Choose and Interpret Bird Distribution, Abundance and Movement Maps

Most bird maps answer only one kind of question. A range map shows where a species occurs in a season. An abundance map, often searched for as a bird density map, shows how many birds a standardized observer would expect to find. A trend map shows how that expected abundance has changed. A migration or movement map shows inferred connections between places across weeks. Pick the map type that matches your question. Then check five things before drawing a conclusion: what the map measures, how it handles observer effort, which area it covers, which dates it reflects, and how uncertain it is.

Match the map to the question

Your question	Map type to use	What it actually shows	Common misreading
Where does this species occur in a given season?	Range or seasonal distribution map	The outer boundary of places with nonzero estimated or reported presence	Treating every point inside the outline as suitable habitat
Where is this species most common?	Modeled relative abundance, such as eBird Status	The expected count on a standardized checklist, by week or season	Reading the value as birds per acre or per square mile
Where have people reported many birds?	Observation-based relative density or recency map, such as those in the Montana Field Guide	Classes of submitted observation totals, or how recently a sighting was reported	Reading it as true population density
Is the species increasing or declining here?	Trend map, such as eBird Trends	Modeled change in relative abundance over a stated season, region and span of years	Comparing two yearly abundance releases instead
Where do birds in one area go next?	Movement model, such as BirdFlow	Probable week-to-week shifts inferred from distribution estimates	Treating a synthetic route as a tracked bird's journey

What a bird density map really measures

Two very different products are often called density maps. Read the legend to see which one you have.

Modeled relative abundance comes from the Cornell Lab of Ornithology's eBird Status and Trends products, which are among the most widely used eBird abundance maps. The 2023 Status estimates do not count birds per unit area. They predict what an expert observer would record on a one-hour checklist covering 2 kilometers under ideal detection conditions. The model produces three quantities. Occurrence is the probability of encountering the species on that standard checklist. Count is the expected number of individuals, given that the species was encountered. Relative abundance is occurrence multiplied by count. Because every estimate uses the same checklist definition, the model controls for observer effort. Values are comparable across places and weeks for the same species. The finest weekly estimates are mapped on 3-kilometer cells.

Observation-based density classes work differently. The Montana Field Guide's relative-density maps sort the totals of submitted observations into classes, and the guide states that these are not population density. The underlying Montana Natural Heritage Program records combine incidental sightings with structured surveys. A high class can therefore reflect heavy reporting as much as a large number of birds. The same guide's recency maps answer a separate question: how long ago the newest observation in an area was reported, shown in age classes.

A worked example of relative abundance

Illustrative example with hypothetical numbers: suppose a cell shows an encounter probability of 0.4 and a conditional count of 3. Relative abundance is 0.4 × 3 = 1.2. That means a standardized one-hour, 2-kilometer checklist there would average about 1.2 individuals, once the checklists that miss the species are included. It does not mean 1.2 birds per square kilometer.

Keep abundance separate from species counts. If you list three different species while planning a visit, your list has three species no matter how many individuals of each you expect. Label that tally as your own species list. It is neither abundance nor any publisher's species-richness product.

Species range maps and seasonal outlines

In eBird Status, seasonal range polygons outline the area where estimated seasonal abundance is above zero, and those outlines may be smoothed. Range polygons from other sources can include unsuitable habitat within their boundaries. A grassland species' outline can therefore cover wetlands, cropland or towns where it would not be expected. Use a range map to judge whether a species is plausible in a region and season. Use abundance or habitat-suitability models to judge where within that region it is likely. If you use a habitat-suitability or combined biodiversity model, read its own metadata. These models are built differently from both range maps and observation records.

Trend maps: reading change correctly

eBird Trends estimates change in relative abundance on 27-kilometer cells. Each estimate covers one season defined for that species, plus a documented region and span of years. Check whether a figure is an annual percentage change or a cumulative change over the whole period, because the two can look very different for the same species. Each estimate comes with an 80 percent interval from the model ensemble. When the interval includes both increases and declines, the direction of change is not settled. Regional trend summaries average the cell trends, weighting each cell by its relative abundance. Cells where the species is common therefore count more than marginal cells.

Do not create your own trend by comparing two Status releases. The 2023 Status products are paired with the 2022 Trends products, and the methods differ between releases. A difference between release years can come from model changes rather than from changes in bird numbers.

Bird migration maps and movement models

Weekly abundance maps from eBird, viewed in sequence, show where a species is concentrated as the seasons change. That is the most common form of an eBird migration map. They show changing distribution, not which individual birds moved where.

BirdFlow goes a step further and infers probable movement between weekly distribution estimates. The routes it generates are synthetic. They are statistical draws, not tracked individuals. The model is memoryless: each next step depends only on the current location, not on where a bird came from. Any errors in the input distribution estimates carry through to the output. BirdFlow cannot resolve daily movements or movements smaller than its grid. Stopover behavior, birds returning to the same sites, and subpopulations that migrate differently are known weak points. Use these routes as hypotheses about connections between regions. They are not evidence that particular birds pass a particular place.

Checks before you draw a conclusion

• Model quality: eBird Status assigns review scores to each species and season. Treat a species-season that failed review as unusable rather than as a rough guide.
• No estimate versus zero: NA means the model made no estimate for that cell, which is different from an estimate of zero birds.
• Uncertainty: eBird abundance estimates include lower and upper ensemble bounds at the 10th and 90th percentiles. Predictive-performance layers at 27 kilometers show where the model predicts well. Keep these beside the estimate instead of reporting the central value alone.
• Effort: confirm whether the map standardizes effort, as modeled abundance does, or summarizes raw submitted observations.
• Dates: note the season definition, the observation years, the release version and the date you retrieved the data.
• Scale: compare the cell size, such as 3 or 27 kilometers, with the size of the place you care about.

Getting bird GIS data

eBird Status and Trends data are available through the ebirdst package for R. Access requires a request for an access key and acceptance of the terms of use, and commercial uses follow a different request process. Researchers who do not use R can download GeoTIFF and GeoPackage files or use an API. For observation records and range or habitat layers from the Montana Natural Heritage Program, read the accompanying metadata for each layer before combining it with modeled data.

Applying maps to northeastern Montana and Medicine Lake

Before you rely on a resource for the Medicine Lake area, fill in a short selection worksheet:

• Question: distribution, abundance, change or movement
• Species or set of species
• Publisher and product name
• Observation record or model
• Metric and its units
• How observer effort is handled
• Season and observation years
• Release date and retrieval date
• Cell size
• Whether the product actually covers northeastern Montana for your species
• Uncertainty measures provided
• Access and use terms

Confirm that your species has a model that passed review for the season you need. Then look at the specific cells that overlap the refuge and surrounding prairie rather than relying on the statewide picture. Even a strong seasonal abundance map tells you where a species is expected under standard conditions. It cannot confirm that the species is present at Medicine Lake right now, and it cannot trace any individual bird's journey. For current presence, use recent observation records and field checks.