About the Data Contributing to RAIL Results

RAIL bird lists:

The RAIL bird list results are generated by drawing from observation data in the AKN that has been designated by its owner as Level 2-5 in the AKN data sharing hierarchy.  This includes a combination of public datasets (such as eBird data and Breeding Bird Survey data) and privately owned survey datasets that are stored in AKN. Review the current list of datasets being used to generate your bird list and abundance and presence levels. 

IMPORTANT NOTES:

1. Do the RAIL tool results incorporate new observation/survey data regularly being added to AKN? Datasets feeding this tool are not currently dynamically updated as users add new data into the AKN. Data from datasets with regular contributors in the AKN were last updated in early 2022.  Data updates to the tools that utilize AKN data is on the list of AKN priorities to implement, so it will be a focus of the AKN partners to fund in the near future. eBird datasets were recently updated up until October 2021, and those updates are now reflected in the RAIL tool results.  

2. Bird list returns offshore. If you’re drawing a polygon off the Atlantic coast your results will be generated based on the Northwest Atlantic Seabird Dataset. This is a limited dataset which included at sea transects or particular bird species between 1978 and 2014, so polygons drawn in this area will likely not return a comprehensive list of all the birds that may exist in this area. For additional details about the relative occurrence and abundance of both individual bird species and groups of bird species within your project area off the Atlantic Coast, please visit the Northeast Ocean Data Portal. The Portal also offers data and information about other taxa besides birds that may be helpful to you in your project review. Alternately, you may download the bird model results files underlying the portal maps through the NOAA NCCOS Integrative Statistical Modeling and Predictive Mapping of Marine Bird Distributions and Abundance on the Atlantic Outer Continental Shelf project webpage.

RAIL bird profiles:

The RAIL bird profiles are pulling graphics and information from six different data sources. Below are the list of sources, along with links to the source websites and information about the specific pieces of information being drawn into the tool from those data sources. 

1. PIF Population Estimates Database (Dynamic updates within the RAIL tool as new information becomes available)

Population Estimate Global 

% Population Estimate USA (including upper and lower bounds)

% Population in BCR (including lower and upper bounds)

% Population in State (including lower and upper bounds)

2. Avian Conservation Assessments Database (Dynamic updates within the RAIL tool as new data becomes available)

Conservation Importance

BBS Half Life

Primary Breeding Habitat

Primary Wintering Geography (only if range maps cannot be linked)

Population Trend – Continental

Population Trend – Regional

3. The Cornell Lab of Ornithology’s All About Birds Database (Static database download in June 2021)

Weight

Length

Habitat category (e.g., “Grasslands”)

Food category (e.g., “Insects”)

Nesting category (e.g., “Building”)

Behavior category (e.g., “Aerial Forager”)

Clutch size

Number of broods

Egg length

Egg width

Incubation period

Nestling period

Egg description

4. The Cornell Lab of Ornithology’s McCauley Library (Static photo capture in 2021)

Bird photos

5. Birdlife International (Dynamic updates within the RAIL tool as updated maps become available)

Profile range maps

6. Spreadsheet with supplemental information compiled by USFWS (Static data capture between June 2020 and June 2021 from various sources)

Alpha code

MBTA listing status

Bird of Conservation Concern (BCC) status

BCRs where BCC listed as breeding and non-breeding

Federally threatened or endangered status

State threatened or endangered status (sourced from state wildlife action plans where available)

Breeding Time-frame (BCC only at this time)

How accurate are the bird list results?

Birds that are included on your bird list have been observed within the 10km grid cell(s) your project area overlaps within the past 10 years (for offshore areas there may be more than 10 years of data being referenced to generate lists due to the availability of less data currently in these areas). So, it’s important to note that your bird list may include birds observed outside the area drawn or selected. Additionally, the results are only as good as the underlying data. Because the AKN relies on observation data from various sources, if there is less data in certain areas than others (e.g. because of their remote location or other factors affecting volume of data such as fewer public contributions to eBird lists), then the lists will be less accurate. Therefore, it is the responsibility of the user to look at the sampling effort in the area. This can be done by examining the survey effort associated with the Phenology graphs in the bird profiles. The survey effort is indicated by the black vertical bar). If this bar is absent, this indicates the absence of survey effort altogether.  A high survey effort is the key component. If the survey effort is high, then the probability of presence score can be viewed as more dependable. In contrast, a low or non-existent survey effort bar means a lack of data and, therefore, a lack of certainty about presence of the species. This list is not perfect; it is simply a starting point for identifying what birds have the potential to be in your project area and the information that’s currently available from a number of dependable resources. Please refer to the websites for the other data sources for more information about the data provided via those tools and websites. 

How often are the data and information in the RAIL tool updated?

The data in the AKN that’s driving the results you get when you specify your project area are updated when new data is input into the AKN by an AKN user and shared at the appropriate data sharing level for use in the tool (Level 2-5) and the data is refreshed by the development team. This was last done in early 2022. Some large datasets that are not owned by any specific users (e.g. eBird and BBS) are static datasets that are not updated regularly. To see the range of years covered by these types of datasets, refer to the current list of datasets being used to generate your bird list and abundance and presence levels. 

The data and information being drawn into the bird profiles from other data sources are either static (they are derived from a one-time download of the data on a specific date in the past), or will be dynamically updated whenever those datasets are updated. Refer to the list above to see which data sources will be dynamically updated. The frequency of updates to the dynamically updated data sources will vary and will only occur as that data provider updates their information. For more information about the frequency of data updates, please refer to the webpages provided above for the data resource and contact the data resource owner directly for more information. 

About the Phenology Graphs on the Bird Profiles

The following information provides a description of how the graphs are calculated and how the data should be interpreted and used. The phenology graphs can be very useful in determining how accurate your bird lists in a given area are, or if a particular birds on the list are uncommon in your selected area. 

How Is Abundance Calculated and What Does It Tell Me?
Each blue bar represents the bird’s abundance score during a particular week of the year (A year is represented as 12 4-week months). A taller bar represents a higher abundance score. The survey effort (see below) can be used to establish a level of confidence in the abundance score. One can have higher confidence in the abundance score if the corresponding survey effort is also high.

How is the abundance score calculated? It is done in 2 steps:

Step 1: The abundance for each week is estimated as the total sum of birds detected, divided by the total number of survey events for that week. For example, if in week 12 a total of 45 Spotted Towhees were detected in 5 survey events, the abundance is then 9 birds/survey event.

Step 2: The abundances across all weeks are smoothed to fill gaps from poorly surveyed weeks amidst weeks with good sampling. The resulting smoothed abundances are binned so that all possible values fall in bins of log-base-2: 0, 1-2 (= 2^1, so the graph bar takes value 1), 3-4 (= 2^2, so bar = 2), 5-8 (bar = 3), 9-16 (bar = 4), and so on. Any weeks with abundances > 1024 (i.e., > bar = 10) are assigned value 11. These values from 0 to 11 represent the abundance scores for the species.

Thus, note that the abundance score is really an index of abundance and should not be taken to mean the absolute abundance of the species that week. See “Proper Interpretation and Use of This Report” below to understand how best to use this abundance score for decision-making.

How Is Relative Probability of Presence Calculated and What Does It Tell Me?
Each light green bar represents the bird’s relative probability of presence in the 10km grid cell(s) your project overlaps during a particular week of the year. (A year is represented as 12 4-week months.) A taller bar indicates a higher relative probability of species presence. The survey effort (see below) can be used to establish a level of confidence in the presence score. One can have higher confidence in the presence score if the corresponding survey effort is also high.

How is the relative probability of presence score calculated? The calculation is done in three steps:

Step 1: The probability of presence for each week is calculated as the number of survey events in the week where the species was detected divided by the total number of survey events for that week. For example, if in week 12 there were 20 survey events and the Spotted Towhee was found in 5 of them, the probability of presence of the Spotted Towhee in week 12 is 0.25.

Step 2: To properly present the pattern of presence across the year, the relative probability of presence is calculated. This is the probability of presence divided by the maximum probability of presence across all weeks. For example, imagine the probability of presence in week 20 for the Spotted Towhee is 0.05, and that the probability of presence at week 12 (0.25) is the maximum of any week of the year. The relative probability of presence on week 12 is 0.25/0.25 = 1; at week 20 it is 0.05/0.25 = 0.2.

Step 3: The relative probability of presence calculated in the previous step undergoes a statistical smoothing to fill gaps from poorly surveyed weeks amidst weeks with good sampling. We then rescale the resulting smoothed relative probabilities so that all possible values fall between 0 and 10, inclusive. This is the relative probability of presence score.

To see a bar’s probability of presence score, simply hover your mouse cursor over the bar.

What is Meant by Survey Effort?
Vertical dark blue lines superimposed on probability of presence bars indicate the number of surveys performed for that species in your selected area. The number of surveys is expressed as a range, for example, 33 to 64 surveys. Ranges follow a Log-base-2 scale: 0, 1-2 (2^1 =”event count” of 1), 3-4 (2^2 =”event count” of 2), 5-8 (2^3 =”event count” of 3), 9-16 (2^4 = “event count” of 4), and so on. The last bin (bars of “event count” value 10) represents number of survey events > 1024 (2^10).

Proper Interpretation and Use of The Phenology Graph Reports
Please be aware that these reports provide the “relative probability of presence” and “relative abundance” of birds within the 10 km grid cell(s) that overlap your project; not your exact project footprint and not exact probabilities or abundances. Thus, the bar graphs are designed to depict how much more (or less) relatively common/abundant a species may be relative to other locations and/or times of the year. Results should not be used as proper probabilities of presence or estimates of abundance for a particular location. On the graphs provided, please also look carefully at the survey effort (indicated by the black vertical bar) and for the existence of the “no data” indicator (lack of a survey effort bar). A high survey effort is the key component. If the survey effort is high, then the probability of presence score can be viewed as more dependable. In contrast, a low or non-existent survey effort bar means a lack of data and, therefore, a lack of certainty about presence of the species. This list is not perfect; it is simply a starting point for identifying what birds have the potential to be in your project area and when they might be there. The list helps you know what to look for to confirm presence, and helps guide you in knowing when to implement measures to avoid or minimize potential impacts from your project activities, should presence be confirmed. 

Data Source Attributions and Contacts

The Rapid Avian Information Locator (RAIL) Tool

Attribution: Avian Knowledge Network.  2021. Rapid Avian Information Locator (RAIL). Available at: https://data.pointblue.org/apps/rail/. Accessed on . 

Contact: Eric Kershner, Eric_Kershner@fws.gov

All Other Contributing Sources

Please refer to specific source data attributions within the species profiles.