Each autumn, several hundred million monarch butterflies emerge from their summer breeding grounds in the northern United States and southern Canada and begin a flight of three thousand kilometres to a small number of mountain forests in central Mexico. The individual butterflies making the journey have never made it before; their parents, grandparents, and great-grandparents were born in the breeding range during the summer and have died without ever leaving it. Yet every autumn, a new generation of adults, which will live for eight months rather than the few weeks of the summer generations, begins the journey and arrives, with remarkable consistency, at the same small area of oyamel fir forest their ancestors four generations back had left the previous spring.

The migration is one of the most studied in insect biology. The butterflies travel in a southwesterly direction during daylight hours, rising on thermal air currents to altitudes above a kilometre and gliding for long distances to conserve energy. They concentrate at well-known stopover sites along the route, sometimes forming roosts of tens of thousands of individuals on a single tree. The final destination, a cluster of mountain forests on the boundary of the Mexican states of Michoacan and Mexico State, was unknown to science until 1975, when it was described by the Canadian biologist Fred Urquhart after nearly forty years of tag-and-recover work by his team and a network of volunteers.

Arriving at the overwintering sites, the butterflies cluster in dense aggregations on the branches of oyamel firs, covering the trees so thoroughly that a single branch can break under the accumulated weight of its occupants. The cold, humid microclimate of the forest allows the butterflies to enter a state of reduced metabolic activity that conserves their energy reserves through the winter. Temperatures are consistently cool but almost never freezing, and the forest canopy provides protection from both strong sunshine and precipitation. In spring, the surviving butterflies mate, fly north to the southern United States, lay eggs on milkweed plants, and die. The three subsequent generations over the summer recolonise the full breeding range, and the fourth autumn generation makes the return journey.

The navigational basis of the migration has been studied in some detail. Like many migratory species, monarchs appear to use a combination of a time-compensated sun compass and the Earth's magnetic field. A butterfly's eye contains specialised ultraviolet-sensitive photoreceptors that can detect the polarisation pattern of skylight, which contains information about the position of the sun even when the sun itself is obscured by clouds. The Canadian entomologist Catherine Lalonde has shown experimentally that disrupting the butterfly's magnetic sense, by placing it briefly in a strong artificial field, disrupts its directional preference for the remainder of the experiment, suggesting that the magnetic component is a significant part of the navigational system rather than a backup.

The fourth generation's physiology differs from that of its siblings. Where the summer generations are reproductive adults with a lifespan of a few weeks, the autumn generation enters a state called reproductive diapause: they do not reproduce until the following spring, and their energy reserves are accumulated as fat for the long flight and winter dormancy rather than spent on current reproduction. The trigger for diapause appears to be the combination of shortening days and cooler temperatures experienced during larval development, which switch the caterpillar's hormonal trajectory onto a different developmental path.

The monarch migration is one of the most celebrated biological phenomena of North America, but it has declined sharply in recent decades. Counts at the Mexican overwintering sites, measured by the area of forest occupied, have fallen by roughly eighty percent since the 1990s. The causes are multiple. Loss of milkweed in the breeding range - driven in part by herbicide-resistant crops that allow weeds, including milkweed, to be cleared from agricultural fields - has reduced the food supply for monarch caterpillars. Climate change has altered the timing of milkweed emergence and the temperature conditions the butterflies require during migration. Illegal logging at the Mexican overwintering sites, although reduced by the establishment of a reserve, has thinned the forest canopy in some areas, removing the microclimatic conditions on which the wintering butterflies depend.

Conservation responses have been varied. Milkweed planting programmes have been organised across the breeding range, with coordinated projects engaging schools, gardens, and roadside plantings. The protection of the Mexican overwintering sites has improved, with the establishment of the Monarch Butterfly Biosphere Reserve in 1986 and subsequent expansions. Scientific monitoring has intensified, with citizen-science networks collecting millions of observations that track the migration in real time. Lalonde has argued, however, that the scale of these interventions has not matched the scale of the threats, and that the monarch's continued decline suggests more ambitious landscape-level measures will be needed.

A particularly difficult problem is that the migratory population is a distinct phenomenon from the species as a whole. Other monarch populations, including a non-migratory population in Florida and several Pacific coast populations that overwinter in California, are in various states of health. The loss of the main eastern migration would not necessarily extinguish the species but would eliminate a biological phenomenon of exceptional richness. Lalonde has described the migration as a cultural heritage of North America, one whose value is not captured by a simple count of individuals. Its continuation into the coming century is not assured, and the measures required to sustain it will depend on coordination among three countries whose policies in other respects are not always aligned.