Predator-prey collapses at the edge of predator distribution: the case of clupeids and common guillemots (Uria aalge) in NW Iberia

Abstract

The spatial structuring of seabird populations makes individual colonies very dependent on regional factors. That is especially the case in small edge populations located far from large colonies. We analysed retrospectively the poorly known collapse, some 50 years ago (around 1962-1973), of a relatively small population of breeding common guillemots (Uria aalge) located at their southernmost limit of distribution in Europe (NW Iberia). We assumed that guillemots behaved locally as facultative specialists in small pelagic fish due to the occurrence of a strong clupeid fishery, and we studied the association between the guillemot collapse and annual regional landings of sardine (Sardina pilchardus) and anchovy (Engraulis encrasicolus), used as a proxy of total stock size. The overall relationship between ln-transformed guillemot counts (May-June) and ln-transformed sardine landings (May-October) was stronger than the fit on untransformed variables (r2 =0.52 vs. 0.27), indicating an exponential relationship between the non-transformed variables. This relationship was somewhat stronger and linear after the collapse, when only a few tens of guillemots remaining (r2 =0.60). A strong regime shift in sardine landings was detected in 1968 and also in anchovy landings in 1969. The overall relationship between guillemot numbers and anchovy landings was linear and strong (r2 =0.72) but completely dependent on the large 1960s estimate of guillemots. However, no relationship was found between guillemot numbers and anchovy landings (April-June) after the guillemot collapse. The most likely period for the guillemot collapse was therefore 1968-1970, as seabird colonies are known to collapse immediately after their staple prey crashes. Local guillemot colonies were not subsequently rescued by immigration and have remained empty or almost empty until present, showing the high sensitivity of edge populations to environmental variability at the regional scale.