Examining the total arrival distribution of migratory birds

Transcription

1 Global Change Biology (2005) 11, 22 30, doi: /j x Examining the total arrival distribution of migratory birds T. H. SPARKS*, F. BAIRLEINw, J. G. BOJARINOVAz, O. HÜPPOPw, E. A. LEHIKOINEN, K. RAINIO, L. V. SOKOLOV} and D. WALKERk *NERC Centre for Ecology and Hydrology, Abbots Ripton, Huntingdon, Cambridgeshire PE28 2LS, UK, winstitute of Avian Research, Vogelwarte Helgoland, An der Vogelwarte 21, D Wilhelmshaven, Germany, zbiological Institute of St-Petersburg State University, Oranienbaumskoje sh. 2, Stary Peterhoff, St Petersburg , Russia, Department of Biology, University of Turku, Turku, Finland, }Biological Station Rybachy, Kaliningrad Region, Rybachy , Russia, kdungeness Bird Observatory, 11 RNSSS Cottages, Dungeness, Romney Marsh, Kent TN29 9NA, UK Abstract This paper reports on the total distribution of spring migration timing of willow warbler, chiffchaff and pied flycatcher at locations in the UK, Germany, Russia and Finland. This is the first time that high-quality data based on known-effort monitoring has been examined on a continental scale. First arrival dates, commonly reported in the literature, were positively correlated with mean arrival dates although they would not make good predictors of the latter. At all locations, at least one aspect of the arrival distribution of each species had got significantly earlier in recent years. The trend towards earliness was associated with warmer local temperatures and more positive winter North Atlantic Oscillation index. In years that were early, the arrival distribution became more elongated and skewed. Researchers should now investigate the consequences of earlier arrival on current and future bird populations. Keywords: climate impacts, Finland, first arrival date, Germany, Migration, NAO, phenology, Russia, spring, UK Received 5 April 2004; revised version received 12 July 2004 and accepted 22 July 2004 Introduction A large body of evidence has shown a trend in the first arrival date (FAD) of migrating birds such that 39% of nearly 1000 data series suggests earlier arrival while only 2% were significantly later (review in Lehikoinen et al., in press). This is surely evidence of a change although it may be that FAD, as an extreme observation in the arrival distribution, can be influenced by such factors as visibility, bird population size, observer activity and effort (e.g. Sparks et al., 2001). However, it is unlikely that the magnitude of observed change results solely from these factors, particularly as many bird populations are in decline in Europe (e.g. Donald et al., 2001 for farmland birds) which, if anything, might be expected to result in later observation. A small number of studies (e.g. Sokolov et al., 1998; Hüppop & Hüppop, 2003; Vähätalo et al., 2004) have overcome this problem by examining average dates at individual Correspondence: T. H. Sparks, fax , ths@ceh.ac.uk locations (either mean or median) that will be unaffected by population size or observer activity. In addition to reported changes to FAD and the central location of the arrival distribution, recent studies on white stork Ciconia ciconia (Ptaszyk et al., 2003) and pied flycatcher Ficedula hypoleuca (Ahola et al., 2004) have suggested a change in the shape of the arrival distribution. A growing literature is starting to examine changes in the timing of bird departure (e.g. Jenni & Kéry, 2003). The obvious causes of the changing arrival dates of migrants are climatic changes encountered in the wintering area, passage route and breeding area of the birds. Changes attributed to climate change in plants, invertebrates and vertebrates are widespread and affect phenology, population size and geographical distribution (e.g. Walther et al., 2002; Parmesan & Yohe, 2003; Root et al., 2003). The current paper examines in detail the arrival distributions of three example species in four European countries to more fully understand the relationships, if any, between parameters of the arrival distribution, 22 r 2005 Blackwell Publishing Ltd

2 THE ARRIVAL DISTRIBUTION OF MIGRATORY BIRDS 23 their changes over time and their relationships, if any, with prevailing weather conditions. This is the first time that multiple parameters of the arrival distribution, based on high-quality data from known-effort monitoring programmes, have been examined at the continental scale. The species selected are willow warbler Phylloscopus trochilus, chiffchaff P. collybita and pied flycatcher F. hypoleuca. Data sources were asked to provide 20 1 years of data on these species, although, in practice, there was some moderate relaxation of this criterion. Materials and methods Dungeness The bird observatory at Dungeness on the Kent coast is located on the SE tip of the UK at N, E. Data on willow warbler and chiffchaff were provided separately from both daily visual observations and from daily mist netting records. There were too few observations of pied flycatcher to provide reliable data. Data covered the period for observations and for caught birds. A cut-off date of June 15 was applied to separate spring migrating birds from residents. Monthly temperature data for the spring months were provided by the Central England Temperature record (Parker et al., 1992). Helgoland The bird observatory is sited on the island of Helgoland in the German North Sea at N, E. Records for all three species were obtained from captures in a Helgoland trap and cover the years Consistent with the earlier published work (Hüppop & Hüppop, 2003) cut-off dates of day 182 (July 1) for willow warbler and pied flycatcher, and day 230 (August 18) for chiffchaff were applied to separate spring passage birds from autumn passage birds. Monthly mean temperature data were obtained from NCEP reanalysis data ( covering a grid square including Helgoland. Rybachy The Rybachy observatory is sited on the Courish Spit in the Russian Baltic Sea at N, E. Records for all three species were obtained from captures in Rybachy-type traps (Payevsky, 2000) and cover the years Temperature data were obtained from the Kaliningrad meteorological station (65 km to the S of Rybachy). For all three species a cut-off date of day 182 (July 1) was applied. Ladoga The observatory is situated on the south-eastern shore of Lake Ladoga, Russia at N, E. Only data for willow warbler were available and were obtained from Rybachy-type traps. Data cover 20 years within the period Temperature data covering for Lodeinoye Pole meteorological station, about 34 km ENE of the observatory, were obtained from NW Meteo Department. A cut-off date of June 10 was applied. All birds with an incubation patch and a large cloacal protuberance (local breeding birds) trapped during the final period of migration were excluded from the analysis. Jurmo The observatory is sited in southern Finland at the southern edge of the Archipelago Sea in the Baltic Sea at N, E. Records on all three species were obtained from daily visual observations (numbers include ringed birds) and cover the period Monthly temperature data for Turku were obtained from the Finnish Meteorological Institute. The cut-off dates for all species were set at day 166 (June 15). North Atlantic Oscillation (NAO) An NAO winter index was calculated as a mean of the monthly values (normalized pressure differences between the Azores and Iceland) for December March (Hurrell, 1995). Data summary and analysis From data provided on individual bird captures/ observations the following were calculated for each year: FAD, mean date, median date, SD and skewness coefficient. These summary measures were related to one another and to temperature and NAO using correlation and regression techniques. Results Arrival distribution summary and trends Table 1 summarizes many features of the data. Larger numbers of birds were involved in observational studies than in those necessitating capture. Within-year variability (SD) was noticeably larger in chiffchaff than in the other two species (except for Rybachy willow

3 24 T. H. SPARKS et al. Table 1 Summary of migration information Site Species Method Years Mean n Mean SD Mean skew Mean, FAD SD Trend, FAD SE Mean, mean SD Trend, mean SE Mean, median SD Trend, median SE Trend in n Dungeness ww O Mar * Apr *** Apr ** * Dungeness ww C Apr Apr * Apr Helgoland ww C Apr *** *** Rybachy ww C *** *** *** 1 Ladoga ww C w Apr ** Jurmo ww O Apr ** May May ** Dungeness cc O Mar *** ** 1 Dungeness cc C Mar * Apr *** Apr *** 1 Helgoland cc C ** *** Rybachy cc C *** Apr *** Apr *** 1 * Jurmo cc O ** May May ** 1 Helgoland pf C Apr * ** Rybachy pf C Apr ** *** ** Jurmo pf O Apr *** Mean number of birds per annum observed (O) or caught (C), within-year standard deviation of arrival dates and within-year skewness coefficient of arrival dates. The subsequent six columns summarize first arrival date (FAD), mean arrival date and median arrival date together with an indication whether they have changed over time. The final column indicates if there has been a trend in the number C/O. *Po0.05, **Po0.01, ***Po w 20 years within this period. ww, willow warbler; cc, chiffchaff; pf, pied flycatcher.

4 THE ARRIVAL DISTRIBUTION OF MIGRATORY BIRDS 25 warbler) suggesting a longer passage period. The smaller SD in Rybachy and Jurmo chiffchaff than in other locations may reflect the different subspecies being recorded, i.e. the eastern population. Correlation of mean SD against mean FAD for the 14 tabled series produced a significant correlation (r , Po0.01) suggesting a longer arrival passage for species and locations where arrival was earliest. All skewness coefficients were positive, although not outstandingly so, suggesting a slight but consistent tendency for longer upper tails in the arrival distribution. Arrivals, as expected, were later at higher latitudes. There were, however, exceptions to this with Helgoland arrivals appearing to be later than might be expected from its latitude, especially in comparison with Rybachy. There was much less difference between sites in arrival dates of pied flycatcher than in the other two species. All but one of the 18 aspects of the arrival of willow warbler had negative trend coefficients. At least one aspect of willow warbler arrival distribution was significantly earlier at each site. Four of the six FADs suggested significantly earlier arrival, as did four of the six mean arrival dates and three of the six median arrival dates. Trends in FAD and mean arrival date are displayed in the left-hand columns of Figs 1 and 2, respectively, which emphasizes the greater magnitude of change at Dungeness. Similarly, all but one of the 15 aspects of the arrival of chiffchaff had negative trend coefficients. Four out of five FADs, three out of five mean arrival dates and four out of five median dates displayed significantly earlier behaviour. Again, at least one aspect of the arrival distribution changed significantly at each site. Trends in FAD and mean arrival date are displayed in the middle columns of Figs 1 and 2, respectively, which also emphasizes the greater magnitude of change at Dungeness. Eight of nine aspects of the arrival of pied flycatcher had negative trend coefficients. Two of three FADs, two of three mean dates and one of three median dates were significantly earlier. At each site at least one aspect of the arrival distribution was significantly earlier. Trends in FAD and mean arrival date are displayed in the right-hand columns of Figs 1 and 2, respectively, which reveals the more compact arrival distribution of pied flycatchers. The interdependencies of different aspects of the arrival distribution FAD and mean or median arrival dates were positively correlated in all instances (Table 2). Twelve and seven of these, out of 14, were significant. The mean and median arrival dates were highly significantly correlated in all cases with a minimum correlation coefficient of In 13 of 14 instances, there was a negative correlation between FAD and SD (significant in six cases) suggesting a tendency for greater SD in years of early first arrival (i.e. a more protracted arrival distribution). There was some suggestion that early arrival was also associated with a tendency to negative skewness in the arrival distribution, although this was only significant in two cases. Ten out of 14 correlations between FAD and sample size were negative, although only two were significant suggesting earlier FAD in more abundant years. Earlier mean dates seemed to be associated with positive skewness (as opposed to earlier FAD with negative skewness) with 10 of 14 significant correlations. However, the relationship between mean date and SD was less clear cut with a balance of both positive and negative effects. The influence of temperature and NAO Table 3 summarizes the correlations between summary measures of the arrival distribution and monthly temperatures for March, April and May (where relevant) and with winter NAO. For eight of the 14 series there was a significant correlation with local temperature for FAD. None of the four (short) series from Dungeness produced a significant correlation between FAD and local temperature. FAD and winter NAO were only correlated at Helgoland and Jurmo. There was more evidence of significant correlations with mean and median dates, with more significant correlations with temperature than with NAO. No significant correlations were present for capture data from Dungeness, but otherwise there was strong evidence supporting the concept that NAO and especially temperature are influencing some aspect of the arrival distribution. Warmer temperatures and higher NAO indices seem to be associated with earlier bird arrival. Although not shown here, the winter NAO index (the mean of 4 months) consistently produced a larger correlation than the individual months NAO values. Discussion This paper gathers together some of the best-quality data available on the phenology of bird migration. The sites provided data of different lengths and with different cut-off dates but it is very evident from these data that one or more aspects of the spring arrival dates of the three examined species have got significantly earlier at each site. However, it should be noted that it is not always the case that both FAD and mean (or

5 26 T. H. SPARKS et al. Willow warbler Chiffchaff Pied flycatcher Jurmo Ladoga Rybachy Helgoland Dungeness Mar 1 5 Mar Mar Mar 1 5 Fig. 1 Patterns in first arrival date. Open symbols indicate observations, solid symbols captures. An unusual record for chiffchaff at Helgoland (31 January 1975) is not displayed on the appropriate graph. median) arrival dates have both changed significantly. The former may be influenced by individual birds responding independently to external stimuli and we consider the latter more reliable, being less influenced by records of rogue birds, population size or recorder effort. The data reported here come from schemes

6 THE ARRIVAL DISTRIBUTION OF MIGRATORY BIRDS 27 Willow warbler Chiffchaff Pied flycatcher Jurmo Ladoga Rybachy Helgoland Dungeness Fig. 2 Patterns in mean arrival date. Open symbols indicate observations, solid symbols captures. where changes in recorder effort are minimal, but this is not the case in all reported examples of FAD, where some judgement is necessary to consider the effect, if any, of an increasing interest in bird migration. A changing mean date implies that a larger proportion of the population is responding than can be inferred from a changing FAD. The length of the series and the size of the studied population also determine the ability to

7 28 T. H. SPARKS et al. Table 2 Correlations (r) between first arrival date (FAD), mean arrival date and various summary features of the arrival distribution Correlation with FAD Correlation with mean arrival date Site Species Method Mean Median SD n Skew Median SD Skew Dungeness ww O 0.68*** 0.66** *** * Dungeness ww C 0.50* *** Helgoland ww C 0.33* *** 0.31* 0.40** Rybachy ww C 0.58*** 0.59*** 0.34* *** 0.33* 0.77*** Ladoga ww C 0.45* * *** ** Jurmo ww O 0.38* ** 0.43* 0.92*** Dungeness cc O * 0.52* *** *** Dungeness cc C 0.60** 0.50* 0.69** *** Helgoland cc C 0.31* * 0.44** 0.44* 0.93*** ** Rybachy cc C 0.55*** 0.36* *** 0.53** 0.41** Jurmo cc O *** *** Helgoland pf C 0.44** *** * Rybachy pf C 0.51*** 0.38** 0.32* *** ** Jurmo pf O 0.56*** 0.46* *** For key to abbreviations see Table 1. Table 3 Significant correlations (r) with local monthly mean temperatures for March (M), April (A) and May (My) and with the December March mean NAO index Correlation with temperatures Correlation with NAO Site Species Method FAD Mean date Median date FAD Mean date Median date Dungeness ww O My* Dungeness ww C Helgoland ww C A** M***A***My** M**A***My** * *** *** Rybachy ww C A** A**My* A*My** ** ** Ladoga ww C M*A*** Jurmo ww O M**A* M* My** * Dungeness cc O A* A* Dungeness cc C Helgoland cc C M***A** M*A*My* My* ** ** * Rybachy cc C A** A* * Jurmo cc O My* A* ** * Helgoland pf C A* M*A**My** M*A*My* * *** *** Rybachy pf C A** My** Jurmo pf O A*** My* * ** ** All significant correlations were negative. For key to abbreviations see Table 1. NAO, north Atlantic oscillation; FAD, first arrival date. detect change and correlations with influencing factors; short series are less statistically powerful while long series may cover periods of both climate cooling and warming and hence give greater confidence to any apparent temperature relationships (e.g. McCleery & Perrins, 1998). The fewer significant correlations at Dungeness and Ladoga may simply be a result of their relatively short time series. Two notes of caution need to be expressed. Firstly, we have made a large number of correlation tests in this paper and the reader should be aware that some of these may have resulted by chance. Secondly, using an arbitrary cut-off date (where it is difficult to separate spring migrants, autumn migrants and local breeding birds) may influence some of the calculated parameters. The influence of the latter point is likely to be small because of the small numbers

8 THE ARRIVAL DISTRIBUTION OF MIGRATORY BIRDS 29 of birds in the tails and the general advancement of the whole distribution. As with plants (Sparks & Menzel, 2002), the early events are more variable than later ones. This may, in part, reflect the greater variability in temperatures earlier in the year; it is not a function of population size. While FAD and mean data are positively, usually significantly, correlated, FAD will not be a very accurate predictor of mean date. Changes in FAD do at least indicate a change in the lower tail of the arrival distribution. An early FAD is associated with a more protracted arrival distribution (SD) and negative skewness while an early mean arrival date was more associated with positive skewness. Thus, years in which there are very early individuals are associated with longer lower tails to the arrival distribution while early arrival of the bulk of the population (indicated by early mean) are associated with longer upper tails. Therefore, it is not feasible to consider the shape and spread of the arrival distribution as static in time; both elements change according to the earliness of the season. Earliness in different aspects of the arrival distribution is associated with higher temperatures and, to a slightly lesser extent, high winter NAO index values. This is highly compatible with the increasing literature on earlier arrival, particularly FAD but also including mean arrival dates (Hüppop & Hüppop, 2003; Sokolov & Kosarev, 2003; Vähätalo et al., 2004). Where migration routes are known, further analysis incorporating migration route temperatures would be desirable. While an analysis of these species shows some broad agreements across sites within Europe, it also demonstrates that change and responses are not homogeneous and should not be considered as such. Here, for example, is solid evidence of the advance in arrival dates of pied flycatcher across northern Europe in contrast with the oft-quoted lack of change in the Netherlands (Both & Visser, 2001). While the data used here are the best available, they demonstrate the difficulty of working with highly mobile species that are not constrained to follow migration routes over fixed point observatories each year. Detection (through capture or observation) must be affected by different weather conditions although we consider that the mean (or median) date will be robust to such events. This paper provides powerful evidence that bird migration phenology is being modified across Europe and that the likely causes of a shift towards earlier arrival are changes in climatic conditions. Changes to the phenology of habitats (e.g. leafing dates) and elements of the food chain (e.g. invertebrates) are well documented (e.g. Walther et al., 2002). While confirmation of change is highly important, it is now fundamental that researchers consider what the implications of earlier arrival are to the population processes of migratory birds. Many, but not all, studies suggest that breeding performance is enhanced in early nesting individuals (e.g. Winkel & Hudde, 1997; Sokolov, 2000; Bairlein & Winkel, 2001). If early arrival results in early and more successful breeding will this result in population improvements or could any such gains be wiped out by adverse wintering and migration conditions? Acknowledgements We are grateful to all those whose hard work resulted in the abundance of data summarized here and to Ian Newton and Piotr Tryjanowski for comments on an earlier version of this paper. References Ahola M, Laaksonen T, Sippola K et al. (2004) Variation in climate warming along the migration route uncouples arrival and breeding dates. Global Change Biology, 10, Bairlein F, Winkel W (2001) Birds and climate change. In: Climate of the 21st Century: Changes and Risks (eds Lozan JL, Gral H, Hupfer P), pp Wissenschaftliche Auswertungen, Hamburg. Both C, Visser ME (2001) Adjustment to climate change is constrained by arrival date in a long-distance migrant bird. Nature, 411, Donald PF, Green RE, Heath MF (2001) Agricultural intensification and the collapse of Europes farmland bird populations. Proceedings of the Royal Society of London B, 268, Hüppop O, Hüppop K (2003) North Atlantic Oscillation and timing of spring migration in birds. Proceedings of the Royal Society of London B, 270, Hurrell JW (1995) Decadal trends in the North Atlantic Oscillation and relationships to regional temperature and precipitation. Science, 269, Jenni L, Kéry M (2003) Timing of autumn bird migration under climate change: advances in long-distance migrants, delays in short-distance migrants. Proceedings of the Royal Society of London B, 270, Lehikoinen E, Sparks TH, Zalakevicius M (2004) Arrival and departure dates. In: The Effect of Climatic Change on Birds. Advances in Ecological Research (eds Moller AP, Fiedler W, Berthold P), Academic Press, New York. McCleery RH, Perrins CM (1998)... temperature and egglaying trends. Nature, 391, Parker DE, Legg TP, Folland CK (1992) A new daily central England temperature series, International Journal of Climatology, 12, Parmesan C, Yohe G (2003) A globally coherent fingerprint of climate change impacts across natural systems. Nature, 421, Payevsky VA (2000) Rybatchy-type trap. In: Bird Station Manual (ed. Busse P), pp Gdańsk University Press, Gdańsk.

9 30 T. H. SPARKS et al. Ptaszyk J, Kosicki J, Sparks TH et al. (2003) Changes in the timing and pattern of arrival of the White Stork (Ciconia ciconia) in western Poland. Journal Für Ornithologie, 144, Root TL, Price JT, Hall KR et al. (2003) Fingerprints of global warming on wild animals and plants. Nature, 421, Sokolov LV (2000) Spring ambient temperature as an important factor controlling timing of arrival, breeding, post-fledging dispersal and breeding success of Pied Flycatcher Ficedula hypoleuca in Eastern Baltic. Avian Ecology and Behaviour, 5, Sokolov LV, Kosarev VD (2003) Relationship between timing of arrival of passerines to the Courish Spit and North Atlantic Oscillation index (NAOI) and precipitation in Africa. Proceedings of the Zoological Institute Russian Academy of Sciences, 299, Sokolov LV, Markovets MY, Shapoval AP et al. (1998) Long-term trends in the timing of spring migration of passerines on the Courish spit of the Baltic Sea. Avian Ecology and Behaviour, 1, Sparks TH, Menzel A (2002) Observed changes in seasons: an overview. International Journal of Climatology, 22, Sparks TH, Roberts DR, Crick HQP (2001) What is the value of first arrival dates of spring migrants in phenology? Avian Ecology and Behaviour, 7, Vähätalo A, Rainio K, Lehikoinen A et al. (2004) Spring arrival of birds depends on North Atlantic Oscillation. Journal of Avian Biology, 35, Walther G-R, Post E, Convey A et al. (2002) Ecological responses to recent climate change (review). Nature, 416, Winkel W, Hudde H (1997) Long-term trends in reproductive traits of tits (Parus major, P. caeruleus) and Pied Flycatchers Ficedula hypoleuca. Journal of Avian Biology, 28,