Background

The study and conservation of the natural world relies on detailed information about species’ distributions, abundances, environmental associations, and population trends over time. For many taxa, this information is challenging to obtain at relevant geographic scales. The goal of the eBird Status and Trends project is to use data from eBird, the global citizen science bird monitoring administered by Cornell Lab of Ornithology, to generate a reliable, standardized source of biodiversity information for the world’s bird populations. To translate the eBird observations into robust data products, we use statistical models to fill spatiotemporal gaps, using local land cover descriptions derived from NASA MODIS and other remote sensing data, while controlling for biases inherent in species observations collected by volunteers.

This data set provides estimates of the year-round distribution, abundances, and environmental associations for a the majority of North American bird species for the year 2019 For each species, distribution and abundance estimates are available for all 52 weeks of the year across a regular grid of locations that cover terrestrial North and South America at a resolution of 2.96 km x 2.96 km. Variation in detectability associated with the search effort is controlled by standardizing the estimates as the expected occupancy rate and count of the species at the optimal time of day, while expending the effort necessary to maximize detection of the species. Predictions have been optimized for search effort and user skill, specific for the given region, season, and species, in order to maximize detection rates.

To describe how each species is associated with features of its local environment, estimates of the relative importance of each remotely sensed variable (e.g. land cover, elevation, etc), are available throughout the year at a monthly temporal and regional spatial resolution. Additionally, to assess estimate quality, we provide upper and lower confidence bounds for all abundance estimates and we provide regional-seasonal scale validation metrics for the underlying statistical models. For more information about the data products see the FAQ and summaries. See Fink et al. (2019) for more information about the analysis used to generate these data.

Data Access

As of July 2021, data access is now granted through a Access Request Form at: https://ebird.org/st/request. Access with this form generates a key to be used with this R package and is provided immediately (as long as commercial use is not requested). Our terms of use have been updated to be more permissive in many cases, particularly academic and research use. When requesting data access, please be sure to carefully read the terms of use and ensure that your intended use is not restricted. Data are no longer publicly available via AWS, aside from an example dataset.

After completing the Access Request Form, you will be provided a Status and Trends access key, which you will need when downloading data. To store the key so the package can access it when downloading data, use the function set_ebirdst_access_key("XXXXX"), where "XXXXX" is the access key provided to you.

Throughout the package vignettes, a simplified example dataset is used consisting of Yellow-bellied Sapsucker in Michigan. This dataset is designed to be small for faster download and is accessible without a key.

library(ebirdst)

# download data
# download a simplified example dataset from aws s3
# example data are for Yellow-bellied Sapsucker in Michigan
# by default file will be stored in a persistent data directory:
# rappdirs::user_data_dir("ebirdst"))
sp_path <- ebirdst_download(species = "example_data")

For a full list of the species available for download, look at the data frame ebirdst_runs, which is included in this package. Data for all other species requires an API key to access.

Data Types and Structure

IMPORTANT. AFTER DOWNLOADING THE RESULTS, DO NOT CHANGE THE FILE STRUCTURE. All functionality in this package relies on the structure inherent in the delivered results. Changing the folder and file structure will cause errors with this package. If you use this package to download and analyze the results, you do not ever need to interact with the files directly, outside of R. If you intend to use the data outside of this package, than this warning does not necessarily apply to you.

Data Types

The data products included in the downloads contain two types of data: a) raster data containing occurrence, count, and abundance estimates at a 2.96 km resolution for each of 52 week across North America, b) non-raster, tabular, text data containing information about modeled relationships between observations and the ecological covariates, in the form of: predictor importances (PIs), partial dependence (PD) relationships, and predictive performance metrics (PPMs). The raster data will be the most commonly used data, as it provides high resolution, spatially-explicit information about the abundance, count, and occurrence of each species. The non-raster data is an advanced, modeling-oriented product that requires more understanding about the modeling process.

Data Structure

Data are grouped by species, using a unique run name that starts with the standard 6-letter eBird species code. A full list of species and run names can be found in the ebirdst_runs data frame.

For each individual species, the data are stored within a sub-folder named with the run name and containing the following files:

  • weekly_cubes/: a directory containing GeoTIFFs of the occurrence, count, and abundance estimates at three resolutions. See below for more details.
  • abundance_seasonal/: a directory containing GeoTIFFs of seasonally averaged abundance estimates at three resolutions. Dates defining the boundary of each season are set on a species-specific basis by an expert reviewer familiar with the species. These dates are available in the ebirdst_runs data frame. Only seasons that passed the expert review process are included.
  • pi-pd.db: an SQLite database containing predictor importance (PI) and partial dependence estimates, accessible with the load_pis() and load_pds() functions, respectively.
  • predictions.db: model predictions for a test dataset held out of the model fitting. These predictions are used for calculating predictive performance metrics (PPMs) using the ebirdst_ppms() function.
  • config.rds: run-specific parameters, mostly for internal use, but also containing useful parameters for mapping the abundance data. These parameters can be loaded with load_config().

Raster Data

eBird Status and Trends abundance, count, and occurrence estimates are currently provided in the widely used GeoTIFF raster format. These are easily opened with the raster package in R, as well as with a variety of GIS software tools. Each estimate is stored in a multi-band GeoTIFF file. These “cubes” come with areas of predicted and assumed zeroes, such that any cells that are NA represent areas outside of the area of estimation. All cubes have 52 weeks, even if some weeks are all NA (such as those species that winter entirely outside of the study area). In addition, seasonally averaged abundance layers are provided. The two vignettes that are relevant to the raster data are the intro mapping vignette and the advanced mapping vignette.

All predictions are made on a standard 2.96 x 2.96 km grid, however, for convenience lower resolution GeoTIFFs are also provided, which are typically much faster to work with. The three resolutions are:

  • High resolution (hr): the native 2.96 km resolution data
  • Medium resolution (mr): the hr data aggregated by a factor of 3 in each direction resulting in a resolution of 8.89 km
  • Low resolution (lr): the hr data aggregated by a factor of 9 in each direction resulting in a resolution of 26.7 km

The weekly cubes use the following naming convention:

weekly_cubes/<run_name>_<resolution>_2019_<layer>.tif

And the seasonal GeoTIFFs use the following convention:

seasonal_abundance/<run_name>_<resolution>_2019_abundance-seasonal_<season>.tif

Where <season> is one of postbreeding_migration, prebreeding_migration, nonbreeding, breeding, or year_round.

Projection

The GeoTIFFs use the same Sinusoidal projection as NASA MODIS data. As part of each data package, a template raster (srd_raster_template.tif) is provided, that contains the spatial extent and resolution for the full Western Hemisphere. Accessing this raster directly through the package is not necessary, and can be applied elsewhere (e.g., other GIS software). Note that this projection is ideal for analysis, as it is an equal are projection, but is not ideal for mapping. See the intro mapping vignette for details on using a more suitable projection for mapping.

Raster Layer Descriptions

Type Measure File Name
occurrence median <run_name>_<resolution>_2019_occurrence_median.tif
count median <run_name>_<resolution>_2019_count_median.tif
abundance median <run_name>_<resolution>_2019_abundance_median.tif
abundance 10th quantile <run_name>_<resolution>_2019_abundance_lower.tif
abundance 90th quantile <run_name>_<resolution>_2019_abundance_upper.tif

occurrence_median

This layer represents the expected probability of occurrence of the species, ranging from 0 to 1, on an eBird Traveling Count by a skilled eBirder starting at the optimal time of day with the optimal search duration and distance that maximizes detection of that species in a region.

count_median

This layer represents the expected count of a species, conditional on its occurrence at the given locatiion, on an eBird Traveling Count by a skilled eBirder starting at the optimal time of day with the optimal search duration and distance that maximizes detection of that species in a region.

abundance_median

This layer represents the expected relative abundance, computed as the product of the probability of occurrence and the count conditional on occurrence, of the species on an eBird Traveling Count by a skilled eBirder starting at the optimal time of day with the optimal search duration and distance that maximizes detection of that species in a region.

abundance_lower

This layer represents the lower 10th quantile of the expected relative abundance of the species on an eBird Traveling Count by a skilled eBirder starting at the optimal time of day with the optimal search duration and distance that maximizes detection of that species in a region.

abundance_upper

This layer represents the upper 90th quantile of the expected relative abundance of the species on an eBird Traveling Count by a skilled eBirder starting at the optimal time of day with the optimal search duration and distance that maximizes detection of that species in a region.

Non-raster Data

The non-raster, tabular data containing information about modeled relationships between observations and the ecological covariates are best accessed through functionality provided in this package. However, in using them through the package, it is possible to export them to other tabular formats for use with other software. For information about working with these data, please reference to the non-raster data vignette for details on how to access additional information from the model results about predictor importance and directionality, as well as predictive performance metrics. Important note: to access the non-raster data, use the parameter tifs_only = FALSE in the ebirdst_download() function.

Vignettes

Beyond this introduction to the eBird Status and Trends products and data, we have written multiple vignettes to help guide users in using the data and the functionality provided by this package. An intro mapping vignette expands upon the quick start readme and shows the basic mapping moves. The advanced mapping vignette shows how to reproduce the seasonal maps and statistics on the eBird Status and Trends website. Finally, the non-raster data vignette details how to access additional information from the model results about predictor importance and directionality, as well as predictive performance metrics.

Conversion to Flat Format

The raster package has a lot of functionality and the RasterLayer format is useful for spatial analysis and mapping, but some users do not have GIS experience or want the data in a simpler format for their preferred method of analysis. There are multiple ways to get more basic representations of the data.

library(raster)

# load median abundances
abd <- load_raster("abundance", path = sp_path)

# use parse_raster_dates() to get actual date objects for each layer
date_vector <- parse_raster_dates(abd)

# to convert the data to a simpler geographic format and access tabularly   
# reproject into geographic (decimal degrees) 
abd_ll <- projectRaster(abd[[26]], crs = "+init=epsg:4326", method = "ngb")

# Convert raster object into a matrix
p <- rasterToPoints(abd_ll)
colnames(p) <- c("longitude", "latitude", "abundance_umean")
head(p)
#>      longitude latitude abundance_umean
#> [1,] -88.56451 48.15386       0.0000000
#> [2,] -88.52431 48.15386       0.0000000
#> [3,] -88.64491 48.12726       0.0000000
#> [4,] -88.60471 48.12726       0.0000000
#> [5,] -88.56451 48.12726       0.2123874
#> [6,] -88.72531 48.10066       0.0000000

These results can then be written to CSV.

write.csv(p, file = "yebsap_week26.csv", row.names = FALSE)

References

Fink, D., T. Auer, A. Johnston, V. Ruiz‐Gutierrez, W.M. Hochachka, S. Kelling. 2019. Modeling avian full annual cycle distribution and population trends with citizen science data. Ecological Applications, 00(00):e02056. https://doi.org/10.1002/eap.2056