A map can make thousands of bird records feel as solid as a census. Dots cluster in cities, colours deepen around wetlands, and whole stretches of countryside stay blank. It is tempting to read the darkest square as “most birds” and the empty square as “no birds.” Neither conclusion follows from the map alone.
A bird record usually tells us that some source reports an observation, specimen or other occurrence associated with a taxon, place and perhaps a date. It does not necessarily tell us how many birds are alive, how hard anyone searched, or what went unreported. Useful maps can still be made from such records. The trick is knowing which question they answer.
I maintain KoreAves, a public reference connecting South Korean bird names with source-backed records. Building it taught me how easily a convincing map can imply more certainty than its data supports.
1. What exactly was counted?
Imagine two parks: one has hundreds of dots, the other only a handful. The first may simply be beside a busy road and visited daily, while the second is rarely surveyed. More reports need not mean more birds.
One bird can generate multiple records across dates and observers. Datasets may contain specimens, photographs, checklists or opportunistic sightings. A map that sums these rows shows reported evidence, not necessarily distinct animals.
Even a checklist count is not automatically a population estimate: it may be one observer’s count on one morning, with birds outside the route never sampled.
The practical question is: What is the unit in the legend: records, checklists, detected individuals, or a statistically modelled estimate? These are different measures.
The distinction has consequences. Reports are excellent for discovering where evidence may be available and for finding source records worth checking. Used without a comparable survey design, they cannot reliably rank places by bird abundance or population trend.
2. Which dates and search efforts lie behind the picture?
For an illustration of why this matters, consider the Oriental Magpie (Pica serica) in a versioned KoreAves release. Its taxon-linked source records and the records eligible for a dated summary are not necessarily identical: some source records lack sufficient temporal information for that view. The difference reflects how records are processed, not a change in the number of birds.
Records excluded from the dated summary do not represent birds disappearing. They mark different processing stages: records mapped to a taxon, then records eligible for a dated summary. “Not suitable for this dated view” must not become “bird absent.”
The KoreAves Oriental Magpie species page shows the dated evidence summary, and the KoreAves methodology explains the project’s broader evidence boundaries. The versioned taxon-release data provides a source-level reference. These are processing summaries, not a claim that KoreAves discovered an ecological trend.
Date quality matters too. The Global Biodiversity Information Facility (GBIF) documents how it interprets occurrence dates and flags invalid or conflicting values. A year-only date is less precise than a day; a multi-month interval cannot become a single-day sighting.
GBIF also explains that time-limited searches exclude records whose known intervals do not fit fully inside the query period. A record spanning 2019–2021, for example, is not safely assignable to a search limited to 2020 alone. Excluding it from that view is a decision about temporal precision, not evidence that the species was absent in 2020.
Search effort matters. A ten-minute hide watch and a three-hour survey offer different chances of detection, even if the species is equally common.
A thought experiment, not measured data: At the same wetland, one volunteer watches for ten minutes and notes three species; another walks for two hours and notes twenty. The difference may reflect search effort, not a sudden gain of seventeen species.
A useful caption should state the date window, known effort and sampling protocol, including what was excluded. Where the source provides no effort information, readers should not assume equal observation coverage between places.
A map can still help locate reports and plan surveys, but it cannot fairly compare abundance simply because every region shares the same colour scale.
3. How precise is “here”?
A pin has coordinates. The bird might not.
Some records have GPS coordinates and uncertainty estimates; others identify only a park, district or municipality. A map may place a broad locality at its centre, but that does not mean the bird was seen at the pin.
Coordinate uncertainty is often expressed in metres around a reported point; its precise interpretation depends on the dataset and the uncertainty definition. Identical decimal precision can therefore represent very different geographic confidence. A map should preserve that uncertainty in its legend, grouping or point style rather than turning every record into an apparently exact dot.
A point or coloured administrative region may show where evidence was assigned, not an exact observation site or a species-range boundary.
GBIF flags zero or invalid coordinates, country mismatches and invalid uncertainty fields. Its data-quality recommendations encourage publishers to retain coordinate precision and uncertainty because these affect what a map can honestly show. The occurrence issues and flags explain the specific checks.
For a reader, the key question is: Is the map showing an exact observation location, a broad area associated with a record, or a modelled geographic estimate?
A blank region may mean no one submitted records, a source was not connected, or records failed date or location rules. Only a properly documented survey can provide evidence of nondetection, and only within its scope and detection limits.
An independent example: what eBird asks its volunteers to record
Cornell Lab of Ornithology’s eBird uses protocols to describe how a checklist was collected. All core protocols require an exact calendar date. For stationary and travelling checklists, observers also record the start time and duration; travelling checklists include the distance covered.
A complete checklist includes every species the observer could identify to the best of their ability, not every individual bird present.
Its Tree Swallow abundance map defines relative abundance as the estimated average count detected during a one-hour, two-kilometre travelling checklist at the optimal time of day for that species. This is a modelled, standardised estimate, not a raw tally of every bird in a place.
The page identifies its input period (2009–2023) and estimate year (2023), details that matter when interpreting the map. This comparison does not make the model a census. Modelled abundance and raw occurrences answer different questions, rely on different assumptions and still require methods and uncertainty disclosures. A statistical surface cannot be reconstructed by adding up sightings-map dots.
A three-question test for any wildlife graphic
Next time you see a striking bird map shared online, try this quick check:
- What was counted? Occurrence records, lists, individual birds observed, or estimates from a defined model?
- When, and with what effort? What dates were accepted, how hard were observers searching, and what observations were left out of the view?
- Where, and how exactly? Are the locations precise observations, approximate areas, sensitive-species generalisations or modelled predictions?
A useful graphic answers these questions in its legend, caption or methods link. Otherwise, treat it as a guide to available evidence, not a stronger claim than its inputs support.
Wildlife maps persuade by simplifying. Their scientific value grows when those limits remain visible: a blank square may invite investigation, while a dark one may reflect attentive observers rather than more birds. Reading what a map actually counts helps us use public biodiversity records more responsibly.
For a related example of how wildlife monitoring informs broader conservation reporting, see Unsustainable’s feature on South Georgia’s wildlife recovery.
Editorial transparency
Jihun Oh is the creator and maintainer of KoreAves, which is discussed as one case study in this article. KoreAves is an independent, noncommercial reference; it is not a field survey and does not produce population estimates.
The submitted manuscript was developed with AI assistance for research, outlining, drafting and editing. Jihun states that he personally reviewed and approved the submitted version, its factual claims, sources and attribution. Unsustainable has made a light editorial pass before publication consideration.
