The Condor IO1 :205-218 0 The Cooper Ornithological Society 1999

MIGRATION ROUTES OF SOOTY IN THE PACIFIC ’

LARRY B. SPEAR AND DAVID G. AINLEY H. T. Harvey & Associates,P.O. Box 1180, Alviso, CA 95002, e-mail: [email protected]

Abstract. During 17 cruises, 1983 to 1991, we recorded flight directions and densities of Sooty Shearwaters(Puf/inus griseus)migrating acrossthe equatorialPacific, between the Americas and 17O”W.Sooty Shearwatersbreed in and in winter (seasons given as “boreal”), migrate to the North Pacific during spring, and return south in autumn. A two-fold increase in numbers seen flying northwest from the Current in spring compared to the number flying southeast on return in autumn, and a six-fold increase in numbers flying southwest towards New Zealand during autumn compared to the number migrating northeast during spring, indicates that many completed a figure-eight route (ca. 40,500 km) each year. This route would involve easterly flight from New Zealand to the Peru Current in winter, northwesterly flight to the western North Pacific in spring, eastward movement to the eastern North Pacific during summer, and southwest flight to New Zealand during autumn. We suggest that most shearwaters using this route are nonbreeders, possibly from both the New Zealand and Chilean populations. Many , probably breeders from both populations, likely use shorter routes to and from the North Pacific (ca. 28,000 to 29,000 km). Annual variation in the number of transcquatorial migrants was positively correlated with a progressive, annual increase in sea-surface temperature, due to large-scale ocean warming in the eastern Pacific. A progressive increase in number of shearwaters migrating to the North Pacific mostly reflected increased migration from the Peru Current, consistent with a concurrent sharp decline of these birds in the California Current. These results indicate a distributional shift in feeding location during the nonbreeding period, from the eastern boundary currents to the central North Pacific, which has been exhibiting a cooling trend. Kev words: El Nitio, migration routes,ocean warming, Pacijc Ocean, griseus, Soot$. y

INTRODUCTION tween Japan and North America (Shuntov 1974, The (PufJinus griseus) is one Gould and Piatt 1993, Warham 1996). Prior to of the most abundant of the Pacific the 1990s very large concentrations summered Ocean, with breeding colonies on islands near in the California Current (Briggs et al. 1987, New Zealand, Chile, and (Everett and Veit et al. 1997). The northward migration oc- Pitman 1993, Warham 1996). The New Zealand curs in March to May, and the southward return population numbers in the millions, including an occurs in September to December. Migration estimated 5.5 million birds in the Snares Islands routes in the Pacific are not well known, al- population alone (Warham and Wilson 1982). though much information was assembled by The only Chilean colony for which a population Shuntov (1974) from studies on seasonal occur- estimate is available (200,000 birds) breeds on rence at-sea, including a lo-year study, 1959- Guafo Island (Clark et al. 1984a), although other 1968, by the Russian Scientific Research Insti- colonies are numerous (Clark et al. 1984b, Ma- tute. One of their findings was that the migration rin 1984). Colonies near Australia are small occurs along a broad front extending 8,000 km (Lane and White 1983); the largest includes an across the between 12O”W and 170”E estimated 2,000 birds (Brothers 1979). (see also King 1967, Pitman 1986). Sooty Shearwaters from New Zealand and It has been suggested that birds from the Chi- Australia (hereafter “New Zealand”) are trans- lean population migrate northward into the equatorial migrants, and during summer (sea- North Pacific along a narrow front off the coast sons refer to the “boreal” time frame through- of the Americas, possibly to as far north as Brit- out), are distributed across the North Pacific be- ish Columbia (Shuntov 1974, Guzman and My- res 1983, Warham 1996). However, this migra- tion route is in question because sightings of this ’ ’ Received 16 May 1998. Accepted 8 January 1999. off Central America are rare (Jehl 1974).

[2051 206 LARRY B. SPEAR ANDDAVID G. AINLEY

(I Tahiti 2o” I 1 P 1 170° 160" 150" 140° 130" 120" 110" 100" 90" 60" FIGURE 1. Studyarea in the equatorial Pacific, including the three designatedsectors. Dots denote noon-day positions during surveysat sea.

Indeed, the possibility that Chilean birds per- the equatorial Pacific to foraging areas in the form transequatorial migrations at all is ques- North Pacific. tionable (Warham 1996) because many feed in We also examined the relationship between the Peru Current during the period of, and just extent of transequatorial movement and the oc- following, the postnuptial migration (March- currence of El Nifio. July; Jehl 1973, Duffy 1981). During at-sea surveys in the equatorial Pacific METHODS from 1983 to 1991 (Fig. l), we gathered infor- STUDYAREAANDSURVEYPROTOCOL mation on migration of Sooty Shearwaters. Our primary objective was to test for differences in The study area included waters lying between flight direction between seasonsand locations to 20”N and the Equator, and from the coast of the provide insight on migration routes used by each Americas west to 17O”W. From 1983 to 1991, population, Chilean and New Zealand. We as- we made 17 cruisesin that area, each lasting one sumed that birds flying northwest had originated to three months. With exception of 1983 (one from the Peru Current, and that flight towards cruise during autumn), cruises were made twice the northeast indicated movement from New each year, one in spring (18 April-27 June) and Zealand. This assumption is valid because (1) one in autumn (22 September-20 November). the two populations are separatedby 8,500 km, During daylight, while the ship was underway, one southeast of our study area and the other we conducted observations from the flying southwest, (2) before and during the postbreed- bridge (14-15 m above sea level), and counted ing migration, Sooty Shearwaters do not occur all Sooty Shearwaters passing within 500-600 in pelagic waters of the temperate to subtropical m (depending on platform height) of the fore- South Pacific that are not on the normal migra- quarter offering the best observation conditions. tion routes (references above; Spear and Ainley, Two observers watched together during 90% of unpubl. data), and (3) they pass quickly across the surveys; one, three, or four observers MIGRATION IN SOOTY SHEARWATERS 207 watched during the remainder. The outer portion will bias analyses. For example, if birds flew of the transect area was scannednearly constant- east and west in equal numbers, unadjusted ly with hand-held 8 to 10X binoculars to detect counts from a ship transiting west would show shearwatersmissed with the unaided eye. greater numbers flying east becausethe observer Behaviors recorded for each sighting were: would count more that were flying east com- (1) resting on the water, (2) foraging (feeding or pared to those flying west. This adjustment is milling over a potential food source), and (3) particularly important for fast fliers, such as flying in a steady direction. For the latter behav- Sooty Shearwaters. We calculated densities for ior, we recorded flight direction to the nearest each transect by dividing the adjusted count by 10”. the number of km* surveyed. Hereafter, all Although surveys were usually conducted shearwater enumerations pertain to adjusted continuously while the ship was underway dur- counts. ing daylight, we divided the survey period into We used log-likelihood G-tests to test for be- 30-min transects. Each half-hour we recorded havioral differences (significance assumed at P ship’s position, speed and course, wind speed < 0.05). We used multiple regression analyses and direction, and sea-surface temperature performed with STATA (STATA Corp. 1995) to (SST). We conducted 1,826.5 hr of surveys dur- test for trends in shearwater densities (the de- ing spring, and 1,660 hr during autumn. The to- pendent variable) as related to year and SST Si- tal area surveyed, calculated for each transect as dak multiple comparison tests (an improved ver- the survey period multiplied by ship speed and sion of the Bonferroni test; SAS Institute 1985) transect width, was 27,282 km* during spring and t-tests were used to compare densities and 25,785 km* during autumn. among sectors and use of different flight direc- We collected 37 Sooty Shearwaters, 11 during tions. The sample unit was one 30-min survey spring and 26 during autumn. Each was transect, each weighted for the area surveyed. weighed, examined for molt, and sexed by ex- We log-transformed density to satisfy assump- amination of gonads. To compare fat loads, L. tions of normality (skewness/kurtosis test for B. Spear scored subcutaneousfat of each after normality of residuals, P > 0.05). Because den- skinning them. Fat was scored as: 0 = no fat or sities (birds 100~ktn*) included values of zero, traces, 1 = light deposit near hind limbs and transformations were calculated as the log (den- abdomen but absent over the pectorals, 2 = light sity + 1). Experimentation with different mod- deposit continuous or mostly continuous over ifications (e.g., log [density + 0.51) demonstrat- the pectoral, 3 = moderate deposit throughout, ed no appreciable effect of choice of modifica- 4 = heavy deposit throughout (see Spear and tions on quantitative outcomes of respective an- Ainley 1998, for validation of this method). alyses. All ANOVAs and t-tests were of the log-transformed density values, although densi- DATA ANALYSES ties presentedin figures are nontransformed data We divided the study area into three sectors: except for figures showing density comparisons eastern, central (western edge of the eastern sec- among years. Normality was not achieved in all tor to 129”59.9’W), and western (13O”W to analyses, but least-squares regression analysis 17O”W, Fig. 1). For each sector, by season and (ANOVA and t-tests) are considered to be very year, our survey effort averaged 1,041 2 758.5 robust with respect to non-normality (Seber km* (n = 51, range 95-3,320 km2). We recorded 1977, Kleinbaum et al. 1988). Although regres- 912 Sooty Shearwaters(see Discussion for eval- sion analyses yield the best linear unbiased es- uation of number recorded relative to the num- timator relating density to independent variables, ber expected). Of these, 783 birds were flying in even in the absence of normally distributed re- a steady direction (migrating), or 429.8 after ad- siduals, P-values at the lower levels of signifi- justing counts for the effect of movement cance must be regarded with caution (Seber relative to that of the ship (Spear et al. 1992; 1977). Therefore, to reduce the chancesof com- using flight speeds for this species given in mitting a Type I error, we assumed significance Spear and Ainley 1997b), with an overall count for ANOVAs and t-tests at P < 0.02. We in- of 558.8 birds. The latter adjustment is required eluded two- and three- polynomials in re- in studies of flight direction from a moving ves- gression analysis to test for curvilinearity. sel because any patterns in bird/ship direction Because Sooty Shearwaters feed little in the 208 LARRY B. SPEAR AND DAVID G. AINLEY

TABLE 1. Behavior of Sooty Shearwaters recorded during at-sea surveys in the equatorial Pacific, with respect to season and ocean sector (definitions given in Methods). “Enroute” denotes actively migrating birds who were flying in a steady direction.

Spnng A”t”lIl”

On water Foraging EnKHlte on water Foraging EIlXXlte S.TUX n (%I n (%I n (%I n (%I n (%) n (%) Eastern 0 (0.0) 6 (37.0) 10.2 (63.0) 1 (34.5) 1 (34.5) 0.9 (31.0) Central 0 (0.0) 13 (21.2) 48.2 (78.8) 3 (9.6) 1 (3.2) 27.1 (87.1) Western 24 (16.2) 11 (7.4) 113.1 (76.4) 61 (20.4) 8 (2.7) 230.3 (76.9) Total 24 (10.6) 30 (13.3) 171.5 (76.1) 65 (19.5) 10 (3.0) 258.3 (77.5) equatorial Pacific (see Results), we did not ex- RESULTS pect them to have responded directly to ocean- ographic conditions in the study area, such as BEHAVIOR AND BODY CONDITION SST, that may relate to food supply (Polovina et al. 1994, Roemich and McGowan 1995). How- During both spring and autumn, 76-78% of the ever, we used SST in the South Equatorial Cur- observations of Sooty Shearwaters involved rent (SEC) as a surrogatefor SST in the eastern birds that were actively migrating (Table 1). boundary currents to the northeast (California However, allocation of behaviors differed sig- Current) and southeast(Peru Current). Like the nificantly between seasons(sectors grouped; G2 SEC, these current systemsare affected strongly = 26.8, P < 0.001). Higher proportions of birds by wind-driven upwelling of cool, deep waters. foraging/feeding were recorded during spring, Processesthat affect upwelling in the SEC are and higher proportions of birds resting on the propagated to the east (then north and south) in water were recorded during autumn. Only 3% of warming events such as El Nifio (Kessler and the shearwaters were foraging during autumn Taft 1987, Philander 1989, McPhaden and compared to 13% during spring. Hayes 1990). Therefore, we confined the regres- During spring, behavior differed between the sion analysis relating shearwater densities to three sectors (Gz = 24.4, P < 0.001, Table 1). SST to that part of the study area lying within The difference resulted from a higher proportion the SEC, i.e., waters between 0” and 4”N (Wyrtki resting on the water in the western sector com- 1966); the purposewas to eliminate confounding pared to a higher proportion foraging in the east- effects that could result from differences in SST in the SEC vs. the North Equatorial Countercur- em and central sectors.Shearwater behavior did rent and North Equatorial Current. The latter not differ significantly between sectors during also lie within the study area, but have warmer autumn (G, = 2.3, P > 0.5). SST than the SEC. We also included “longi- Of the 40 birds recorded as foraging, 5% were tude” in these models to control for the east-to- scavenging dead and 95% were pursuit west increase in SST in the SEC. We tested the plunging for live prey in mixed-species flocks null hypothesis that shearwater abundance did over . not increase with increase in SST, where rejec- Weight and fat load of both males and females tion would indicate that more shearwaters mi- were significantly greater during autumn com- grated to the North Pacific during periods of pared to spring (Table 2). None of the 37 spec- ocean warming in the eastern Pacific. imens were molting body or flight .

TABLE 2. Body mass (g) and fat score (mean ? SE), by season, of female and male Sooty Shearwaters collected in the equatorial Pacific. Comparisons by t-tests. See Methods for scoring fat load. Sample sizes: spring, 2 females and 9 males; autumn, 9 females and 17 males.

Mass Fat score

Sprmg Autumn P Spring Autumn P

Females 603 k 43 778 k 17

FLIGHT DIRECTION 1.0. A The proportion of shearwatersflying in a steady 0.8- direction (i.e., migrating vs. resting on the water and foraging) differed little between seasons 0.6- (Table 1). Therefore, between-season compari- son of densities of birds hying in different di- 0.4- I rections was a valid assessmentof seasonaldif- ferences in the use of different migration routes. Nevertheless, density values used in these ana- ; ::j f , ( : : : lyses are not absolute, because migrants resting Est Cen Wes Est Cen Wes on the water or foraging were excluded. & Northwest Northeast During spring, 87.0% of the shearwaters re- bound bound corded as flying in a steady direction were head- ed northwest (300” to 350”) or northeast (10” to i% 60”). During autumn, 79.9% of the directional B flyers were headed southwest (190” to 240”) or southeast(120 ” to 170”). During spring, density indices of shearwaters 0.6-1 migrating northwest increased from east to west 0.4 -I t (Sidak test comparing densities among sectors, all P < 0.001, Fig. 2A). Densities of those mi- grating northeast increased in a similar way (Si- dak test, P < 0.01, Fig. 2A), except that the ::]Ify , : y , difference between the eastern and central sec- Est Cen Wes Est Cen Wes tors was nonsignificant (P > 0.3). These results Southeast Southwest indicate that birds from the Peru Current flew to bound bound the central and western North Pacific, and that FIGURE 2. Density indices (mean ? SE) of Sooty the New Zealand birds recorded during spring Shearwaters seen flying in different directions in the were migrating to the eastern North Pacific. eastern (Est), central (Cen), and western (Wes) sectors of the eastern tropical Pacific during (A) spring and During autumn, density indices of shearwa- (B) autumn, 1983-1991. Number of transects were: ters headed southwest increased from east to 629, 1,266, and 1,758 for eastern, central, and western west (Sidak test, all P < 0.01, Fig. 2B), indi- sectors, respectively, in spring; 464, 1,026, and 1,830, cating that they were enroute from the eastern respectively, during autumn. Sector locations shown in Figure 1. North Pacific to the vicinity of New Zealand. Density indices of birds flying southeasttowards the Peru Current also increased from east to cantly greater among those headed southwest west, indicating that they had come from the than birds headed southeastin the western sector central or western North Pacific (Sidak test, all (Sidak test, P < 0.001, Fig. 2B), but not the P < 0.01, Fig. 2B). central or eastern sectors(both P > 0.6). When During spring, density indices of shearwaters grouping the sectors, densities of shearwaters compared within each sector were significantly migrating southwest in autumn were 1.8 times greater among those headed northwest than birds greater than those of birds moving southeast headed northeast (Sidak test, all P < 0.02, Fig. (Fig. 3). These results indicate that the majority 2A). In fact, when grouping the three sectors, of birds crossing the Equator in autumn were densities of shearwaters moving northwest in migrating toward New Zealand. spring were 1.5 times greater than densities of The density index for shearwaters migrating birds moving northeast (Fig. 3). These results on a northwest heading from the southeastPa- indicate that, within the study area, the vast ma- cific during spring was 2.0 times greater than jority of birds crossing the Equator in spring that recorded for birds returning on a southeast originated from the Peru Current. heading during autumn (unpaired t6,97,= 6.0, P During autumn, density indices of shearwa- < 0.001, Fig. 3). In contrast, the densities re- ters compared within each sector were signifi- corded for shearwatersmigrating on a northeast 210 LARRY B. SPEAR AND DAVID G. AINLEY

B 0 North bound 1 South bound and 1990 were the only years in which appre- &- 0.61 ciable numbers were seen (Fig. 5). Thus, in- creased movement in the western sector, where most shearwaters were flying northwest (Fig. 2A), was primarily responsible for the overall increase in density with year during spring. During autumn, densities of Sooty Shearwa- ters increased significantly with year in the cen- tral sector (Table 3, Fig. 5). The relationship was curvilinear due to low, stable densities during 1983 to 1988, followed by a marked and pro- Origin:southeast Origin:southwest gressiveincrease during 1989 to 1991. No shear- 5 Pacific Pacific waters were recorded in the central sectorduring FIGURE 3. Density indices (mean + SE) of Sooty autumn 1984, 1985, 1987, or 1988. There were Shearwaters migrating from the southeastand south- no significant relationships between densities west Pacific to the North Pacific, and returning from the North Pacific to respective locations in the South and year in the eastern and western sectors(Ta- Pacific. Indices of shearwatersmigrating from, and re- ble 3, Fig. 5). During autumn, no shearwaters turning to, the southeastPacific are those migrating were recorded in the eastern sector during six of northwest during spring, and southeastduring autumn, the nine years. Thus, the annual increase during respectively. Indices of shearwaters migrating from, autumn was primarily due to increase in the cen- and returning to, the southwest Pacific are those mi- grating northeastduring spring, and southwestduring tral sector, where densities of birds moving autumn, respectively. Density values are not absolute southeastand southwestwere similar (Fig. 2B). values because, when seen, somebirds in migration were sitting on the water, foraging, or flying in other SHEARWATER DENSITIES IN RELATION TO directions, and were not included in the calculations. SEA-SURFACE TEMPERATURE (SST) Samples sizes (number of transects) were 3,653 tran- sects in spring and 3,320 in autumn. There were significant, positive relationships be- tween shearwater densities and SST in the SEC during both spring and autumn (Table 4, Fig. 6). heading from the southwest Pacific during Thus, for a given longitude, densities increased spring, was 12 times less than densities returning with SST. Warmest SSTs occurred during El on a southwest heading in autumn (tG9,, = 8.6, Nifio 1986-1987 and 1990-1991 (El Nifio tim- P < 0.001; Fig. 3). These results indicate that, ing in Trenberth and Hoar 1996, Trenberth within the study area, more birds migrated from 1997), and during non El Nhio 1989 (Fig. 6). the Peru Current to the North Pacific in spring than returned to the Peru Current in autumn, and DISCUSSION the opposite among birds from New Zealand. Our total count of 912 Sooty Shearwaters may seem low, considering the abundance of this ANNUAL TRENDS IN NUMBER OF MIGRANTS species and our claim that we surveyed a large Densities increased significantly with year dur- section of the migration front. However, a more ing both spring and autumn when analyzed accurate perspective is obtained when consider- across sectors (see “All sectors” in Table 3, Fig. ing the shearwater count in relation to the width 4). During spring, the relationship was curvilin- of the transect strip (500-600 m) and the total ear due to a slight decline in density during 1984 area surveyed (53,067 km*), relative to the area to 1986, followed by leveling to 1988, and then over which the shearwatersmigrated (study area an abrupt increase during 1989 to 1991. = 19.25 million km*). Hence, with an average During spring, a curvilinear relationship be- (corrected) density of 0.00827 birds km-* re- tween shearwater densities and year, similar to corded during spring, an estimated average of that of the combined data, was evident in the 3.8 to 5.7 million Sooty Shearwaterscrossed the western sector (Table 3, Fig. 5). There also was study area each spring, assuming that the migra- a significant linear increase in density with year tion period lasted 60 to 90 days (see Introduc- in the central sector, but not in the eastern sector. tion), and that, with an average flight speed of No shearwaters were observed in the eastern 47 km hr-l (Spear and Ainley 1997b), 2.5 days sector during spring 1985 and 1988; spring 1986 were required for a shearwaterto crossthe study MIGRATION IN SOOTY SHEARWATERS 211

TABLE 3. Regression analyses for annual trend in density Higher feeding incidence in eastern and cen- of Sooty Shearwatersmigrating across three sectors in the tral sectors was probably related to the increased equatorial Pacific, 1983-1991. For these analyses, the de- pendent variable was log-density, or log-birds per 100 km2. productivity west-to-east in the equatorial Pacif- Year was analyzed as a continuous term. See Figure 1 for ic (Fiedler et al. 1991). Thus, feeding opportu- sector locations; Figure 4 for sample sizes. nities were likely to have occurred more often in the more eastern regions. The fact that more Regression coefficient * SE F-VdlK! P shearwaters migrated through the western sector where feeding incidence was lowest, also indi- Spring cates that feeding was not a high priority during Eastern sector migration through the equatorial Pacific. year -0.005 2 0.007 0.0 >0.9 Central sector year 0.018 t 0.007 5.8 co.02 MIGRATION ROUTES Western sector Consistent with Shuntov (1974), our results in- year, linear 0.044 2 0.008 29.9 0.2 spring migrants from New Zealand flew to the Central sector year, linear 0.033 -c 0.005 41.1

0.8 - SPRING AUTUMN Eastern 0.6 - 0.6-

0.4 - 0.4- 91 94 0.2 - 0.2 - 61 71 82 185 84 79

14 31 o.o-.f 46l 1 “r9 29 26 21 11 29 0.0 _ 1 1.1 = I / I I I I I I I 84 Si 86 87 88 89 90 91 83 84 85 86 8; ;S S'S ;O ;l

0.8 - 0.8 Central 1 0.6 - 0.6

i

0.4 - 346

0.2 -

0.0 I 1 I I -_84 85 56 87 i& 89 90 91

0.8 - 0.8

0.6 261 121 394 o.4- 1 176 188 262 102 105 383 0.2 - 1 f I I o.o-( , , , , , , , , 84 85 86 87 88 89 90 91 83 84 85 86 87 SS S9 90 91 Year FIGURE 5. Annual density (mean k SE, log-transformed) of Sooty Shearwaters, shown by season and sector. Numbers adjacent to data are sample sizes. See Figure 1 for sector locations. MIGRATION IN SOOTY SHEARWATERS 213

TABLE 4. Regression analyses for the relationship be- population also may have used the “figure- tween Sooty Shearwater density (log-shearwaters per 100 eight” route proposed above for the New Zea- km*) and sea-surface temperature (SST), controlling for longitude. This analysis pertains only to data collected land population. Serventy (1953) suggestedthat within the South Equatorial Current. The sample in units the closely related Short-tailed Shearwater (P. of 30.min survey transects was 1,709 for springand 1,291 tenuirostris) also uses a figure-eight route as a for autumn. All numerator df = 1. Spring:model F2,1706 consequenceof wind regimes (see below), albeit = 16.1,P = < 0.001;Autumn: model F2,t2a8 = 13.0. P a different route than the one we suggestfor the < 0.001. Sooty Shearwater. Movement of Sooty Shearwaters from New Zealand to the Peru Current would not be sur- Spring prising. The Peru Current is one of the world’s SST 0.022 -c 0.008 7.3 co.01 most productive ocean systems (Glantz and Longitude 0.003 2 0.001 25.2

The idea that the Chilean birds migrate north Spear and Ainley (1997a) found that cross- along the coast of the Americas to the California wind flight was preferred by this species,which Current, and then back to the Peru Current also flew regularly into the wind and across (Shuntov 1974, Warham 1996), is not supported headwinds; tailwind and across-tailwind flight by our results. First, the nearly complete absence was avoided. In that earlier study, in which most of Sooty Shearwatersmigrating south along the data were gathered in the California and Peru coast of Central America during autumn indi- Currents where the shearwaters were foraging, cates that few breeders from the Chilean popu- we concluded that headwind flight maximized lation occurred in the California Current. Sec- the chance of detecting prey. Specifically, be- ond, densities of shearwaters moving north in cause air speed for downwind flight must be the eastern sector were low during spring. These greater than wind speed (especially in birds like results indicate that most Sooty Shearwaters Sooty Shearwaters, which must fly fast due to found in the California Current had flown there high wing loading; Pennycuick 1989, Spear and from the western or central North Pacific, and Ainley 1997a), ground speeds obtained when that most Sooty Shearwaters found off the west flying downwind are likely too great to allow coast of North America originate from New Zea- adequate time to detect and capture mobile prey land. on the flight path. Alternatively, the higher These results also indicate that Chilean birds ground speedsobtained during flight acrosstail- that migrated in spring to the central or western winds increases the migration rate over waters North Pacific remained there until autumn (i.e., unfavorable for feeding (such as the ). they did not move to the eastern North Pacific Consistent with this idea, our results indicate before returning to Chile). However, because that headwind flight occurred among migrating subadults remain at sea for several years before Sooty Shearwatersonly during autumn, and pri- returning to the natal colony (reviewed in War- marily among birds of Chilean origin. This, in ham 1996), there is a possibility that some Chi- addition to the fact that the Peru Current is, in lean birds also used the figure-eight route ap- most years, a highly productive system where parently used by the New Zealand population feeding opportunities are usually favorable, (Figs. 7A, B). This would be consistent with fac- might explain the large number of Sooty Shear- tors related to wind regimes (see below). waters who spend the nonbreeding seasonin the Peru Current (Bourne and Dixon 1973, Jehl MIGRATION ROUTES IN RELATION TO WIND 1973, Duffy 198 1, Spear and Ainley, unpubl. DIRECTION data). The headwind flights during the prenuptial Based on wind regimes in the Pacific Basin, and migration in autumn could be facilitated by the using the proposed migration routes, Sooty larger body mass and, thus, wing loading (22.5% Shearwaters migrating north from New Zealand and 19.2% increase among females and males, would be oriented across-wind, or across tail- respectively; Table 2) of breeders, compared to winds (i.e., with the wind on one rear quarter; post-breeders migrating in spring. Higher wing Figs. 7A, C). The same would be true during loading enhances capabilities for headwind spring for birds migrating to the North Pacific flight (Pennycuick 1989). from the Peru Current (Figs. 7B, C). During au- tumn, however, birds returning to colonies near MIGRATION DISTANCE COMPARED WITH Chile would be headed into the wind, or across THAT OF OTHER AVIAN SPECIES headwinds, after crossing the Equator. This The distancesflown by Sooty Shearwatersusing would not be true for birds returning to New the figure-eight route (ca. 40,500 km) would Zealand, which always would be hying with probably differ little from migration distances quartering tail-winds or across-wind. flown by Arctic Terns (Sterna parudisaea), the

FIGURE 7. Suggested migration routes of Sooty Shearwaters from colonies near (A) New Zealand and Aus- tralia, and (B) Chile. Sizes of vectors reflect differences in the number of shearwaters, as suggested by Shuntov (1974; Figs. 17, 18) and densities recorded in this study for various sectors and seasons in the equatorial Pacific (see Fig. 2). (C) Wind regimes of the Pacific Ocean during the two seasonal periods (from Gentilli 1966). MIGRATION IN SOOTY SHEARWATERS 215

Spring - Summer Autumn -Winter

160-W

16oW’ 216 LARRY B. SPEAR AND DAVID G. AINLEY species thought to have the longest migrations pelagic waters of the western and central North of any avian species (e.g., Gill 1995). The Arctic Pacific. This pattern would be consistent with a Tern breeds in the arctic to 80”N (Harrison noted preference of Sooty Shearwaters for cold- 1987), and winters in the Southern Ocean to water areas (Allen 1994, Warham 1996, Oede- about 75’S (Ainley et al. 1984, Gudmundsson et koven 1997), and the compensatory cooling of al. 1992, Stahl et al. 1996). Although migration the western and central North Pacific as the routes are not well known, based on band re- boundary currents warmed (McGowan et al. turns and observations at sea, Arctic Terns 1998). breeding in Scandinavia likely complete a route In conclusion, migration routes of Sooty which, at maximum, includes flight to and from Shearwaters are more complex than previously waters south of Australia and New Zealand (de- suspected, and likely differ depending on pop- tails in Stahl et al. 1996). The round-trip flight ulation, breeding status, and environmental fac- would be about 46,000 km, or 5,500 km (12%) tors. This idea requires further investigation. We longer than the figure-eight proposed for Sooty suspect, however, that the proposed figure-eight Shearwaters. route, which is about 1.5 times longer than the relatively direct routes between the North Pacific MIGRATION IN RELATION TO and New Zealand (round-trip, ca. 29,000 km) or OCEANOGRAPHIC FACTORS Chilean (ca. 28,000 km) breeding colonies (Figs. Shearwater densities were correlated with SST 7A, B), is used primarily by nonbreeders, per- in the SEC, where changes in oceanographic cli- haps from both populations. We also suspect that mate foretell El Nifio conditions in the Peru and the proposed direct routes are used primarily by California Currents (Philander 1989, McGowan breeders. et al. 1998). A very strong El NiIio occurred in The positive correlation between SST and 1990-1991 (Trenberth 1997). In concert, a sig- density of shearwaters migrating to the North nificant increase occurred in the number of Pacific per annum has strong implications re- transequatorial migrant shearwaters. Interesting- garding the effects of large-scale ocean warm- ly, the increase mostly involved migrants from ing, particularly in the context of the four back- the Peru Current, but likely included birds from to-back El Niiios that occurred since 1990 (Tren- the Chilean and New Zealand populations (see berth 1997). The results, indicating a major de- above). The pattern is consistent with that of cline in numbers of Sooty Shearwaters using Sooty Shearwaters in the California Current. historically important foraging areas in the Peru Previously, 3-5 million Sooty Shearwaters sum- and California Currents, also have strong impli- mered in that current system (Chu 1984, Briggs cations regarding possible negative impacts by et al. 1987), but they declined by 90% during the recent, extended period of ocean warming 1985 to 1994, primarily after 1989 (Ainley et al. on populations of species endemic to the eastern 1995, Oedekoven 1997, Veit et al. 1997). The boundary currents. However, adequate monitor- decline also correlated with ocean warming and ing programs for such species are sparse (Cali- decline in (Roemich and McGowan fornia Current: Carter et al. 1992, Oedekoven 1995, McGowan et al. 1996). Our failure to de- 1997; Peru Current: Crawford and Jahncke, in tect a decrease in numbers of transequatorial mi- press.) grants indicates that the decrease in the Califor- nia Current was not due to a major population ACKNOWLEDGMENTS decline. The California and Peru Currents, both east- We thank the staff of the National Oceanic and At- mospheric Administration (NOAA) vessels Discover- em boundary currents, are very productive but er, Malcolm Baldrige, and Oceanographer. Cruises are strongly affected by El Niiio (McGowan were made possible by the Pacific Marine Environ- 1990). The decline of Sooty Shearwaters in the mental Laboratories and Atlantic Marine Oceano- California Current, concurrent with increased graphic Laboratories.We thank Tom Haney for apply- migration from the Peru Current, indicates anal- ing his graphics expertise to Figure 7, and three anon- ymous reviewers for constructive comments. Ronald ogous change in both systems. We pro- Cole, Curator of the Museum of Wildlife and pose that shearwaters have redistributed them- Biology, University of California-Davis, provided in- selves in the Pacific Basin, including a shift from structionsfor scoring fat load. Partial funding was by prevalence in eastern boundary currents to more NSF grants OCE8.515637and OCE8911125,-and Na- MIGRATION IN SOOTY SHEARWATERS 217 tional Geographic Society grants 3321-86 and 4106. GILL, E B. 1995. . W. H. Freeman, New 89. York. GLANTZ,M. H., AND J. D. THOMPSON.(EDS.). 1981. LITERATURE CITED Resource management and environmental uncer- tainty: lessons from coastal upwelling fisheries. AINLEY,D. G., E. E OCONNER,’ AND R. J. BOEKELHEI- John Wiley and Sons, New York. DE. 1984. The marine ecology of birds in the Ross GOULD,I? J., AND J. E PIATT. 1993. Seabirds of the Sea, . Ornithol. Monogr. 32. central North Pacific, p. 27-38. In K. Vermeer, K. AINLEY,D. G., R. R. VEIT, S. G. ALLEN,L. B. SPEAR, T. Briggs, K. H. Morgan, and D. Siegel-Causey AND P PYLE. 1995. Variation in marine bird com- [eds.], The status, ecology, and conservation of munities of the California Current, 1986-1994. marine birds of the North Pacific. Special Publ. Calif. Coop. Ocean. . Invest. Rep. 36:72-77. Can. Wildl. Serv., Ottawa, Canada. ALLEN,S. G. 1994. The distribution and abundanceof GUDMUNDSSON,G. A., T ALERSTAM,AND B. LARSSON. marine birds and in the Gulf of the Far- 1992. Radar observations of northbound migra- allones and adjacent waters, 1985-1992. Ph.D. diss., Univ. California, Berkeley, CA. tion of the Arctic Tern Sterna paradisaea, at the Antarctic Peninsula.Antarc. Sci. 4: 163-l 70. BOURNE,W. R. I?, AND J. DIXON. 1973. Observations of seabirds, 1967-1969. Sea Swallow 22:29-60. GUZMAN,J. R., AND M. T MYERS. 1983. The occur- BRIGCS,K. T., W. G. TYLER,D. B. LEWIS,AND D. R. rence of shearwaters(Pufinus spp.) off the west CARLSON.1987. Bird communities at sea off Cal- coast of Canada. Can. J. Zool. 61:2064-2077. ifornia, 1975-1983. Stud. Avian Biol. ll:l-74. HARRISON,I? 1987. A field guide to the seabirdsof the BROTHERS,N. F! 1979. Seabird islandsNo 73, Tasman world. Stephen Greene Press, Lexington, MA. Island, .Corella 3:55-57. HOWELL,S. N. G., D. G. AINLEY,S. WEBB,B. D. HAR- CARTER.H. R.. G. J. MCCHESNEY.D. L. JAOUES.C. S. DESTY,AND L. B. SPEAR.1996. New information STRONG,h. W. PARKER,J. I?. TAKE&wA,‘D. L. on the distribution of three species of Southern JORY,D. L. WHITWORTH,POINT REYES BIRD OB- Ocean gadfly petrels (Pterodroma spp.). Notornis SERVATORYAND CHANNEL ISLANDS NATIONAL 43:71-78. PARK. 1992. Breeding populationsof seabirdson HOWELL,S. N. G., AND S. WEBB. 1995. Noteworthy the northern and &&al California coasts in bird observationsfrom Chile. Bull. Br. Ornithol. 1989-1991. US Deot. Inter.. Minerals Manap. Club 115:57-66. Ser., Los Angeles. ’ JEHL,J. R., JR. 1973. The distribution of marine birds CHU, E. W. 1984. Sooty Shearwaters off California: in Chilean waters in winter. Auk 90:114-135. diet and energy gain, p. 64-71. In D. N. Nettle- JEHL,J. R., JR. 1974. The near-shoreavifauna of the ship, G. A. Sanger, and P E Springer [eds.], Ma- middle American west coast. Auk 91:681-699. rine birds: their feeding ecology and commercial KESSLER,W. S., AND B. A. TAFT. 1987. Dynamic fisheries relationships. Can. Wild]. Serv., Dart- heights and zonal geostrophic transports in the mouth, Nova Scotia. central tropical Pacific during 1979-1984. J. Phys. CLARK,G. S., A. J. GOODWIN,AND A. P VON MEYER. Oceanogr. 17:97-122. 1984b. Extension of the known range of some sea- KING, W. B. 1967. Seabirds of the tropical Pacific birds on the coast of southernChile. Notornis 3 1: Ocean. Smithson. Inst. Press, Washington,DC. 320-324. KLEINBAUM,D. G., L. L. KUPPER,AND K. E. MULLER. CLARK,G. S., A. I? VON MEYER, J. W. NELSON,AND J. 1988. Applied regressionanalysis and other mul- N. WAN. 1984a. Notes on Sooty Shearwatersand tivariable methods. PWS-KENT Publishing, Bos- other avifauna of the Chilean offshore island of ton. Guafo. Notornis 3 1:225-23 1. LANE,S. G., ANDG. WHITE. 1983. Nesting of the Sooty CRAWFORD,R. J., AND J. JAHNCKE.In press. Compari- Shearwater in Australia. Emu 8:117-l 18. son of trends in abundanceof guano-producing MARIN, A. M. 1984. Breeding record for the Sooty seabirds in Peru and southern Africa. South Af- Shearwater (Pufinus griseus) from Chiloe Island, rican J. Marine Sci. Chile. Auk 101:192. DUFFY, D. C. 1981. Seasonal changes in the seabird fauna of Peru. Ardea 69:109-l 13. MCGOWAN,J. A 1990. Climate and change in oceanic ecosystems:the value of time-series data. Trends EVERETT,W. T, AND R. L. PITMAN.1993. Status and conservationof shearwatersin the North Pacific, Ecol. Evol. 5:293-299. p. 93-100. In K. Vermeer, K. T Briggs, K. H. MCGOWAN,J. A., D. R. CAYAN, AND L. M. DORMAN. Morgan, and D. Siegel-Causey [eds.], The status, 1998. Climate-ocean variability and ecosystemre- ecology, and conservation of marine birds of the sponse in the northeasternPacific. Science 281: North Pacific. Special Publ. Can. Wildl. Serv., Ot- 210-217. tawa, Canada. MCGOWAN,J. A., D. B. CHELTON,AND A. A. CONVERSI. FIEDLER,I? C., V. PHILBRICK,AND E I? CHAVEZ.1991. 1996. patterns, climate, and change in Oceanic upwelling and productivity in the eastern the California Current. Calif. Coop. Ocean Fish. tropical Pacific. Limnol. Oceanogr. 36: 1834- Investig. Rep. 37:45-68. 1850. MURPHY,R. C. 1936. Oceanic birds of South America. GENTILLI,J. 1966. Wind principles, p. 989-993. In R. Macmillan, New York. W. Fairbridge [ed.], The encyclopedia of ocean- OEDEKOVEN,C. S. 1997. Seabird abundancesand dis- ography. Reinhold, New York. tribution in the California Current in relation to 218 LARRY B. SPEAR AND DAVID G. AINLEY

changes in the marine environment, 1985-1997. SPEAR,L. B., AND D. G. AINLEY. 1998. Morphological M.Sc. thesis, Univ. Bonn, Bonn, Germany. differences in relation to ecological segregation PENNYCUICK,C. J. 1989. performance. Ox- among petrels (: Procellariidae)of the trop- ford Univ. Press, New York. ical Pacific and Southern Ocean. Auk 115:1017- PHILANDER,S. G. 1989. El Nifio, La Nina, and the 1033. SouthernOscillation. Academic Press, New York. SPEAR,L. B., D. G. AINLEY,AND S. WEBB. 1995. [Ab- PHILLIPS,J. H. 1963. The pelagic distribution of the stract]. Distribution and abundance of Buller’s, Sooty Shearwater.Ibis 105:340-353. Chatham Island, and Salvin’s off PITMAN,R. L. 1986. Atlas of seabird distribution and Chile and Peru: potential interactionwith longlin- relative abundancein the eastern tropical Pacific. et-s.Proc. First Int. Conf. on the Biology and Con- Admin. Rep. LJ-86-02C. Southwest Fisheries servation of Albatrosses,28 August-l September Center, LaJolla, CA. 1995, Hobart, Australia. PITMAN,R. L., AND L. T. BALLANCE.1992. Parkinson’s SPEAR,L. B., N. NUR, AND D. G. AINLEY. 1992. Esti- Petrel distributions and foraging ecology in the mating absolutedensities of flying seabirdsusing eastern tropical Pacific: aspects of an exclusive analysesof relative movement. Auk 109:385-389. feeding relationship with . Condor 94: STAHL,J. C., J. A. BARTLE,P JOUVENTIN,J. I? Roux, 825-835. AND H. WEIMERSKIRCH.1996. Atlas of seabirddis- POLOVINA,J. J., G. T MITCHUM,N. E. GRAHAM,M. I? tribution in the south-west . Centre National de la Recherche Scientifique. Villiers en CRAIG,E. E. DEMARTINI,AND E. N. FLINT. 1994. Bois, . Physical and biological consequencesof a climate STATA CORPORATION.1995. STATA reference manu- event in the central North Pacific. Fish. Oceanogr. al. Release 3.1. 6th ed. Stata Corn.. College Sta- 3:15-21. tion, TX. ROEMICH,D., AND J. A. MCGOWAN. 1995. Climate TRENBERTH,K. E. 1997. The definition of El Nina. J. warming and the decline of zooplankton in the Am. Meteorol. Sot. 78:2771-2777. California Current. Science 267:1324-t 326. TRENBERTH,K. E, AND T. J. HOAR. 1996. The 1990- SAS INSTITUTE.1985. SAS user’s guide: statistics,5th 1995 El Nitio-Southern Oscillation event: longest ed. SAS Institute Inc., Cary, NC. on record. Geophys. Res. Lett. 23:57-60. SEBER,G. A. I? 1977. Linear regressionanalysis. John VEIT, R. R., J. A. MCGOWAN,D. G. AINLEY, T R. Wiley and Sons, New York. WAHL, AND P PYLE. 1997. Aoex marine nredator SERVENTY,D. L. 1953. Movements of pelagic sea-birds declines ninety percent in associationwith chang- in the Indo-Pacific region. Proc. 7th Pacific Sci. ing oceanic climate. Global Change Biol. 3:23- Congr. 4:394-407. 28. SHUNTOV,V. E?1974. Seabirdsand the biological struc- WARHAM,J. 1996. The behaviour,population biology ture of the ocean. [Translatedfrom Russian.] US and physiology of the petrels. Academic Press, Dept. Comm., Springfield, VA. New York. SPEAR,L. B., AND D. G. AINLEY. 1997a. Flight behav- WARHAM,J., AND G. J. WILSON.1982. The size of the iour of seabirdsin relation to wind direction and Sooty Shearwater population at the Snares Is- wing morphology. Ibis 139:221-233. lands, New Zealand. Notornis 29:23-30. SPEAR,L. B., ANDD. G. AINLEY. 1997b. Flight speed WYRTKI,K. 1966. Oceanographyof the eastern equa- of seabirdsin relation to wind speedand direction. torial Pacific Ocean. Oceanogr.Mar. Biol. Annu. Ibis 139:234-251 Rev. 4:33-68.