Back

Flight crashes have demographic consequences in a long-lived seabird

Barnett, S.; Bull, J. C.; Ross, A. N.; Wanless, S.; Shepard, E. L.

2026-01-09 zoology
10.64898/2026.01.09.698620 bioRxiv
Show abstract

Risk is a key currency in animal ecology, yet studies of risk have almost exclusively focused on predation. Accidents, defined as a momentary loss of control, represent another potential source of injury and mortality. Such events are seldom documented and are generally assumed to be vanishingly rare in natural systems. Nonetheless, regular crash-based mortality has been documented in a population of northern gannets through monthly surveys conducted over three years in the 1970s. We revisit these data, using hindcasting, environmental records and computational fluid dynamics (CFD) models to investigate the environmental drivers of gannet crashes, and matrix population models to examine their demographic consequences under a range of environmental scenarios. Wind direction emerged as the sole predictor of crash-based mortality, with the probability of crashes increasing in north-westerly winds. CFD models revealed that north-westerlies are associated with a marked increase in turbulent kinetic energy along the breeding cliffs, relative to the other modal wind direction, which likely challenges flight control. A total of 367 crashes accounted for 5.4% of annual adult mortality. Removing this mortality led to a projected increase in population of 23.9% over 50 years, equivalent to an additional 10,583 individuals. Overall, this suggests that turbulence close to the substrate can result in fatal losses of flight control that can impact population-level processes. Our results also highlight the need for greater understanding of how airflows influence the risk of accidents across flying animals, both in current and changing wind regimes.

Matching journals

The top 11 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.