-absence data you'll need to include the sampling
    event data file in the `auk_ebd()`, subset both the EBD and sampling
    event data files separately to points within the polygon, the
    combine them together and zero-fill with `auk_zerofill()`.

auk users are welcome to submit issues on GitHub with bug reports or
feature requests, or to [email me
directly](https://github.com/CornellLabofOrnithology/auk/blob/88cf9a370adb8d7a106f07cd45003033c73edc8e/DESCRIPTION#L8).
Should you be interested in contributing to auk, there is also a
[vignette specifically for adding functionality to the
package](https://docs.ropensci.org/auk/articles/development.html).


One of the first things I took on when I started at the Cornell Lab of Ornithology was creating the auk R package for accessing eBird data. The entire eBird dataset can be downloaded as a massive text file, called the eBird Basic Dataset (EBD), and auk pulls out manageable chunks of the dataset based on various spatial, temporal, or taxonomic filters.

Extracting eBird Data from a Polygon

Maëlle Salmon

Thanks to the [first post of theseries](/blog/2018/08/14/where-to-bird/)
we knowwhere to observe birds near Radolfzell's Max Planck Institute
forOrnithology, so we could go and do that! Or we can stay behind
ourlaptops and take advantage of [eBird](https://ebird.org/home),
afantastic bird sightings aggregator! As explained by Matt
Strimas-Mackeyin [his recent blog post](/blog/2018/08/07/auk/),"The
eBird database currently contains over 500 million records of
birdsightings, spanning every country and over 98% of species, making it
anextremely valuable resource for bird research and conservation.".

Luckily for us, there are no less than *two* rOpenSci packages giving
usaccess to eBird data! In this blog post, I shall play with both of
them,highlighting their respective strengths, while discovering what
birdsare observed in the area.

## How to access eBird data?

There are two ways to access eBird data with an R package for each
ofthese methods,

- whole dataset download via `auk` maintained by
  [MattStrimas-Mackey](http://strimas.com/),

- APIs, via `rebird` maintained by
  [SebastianPardo](https://sebpardo.github.io/).

Your use case will help you decide which entry point is the
mostappropriate for your use case. *Note that both packages have
documentedtheir respective applications in order to help potential
users:[`rebird`
README](https://github.com/ropensci/rebird#auk-vs-rebird),[`auk`README](https://github.com/CornellLabofOrnithology/auk#auk-vs-rebird).*

- You want to study a region, or a bird, quite deeply and you evenwant
  absence/presence data, not only presence data. Use `auk`!

- You want to build a tool based on recent observations only or youwant
  to get a quick taste of eBird's data. Use `rebird`!

- A bit provocatively, do you want birds data only? If not, maybeyou'll
  need a combination of `auk`/`rebird` and another package.Check out
  this list of [data providers covered
  by`spocc`](https://github.com/ropensci/spocc#spocc), umbrella
  packagefor rOpenSci's packages accessing occurrence data. Many data
  sourcesactually end up in
  [GBIFdatasets](https://www.gbif.org/dataset/search), eBird seems
  toupload their data there once a year.

- You want to analyze *your* eBird's sightings? Check out
  thework-in-progress [`myebird`](https://github.com/sebpardo/myebird)
  by[Sebastian Pardo](https://sebpardo.github.io/),
  `rebird`'smaintainer, and [this app](http://www.ebirdedgeapp.io/) by
  [SimónValdez-Juarez](https://www.sfu.ca/biology/people/profiles/svaldez.html)and
  [Sebastian Pardo](https://sebpardo.github.io/) highlighting themost
  endangered species you observed.

- You're writing a birder's guide to rOpenSci? Use both `rebird`
  and`auk` to show them off!

## How to get access to eBird's data

### Whole eBird dataset, quarterly updated

One needs to first [create an
eBirdaccount](https://secure.birds.cornell.edu/cassso/login) and
[thenrequest access to the data](https://ebird.org/data/request). Once
onehas gotten green light from eBird (in my case a few days following
myrequest), after a small dance of joy it's time to head to
eBird's[download page](https://ebird.org/data/download/ebd). If one
doesn'twant nor need to download the whole eBird Basic Dataset (EBD),
one canrequest a custom download, which I did, asking for only the data
forGermany which I got after a few days (the time to receiving the link
todownload a custom dataset is variable). While waiting, I worked on
the`rebird` part of this post, among other things.

### API key? Not yet

At the moment, `rebird` interfaces the version 1.1 eBird APIs that
willbe retired "at some point in thefuture".When this happens, the
`rebird` package [will use the
newAPI](https://github.com/ropensci/rebird/issues/58) which will
meanyou'll need an API key. Currently, though, you don't need
anyauthentication to use `rebird`.

## Using rebird while waiting for the eBird's full dataset

In the following, we'll use the rOpenSci's package `rebird` to get
andmap all observations in the last 30 days near Radolfzell in Germany.

The *Radolfzell* part of that sentence is a bit different than in
[thelast post about finding bird hides near the MPI institute
forornithology](/blog/2018/08/14/where-to-bird/): Iwant all observations
inside the polygon of the district of Constance(Landkreis Konstanz,
including Radolfzell... and a protected naturalarea!) so I'll first need
to get it. For doing that I'll use`osmdata::getbb`, that uses the free
Nominatim API provided byOpenstreetmap.

::: {.highlight}
``` {.chroma tabindex="0"}
library("sf")landkreis_konstanz <- osmdata::getbb("Landkreis Konstanz",                             format_out = "sf_polygon")plot(landkreis_konstanz)
```
:::

![Limits of the County
ofConstance](/img/blog-images/2018-08-21-birds-radolfzell/lk-1.png)

Neither `rebird` nor `spocc` currently offer built-in trimming
ofoccurrence data to a polygon (whereas `osmdata` does). A
furtherdifficulty created by eBird's API is that it doesn't allow for
the useof a bounding box, but instead demands a `lat`, `lng` and a
`dist`defining the radius of interest from given `lat`/`lng` in
kilometers.Thanks to [Marco Sciaini](https://marcosci.rbind.io/) for
providing mewith an easy way to compute `dist`, using the `sf` package.

::: {.highlight}
``` {.chroma tabindex="0"}
coord <- sf::st_coordinates(landkreis_konstanz)bbox <- c(x1 = min(coord[, "X"]),          x2 = max(coord[, "X"]),          y1 = min(coord[, "Y"]),          y2 = max(coord[, "Y"]))center <- c(x = (bbox["x1"] + bbox["x2"])/2,            y = (bbox["y1"] + bbox["y2"])/2)dist <- landkreis_konstanz %>%  sf::st_cast("POINT") %>%  sf::st_distance() %>%   max() * 0.5dist
```
:::

    ## 24129.15 m

Now, we can make the query.

::: {.highlight}
``` {.chroma tabindex="0"}
birds <- rebird::ebirdgeo(species = NULL,                          lng = center["x.x1"],                          lat = center["y.y1"],                          back = 30,                          dist = as.numeric(                            units::set_units(dist, "km")))nrow(birds)
```
:::

    ## [1] 55

::: {.highlight}
``` {.chroma tabindex="0"}
str(birds)
```
:::

    ## Classes 'tbl_df', 'tbl' and 'data.frame':    55 obs. of  12 variables:##  $ lng            : num  8.94 8.94 8.94 8.94 8.94 ...##  $ locName        : chr  "Radolfzeller Aachmündung (Bodensee)" "Radolfzeller Aachmündung (Bodensee)" "Radolfzeller Aachmündung (Bodensee)" "Radolfzeller Aachmündung (Bodensee)" ...##  $ sciName        : chr  "Chroicocephalus ridibundus" "Motacilla alba" "Rallus aquaticus" "Aythya fuligula" ...##  $ obsValid       : logi  TRUE TRUE TRUE TRUE TRUE TRUE ...##  $ locationPrivate: logi  FALSE FALSE FALSE FALSE FALSE FALSE ...##  $ obsDt          : chr  "2018-08-08 13:30" "2018-08-08 13:30" "2018-08-08 13:30" "2018-08-08 13:30" ...##  $ obsReviewed    : logi  FALSE FALSE FALSE FALSE FALSE FALSE ...##  $ comName        : chr  "Black-headed Gull" "White Wagtail" "Water Rail" "Tufted Duck" ...##  $ lat            : num  47.7 47.7 47.7 47.7 47.7 ...##  $ locID          : chr  "L3314048" "L3314048" "L3314048" "L3314048" ...##  $ locId          : chr  "L3314048" "L3314048" "L3314048" "L3314048" ...##  $ howMany        : int  NA 2 1 1 NA 3 NA NA 8 20 ...

Now that we have the occurrence data, let's plot it to see
whethertrimming is required.

::: {.highlight}
``` {.chroma tabindex="0"}
crs <- sf::st_crs(landkreis_konstanz)birds_sf <- sf::st_as_sf(birds,                         coords = c("lng", "lat"),                          crs = crs)library("ggplot2")ggplot() +  geom_sf(data = landkreis_konstanz) +  geom_sf(data = birds_sf) +  theme(legend.position = "bottom") +  hrbrthemes::theme_ipsum() +  ggtitle("eBird observations over the last 30 days",          subtitle = "Observations within a circle around the County of Constance")
```
:::

![map of raw observations within
acircle](/img/blog-images/2018-08-21-birds-radolfzell/map_circle-1.png)

Yes, trimming is required! It'd have been too bad not to learn how to
doit, anyway. We also add the MPI to the map.

::: {.highlight}
``` {.chroma tabindex="0"}
# which parts of the oject are in the countyin_indices <- sf::st_within(birds_sf, landkreis_konstanz)# filter themtrimmed_birds <- dplyr::filter(birds_sf,                               lengths(in_indices) > 0)# summarize to get no. of birds by  locationsummarized_birds <- trimmed_birds %>%  dplyr::group_by(locName) %>%  dplyr::summarise(n = n())# MPI mpi <- opencage::opencage_forward("Am Obstberg 1 78315 Radolfzell",                                   limit = 1)$resultscoords <- data.frame(lon = mpi$geometry.lng,                     lat = mpi$geometry.lat)crs <- sf::st_crs(landkreis_konstanz)mpi_sf <- sf::st_as_sf(coords,                       coords = c("lon", "lat"),                        crs = crs)# Map!ggplot() +  geom_sf(data = landkreis_konstanz) +  geom_sf(data = summarized_birds,          aes(size = n), show.legend = "point") +  hrbrthemes::theme_ipsum() +  ggtitle("eBird observations over the last 30 days",          subtitle = "County of Constance, MPI as a triangle") +  geom_sf(data = mpi_sf,          shape = 2) 
```
:::

![trimmed observations in
thecounty](/img/blog-images/2018-08-21-birds-radolfzell/trimmed1-1.png)

We got 49 observations (`nrow(trimmed_birds)`) of 49
species(`length(unique(trimmed_birds$comName))`), over 2
places(`length(unique(trimmed_birds$locName))`) during 5 observation
sessions.Hopefully merely an appetizer to what we can get from using the
fulleBird dataset in the next section...

Note that the initial query could have been made with `spocc` whichwould
have helped using the rOpenSci occurrence suite.

::: {.highlight}
``` {.chroma tabindex="0"}
birds2 <- spocc::occ(from = "ebird",                     ebirdopts = list(method = "ebirdgeo",                                      species = NULL,                                      lng = center["x.x1"],                                      lat = center["y.y1"],                                      back = 30,                                      dist = as.numeric(                                        units::set_units(dist, "km"))))                     mapr::map_leaflet(birds2)
```
:::

<figure>
<img src="/img/blog-images/2018-08-21-birds-radolfzell/mapr-1.png"
alt="mapr leaflet map of observations locations" />
</figure>

Quite handy!

Now, let's explore the whole eBird dataset for Germany.

## Using auk to process EBD dataset for Germany

After getting access to a custom dataset corresponding to the EBD
forGermany only, I used `auk`'s documentation and
[thispost](/blog/2018/08/07/auk/) to learn how to processit. Since I
wasn't planning on zero-filling the data to getpresence/absence counts,
I was able to ignore the sampling event datathat contains the
checklist-level information (e.g. time and date,location, and search
effort information). For an example of a moreadvanced `auk` workflow
involving the **full** EBD, and sampling data,refer to [Matt
Strimas-Mackey's own blog post about hispackage](/blog/2018/08/07/auk/).

### Preparing the dataset

Here, the workflow is to *clean* the data and to *filter* it using oneof
`auk`'s built-in filters and then polygon filtering as earlier inthis
post. All steps are quite fast, because the custom dataset forGermany
isn't too big (a few hundred megabytes).

Cleaning happens in the following:

::: {.highlight}
``` {.chroma tabindex="0"}
ebd_dir <- "C:/Users/Maelle/Documents/ropensci/ebird"f <- file.path(ebd_dir, "ebd_DE_relMay-2018.txt")f_clean <- file.path(ebd_dir, "ebd_DE_relMay-2018_clean.txt")auk::auk_clean(f, f_out = f_clean, remove_text = TRUE)
```
:::

Then one can filter the data. Note that the `auk_extent` function
thatonly retains observations within a bounding box has been
renamed`auk_bbox` in the dev version of `auk`, the old name will be
deprecatedsoon.

::: {.highlight}
``` {.chroma tabindex="0"}
ebd_dir <- "C:/Users/Maelle/Documents/ropensci/ebird"f_in_ebd <- file.path(ebd_dir, "ebd_DE_relMay-2018_clean.txt")library("magrittr")landkreis_konstanz_coords <- sf::st_coordinates(landkreis_konstanz)ebd_filter <- auk::auk_ebd(f_in_ebd) %>%   auk::auk_extent(c(min(landkreis_konstanz_coords[, "X"]),                    min(landkreis_konstanz_coords[, "Y"]),                     max(landkreis_konstanz_coords[, "X"]),                     max(landkreis_konstanz_coords[, "Y"])))ebd_filter
```
:::

    ## Input ##   EBD: C:\Users\Maelle\Documents\ropensci\ebird\ebd_DE_relMay-2018_clean.txt ## ## Output ##   Filters not executed## ## Filters ##   Species: all##   Countries: all##   States: all##   BCRs: all##   Spatial extent: Lon 8.6 - 9.2; Lat 47.7 - 47.9##   Date: all##   Start time: all##   Last edited date: all##   Protocol: all##   Project code: all##   Duration: all##   Distance travelled: all##   Records with breeding codes only: no##   Complete checklists only: no

::: {.highlight}
``` {.chroma tabindex="0"}
fs::dir_create("ebird")f_out_ebd <- "ebird/ebd_lk_konstanz.txt"f_out_sampling <- "ebird/ebd_lk_konstanz_sampling.txt"ebd_filtered <- auk::auk_filter(ebd_filter, file = f_out_ebd,                                overwrite = TRUE)
```
:::

On top of this filtering with `auk`, after loading the data we
filterobservations inside the polygon of the county.

::: {.highlight}
``` {.chroma tabindex="0"}
crs <- sf::st_crs(landkreis_konstanz)ebd <- auk::read_ebd(f_out_ebd) %>%  sf::st_as_sf(coords = c("longitude", "latitude"),                 crs = crs) in_indices <- sf::st_within(ebd, landkreis_konstanz)ebd <- dplyr::filter(ebd, lengths(in_indices) > 0)ebd <- as.data.frame(ebd)
```
:::

### What are the observed birds?

Before looking at species names, let's have a brief look at the size
andtemporal extent of the data.

::: {.highlight}
``` {.chroma tabindex="0"}
library("ggplot2")dim(ebd)
```
:::

    ## [1] 10156    41

::: {.highlight}
``` {.chroma tabindex="0"}
ebd %>%  dplyr::mutate(year = lubridate::year(observation_date)) %>%ggplot() +  geom_bar(aes(year))  +  hrbrthemes::theme_ipsum(base_size = 12, axis_title_size = 12, axis_text_size = 12) +  ylab("No. of eBird observations") +  xlab("Time (years)") +  ggtitle("Full eBird dataset for the County of Constance")
```
:::

![No. of eBird observations over
theyears](/img/blog-images/2018-08-21-birds-radolfzell/barplot-1.png)

eBird started in 2002 but only became global in 2010. It allows peopleto
enter older observations, though.

Now we can look at what birds have been reported the most.

::: {.highlight}
``` {.chroma tabindex="0"}
ebd %>%  dplyr::filter(approved) %>%  dplyr::count(scientific_name, common_name) %>%  dplyr::arrange(- n) %>%  head(n = 10) %>%  knitr::kable()
```
:::

| scientific_name            | common_name         |   n |
|:---------------------------|:--------------------|----:|
| Corvus corone              | Carrion Crow        | 288 |
| Turdus merula              | Eurasian Blackbird  | 285 |
| Anas platyrhynchos         | Mallard             | 273 |
| Fulica atra                | Eurasian Coot       | 268 |
| Parus major                | Great Tit           | 266 |
| Podiceps cristatus         | Great Crested Grebe | 254 |
| Ardea cinerea              | Gray Heron          | 236 |
| Cygnus olor                | Mute Swan           | 234 |
| Cyanistes caeruleus        | Eurasian Blue Tit   | 233 |
| Chroicocephalus ridibundus | Black-headed Gull   | 223 |

I had to google most of them, but only because I didn't know
thescientific and English names of these birds: they're birds even I,
not abirder, know, probably because they're also common in Brittany
where Igrew up.

We can also look at birds whose observation was rejected. Out of
10156observations only 64 were reviewed, and only 5 were not approved.

::: {.highlight}
``` {.chroma tabindex="0"}
ebd %>%  dplyr::select(scientific_name, common_name,               approved, reviewed, reason) %>%  dplyr::filter(!approved) %>%  knitr::kable()
```
:::

| scientific_name     | common_name       | approved | reviewed | reason                    |
|:--------------------|:------------------|:---------|:---------|:--------------------------|
| Cygnus atratus      | Black Swan        | FALSE    | TRUE     | Species-Introduced/Exotic |
| Cygnus atratus      | Black Swan        | FALSE    | TRUE     | Species-Introduced/Exotic |
| Cygnus atratus      | Black Swan        | FALSE    | TRUE     | Species-Introduced/Exotic |
| Oxyura leucocephala | White-headed Duck | FALSE    | TRUE     | Species-Introduced/Exotic |
| Mareca sibilatrix   | Chiloe Wigeon     | FALSE    | TRUE     | Species-Introduced/Exotic |

Black Swans [are mostly present in Australia, imported and escaped in
afew other places](https://en.wikipedia.org/wiki/Black_swan) but
eBirdmostly doesn't accept the entry of exotic species [although
it'sdebated](https://help.ebird.org/customer/portal/questions/5390335-exclusion-of-not-native-birds-in-lists).In
any case, eBird's curation of the data entered is quite admirable.

### Who observed birds?

In [one of his latest
blogposts](https://recology.info/2018/06/butte-county/) Scott
Chamberlainmentioned the legendary Lowell Ahart, super plant collector
in ButteCounty, California. Does the county of Constance have a super
birder?

::: {.highlight}
``` {.chroma tabindex="0"}
(first_birder <- ebd %>%  dplyr::count(observer_id) %>%  dplyr::arrange(- n) %>%  head(n = 1) )
```
:::

    ## # A tibble: 1 x 2##   observer_id     n##   <chr>       <int>## 1 obsr457108   3551

::: {.highlight}
``` {.chroma tabindex="0"}
(proportion <- round(first_birder$n/nrow(ebd),                    digits = 2))
```
:::

    ## [1] 0.35

Wow, that person made 35% of eBird observations in the county! The EBDno
longer provides names (consequence of the [EU General Data
ProtectionRegulation](https://www.eugdpr.org/)) but from the checklist
ID one canget access to the checklist page e.g
[thisone](https://ebird.org/view/checklist/S42391392) where the name of
theobserver is present. The super birder of the County of Constance
is[Antonio Anta Bink](https://ebird.org/profile/NDU3MTA4/DE).

## Conclusion

### R packages for occurrence data

In this post I gave a rough view of what birds are present in the
countyaround Radolfzell: Eurasian Blackbirds, Carrion Crows, Great
Tits... butnot Black Swans in eBird's data. We mostly illustrated the
use of two Rpackages accessing eBird's data:

- `auk` for processing the gigantic whole eBird's dataset.

- `rebird` for getting access to recent data via an API. `rebird` ispart
  of a larger collection of packages for occurrence data
  withinrOpenSci's suite, with `spocc` being an umbrella package
  accessingseveral data sources; `scrubr` a helper for cleaning data
  obtainedthis way; and `mapr` a utility package for mapping such data.

Explore these packages, and more of rOpenSci's suite, by checking outour
[packages page](/packages/)!

### More birding soon!

Stay tuned for the next post in this series, that'll mark a break
frommodern data since we'll try to extract information from old
naturalhistory bird drawings! After that, in a following post we'll come
backto the occurrence data obtained from eBird in order to complement
itwith open taxonomic and traits data. In the meantime, happy
(e)birding!


Thanks to the first post of theseries we knowwhere to observe birds near Radolfzell’s Max Planck Institute forOrnithology, so we could go and do that! Or we can stay behind ourlaptops and take advantage of eBird, afantastic bird sightings aggregator!

Postagens de Rogue Scholar

Extracting eBird Data from a Polygon

What birds are observed near Radolfzell? Bird occurrence data in R