Behavioural responses of a trans-hemispheric migrant to climate oscillation

Abstract

Large-scale climatic fluctuations, such as the El Niño-Southern Oscillation, can have dramatic effects on ocean ecosystem productivity. Many mobile species breeding in temperate or higher latitudes escape the extremes of seasonal climate variation through long-distance, even trans-global migration, but how they deal with, or are affected by, such longer phased climate fluctuations is less understood. To investigate how a long-lived migratory species might respond to such periodic environmental change we collected and analysed a 13 year biologging dataset for a trans-equatorial migrant, the Manx shearwater (Puffinus puffinus). Our primary finding was that in El Niño years, non-breeding birds were at more northerly (lower) latitudes than in La Niña years, a response attributable to individual flexibility in migratory destinations. Daily time spent foraging varied in concert with this latitudinal shift, with birds foraging less in El Niño years. Secondarily, we found that in subsequent breeding, a hemisphere away, El Niño years saw a reduction in foraging time and chick provisioning rates: effects that could not be attributed to conditions at their breeding grounds in the North Atlantic. Thus, in a highly migratory animal, individuals may adjust to fluctuating non-breeding conditions but still experience cascading carry over effects on subsequent behaviour.

Keywords: El Niño–Southern Oscillation; behaviour; biologging; climate; migration; seabird.

Figure 1.

A diagram illustrating path analysis correlations between environmental, behavioural and phenological factors. All significant effects are represented in green and non-significant paths in black. Dotted lines indicate negative relationships and path estimates (ß ± s.e.) are given for each significant path.

Figure 2.

The effect of the ENSO index on non-breeding latitude. (a) Variation in mean non-breeding latitude (n = 422) between years with 95% confidence intervals, where each point represents a mean of all individuals. (b) Smoothed January latitude (°) and longitude (°) for all individuals are plotted for a strong El Niño year in orange (2009) (n = 36) and a strong La Niña year in blue (2014) (n = 37). (c) Variation in mean non-breeding latitude with the SOI index [28]. Grey lines connect individuals tracked over multiple years to visualize individual adaptation to varying ENSO conditions. The regression line is derived from the path analysis model. (d) The relationship between the latitude at which the maximum chlorophyll was centred for a given year against mean January latitude. Chlorophyll data were taken from the Aqua-MODIS project [47]. The regression line is derived from the mixed effects model.

Figure 3.

The correlative effects of January latitude on non-breeding and breeding foraging activity. (a) The proportion of the day spent foraging in January plotted against mean January latitude (n = 226). (b) The proportion of the day spent foraging in August during chick rearing plotted against the previous January’s foraging time (n = 87). For both, proportions are derived from foraging hours divided by the available daylight hours at the foraging site.

Figure 4.

The correlative effects of foraging activity on colony visitation and chick peak mass. (a) The proportion of the day spent foraging in August (foraging hours divided by the available daylight hours at the foraging site) plotted against the number of colony visits in year 0 (n = 214). (b) The number of colony visits plotted against chick peak mass for Skomer birds only (n = 63). Regression lines are derived from (a) path analysis and (b) a mixed effects model.

Figure 5.

Correlations between southbound (autumn) and subsequent northbound migration (spring). (a) Northbound/spring migration start date plotted against previous southbound/autumn migration start date (n = 421). (b) Northbound/spring migration end date plotted against northbound/spring migration start date (n = 419). Regression lines are derived from the path analysis model.