SSS

General

Updated: November 2022

The annual update as of November 2022 includes updated visualizations and data products for 1200 species, with 586 of those species having new Trends estimates. Both Status and Trends products use eBird data from 2007 through 2021. Status products are estimated for the year 2021 and Trends estimates represent the change from 2007 through 2021. In December 2022, we expect to release another ~1000 species of Status products, but no additional species with Trends estimates.

What has changed in the latest version?

The most significant change in this latest version is that Trends have been added. These show the change in relative abundance from 2007 through 2021 at 27km x 27km pixels for either the breeding or non-breeding season. Read more in the Trends FAQs section. The visualizations and products related to habitat have been removed. These have gotten difficult to maintain, were not heavily used, and may become impossible to produce in the near future. For analysts, the detailed underlying data used to produce habitat plots are still available this year through the ebirdst R package. However, that underlying data is likely to be removed next year. From a modeling perspective, one of the biggest changes is that we have doubled the number of models in the ensemble for each species (from 100 to 200). As a result, a number of species in sparse regions such as Central Asia, Russia, and Africa have seen a dramatic improvement in coverage. On seasonal maps, the representation of year-round is now based on a 0.1% overlap between breeding and non-breeding seasons, resulting in a better depiction of stationary, resident populations within migratory species. For a detailed list of all changes and explanations, see the changelog for this version.

What are eBird Status and Trends Data Products and visualizations?

The eBird Status and Trends Data Products provide basic ecological information for more than 1100 species globally, describing their ranges, abundances, and trends. To generate the visualizations, we use statistical and machine learning analyses designed to combine eBird data with a range of environmental data. The analyses are used to predict the occurrence and abundance of species across the globe at weekly intervals. These predictions are the cornerstone of the Status and Trends Data Products and are summarized in several ways to produce the different visualizations.

Currently available eBird Status and Trends visualizations:

Weekly Abundance represents weekly relative abundances, revealing movements of a population throughout the year.
Seasonal abundance maps indicate the average relative abundance of a species in each season of their annual cycle.
Range maps show species seasonal range boundaries, similar to traditional range maps.
Regional stats tables show mean relative abundance, percentage of the seasonal population, percentage of the region occupied, percentage of the range in region, and days of occupation in region for countries, territories, and dependencies, and subregions within.
Trends show the change in relative abundance from 2007 through 2021 at 27km x 27km pixels for either the breeding or non-breeding season.

Different visualizations require different volumes of data and the most data-intensive ones are only available for some of the species and seasons. We only present visualizations for species and seasons that have passed analytical and expert quality review tests.

Can I contribute data to the eBird Status and Trends project?

Yes, any eBirder can contribute! eBird Status and Trends is only possible thanks to the eBird submissions of hundreds of thousands of eBird users. Our ability to update and improve the Status and Trends Data Products in the future continues to depend on the contributions of eBirders like you!

If you submitted checklists that meet all the requirements answered under “Which eBird data were used to generate the Status and Trends Data Products?“, then you have already contributed data to the Status and Trends! Any future checklists you submit that meet these requirements will automatically be included in analyses for the updated Status and Trends Data Products.

Remember, any observation is useful, whether it is from today or your field notebooks from 15 years ago. Whether it is from a hotspot with amazing birds, or a place with few species – all checklists are valuable. To ensure your eBirding checklists are most useful to scientific efforts like this, you can make your checklists:

Complete checklists (i.e., record all species you were able to identify),
Provide a count or estimate of the number of individuals for each species
Use one of these protocols: traveling or stationary count.
Provide information on the start time, duration, number of observers, and distance traveled (The eBird mobile App now does many of these automatically with the new tracks)
Provide documentation of unusual sightings with descriptions or photos.

See our article on how to make your eBird checklists more valuable.

Abundance

How is relative abundance defined?

Relative abundance is the count of individuals of a given species detected by an expert eBirder on a 1 hour, 1 kilometer traveling checklist at the optimal time of day. Relative abundance predictions have been optimized for user skill and hourly weather conditions, specific for the given region, season, and species, in order to maximize detection rates.

For each species, relative abundance was estimated for all 52 weeks of the year across a regular spatial grid with a density of one location per 2.96 km × 2.96 km. Estimates at each location and date were made based on the local habitat, elevation, and topography.

Because detecting birds in the environment can be difficult, we know that there are always some individual birds that are missed by eBirders. For this reason, we refer to the quantity estimated as a relative measure of abundance. Although the relative abundance estimates will underestimate the true abundance, they do provide a standardized index that can be used to compare abundance in different regions. For example, if relative abundance is 10 in one area and 5 in another area, then we would estimate abundance is twice as high in the first area, even if we’re not sure of the actual number of individuals in the area.

See featured examples of relative abundance

How is “Percentage of total population in region” calculated?

For each species and season, we summed the relative abundance estimates across the selected region and then divided it by the sum of the relative abundance estimates across the entire seasonal range. The result is presented as a percentage.

This will be a reasonable estimate if the whole population is within the “modeled area.” For this reason, the reporting of this value was contingent on the majority of a species’ known seasonal range being within the modeled area. If the percentage of total population value is missing, but the season was mapped in other visualizations, it is because an expert reviewer believed that a significant portion of the species’ seasonal range was not included, but that the distribution within the modeled area was correct.

See featured example of regional stats

Range

How are species’ ranges defined?

Species’ ranges are defined as the areas where the species is expected to occur on at least one out of seven predicted hypothetical checklists in a given week. This is equivalent to a single skilled eBirder starting at the optimal time of day, expending the effort necessary to maximize detection of the species for each day of the week, and detecting the species on at least ten percent of the checklists within a week. Each species’ range was estimated for all 52 weeks of the year at 2.96 km × 2.96 km grid cell locations. To create easy-to-read range boundaries, the 2.96 km grid data were aggregated to an 8.89 km grid and spatially smoothed. Both the smoothed, aggregated seasonal boundaries and the raw 2.96 km grid cell boundaries are available for download.

See featured examples of eBird Status and Trends range maps

How is “Percentage of region occupied” calculated?

Percentage of the region occupied is calculated as the percent of the selected region that is covered by the range of the species.

How is “Percentage of range in region” calculated?

Percentage of range in a region is calculated as the fraction of a species’ range that falls within the selected region.

See featured example of regional stats

How is “Days of occupation in region” calculated?

Days of occupation in a region is the number of days that a species is present in the selected region. A species is defined to be present in a region when at least 5% of the region was within the species range during the given season.

Trends

How are Trends defined?

eBird Trends maps show the cumulative percent change (or trend) in relative abundance across a grid of 27 x 27 km pixels from 2007-2021 represented by blue (increasing) and red (decreasing) pixels. The darker the color, the stronger the trend for each pixel. Each pixel is a cumulative trend that represents an average rate of change from 2007 to 2021 and does not represent annual fluctuations in populations. White circles represent trend estimates that are not significantly different from zero. These are regions where uncertainty about the estimated trend is relatively high (read more about confidence here).

To emphasize where overall population change was greatest, each 27 x 27 pixel is also sized by the estimated relative abundance in 2014, the mid-point of the study period.

The example below highlights how to interpret the Trends maps.

The range-wide view of the Brown Thrasher Trends map shows clear regional differences—the species is increasing in some areas and decreasing in others. In parts of Tennessee, North Carolina, and Georgia Brown Thrashers are increasing, but also note that the circles get smaller north of Athens, Georgia and east of Knoxville, Tennessee indicating lower abundances in the higher elevation portions of the Great Smoky Mountains.

In Oklahoma and Nebraska, Brown Thrasher abundance is decreasing. Again, note that the circles are smaller between Wichita and Dallas indicating lower abundance of Brown Thrasher, especially approaching the edge of the species range near Dallas, Texas.

White pixels on the map indicate uncertainty in the direction of the trend. Because of this uncertainty we do not display the results within the white pixels by default. To see the direction of the uncertain modeled estimates toggle the bar under the trend abundance legend.

Move the mouse over the map to get detailed information including relative abundance trends and confidence intervals. In this example, the pixel is showing a 22% decline over the period of 2007-2021. Mouse over pop-ups also include the 80% confidence interval, or range of values where the model is fairly confident the true trend could be, with it being most likely near the median value. In this example, the confidence interval ranges from -5.5% to -31%, indicating a higher confidence that the true trend at this location is decreasing. The pop-up also provides the value for relative abundance at that location in 2014, making it easier to compare abundances numerically between locations.

How are Trends estimated?

Each eBird participant determines where, when, and how they go birding. Over time not only do eBirders change how they bird, but they become more efficient at detecting and identifying species. These changes make it difficult to estimate population trends because changes in bird populations can be confounded with changes in the participants’ observation process. These confounding factors present a fundamental analytical challenge for estimating trends from eBird data.

To address this challenge, we adopted a novel modeling approach designed specifically to estimate species population trends while controlling for the interannual confounding factors. The approach is based on a statistical framework that uses double machine learning to estimate how participants change their search patterns over the years, and then use this information to isolate changes in species’ population from change in the observation process itself. To learn more about this analytical methodology, see the manuscript.

An important part of generating estimates of population trends is to assess the robustness and reliability of the resulting estimates. We do this in two ways. First, we assess the robustness of the analytical approach for each species by testing it with numerous different types of population change (differences in the direction, strength, and spatial patterns of population change). Good performance across these tests gives us confidence that the method can adapt to a wide variety of realistic types of population change. Second, we estimate the uncertainty of each trend map to assess the reliability of the estimates. Understanding the level of uncertainty is a critical component for interpreting the trend estimates because it provides a guide for separating the signal from the noise, helping to identify where we can be most certain about the estimated change in species populations. Learn more about this analytical methodology in the manuscript.

What are trends “not significantly different from zero”?

Within each 27km x 27km pixel the analysis produces an estimate of the expected trend, represented by a colored dot, and the uncertainty of that estimate, represented by a 80% confidence interval. The 80% confidence interval indicates a range of values expected to contain the true (but unknown) trend that is being estimated.

If an interval contains zero, then there is low confidence in the estimated direction of the trend (positive or negative). By default, circles with low confidence in the estimated direction are shown in white. However, for some applications understanding the full spatial pattern of population change can also be valuable to interpret the trends. To see the trends across all locations toggle the “Show All Trends” in the legend.

Why does the size of the circles vary on the Trends maps?

On the eBird Trends maps, 27km x 27km regions or pixels are represented by circles. To emphasize the important, high abundance portions of species ranges, these circles have been sized according to relative abundance (see here for more about abundance) at the midpoint (2014) of the trends analysis period (2007-2014).

Why are trends available for only one season?

For each species, we’ve chosen the season that was most likely to give us reliable trends estimates for as much of the population as possible. For trends analysis, a significant amount of data is required and the estimates are more reliable when the species is more easily detected. While many species are more detectable when vocalizing during their breeding season, many species breed far from our highest data density areas and would produce poorer quality estimates. In those cases, we’ve chosen to generate estimates for the non-breeding season. While it would be possible for many species to generate reliable trends at both breeding and non-breeding seasons, comparing the estimates (especially the spatial patterns) from both seasons can be complicated and may cause confusion, so we have chosen a single, representative season for each species.

Why are trends not available for all species?

Our goal is to showcase trends for as much of the North American avifauna as our analysis will reliably allow while showing examples from other parts of the world where the volume and longevity of eBird data can be supported by the analysis. We intend to increase the number of species we generate trends estimates annually. This link provides a list of all species with trends available.

Why are trends limited to a portion of a species range?

Our goal is to produce trends for regions where the volume and longevity of eBird data can support the trends analysis. These regions are: North America, Central America, South America, the Iberian Peninsula, India, New Zealand, and South Africa. In many of these regions, such as India and South Africa, adjacent countries and the resulting portions of species’ ranges nearby lack sufficient data for the trends analysis and would perform poorly, degrading the overall trend estimates. Accordingly, we have masked species ranges to the above regions, as these are the regions most likely to currently produce robust estimates of trends. The light gray background polygon shows the full range of the species, including areas where we did not produce trend estimates. See the section on modeled seasonal ranges to learn more.

What are the “modeled seasonal ranges” for eBird Trends?

On Trends maps, the areas inside a species modeled range have a light gray background, while those areas outside of a species modeled range have a dark gray background. The notion of a “modeled range” refers to the range boundary defined in our Range products. Sometimes, the modeled range is incomplete compared to reality and these incomplete areas will be represented with a dark gray background. Within the species modeled range, we have produced trend estimates within particular regions (see list here). For some species, such as the American Kestrel in South America, there is a pale gray background, but no circles representing trends estimates. This happens when our Range products have indicated the species is present, but we have not attempted to estimate trends within that region.

Technical

What modeling methods were used?

The bird observation data that are the backbone of eBird Status and Trends Data Products and visualizations. The total dataset consists of 44 million eBird checklists (sample size) from 14 million unique locations collected from 2007 through 2021 across the hemisphere.

To generate estimates of relative abundance the ecological data science team created state-of-the art statistical and machine learning models (Fink et al. 2019). The models include three classes of predictor variables that account for variation in ecological and citizen science data. The predictors include: (1) five search-effort variables and 12 hourly weather variables to account for variation that affects how well a birder can detect a species in a region if it is present, (2) three variables to account for variation during different time periods, and (3) sixty environmental descriptors from remote sensing data to capture associations of birds with a variety of landscapes across the hemisphere. The search-effort variables are: (1) the time spent searching for birds, (2) whether the observer was stationary or traveling, (3) the distance traveled during the search, (4) the number of people in the search party, and (5) a standardized measurement to account for differences in behavior among eBirders (Kelling et al. 2015; Johnston et al. 2018). The 12 hourly weather variables come from the Copernicus ECMWF hourly reanalysis product (Hersbach et al. 2018) and help account for both an observer’s availability to detect species and how easy a species is to detect. The observation time of the day is used to account for variation in bird behavior throughout the day. The day of the year (1-366) and year on which the search was conducted are used to capture intra- and inter-annual variation. To describe the local landscape where eBirders went birding, variables describing elevation (Becker et al. 2009), topography (Amatulli et al. 2017), shorelines (Carroll et al. 2017; Murray et al. 2019), islands (Sayre et al. 2018), land cover, land use, & hydrology (Friedl & Sulla-Menashe 2019), and nighttime lights (Cao et al. 2014) are included in the model. From these variables we generate a suite of predictors that describe the average feature value and how much it varies across the local landscape, defined here as a 3km pixel.

The statistical model aims to generate accurate predictions of each species’ occurrence and abundance while dealing with the inherent challenges of abundance estimation based on citizen science data (Fink et al. 2019). We use a two-step hurdle model based on Random Forests (RFs) (sensu Johnston et. al. 2015) to incorporate the list of predictor variables (above) while accounting for a large number of observations with zero counts of a species. To alleviate site selection biases, common in citizen science data sets, we randomly subsample the training data to balance spatial and temporal coverage. By including search-effort and weather predictors in the RF models, we can control for important sources of variation in detectability when making predictions. We also case-balanced the first step occurrence model (Chen et al. 2004, Robinson et al. 2018) to improve performance for rare and/or hard to detect species and we calibrate occurrence rate predictions to ensure the probabilistic quality of the resulting estimates (Dormann 2020).

To scale-up the relative abundance base model across global-year-round spatio temporal extents while preserving fine-scale information we use a divide-and-recombine strategy based on the Adaptive Spatio-Temporal Exploratory Model (AdaSTEM; Fink et al. 2013, Fink et al. 2014). The AdaSTEM framework creates and trains an ensemble of spatiotemporally overlapping base models that are subsequently recombined based on shared locations and dates. The ensemble is constructed by partitioning the study extent using a randomly located and oriented spatiotemporal grid. Each partition cell is a spatiotemporal block called a stixel. Each stixel defines the spatiotemporal extent for a single base model that is independently trained using data that falls within that stixel. The stixels’ temporal width is set to 28 days and the spatial stixel dimensions were adaptively sized to generate smaller stixels in regions with higher data density, using QuadTrees (Samet 1984), a recursive partitioning algorithm. To generate independent, overlapping base models, the randomized partitioning process is repeated 200 times and in each partition training data is subsampled. Finally, to make ensemble predictions at a given location and date, the 200 overlapping base model predictions are averaged.

The AdaSTEM-based Status workflow generates five data products: estimates of species’ 1) occurrence rates, 2) abundances, 3) ranges, 4) model validation metrics, 5) habitat importance.

Which eBird data were used to generate the Status and Trends Data Products and visualizations?

Checklists used in Status and Trends Data Products must meet the following conditions to be included in analyses.

Submitted as of 10 February 2022
Observation dates from 1 January 2007 through 31 December 2021
Complete checklists (all bird species detected and identified were included)
The primary checklist in a shared checklist
Checklists that used the generic traveling or stationary protocols (i.e., not incidental protocol)
If traveling checklists, were not longer than 10 kilometers
Not longer in duration than 24 hours
Contained information on: start time, duration, protocol, number of observers, and distance traveled.
Counts of species were available (i.e., not just ‘present’)

See the eBird Help Center for more information about eBird checklists.

What environmental data were used for the Status and Trend Data Products and visualizations?

The analyses used to produce the Status and Trend products rely on matching bird observations with characteristics of the local environment. We used data on elevation, topography, and habitat to describe the local landscapes where eBirders searched for birds. Each checklist location is matched to the environmental data within approximately a 1.5 km radius around the location. For elevation and bathymetry, we calculated the mean and standard deviation of each within the checklist radius using the SRTM15+ product (Tozer et al. 2019). For topography the aspect and slope within the checklist radius were calculated as the mean and the standard deviation (Amatulli et al. 2018). Habitat was described using the MCD12Q1 dataset from NASA and the FAO-Land Cover Classification System, which includes classes for land cover, land use, and hydrology by year. Water cover was described by the Aster Global Water Bodies Dataset, covering 2000-2013 at 30 meter spatial resolution, and with ocean, river, and freshwater categories (NASA/METI/AIST/Japan Spacesystems 2019). Islands were identified using the global shoreline dataset (Sayre et al. 2019) and as of the November 2021 release, continents now have unique identifiers (previously all mainland had the same value). To describe tidal mudflats, we used a high resolution intertidal change dataset (Murray et al. 2019). The land cover, water cover, and intertidal datasets were summarized as the percentage of land cover and edge density within the 1.5 km radius of each checklist. We used the EOG Annual VNL v2 product for nighttime lights, year matching for 2014-2020. The nighttime lights values were calculated as the mean and standard deviation within the checklist radius. We summarized road density for five types using the GLOBIO Global Roads Inventory Project (GRIP) (Meijer et al. 2018). Finally, hourly weather variables have been assigned at 30 km spatial resolution based on the Copernicus ERA5 reanalysis product (Hersbach et al. 2021). Because of the coarse resolution, these were not used as spatial predictors in the estimates for a given 3km grid cell, instead we predicted to a multivariate-optimized set of weather conditions that maximized the occurrence estimate to the 80th percentile within the region and a one-month window.

How are seasons defined for each species? Why are there gaps between seasons?

Breeding and non-breeding season dates are defined for each species as the weeks when the species’ population does not move. For this reason, these seasons are also described as stationary periods. The dates were defined by experts in the status and distribution of birds based on the weekly abundance maps. The selected dates were then checked to make sure that they generally matched expected patterns of phenology for the species.

Migration periods are defined as the periods of movement between the stationary non-breeding and breeding seasons. Note that for many species these migratory periods include not only movement from breeding grounds to non-breeding grounds, but also post-breeding dispersal, molt migration, and other movements. For some species, the transition between stationary and migratory seasons is not clear. Both breeding and non-breeding ranges are often represented within the migratory seasons since some individuals will have arrived in those areas while other individuals of the species are still migrating. In these cases transitional weeks were excluded to provide the clearest picture of individual seasons. For some species, this resulted in seasons that appear shorter than expected, especially when considered within specific regions.

Season dates are defined specifically to be used with eBird Status and Trends Data Products. These dates should not in general be used to delineate the migration and breeding phenology of species, although in many cases Status and Trends dates may approximate these phenological dates. In addition, the dates used for Status and Trends are distinct from the corresponding seasonal dates defined in Birds of the World.

Why are pre-breeding and post-breeding migrations sometimes separated?

Some species have pre-breeding and post-breeding migration seasons combined into a single migratory season. These species (e.g., Magnolia Warbler, Black-throated Gray Warbler) use fairly similar areas for both their migrations. However, some species such as Rufous Hummingbird use different paths for their two migrations. For these species we split the map to show pre-breeding migration (green) and post-breeding migration (yellow) separately. If at least 40% of the area used for one migration season is not covered by the other migration season, then we show them as distinct colors.

What is the difference between “modeled area” and “no prediction”?

On the relative abundance and range maps the light gray shows the “modeled area,” where there was sufficient data to run a model, but the species was predicted to be absent. Sufficient data required there to be, on average, at least 1% spatial coverage of 3 kilometer grid cells within the region for a given week. The dark gray refers to areas of “no predictions” where there was insufficient data to assess whether the species was present or absent. That is, there was less than 1% spatial coverage of 3 kilometer grid cells within the region for a given week.

Why are some islands areas of “No Prediction”?

Some islands have insufficient data to predict whether a species is present or absent (see above in What is the difference between “modeled area” and “no prediction”?). In eBird Status and Trends, we use an island dataset (Sayre et al. 2019) that enables us to distinguish between islands with and without a particular species. The benefit of this method is that the models can, in effect, distinguish the geographic barriers relevant to islands, constraining species to or excluding species from specific islands. The distributions in the Caribbean for both White-winged Dove and Mourning Dove are good examples of this method in action. However, a consequence is that for species that show variation between islands, each island now needs more information before we can make predictions. As a result, many islands that are not frequently eBirded, such as a number of the islands in British Columbia, are often represented as “No Prediction.”

Some of the maps have errors? Why does this happen?

Like any predictive models, the Status and Trends models make errors when predicting species ranges and abundance. When predicting ranges there are two types of errors: predicting species absence in areas that are actually occupied and predicting species presence in areas that are actually unoccupied. There can also be errors in the estimates of relative abundance, with estimates that are higher or lower than the actual counts.

The predictive models used to generate the Status and Trends Data Products account for gaps in eBird data by sharing information from nearby areas. This works well when (1) there is sufficient eBird data to capture patterns of species’ occurrence and abundance and (2) when the environmental data together with the other predictors used in the models do a reasonably good job describing the ecological characteristics that are important to the species.

Error rates generally increase in regions when one or both of the above conditions is not met. First, in regions where the density of checklists is low (e.g. central Canada and the Amazon basin) there is little information to learn patterns of species’ occurrence and abundance. In these areas, incorrect extrapolation can be a risk. For example, species that occur on the coast of Greenland, such as Iceland Gull, are often extrapolated inland, where they likely do not occur, but where there is no data to inform their absence.

Second, error rates also tend to be higher when species’ detection rates are low. Even if there are many checklists in a region, having very few detections of a species limits the amount of information available to characterize the environment that the species is associated with. Third, error rates increase in regions where the environmental data fail to describe important ecological features for each species.

Why do some products show as “Unavailable” or cannot be clicked on?

Species products may be missing for various reasons:

The absence of range, abundance, or trends maps for a species indicates that the model was poor and expert review indicated that the product and/or season should be excluded from visualizations and analysis. See section Some of the maps have errors? Why does this happen? For more information about prediction errors.
The “Percentage of total population in region” and “Percentage of range in region” statistics were excluded for species when experts judged that more than 25% of the range occurred outside the “modeled area.”

Can I download the data?

Data Products are available via our Data Access Request form and can be downloaded either directly through the website or by using the ebirdst R package, which can be used to access, manipulate, and analyze these data.

Trends data are not currently available for download.

If you are looking to use the data primarily in GIS software, you can directly access GeoTIFF files via the Download link at the top right of each species’ page. The download page then lists all downloads available for the species.

Spatial data for range boundaries (for November 2021 release) can be downloaded as Geopackage (GPKG) files. The boundaries are available both as raw (directly from the analysis) and smoothed (as seen in the visualizations) range boundaries. See the Range section below to learn more.

All images of the maps available on the eBird Status and Trends website can be downloaded and used for presentations and display. Please see our Terms of Use when using more than 5 of these.

The complete set of regional abundance and range statistics are available as a CSV file for download.

All downloads are available through or on the Download page. All downloadable visualizations can be used for research and presentations provided they are for non-commercial purposes and properly attributed; see our Recommended citations below.

Recommended citation when using visualizations from the website or the range boundary spatial data: Fink, D., T. Auer, A. Johnston, M. Strimas-Mackey, O. Robinson, S. Ligocki, W. Hochachka, L. Jaromczyk, C. Wood, I. Davies, M. Iliff, L. Seitz. 2021. eBird Status and Trends, Data Version: 2020; Released: 2021. Cornell Lab of Ornithology, Ithaca, New York. https://doi.org/10.2173/ebirdst.2020

Recommended citation when using the data accessed through the ebirdst R package: Fink, D., T. Auer, A. Johnston, M. Strimas-Mackey, O. Robinson, S. Ligocki, W. Hochachka, C. Wood, I. Davies, M. Iliff, L. Seitz. 2020. eBird Status and Trends, Data Version: 2019; Released: 2020. Cornell Lab of Ornithology, Ithaca, New York. https://doi.org/10.2173/ebirdst.2019

How do we ensure that the eBird data are accurate?

There are more than 5,000 automated filters that are active during the submission process for every checklist submitted in eBird. The common filters that may be triggered are for species that are rare for a region and/or season and abnormally high counts of a species for a region and/or season. eBird will ask that additional information (description of the bird(s), counting process, pictures, sound recordings, etc.) be provided for these observations. There are also more than 2,000 eBird reviewers worldwide that examine these checklists as they are submitted. These reviewers work to make sure that each rare sighting is validated before the data from each checklist is available to be used in any analyses.

What happened to the eBird Status and Trends Data Products from previous years?

Previous versions of the Data Products have been permanently archived at the Cornell Lab of Ornithology. If you have accessed and used previous versions and/or may need access to previous versions for reasons related strictly to reproducibility in publications, please contact ebird@cornell.edu and your request will be considered.

Read this if you are thinking of comparing estimates across years: Every year, there are a number of important changes made to the estimates. These changes were made to improve the quality and scope of the Data Products, taking advantage of increased data volume and quality as well as methodological improvements. These changes were not designed to facilitate cross-year comparisons. For this reason, we do not suggest using individual yearly estimates for direct comparisons.

How do the maps match what we know about each species’ biology?

Each species map is reviewed, week-by-week, by species distribution experts. Individual seasons exhibiting serious inaccuracies are rejected. Species distribution experts evaluate the information over the entire range and if most of it is accurate, the species and season combo is retained. Small regions of false positives or false negatives are acceptable (and highlight areas where more data is needed). Any season (for a single season or all seasons for a single species) that is rejected by an expert reviewer is excluded from all Status and Trends Data Products, including the visualizations online and the downloadable data.

Why does an eBird Status and Trends taxonomic concept not match with elsewhere on the eBird website?

eBird Status and Trends products are generated on an annual basis but due to limitations in the availability of remotely-sensed satellite data the products take months to generate. As a result, the taxonomy used for eBird Status and Trends Products is always one year or version behind the current eBird Taxonomy. As a result, there are often discrepancies in taxonomy. The two most common cases are: a) splits and lumps where the new taxonomic entry retains the same name it previously had, but the eBird Status and Trends now represents either multiple species that have been split off (e.g., Eastern Meadowlark included both Eastern and Chihuahuan Meadowlark in the modeling) or multiple subspecies lumped, including ones that were not included in the modeling process for that species (e.g., Common Reed Warbler was modeled as Eurasian Reed Warbler and did not include data for African Reed Warbler), and b) slashes where eBird Status and Trends modeled two taxonomic entries as one previous, species (e.g., Ring-necked/Green Pheasant, where it had been modeled as just Ring-necked Pheasant). For the 2021 version of eBird Status and Trends, the following species have taxonomic discrepancies:

Bicolored Hawk was modeled with data from both Bicolored and Chilean Hawk.
Broad-billed Hummingbird was modeled with data from both Broad-billed and Tres Marias Hummingbird.
Black-throated Mango was modeled without the newly assigned subspecies A. n. iridescens.
Collared Inca was modeled with data from Collared, Green, and Gould’s Inca.
Eastern Meadowlark was modeled with data from Eastern and Chihuahuan Meadowlark.
Eurasian Collared-Dove was modeled with data from Eurasian and Burmese Collared-Dove.
Common Reed Warbler was modeled with data from only Eurasian Reed Warbler, not including African Reed Warbler. No parent taxa representing both exists in the new taxonomy.
Great Horned Owl was modeled with data from Great Horned and Lesser Horned Owl.
Morepork was modeled with data from Morepork and Tasmanian Boobook.
Northern Wheatear was modeled with data from Northern and Atlas Wheatear.
Olive-striped Flycatcher was modeled with data from Olive-striped and Olive-streaked Flycatcher.
Bran-colored Flycatcher was modeled with data from Mouse-gray Flycatcher and Rufescent Flycatcher.
Three-striped Warbler was modeled with data from Three-striped Warbler and Yungas Warbler.
Asian Pied Starling was modeled with data from Indian, Siamese, and Javan Pied Myna all together.
Common Buzzard was modeled without data from Cape Verde Buzzard.
Black-faced Bunting was modeled with data from both Black-faced and Masked Bunting.
Green Imperial-Pigeon was modeled with data from Enggano Imperial-Pigeon.
Wedge-tailed Sabrewing was modeled without data from Long-tailed Sabrewing.
White-rumped Shama was modeled with data from White-crowned Shama.

References

Amatulli, G., S. Domisch, M.-N. Tuanmu, B. Parmentier, A. Ranipeta, J. Malczyk, and W. Jetz (2018). A suite of global, cross-scale topographic variables for environmental and biodiversity modeling. Scientific Data 5:180040.

Becker, J. J., D. T. Sandwell, W. H. F. Smith, J. Braud, B. Binder, J. Depner, D. Fabre, J. Factor, S. Ingalls, S.-H. Kim, R. Ladner, et al. (2009). Global Bathymetry and Elevation Data at 30 Arc Seconds Resolution: SRTM30_PLUS. Marine Geodesy 32:355–371.

Cao, C., F. J. De Luccia, X. Xiong, R. Wolfe, and F. Weng (2014). Early On-Orbit Performance of the Visible Infrared Imaging Radiometer Suite Onboard the Suomi National Polar-Orbiting Partnership (S-NPP) Satellite | IEEE Journals & Magazine | IEEE Xplore. IEEE Transactions on Geoscience and Remote Sensing 52.

Carroll, Mark, DiMiceli, Charlene, Wooten, Margaret, Hubbard, Alfred, Sohlberg, Robert, and Townshend, John (2017). MOD44W MODIS/Terra Land Water Mask Derived from MODIS and SRTM L3 Global 250m SIN Grid V006. [Online.] Available at https://lpdaac.usgs.gov/products/mod44wv006/.

Chen, C. (2004). Using Random Forest to Learn Imbalanced Data. University of California Berkeley 666.

Dorman, C. F. (2020). Calibration of probability predictions from machine‐learning and statistical models – Dormann – 2020 – Global Ecology and Biogeography – Wiley Online Library. Global Ecology and Biogeography 29:760–765.

Fink, D., T. Auer, A. Johnston, V. Ruiz-Gutierrez, W. M. Hochachka, and S. Kelling (2020). Modeling avian full annual cycle distribution and population trends with citizen science data. Ecological Applications 30:e02056.

Fink, D., T. Damoulas, N. Bruns, F. La Sorte, W. Hochachka, C. Gomes, and S. Kelling (2014). Crowdsourcing Meets Ecology: Hemisphere-Wide Spatiotemporal Species Distribution Models. AI Magazine 35:19–30.

Fink, D., T. Damoulas, and J. Dave (2013). Adaptive Spatio-Temporal Exploratory Models: Hemisphere-wide species distributions from massively crowdsourced eBird data. Proceedings of the AAAI Conference on Artificial Intelligence 27:1284–1290.

Fink, D., W. M. Hochachka, B. Zuckerberg, D. W. Winkler, B. Shaby, M. A. Munson, G. Hooker, M. Riedewald, D. Sheldon, and S. Kelling (2010). Spatiotemporal exploratory models for broad-scale survey data. Ecological Applications: A Publication of the Ecological Society of America 20:2131–2147.

Friedl, M. A., D. Sulla-Menashe, A. Tan, N. Schneider, A. Ramankutty, A. Sibley, and X. Huang (2010). MODIS Collection 5 global land cover: Algorithm refinements and characterization of new datasets, 2001-2012, Collection 5.1. IGBP Land Cover, Boston University, Boston, MA, USA.

Friedl, M., and D. Sulla-Menashe (2019). MCD12Q1 MODIS/Terra+Aqua Land Cover Type Yearly L3 Global 500m SIN Grid V006 [Data set]. NASA EOSDIS Land Processes DAAC.

Hansen, M. C., R. S. Defries, J. R. G. Townshend, and R. Sohlberg (2000). Global land cover classification at 1 km spatial resolution using a classification tree approach. International Journal of Remote Sensing 21:1331–1364.

Hersbach, H., B. Bell, P. Berrisford, G. Biavati, A. Horány, J. Muñoz Sabater, J. Nicolas, C. Peubey, R. Radu, I. Rozum, D. Schepers, et al. (2018). ERA5 hourly data on single levels from 1979 to present.

IUCN (2001). IUCN Red List categories and criteria, version 3.1, second edition. IUCN Species Survival Commission.

Johnston, A., D. Fink, W. M. Hochachka, and S. Kelling (2018). Estimates of observer expertise improve species distributions from citizen science data. Methods in Ecology and Evolution 9:88–97.

Johnston, A., D. Fink, M. D. Reynolds, W. M. Hochachka, B. L. Sullivan, N. E. Bruns, E. Hallstein, M. S. Merrifield, S. Matsumoto, and S. Kelling (2015). Abundance models improve spatial and temporal prioritization of conservation resources. Ecological Applications 25:1749–1756.

Kelling, S., A. Johnston, W. M. Hochachka, M. Iliff, D. Fink, J. Gerbracht, C. Lagoze, F. A. L. Sorte, T. Moore, A. Wiggins, W.-K. Wong, et al. (2015). Can Observation Skills of Citizen Scientists Be Estimated Using Species Accumulation Curves? PLOS ONE 10:e0139600.

Meijer, J. R., M. A. J. Huijbregts, K. C. G. J. Schotten, and A. M. Schipper (2018). Global patterns of current and future road infrastructure. Environmental Research Letters 13:064006.

Murray, N., S. Phinn, M. DeWitt, R. Ferrari Legorreta, R. Johnston, M. Lyons, N. Clinton, D. Thau, and R. Fuller (2019). The global distribution and trajectory of tidal flats. Nature 565:1.

NASA/METI/AIST/Japan Spacesystems, and U.S./Japan ASTER Science Team (2019). ASTER Global Water Bodies Database V001 [Data set]. NASA EOSDIS Land Processes DAAC.

Robinson, O. J., V. Ruiz-Gutierrez, and D. Fink (2018). Correcting for bias in distribution modelling for rare species using citizen science data. Diversity and Distributions 24:460–472.

Samet, H. (1984). The quadtree and related hierarchical data structures. ACM Computing Surveys (CSUR) 16:187–260.

Sauer, J. R., J. E. Fallon, and R. Johnson (2003). Use of North American Breeding Bird Survey data to estimate population change for bird conservation regions. Journal of Wildlife Management 67:372–389.

Sauer, J., R., D. K. Niven, J. E. Hines, D. J. Jr. Ziolkowski, K. L. Pardieck, J. E. Fallon, and W. A. Link (2017). The North American Breeding Bird Survey, Results and Analysis 1966 – 2015. Version 2.07. USGS Patuxent Wildlife Research Center, Laurel MD.

Sayre, R., S. Noble, S. Hamann, R. Smith, D. Wright, S. Breyer, K. Butler, K. Van Graafeiland, C. Frye, D. Karagulle, D. Hopkins, et al. (2019). A new 30 meter resolution global shoreline vector and associated global islands database for the development of standardized ecological coastal units. Journal of Operational Oceanography 12:S47–S56.

Sullivan, B. L., J. L. Aycrigg, J. H. Barry, R. E. Bonney, N. Bruns, C. B. Cooper, T. Damoulas, A. A. Dhondt, T. Dietterich, A. Farnsworth, D. Fink, et al. (2014). The eBird enterprise: An integrated approach to development and application of citizen science. Biological Conservation 169:31–40.

Tozer, B., D. T. Sandwell, W. H. F. Smith, C. Olson, J. R. Beale, and P. Wessel (2019). Global Bathymetry and Topography at 15 Arc Sec: SRTM15+. Earth and Space Science 6:1847–1864.

SSS

Table of Contents