← Where animals go
Open animal-tracking corpus · 1988–2026

What we know about
where animals go

A global movement corpus assembled from every openly-licensed tracking repository we could reach, and an honest account of the questions it can, and provably cannot, answer.

338species
18,424animals
4.44Mfixes
38years
7repositories
The map

Where nobody has ever watched

Every cell is 5° of the Earth. Colour is how many species have been tracked through it; grey marks land where no tracked animal has ever been recorded. This is not a map of animals. It is a map of where science has pointed its antennas.

untracked land 1–2 species 3–7 8–19 20+

Coverage counts are a trap: a 5° cell is roughly 550 km across, so a single stork crossing the Sahara once makes it look “covered”. Measured by intensity instead, the picture inverts.

Land zoneCellsMedian speciesMedian fixesWell studied
Tropical >20°C238538153%
Subtropical 10–20°C11971,28773%
Temperate 0–10°C128787871%
Polar <0°C43311126%

The tropics are known roughly 3.4× more thinly than the subtropics. But the truly dark continent is polar: a median of one species and eleven fixes per cell, across the largest land zone on the map.

Findings

Four things the data says

Isotherm-following is universal. Circumnavigation is not.

strong

Animals ride narrow temperature bands at every latitude, but only the Southern Ocean’s unbroken circumpolar current lets one ride a band the whole way around the world. In the north, continents cut the highway short.

SpeciesLongitude roamedThermal bandRealm
Onychoprion fuscatus · sooty tern160°3.4°Ctropical
Thalassarche chrysostoma · grey-headed albatross280°10.1°CAntarctic
Fulmarus glacialis · northern fulmar205°9.6°CArctic
Phoebetria palpebrata · light-mantled albatross355°12.3°CAntarctic
Limosa lapponica · bar-tailed godwit359°19.6°Cland-breeding

The godwit is the counter-case: it roams just as far but through a band twice as wide, because it commutes from Arctic tundra to tropical coast. Ocean specialists follow isotherms; land-breeding migrants cross them.

The white stork is arriving earlier, but not because of warmth.

strong

Our best-sampled species (24 years, 498 animal-years) advances its spring migration by −2.3 days per decade, matching the published literature. Yet its arrival has no relationship to spring temperature on its breeding grounds: r = −0.03. That is flat, not weak. A climate-shaped trend with a non-thermal cause, and this corpus cannot say which one. Of nine species tested, only the lesser black-backed gull showed a clear thermal response (−3.4 days/°C).

The planet has two different migration rhythms.

solid

Northern-hemisphere animals show the textbook double pulse, a spring surge (March–May) and a larger autumn one (September–October, ~20 km/day), with lulls for midsummer breeding and deep winter. Southern-hemisphere animals show no seasonal pulse at all: a steady 30–40 km/day year-round, because that data is dominated by pelagic seabirds and seals that simply never stop.

Records, once the corrupt tracks are removed.

verified subset
Northernmost fix87.0°NVulpes lagopus · arctic fox, on the polar pack ice
Southernmost fix81.8°SDiomedea exulans · wandering albatross
Longest journey362,017 kmDiomedea exulans · roughly the Earth–Moon distance
Runner-up229,791 kmPagophila eburnea · ivory gull
Limits

What this corpus provably cannot answer

This is the unusual part, and the reason to trust anything above it. Each limit below was learned by being refuted, a confident result that fell apart under checking.

Cross-species geography is a map of research funding

Asked where the planet’s migration converges, the corpus answers “Belgium,” the home institute of its single largest data source. Three separate methods (hotspots, co-occurrence, network centrality) each rediscovered the funders rather than the biology. Betweenness centrality was worse than useless: one cell scored 1.1M while containing a single species, because the metric rewards chain-like graph shape, not ecological importance.

A trend is only visible if it beats the spread between individuals

Same species, same years, opposite outcomes. Arrival date varies about 13 days between storks, and the trend is ~5.5 days over 24 years, marginal but detectable. Wintering latitude varies by 70° between individuals, so a 1–3°/decade shift is permanently invisible. Not biology. Variance.

Presence is not knowledge; extremes always lie

Eighty percent of land cells contain a tracked animal, which sounds like broad coverage and means almost nothing. Meanwhile every metric built on a raw minimum or maximum produced a falsehood: a killifish at the North Pole, a pintail credited with 430,000 km, and five phantom “circumnavigators” that were really Bering Strait animals straddling the antimeridian. Percentiles and eligibility gates, not extremes.