The Auk 111(2):285-291, 1994 OLFACTORY BEHAVIOR OF FORAGING PROCELLARIIFORMS CHRISTOPHEVERHEYDEN AND PIERREJOUVENTIN Centred'Etudes Biologiques de Chiz•,Centre National de la RechercheScientifique, 79360 Beauvoir-sur-Niort, France ABSTRACT.--OIfactoryforaging, although very rare among birds, is frequently found in membersof the Procellariiformes;this finding is basedon a small number of field studies using a standardized method (i.e. raft tests). Reactions of seven speciespreviously tested under artificial conditionswere testedagain under natural feeding conditions(fish-oil slicks) to check validity. Concurrently, we comparedthe flight behavior of two groups of species (with and without olfactory capacities) when approaching an odor source. A large-scale experiment was then conductedin pelagic waters to test the reaction of a community of procellariiforms (15 species)to a food-related odor diffusing within a principal feeding area. We observed the same reactions (attraction or indifference) to oil slicks as to test rafts in all speciesevaluated. Results obtained with the standardized method thus hold under natural conditions.Species guided by oilaction approachedthe odor sourceby flying against the wind very dose to (< 1 m) the surface, whereas other speciesapproached from a direction independentof wind direction and from a greater height (>6 m). Thus, specificsearching behavioris associatedwith olfactoryforaging and we found it to be closelyrelated to direction, height, and speed of odor diffusion by wind. Reactionto the odor test varied accordingto families or subfamilies,some taxa showing consistentresponses (attraction or indifference) to several experimentsand some taxa showing conflicting reactions.We obtained some ev- idence that olfactory behavior may differ before and after locating odor sources,as well as vary according to oceanic zones (coastalvs. pelagic). We discussthe hypothesisthat certain speciesrely mainly on visual cues,recognizing and following speciesthat are tracking food- related odors. Finally, we proposesome new ideasabout the evolution of oilaction in birds. Received3 August 1992, accepted25 November1992. ALTHOUGHOLFACTORY sensitivity in birds is from a small number of field studies in the North now documented in a growing number of spe- Atlantic (Grubb 1972), the North Pacific (Hutch- cies (seea review in Waldvogel 1989), olfactory ison and Wenzel 1980, Hutchison et al. 1984), foraging remains unusual. In terrestrial envi- the Antarctic (Jouventin and Robin 1983), and ronments, only two specieshave been shown the South Indian Ocean (Lequette et al. 1989). to locate their food by smell: the Kiwi (Apteryx These studies used a standardized method un- australis;Wenzel 1971) and the Turkey Vulture der artificial conditions (rafts) and showed that (Cathartesaura; Stager 1964). Some others have olfactory capacities are partly related to phy- shown good capacities under artificial condi- logeny and anatomy. Feeding ecology, partic- tions, including honeyguides(Indicator indicator ularly diet and feeding techniques,also play a and 1.minor; Stager 1967) and Black-billedMag- role in olfactory ability (Lequette et al. 1989). pies (Picapica; Buitron and Nuechterlein 1985). Seabirdsusing olfaction employ searchpatterns In contrast, there is evidence that at least 22 that seem to depend on wind conditions procellariiform speciesmay use olfactionwhen (Hutchison and Wenzel 1980, Hutchison et al. foraging (reviewed in Lequette et al. 1989). The 1984). Nonetheless, important questions re- existence of such a widespread ability and the main unsolved. First, can we extend the results extreme development of olfactory structures obtained under the artificial conditions of the (Wood Jones1937, Cobb 1960,Bang 1966, 1971, standardized method to natural feeding situa- Bang and Cobb 1968) make this order unique tions? Second, are there specific searchbehav- among birds, and may explain its successin iors in those specieswhich do rely on olfactory exploiting the pelagic environment. Very few cuesand, if so, how are they related to wind? laboratory studieshave been conducted(Wen- We report the results of field experiments de- zel 1967, Wenzel and Sieck 1972, Jouventin signed to answer these questions,and provide 1977). Our present knowledge comes mainly new data for 15 speciesof procellariiforms. We 285 286 VERH•DENAND JOUVENTIN [Auk, Vol. 111 give details about the behavior of olfactoryfor- represent the main feeding area of most procellari- aging, and describethe role olfaction plays in iforms, unlike coastal waters where most previous the foraging strategiesof these seabirds. studies have been conducted. This choice also per- mitted the exclusion of larids that may attract other seabirds(including procellariiforms) to the yacht (Ha- STUDY AREA AND METHODS ney et al. 1992). We operated only when the boat was Two series of experiments were conducted from sailing close (+ 40ø) to the wind at speedsvarying from 5.3 to 9.7 knots in order to create a linear odor August 1991 to March 1992 on board a 36-m yacht during a cruise from the American to the African trail behind us. Immediately after closing the box, sector of the Southern Ocean (35-64•S, 70øW-10øE). birds flying within a radius of 300 m around the yacht Whole cod-liver oil was used as an odor stimulus since were identified to speciesand counted for 10 min. it has been shown to be attractive in previous field Each period was divided into five circular counts of studies of procellariiforms. 2 min each to reduce the risk of counting the same Fine-scaleexperiment.--The odor stimulus was pre- bird several times. The highest count for eachspecies sented for 15 min by spreading 0.5 L on the ocean in a unit period was retained to compensatefor po- surface,thereby creating isolated surface slicks. Ex- tential underestimates.We separatedbirds flying up- periments were attempted only when the yacht was wind in our wake (within 45ø on either side of the anchored near the shore (500-1,000 m) of an open vesselaxis) from thoseflying in other directions.Con- bay in order to allow prolonged observations,and trol conditions were based on a 10-min count prior only when the wind was blowing from land to the to opening the box, and another count (also for 10 mouth of the bay so as to enhance the diffusion of min) 50 min after the closing of the box. We con- the odor towards the open ocean. Under these con- ducted a total of 26 trials, including 10 around the ditions, the slick tended to float away from the boat ScotiaSea and Drake Passage,and 16 during a cruise (maximum distance 100 m) and closely resembled in the South Atlantic from Tierra del Fuego to Cape Town. slicks frequently observed at sea (oily material orig- inating from dead animals, excreta, krill swarms) Comparisons.--Inboth experiments, we compared where many birds feed. Each bird sighted within 50 groupsof birds (e.g. having vs. not having olfactory m of the slickwas identified to specieswith binoculars capacities),experimental conditions (control vs. test), and followed by eye until it left the spot. Each bird and bird distributions (observed vs. expected) using was counted as a single event, including those that chi-square statistics (when n > 5 in all classes) or sometimesapproached repeatedly. Directions of ap- G-test(when n < 5 at leastin one class).Contingency proach were recordedas upwind approachesif they tests(chi-square with Yate's correction or G-test) were occured within 45 ø on either side of the wind source, applied to 2 x 2 tables.Our basicnull hypothesiswas and as indifferent approach if they did not. Flight that birds were distributed in proportion to the num- heightswere measuredwith graduatedbinoculars and ber of classesthat were compared. assignedto one of three classes(<1 m, 1-6 m, >6 m). Special behaviors such as landing or feeding on the RESULTS slick also were recorded. Control conditions consisted of a period of 15 min before spreadingthe oil, during Fine-scaleexperiment.--When confronted with which all the birds present within 50 m of the future a simulatedfeeding situation(oil slick),the four location of the slick were identified and counted. We speciesthat had previously been shown to re- also took into accountbirds present within 500 m of spond positively with the standard method (Gi- the spotduring the sameperiod. Ambient conditions on the ocean surface, such as wind direction and ve- ant Petrel [Macronectesgiganteus], Antarctic Ful- locity, as well as wave direction and height, were mar [Fulmarusglacialoides], Cape Pigeon [Daption recorded at the beginning of each trial (control + capense]and Wilson's Storm-Petrel [Oceanites odor test). oceanicus])also showed a positive response;their We conducted 16 experiments in several locations: frequency of occurrence(32.8%) and total num- CapeHorn and Diego RamirezIslands (3 tests);South- bers (n = 65) around slicks were, respectively, Shetlands (6 tests); and Western Antarctic Peninsula 5 and 13 times those recorded in the sameplace (7 tests).These locations were chosen for the presence free from oil (frequency = 6.2%, n = 5). This of procellariiform and nonprocellariiformspecies that distribution is significantly different (X 2 = 29.4, we had already tested in previous work using the P < 0.001) from that expectedif birds were not standard method (Lequette et al. 1989). attracted to the slick. Large-scaleexperiment.--The odor stimulus (0.2 L) was kept in an airtight metal box (diameter 12 cm, Three speciesknown as indifferent to odors height 19 cm) fixed on the back of the yacht and was (Kelp Gull [Larus dominicanus],Antarctic Tern releasedin the air for 50 min after opening the box. [Sternavittata] and Imperial Shag [Phalacrocorax Trials were conductedin pelagic waters becausethey atriceps])showed no interest in oil slicks.Their April1994] ProcellariiformForaging and Olfaction 287 frequencies of occurrence (25%) and numbers TABLE1. Flight features(percent) of two groups of (n = 22) around slicks were close to those ob- seabird speciesdiffering in their olfactory capaci- ties and way of approaching an oil slick: (group 1) served under control conditions (frequency = attracted by odor (Daption capense,Fulmarus glaci- 20.8%, n = 16).