SHORT COMMUNICATIONS 153

The Condor 101:153-156 0 The Cooper Ornithological Society 1999

AUTUMN MIGRATION SPEED OF JUVENILE REED AND SEDGE WARBLERS IN RELATION TO DATE AND FAT LOADS’

STAFFAN BENSCH Department of Ecology, Ecology Building, Lund Universit;v, S-223 62 Lund, . e-mail: [email protected]

Bo NIELSEN Linderotksgatan 28, 703 67 iirebro, Sweden

Abstract: We analyzed speed of migration in two fattening, and the quality of these fattening sites do congeneric warblers, the Reed and the , not deteriorate in the course of the season as much as Acrocepkalus scirpaceus and A. sckoenobaenus. Sedge is the case for the aphid sites used by Sedge Warblers Warblers migrated at a higher speed than Reed War- (Bibby and Green 1981). Because Sedge Warblers blers. The two showed similar rates of fat dep- seem more time stressed, we expected them to migrate osition at our Swedish study site, although Sedge War- at a higher speed than Reed Warblers. blers departed with lower fat loads. The higher speed Apart from the difference in diet, the two species of migration in Sedge Warblers and their lower depar- exhibit other ecological differences that might be ture fat loads suggest that they encounter stopover sites linked to different migratory adaptations: Sedge War- which offer higher relative fat deposition rates farther blers lay more per clutch (Bibby 1978) and have south. The amount of visible fat at the time of banding a shorter life-snan (Peach et al. 1990) than Reed War- was positively related to the speed of migration. Esti- blers. The two’studies that used banding recoveries did mates of speed of migration for the two species sug- not confirm statistically that Sedge Warblers migrated gest that the recoveries were situated on average 76- at a higher speed than Reed Warblers, although both 111 km farther south per increase in fat score, corre- studies reported higher average migration speeds in sponding to 58-85% of the expected distance a Sedge Warblers (Hildtn and Saurola 1982, Ellcgren can cover by using the fuel of one unit of fat score. 1993). Here, we use banding recoveries and employ mul- Key words: schoenobaenus, Acro- tiple regression analyses to examine whether Reed cephalus scirpaceus, fat loads, Reed Warbler, ringing Warblers and Sedge Warblers show different speeds of recoveries, Sedge Warbler, speed of migration. migration. In contrast to earlier studies, WC USCa more homogeneous data set because we limit the analyses For migratory , the speed of migration is set by to juvenile birds banded at one site in south central the total time it takes to cover the distance between Sweden. We also examine whether fat load at the time summer and winter destinations, including stopover of banding is correlated with the speed of migration. time. Speed of migration is expected to vary between different ecological situations depending upon food METHODS availability and whether the bird is minimizing the We used banding recoveries of birds banded at Kvis- time, total energy expenditure, cost of transport, or risk mare Bird Observatory (59”10’N, 15”25’E) between of predation (Lindstriim and Alerstam 1992). 1982 and 1995. Each year the banding was conducted Reed Warblers (Acrocepkalus scirpaceus) and between late June and late September at two to four Sedge Warblers (A. sckoenobaenus) are two closely different sites in the Kvismaren-Hjalmaren area (max- related species with similar morphology and geograph- imum distance between sites 15 km). Birds were aged ical distribution of both breeding and winter quarters according to Svensson (1992). The amount of visible (Cramp 1992). It has been suggested that Sedge War- fat in the furcular cavity and belly was scored on a blers migrate from Northern at a higher speed scale between 0 and 6 (Pettersson and Hasselquist than Reed Warblers in order to reach important sites 1985). for fattening (Bibby and Green 1981). At these sites, In total, 20,661 juvenile Reed Warblers and 8,354 situated north of the , high abun- juvenile Sedge Warblers were banded. Our banding ef- dance of reed aphids enable Sedge Warblers to build fort resulted in 111 and 36 recoveries, respectively, up fat stores sufficient for nonstop flights to tropical during the birds’ first autumn migration through Eu- (Bibby and Green 1981). In contrast, Reed War- rope. In order to exclude birds not yet migrating, we blers, which feed on Diptera, use areas closer to the only included birds recovered south of the banding including southern Europe and north Africa for site, at a distance of at least 50 km and with a migra- tion speed exceeding 10 km day-’ (Hilden and Saurola 1982, Ellegren 1993). We also restrict analyses to birds I Received 29 December 1997. Accepted 24 Sep- that were scored for fat loads at the time of banding. tember 1998. This reduced data set consists of 85 Reed Warbler and 154 SHORT COMMUNICATIONS

3 1 Sedge Warblers. Most recoveries (95%) involve re- vation probably arises due to two circumstances capturesby other banders. (Fransson 1995). Many birds banded early in the sea- Statistical tests were done in SYSTAT (Wilkinson son may not have startedmigration and will thus con- 1992), and all P-values are two-tailed tests. When we tribute to the initially low recordedspeed of migration. applied multivariate statistics,residuals were approxi- Late in the season,on the other hand, many birds will mately normally distributed. We used distance from be capturedwith filled fat loads, and if recoveredsoon the banding site as the dependentvariable and date of after banding, they may have covered a considerable banding and recapture as independent variables. Re- distance. The recorded asymptotic speed of migration sidual distance should hence be analogousto speed of will be reached when one or several full migratory migration. We chose to use distance as the dependent episodes (stopover time plus migratory flight) have variable rather than speed becausethe latter is a ratio been completed. (distance/time) showing nonlinear relations with both The multiple regressionmodel described above can banding and recapturedate. If not otherwise stated,we be used to control for the effect of banding and recap- report mean ? SD. ture date on how far south the birds were recovered (residual distance). By adding fat score to this model, RESULTS we can examine whether the amount of fuel the birds The average speed of migration was 55.3 2 30.7 km carry at the time of banding may further explain the day-’ for Sedge Warblers (n = 31) and 38.8 ? 24.4 distance to the recovery site. For Reed Warblers, fat km day-’ for Reed Warblers (n = 85), and the differ- score significantly increased(P = 0.03) the proportion ence is significant (ANOVA, F, ,,4 = 8.94, P < 0.01). of variance explained (R* = 0.47 for the model in- The two species did not differ ‘in date of banding (7 cluding fat score). The slope of the regression(mean versus 11 August, t-test, t,,, = 1.24, P > 0.2) or in the 2 SE) suggeststhat the distance to the recovery site number of days elapsed between banding and recap- increasedby 76.0 + 34.1 km per fat score. Similarly, ture (20.1 +- 8.7 versus 20.5 2 9.4 days, t,,, = 0.19, when we added fat score of Sedge Warblers to the P > 0.5). However, the distance to the recovery site multiple regressionmodel, it tended (P = 0.07) to in- differed between species (Sedge Warblers: 1,078 2 crease the explanation of the variance in distance (R* 607 km versus 770 5 519 km for Reed Warblers, t,,, = 0.59 for the model including fat score). The slope suggeststhat the birds were recovered 110.9 + 59.3 = 2.71, P < 0.01). For both Sedge and Reed Warblers, we found that km farther south per fat score. the date of banding and the date of recovery had sig- DISCUSSION nificant effects on the distanceto the recovery site (all Ps < 0.001, based on multiple regressionanalysis). In SPEEDOF MIGRATION neither of the specieswas the interactionterm between The present study confirms the suggestionthat Sedge date of banding and date of recovery significant, sug- Warblers migrate at a higher speedthan Reed Warblers gesting that the effect of recovery date on distance is (Bibby and Green 1981). The reason why Ellegren the same, independentof when the banding was done, (1993) did not find a significant difference in speed of and vice versa. migration between the two specieseven though he an- In Reed Warblers, the distance to the recovery site alyzed a larger data set, which partly included the one decreasedby 32.7 km day-’ of banding and increased we examined here, might be that he combined recov- by 37.8 km day-’ of recovery. In other words, for a eries from several banding sites. This possibly increas- given recapturedate, birds will on average be encoun- es the variance of the means, becauselocales may dif- tered 32.7 km closer to the banding site for each day fer in the proportion of birds molting/staging and mi- later in the seasonthey were banded. Similarly, for a grating. given banding date, birds will on average be encoun- In general, the speed of migration increaseswith the tered 37.8 km farther south of the banding site for each migratory distance (Ellegren 1993). At Kvismaren, ju- day later in the seasonthey were recaptured.The cor- venile Sedge Warblers have a more restrictedfirst pre- responding values for Sedge Warblers were 43.4 and basic molt than Reed Warblers which allows them to 63.1 km day-‘, respectively. leave the breeding ground at an earlier date (Hall The variables banding date (ANCOVA; F,,,,, = 1996). The difference in speed of migration between 60.7, P < O.OOl), recapture date (F,,,,, = 76.9, P < birds banded late in the season(Fig. l), suggeststhat 0.001) and species(F ,,,,0 = 4.67, P < 0.05) were sig- the observed higher speed of migration in Sedge War- nificant, as was the interaction between species and blers is not a consequenceof a later start of migration recapturedate (F,,,,, = 4.86, P < 0.05). This supports in Reed Warblers. How do Sedge Warblers accomplish the conclusionthat Sedge Warblers migrated at a high- migration at a higher speed than Reed Warblers? er speed than Reed Warblers. The flight speeds of the two species are probably Figure 1 graphically depicts the relation between very similar because they differ -very little-in wing predicted speed of migration and date of recovery for mornholonv and mass (Hedenstrom and Moller 1992). birds banded at three different dates. For both species, the most i;portant parametersdetermining flight speed the speed of migration seems to increase the later the (Pennycuick 1989). Given that the two species are bird has been banded. Birds banded early have an ini- equally good at selecting favorable winds, difference tially low speed of migration which later increasesun- in speed of migration is likely to reflect differences in til it reaches an asymptotic speed. Conversely, birds time spent at stopover sites (Lindstrom and Alerstam banded late have an initially high speed of migration 1992). There are at least two mechanismsby which which later drops to an asymptotic level. This obser- total stopover time might be shorter in Sedge than in SHORT COMMUNICATIONS 155

100 I I I I a

80

L 60 $ E Z $ 40 ln w 20

a0

0 5 15 25 35 45 Day from ringing

FIGURE 1. The numerical solutions of the models (see text) showing the relationship between speed of migration and number of days between banding and recovery for (a) Reed Warblers and (b) Sedge Warblers banded at three different dates (22 July, 11 August, and 31 August). The relationships have been obtained by dividing the expected values of models by time between banding and recapture.

Reed Warblers: Sedge Warblers may be more efficient Warblers if their stopover sites south of Kvismaren al- in finding good feeding sites and, having found a feed- lowed them to keep a relatively higher fat deposition ing site, may have a higher rate of fat deposition than rate. Both the time-minimization hypothesis and total Reed Warblers. energy minimization hypothesis (Lindstrom and Aler- The first mechanism is difficult to test because it stam 1992) predict that Sedge Warblers should leave requires knowledge of how much time it takes until a Kvismaren with lower fat loads than Reed Warblers if bird has found the site at which it stays for fat depo- they expect to encounter relatively higher fat deposi- sition. However, because the two species use similar tion rates farther south along the migratory route. In habitats for stopover (Cramp 1992), there is no reason agreement with these two hypotheses, Sedge Warblers to expect that it would take longer for Reed Warblers depart from Kvismaren with lower fat loads than Reed to find a suitable stopover site. Warblers (Hall 1996). Unfortunately, there are only At our Swedish study site, Reed and Sedge Warblers limited data available on fat deposition rates along the appear to show the same rate of body mass change, migratory route of Swedish Reed and Sedge Warblers. indicating a similar fat deposition rate (Hall 1996). However, Sedge Warblers of Scandinavian origin do Hence, data on fat deposition rates at Kvismaren can- not put on fat in northern (Basciutti et al. 1998), not explain the difference in speed of migration be- suggesting that the presumably important areas for fat- tween the two species. Sedge Warblers would manage tening could be located farther south in Europe or in to maintain a higher speed of migration than Reed North Africa. 156 SHORT COMMUNICATIONS

FAT LOADS passerinesin West Africa-an adaptation to the In general, birds with large fat loads are capable of &coming dry season?Ibis 133:47-52. flying longer distancesthan birds with smaller fat loads BIBBY. C. J. 1978. Some breeding statistics of Reed (Pennycuick 1989). At present,there are no data avail- and Sedge Warblers. Bird Sturdy25:207-222. able on how our fat scores relate to mass of fat in BIBBY, C. J., AND R. E. GREEN.1981. Autumn migra- Acrocephalus warblers. For Willow Warblers (Phyllos- tion strategiesof Reed and Sedge Warblers. Omis copus frochilus) in autumn, one unit of fat score cor- Stand. 12:1-12. respondsto 0.26 g of fat (Lundgren et al. 1995). Lean CRAMP,S. 1992. The birds of the Western Palearctic. massof Reed and SedgeWarblers at Kvismaren is 10.8 Oxford Univ. Press, Oxford. g for both species (Hall 1996), whereas lean mass of ELLEGREN,H. 1993. Speed of migration and migratory Willow Warblers is 7 8. Bv scaling LID.one unit of fat flight lengths of passerinebirds ringed during au- score in the Acroceph&s.warbleri &uld correspond tumn migration in Sweden. Ornis Stand. 24:220- to 0.40 g of fat. One problem with converting fat 228. scores to actual fat is that the relationship is likely to FRANSSON,T 1995. Timing and speed of migration in be nonlinear (Kaiser 1993). For simplicity however,we North and West European populations of Sylvia assumethat one unit of fat score correspondsto 0.4 g warblers. J. Avian Biol. 26:39-48. of fat in both species. HALL, S. 1996. The timing of post-juvenile moult and Passerinesare predicted to consumefat at a rate of fuel deposition in relation to the onset of autumn about 1% of their body mass per hour in migratory migration in Reed Warblers Acrocephalus scir- flight (Nisbet et al. 1963). One unit of fat score should paceus and Sedge Warblers Acrocephalus schoe- then be enough for 3.7 hr of flight. Assuming a ground nobaenus. Ornis Svecica 6:89-96. speed of migration of 35 km hr-’ (Alerstam 1982), the HEDENSTR~M,A., AND A. P MILLER. 1992. Morpho- bird might be able to cover a distance of 130 km in logical adaptations to song flight in calm conditions.We found that recoveriesof Reed and birds: a comparative study. Proc. R. Sot. Lond. B Sedge Warblers on average were situated 76 and 111 247:183-187. km farther south, respectively,per increasein fat score HILDBN,O., AND I? SAUROLA.1982. Speed of autumn at banding. These values correspondto 58% and 85%, migration of birds ringed in . Omis Fen- respectively, of the expected distancea bird can cover nica 59:140-143. by using the fuel of 0.4 g fat. We acknowledge that KAISER,A. 1993. A new multi-category classification the above calculationsare approximate, but neverthe- of subcutaneousfat depositsof .J. Field. less, our results suggest that of two birds being at a Ornithol._64:24&255. stopover site on any given day, the one with the larger LINDSTR~M,A., AND T ALERSTAM.1992. Optimal fat fat loads will migrate aheadof the leaner bird. For both loads in migrating birds: a test of the time-mini- these specieswintering in West Africa, this difference mization hypothesis.Am. Nat. 140:477-491. may have fitness consequencesas early arrival to the LUNDGREN,B., A. HEDENSTR~M,AND J. PETTERSSON. winter quartersallows an early start of the prealternate 1995. Correlation between some body compo- molt, enabling birds to complete molt before the on- nents and visible fat index in the Willow Warbler coming dry season(Bensch et al. 1991). Phylloscopus trochilus (L.). Omis Svecica 5:75- We are most grateful to the hundredsof bandersand 79. assistantswho have participated with the banding at NISBET,I. C. T., W. H. DRURY,AND J. BAIRD. 1963. Kvismare Bird Observatory over the 14 years of this Weight loss during migration. Bird Banding 34: study. The manuscriptbenefited from comments by F 107-159. Fransson,D. Hasselquist,A. HedenstrGm,D. Irwin, A. PEACH,W. J., S. T BUCKLAND,AND S. BAILLIE.1990. LindstrGm, W. Peach, E Spina, and two anonymous Estimating survival rates using mark-recapture reviewers. This is reuort No. 103 from Kvismare Bird data from multiple ringing sites. Ring 13:87-102. Observatory. _ PENNYCUICK,C. J. 1989. Bird flight performance: a practical calculation manual. Oxford Univ. Press, LITERATURE CITED Oxford. ALERSTAM,T. 1982. Figelflyttning. Signum, Lund, PETERSSON, J., AND D. HASSELQUIST.1985. Fat depo- Sweden. sition and migration capacity of Robins Erithacus BASCIUTTI,I?, 0. NEGRA,AND E SPINA.In press. Au- rubecula and Goldcrest Regulus regulus at Otten- tumn migration strategies of the Sedge Warbler by, Sweden. Ring. Migr. 6:66-76. Acrocephalus schoenobaenus in northern Italy. SVENSSON,L. 1992. Identification guide to European Ring. Migr. .Lars Svensson,Stockholm. BENSCH,S., D. HASSELQUIST,A. HEDENSTR~M,AND U. WILKINSON,L. 1992. SYSTAT for Windows 5. SYS- Orrosso~. 1991. Rapid moult among palaearctic TAT Inc., Evanston, IL.