Abstract

Disentangling causation from correlation is the foundation of the scientific method. Yet, growing evidence suggests that much observational ecology research has not correctly made this distinction due to inappropriate statistical modelling and unappreciated time-delays. Here, we apply time-lagged causal inference techniques to assess the drivers of seabird declines, using multi-decadal North Atlantic seabird data across the behaviour, mass, survival, reproduction and population size of two species' ecology (Atlantic puffin, Fratercula arctica, and black-legged kittiwake, Rissa tridactyla). We demonstrate that both climate and anthropogenic activity can suppress breeding success and cause population declines. Moreover, population size is specifically impacted by delayed recruitment effects where both species decline after a lag corresponding to their estimated age of first reproduction. These North Atlantic seabirds are therefore at risk from future environmental and anthropogenic changes, as time-delays may result in populations already on an extinction trajectory prior to changes being detectable in their abundance.