Anosmia impairs homing orientation but not foraging behaviour in free-ranging shearwaters

Shearwaters deprived of their olfactory sense before being displaced to distant sites have impaired homing ability but it is unknown what the role of olfaction is when birds navigate freely without their sense of smell. Furthermore, treatments used to induce anosmia and to disrupt magneto-reception in displacement experiments might influence non-specific factors not directly related to navigation and, as a consequence, the results of displacement experiments can have multiple interpretations. To address this, we GPS-tracked the free-ranging foraging trips of incubating Scopoli’s shearwaters within the Mediterranean Sea. As in previous experiments, shearwaters were either made anosmic with 4% zinc sulphate solution, magnetically impaired by attachment of a strong neodymium magnet or were controls. We found that birds from all three treatments embarked on foraging trips, had indistinguishable at-sea schedules of behaviour and returned to the colony having gained mass. However, we found that in the pelagic return stage of their foraging trips, anosmic birds were not oriented towards the colony though coastal navigation was unaffected. These results support the case for zinc sulphate having a specific effect on the navigational ability of shearwaters and thus the view that seabirds consult an olfactory map to guide them across seascapes.

. Exploratory statistical analysis between magnetically manipulated birds that retained their magnets (n = 4) until recapture and those that did not (n = 6). The statistical methodology was the same as the main analyses described in the main text.

Nearest neighbour method
The nearest neighbour distance (NND) gives a metric, in km, which is the average spatial distance between two tracks and therefore relates to the fidelity of two tracks. NND between two tracks, a focal track and a comparison track, is calculated by taking in turn each GPS fix on the focal track and finding the nearest (in space) GPS location on the comparison track. The distance between these two nearest neighbours is measured; giving each GPS fix on the focal track an NND to its nearest neighbour. The mean NND for the focal track is then taken as the nearest neighbour distance for that track from the comparison track. The distance varies slightly depending on which track is the focal and comparison track and so for the two tracks, the process is done reciprocally and averaged.

Fidelity of homing route to outbound route within each trip among treatments
To test whether treatment had an effect on birds' propensity to retrace their outbound route as they returned to the colony, we computed nearest neighbour distances (as above) between the outbound and homing sections of each trip. The homing section was defined as for other analyses by our Douglas-Peucker algorithm. The outbound route began at the colony. However, when choosing a point to define as the end of the outbound section, we wanted to avoid foraging behaviour close to the onset of homing whilst capturing the route that the shearwater had taken leading up to its position at the start of homing.
To do this, we chose the first time that the outbound track came within a threshold distance of the position that the bird started homing (shown in figure S2). For the threshold, we chose 10% of the distance between the colony and the start of homing for each trip since most behaviour outside of this threshold is active flight, rather than foraging or resting. However, since this choice is largely arbitrary, we conducted the analysis at 0% (i.e. all track before the start of homing was compared with all track after the start of homing), 5%, 20% and 50%. We chose to only analyse fixes labelled as in the flying state so that average distances were not heavily weighted by where birds chose to rest. Since the data comprised repeated measures from individual birds, we used linear mixed models to test for an effect of treatment on the NND (fidelity) between the outbound and inbound sections of track. The results are shown in table S3.

Repeat trip fidelity within birds
For individual birds that underwent multiple trips during the experiment, we measured the NND pairwise among each bird's trips. For example, a bird which underwent three trips would have the NND for trip 1 and 2, 2 and 3 and 1 and 3 calculated in both directions (the reciprocal) and averaged to give a measure of that bird's repeat trip fidelity. This was done for the outbound sections of trips, from leaving the colony to the furthest point reached from the colony on that trip since this would capture fidelity in foraging destination. This gave within-bird NND measures that could be compared among treatments. Because only 10 birds underwent multiple trips during the experiment, these are exploratory statistics on a small data set (n = 13) but show no evidence of any difference in fidelity across treatments that might have been caused by anosmia. Table S3. Nearest neighbour distances (NND) are shown for repeat tracks from the same birds, and compared among treatments. There was no difference among treatments (GLM: F = 0.10, df = 2,10, p = 0.91) Mean ± s.e.