Bird-Banding Records Reveal Changes in Avian Spring and Autumn Migration Timing in a Coastal Forest Near Niigata

Home , Eyebrowed thrush, Japanese leaf warbler, Prunus mume

Ornithol Sci 19: 41 – 53 (2020)

ORIGINAL ARTICLE Bird-banding records reveal changes in avian spring and autumn migration timing in a coastal forest near Niigata

Alima DORZHIEVA1,*, Makoto NAKATA2,#, Keisuke TAKANO2, Youki FUJIHIKO1, Yasuo ITO3, Kiyoshi AKAHARA3, Katsuyoshi TACHIKAWA3, Yasuko ICHIMURA3, Yaeko FURUKAWA3, Hiroshi SATO3, Mikiko FUJISAWA3, Mika OKAMOTO3 and Takehiko SHIMIZU3

1 Graduate School of Science and Technology, Niigata University, 2–8050, Ikarashi, Nishi-ku, Niigata 950–2181, Japan 2 Faculty of Agriculture, Niigata University, 2–8050, Ikarashi, Nishi-ku, Niigata 950–2181, Japan 3 Niigata Group, Japanese Bird Banding Association, 1–13–11, Teraohigashi, Nishi-ku, Niigata 950–2054, Japan

ORNITHOLOGICAL Abstract Changes in the timing of bird migration in spring and autumn in a coastal forest near the city of Niigata, central Honshu, Japan, were analyzed based on 27 SCIENCE years of bird-banding records. Half of the bird species studied, including all migra- © The Ornithological Society tory types except residents, arrived or departed significantly earlier in spring due to an of Japan 2020 increase in spring temperatures. The rate of change we observed in spring migration timing due to changes in temperature was identical to or slightly greater than those reported in studies from other countries. The spring arrival of the Narcissus Flycatcher Ficedula narcissina and the Japanese Thrush Turdus cardis, both long-distance summer migrants to the nearby mountains, became earlier (advanced), however, for reasons that remained unclear. Median capture date in autumn was significantly associated with year for five species. Of these, the median capture date of the Japanese White-eye Zosterops japonicus, a resident and wandering bird, and the Black-faced Bunting Emberiza spodocephala, a wandering bird, advanced annually, while for the Japanese Robin Luscinia akahige and two other species (all long-distance migrants), it was delayed. We hypothesize that forest succession from a simple pine forest to a mixed forest with well-developed sub-canopy and shrub layers may have strongly influenced the Japanese White-eye and the Black-faced Bunting due to changes in population structure in the study area, resulting in an earlier median autumn capture date. Forest succession may also have influenced the Japanese Robin’s food resources, enabling it to stay longer in the study area and resulting in a delay in autumn departure date. Thus, changes in bird migration timing differ according to different environmental factors in spring and autumn.

Key words Bird-banding records, Climate change, Forest succession, Median capture date, Spring temperature

Climate change, caused mainly by an increase have caused significant environmental changes such in concentration of greenhouse gases in the atmo- as glacial melting and sea level rise, natural disas- sphere, has had a global effect (Ministry of the ters, and distinctive changes in weather conditions Environment 2017). The Fifth Assessment Report affecting the phenology of flora and fauna (Lawler et published by the Intergovernmental Panel on Cli- al. 2009; Acharia & Chettri 2010). Plant phenology mate Change, reported a 0.85°C increase over 132 (e.g. flowering and leaf budding) has changed consis- years (IPCC 2013). Changes in global temperature tently with climate change (Cleland et al. 2007). In central Asia, the growing season has lengthened by (Received 4 January 2019; Accepted 15 July 2019) approximately 10 days per decade due to both earlier # Corresponding author, E-mail: [email protected] * Present address: The Buryat State Academy of Agriculture, 8 spring onset and later leaf fall in autumn (Zhou et al. Pushkin Street, Ulan-Ude 670024, Russia 2001). The timing of bird arrivals has been increas-

41 A. DORZHIEVA et al. ingly studied over the past decades. Birds’ responses periods. Short and long-distance migrants arrived to climate changes have manifested in a shift in in Niigata Prefecture significantly earlier in spring timing of arrival and the beginning of breeding. In by 2.4 and 2.7 days per degree (d/°C), respectively warm spring years, most birds migrate to breeding (Nakata et al. 2011a). Similar patterns were recorded grounds early (Palm et al. 2009; Dunn & Winkler in the Yokohama area, Kanagawa Prefecture. On 2010; Charmantier & Gienapp 2013). Arrival tim- average, winter visitors departed 15 days earlier and ing of birds is also affected by other climatic- fac arrived nine days later in years of warm springs and tors such as wind speed along migration routes or autumns, respectively (Kobori et al. 2012). However, at wintering grounds, which may to some extent be research investigating the relationships between bird independent of changes in temperature (Gordo 2007; migration timing (particularly autumn migration), and Møller 2013). On the other hand, researcher opin- environmental factors, is limited in Japan. In order to ions differ regarding long-term changes in the timing assess the impact of climatic and other environmental of autumn bird migration. Some researchers suggest factors on the timing of bird migration in spring and that, in recent decades, climate warming in western autumn, we analyzed long-term bird-banding records Europe has resulted in both short and long-distance from the coastal forest of Sekiya near the city of migrants arriving much later (Cotton 2003). Obser- Niigata, central Honshu, Japan. vations in northern North America showed that the autumn migration of long-distance migrants peaked MATERIALS AND METHODS significantly earlier, while, in contrast, most of the short-distance migrants arrived later (Mills 2005). 1) Study area Furthermore, researchers suggest that the arrival time The research site was in the coastal forest planta- of birds in autumn is governed by the end of the tion of Sekiya (37°55′N, 139°01′E) in the city of reproductive period, the conditions in the breeding Niigata, central Honshu, Japan. The coastal forest grounds after the breeding season, and the expected extends for approximately 5 km and is 100–200 m conditions during the migration passage and in the wide, and is dominated by Japanese Black Pine Pinus wintering grounds (Jenni & Kery 2003). It is thought thunbergii, which was planted to prevent sand drift that autumn migration happens later during warm caused by the strong northwest monsoon in winter. years when birds stay longer in their breeding areas Due to lack of management of the pine forest, suc- (Sparks & Mason 2001; Carey 2009; Charmantier & cession has occurred from a simple pine forest to Gienapp 2013). a mixed forest with multiple strata (Yamaguchi & Japan’s climate is sensitive to global warming in Nakata 2008). Nakata et al. (2011a) described the several respects (Ministry of the Environment 2017). vegetation characteristics of the research site. Locust The average temperature increase in Japan over the Tree Robinia pseudoacacia and conifer mixes make past 100 years has been 1.19°C (Japan Meteorological up the canopy layer. Evergreen broad-leaved tree Agency 2018a). Climate change in Japan has affected species such as Persea thunbergii, Ilex integra, and phenological events in plants and animals. Thus, Neolitsea sericea, and deciduous broad-leaved tree the increase in spring temperature during the last species such as the Chinese Hackberry Celtis sinen- decade led to early flowering of the Japanese Apri- sis, Prunus speciosa, and Prunus verecunda make up cot Prunus mume by 0.8 days, the Yoshino Cherry the sub-canopy layer. The shrub layer is dominated Prunus×yedoensis by 0.7 days, and the Japanese by evergreen species, such as the Japanese Aucuba Wisteria Wisteria floribunda by 0.7 days (Ogawa- Aucuba japonica, the Japanese Spindle Euonymus Onishi & Berry 2013). Bird arrival times also have japonicas, and Ligustrum japonicum, and decidu- changed in response to an increase in temperature. ous species such as communities mixed with Morus Koike and Higuchi (2002) reported a trend of earlier australis. The herb layer is generally poor due to breeding and larger clutches over 21 years (1978– reduced light intensity caused by the dense growth 1998) in the Red-cheeked Starling Sturnia philippen- of the sub-canopy and shrub layers, however scat- sis in Niigata Prefecture, central Honshu, Japan. The tered growth of Persicaria filiformis and Oplismenus timing of breeding in Barn Swallows Hirundo rustica undulatifolius occurs. was about half a month earlier in the 2000s than in the 1960s in Japan (Deguchi et al. 2012), however 2) Bird banding data their departure timing did not change between these The Niigata Group of the Japanese Bird Band-

42 Migration timing based on bird banding ing Association has been banding birds during the to lowlands or the low mountains south of central spring and autumn migrations in the study area since Honshu (including Niigata Prefecture) in winter. This 1987. Bird banding was conducted in the coastal for- category also includes species that breed in northeast- est from just after sunrise to just before sunset at two ern China and the Korean Peninsula, but migrate to main research sites of ca. 1 ha each and approxi- Japan in winter. mately 350 m apart. Seven to 20 mist nets were used, (e) Long-distance migrants (summer visitors) to depending on the number of researchers or weather the nearby mountains: come from southern China or conditions and were checked at hourly intervals. The Southeast Asia in spring and breed in Japan (in the nets were 12 m long, 0.3–2 m high, and had mesh mountains in Niigata Prefecture). sizes of 24, 30, or 36 mm. Bird banding research was (f) Long-distance migrants (winter visitors): have not conducted on bad-weather days. their main breeding areas in the far north of Hokkaido To obtain uniform study periods for the spring and and the Korean Peninsula, and migrate to mainland autumn migration seasons, we used bird-banding Japan in winter. data from 28 March to 11 June in spring and from (g) Passage visitors: only pass through in spring 25 August to 17 November in autumn for 27 years and autumn, and do not stay in Niigata Prefecture in (from 1991 to 2017). Bird species were selected for summer or winter. analyses only if at least ten different individuals were From the data for the period 28 March to 11 June captured in at least ten years (Buskirk et al. 2009). from 1991 to 2017, 28 bird species were selected for Species were selected separately for spring and analysis, including ten long-distance migrants (sum- autumn. We counted the number of captured individ- mer visitors) to the nearby mountains, seven resi- uals, including those newly caught, repeatedly caught dents, four resident and wandering birds, four short- (recaptured within six months), returned (recaptured distance migrants (winter visitors), two wandering after six months), and recovered (released from other birds, and one long-distance migrant (winter visitor; banding sites) within seasons and years. To assess Table 1). From the data for the period 25 August migration timing, we calculated the median capture to 17 November from 1991 to 2017, 26 bird spe- date using the total number of individuals newly cies were selected for analysis, including ten long- caught, returned, and recovered, but not repeated distance migrants (summer visitors) to the nearby captures, for each species (Marra et al. 2005; Nakata mountains, five short-distance migrants (winter visi- et al. 2011a). Although Nakata et al. (2011a) used the tors), four residents, three resident and wandering term “median arrival date”, we use “median capture birds, two long-distance migrants (winter visitors), date”, since we included both migratory and resident one wandering bird, and one passage visitor (Table birds in this study. 1). Species names and scientific names follow The We categorized seven types of bird migration Ornithological Society of Japan (2012) nomenclature. around the study area (mainly in Niigata City) based on the Wild Bird Society of Japan (2008), the Niigata 3) Climatic data Prefectural Branch of the Wild Bird Society of Japan Weather conditions recorded over 40 years (1971– (2010), and our field observations (Table 1). These 2010) at the Niigata Local Meteorological Office, 3 categories are defined as follows: km east of the research site, revealed a mean annual (a) Residents: mainly occur around the study area temperature of 13.7°C, 1,796 mm of mean annual throughout the year. However, some individuals visit total precipitation, and 37 cm of mean maximum the study area only in spring and autumn. snow depth (Japan Meteorological Agency 2018b). (b) Resident and wandering birds: commonly We used the weather data to analyze the relationship regarded as resident, although this category includes between bird median capture date and climatic fac- some populations that only visit in spring and autumn, tors. The spring and autumn mean temperatures in or move down from the mountains, especially in win- each year were calculated for the periods used for ter. analyzing bird-banding data (from 28 March to 11 (c) Wandering birds: breed in the mountains north June and from 25 August to 17 November in spring of central Honshu (including Niigata Prefecture), and and autumn, respectively). migrate to warmer regions or lowlands in winter. We calculated tailwind assistance (TWA) in spring (d) Short-distance migrants (winter visitors): breed and autumn for each year to express wind data (direc- in the mountains north of central Honshu and migrate tion and intensity) as a single biologically meaning-

43 A. DORZHIEVA et al. 1) 11.1 3.9 13.0 4.0 −− −− −− −− −− 4-Sep 4.6 1-Sep 19-Oct 27-Oct 4.5 19-Oct 14-Oct 6.4 30-Sep 5.6 19-Sep 6.6 Average ±SD Median capture date 1) 6.9 5.5 25-Oct 3.4 6.3 55.5 20-Oct 8.0 251.6 24-Oct 3.3 Autumn 19.9 96.4 16.0 5.9 60.6 32.1 28-Oct 8.4 19.1 21.0 8.7 27.5 30.5 29-Oct 12.3 50.6 52.5 30-Oct 641.6 No. of birds captured Average ±SD − −− − −− − −− − −− − −− 11 11 11 2610 33.8 15.7 13.7 27-Oct 3.1 14 16 15 20 39.4 22.8 5-Oct 7.2 analyzed No. of years 1) 4.6 2.3 10 20.0 8.6 27-Oct 6.3 3.5 3.6 27 42.5 15.4 3.9 26 28.3 17.3 16.4 4.6 3.9 −− −− −− 8-Apr 3.0 4-May 6-May 6-May 11-Apr 19-May Average ±SD 1) 9.2 7.6 5.1 26-Apr 3.4 21 9.7 28-Apr 5.1 27 76.2 36.4 18-Oct 2.8 20.7 30-Apr 5.9 27 48.1 26.7 22-Oct 8.6 9.2 27-May 5.4 4.0 7.8 13-Apr 6.3 Spring −− −− −− 21.5 12.7 13.8 43.9 27.1 23-Apr 10.6 36.2 23.8 28-Apr 61.4 4.0 12 29.0 21.2 13-Oct 8.9 16.1 4.9 19.9 Average ±SD No. of birds captured Median capture date − − − 11 11 11 26 45.0 21.9 25-Apr 2726 89.6 27.726 46.2 14.0 21-Apr 27.6 23-Apr 2.7 11.3 3.9 25-Apr 27 27 4.5 68.0 116.5 27 42.4 95.3 5-Nov 26.7 4.8 13.1 27-Sep 5.7 24 20.3 23 23.1 14 2727 25.4 57.3 23.1 27 155.5 57.5 24-Apr 2.9 27 2725 205.0 30.4 146.8 14.5 17-Apr 18-Apr 9.7 5.9 27 296.6 186.9 2314 27 52.5 19.6 2727 33.727 383.9 188.5 12-May 198.5 94.5 64.3 21-Apr 87.7 22-Apr 4.5 3.5 3-May 7.0 27 27 2412.8 1791.6 26-Oct 6.2 27 95.8 48.9 7-May 23 24.1 11.8 8-May 3.9 15 16.0 3.3 27-Sep 5.2 12 2610 32.1 14.8 13.7 10-May 3.5 12 17.8 7.9 7-Sep 16.5 2713 50.5 40.3 3-Jun 2.5 27 72.0 35.4 26-Sep 3.6 analyzed No. of years Scientific name Regulus regulus Emberiza variabilis Emberiza elegans Tarsiger cyanurus Tarsiger pallidus Turdus squameiceps Urosphena Turdus chrysolaus Turdus Cyanoptila cyanomelana Luscinia akahige Turdus cardis Turdus Phylloscopus borealoides Emberiza spodocephala Parus minor Lanius bucephalus Periparus ater Poecile varius Passer montanus Chloris sinica Spodiopsar cineraceus japonicus Zosterops Cettia diphone Hypsipetes amaurotis Phylloscopus coronatus Ficedula narcissina Troglodytes troglodytes Troglodytes Luscinia cyane Muscicapa dauurica Fringilla montifringilla Phylloscopus xanthodryas Carduelis spinus Carduelis Turdus obscurus Turdus Bird migration types, number of years analyzed, number of birds captured, and median capture date for each species analyzed for spring and autumn. species analyzed date for each capture number of birds captured, and median types, number of years analyzed, Bird migration Species Table 1. Table Pale Thrush Brown-headed Thrush Blue-and-white Flycatcher Japanese Robin Japanese Thrush Black-faced Bunting Japanese Tit Bull-headed Shrike Coal Tit Eurasian Tree Sparrow Eurasian Tree Japanese Bush Warbler Japanese White-eye Brown-eared Bulbul Narcissus Flycatcher Eurasian Wren Siberian Blue Robin Eurasian Siskin Eyebrowed Thrush Varied Tit Varied White-cheeked Starling Yellow-throated Bunting Yellow-throated Asian Stubtail Asian Brown Flycatcher Standard deviation Short-distance migrants (winter visitors) Goldcrest Grey Bunting

Red-flanked Bluetail Long-distance migrants (summer visitors) to the nearby mountains

Sakhalin Leaf Warbler Wandering birds Wandering

Oriental Greenfinch

Resident and wandering birds Eastern Crowned Leaf Warbler

Residents

Long-distance migrants (winter visitors) Brambling Japanese Leaf Warbler

Passage visitor 1)

44 Migration timing based on bird banding ful variable. TWA was calculated by the following tured among the total number captured in autumn for formula (Sapir et al. 2011); the Japanese Robin (this will be discussed in detail below). We also calculated the average length of stay TWA = V · cos ((θ +180°)−θ ) (1) w w b for this species in autumn, using the interval between where Vw is the measured wind velocity in meters the first and last capture dates in the same season for per second, θw is wind direction (i.e. the direction individuals that were caught repeatedly. However, from which the wind was blowing) in degrees, and data for the Eastern Crowned Leaf Warbler Phyl- θb is the presumed direction of the birds. Here, we loscopus coronatus and the Eyebrowed Thrush Tur- assumed θb is northeast (45°) in spring and southwest dus obscurus were analyzed differently from those (225°) in autumn, based on the direction of the coast- for the Japanese Robin because insufficient data were line in Niigata Prefecture. TWA is simply the tailwind obtained (this will be discussed in detail below). component of wind velocity in meters per second or the wind assistance vector at the presumably pre- RESULTS ferred direction of the bird. Wind data at the Niigata Local Meteorological Office were available for each 1) Changes in the timing of spring migration five-day interval in the month (days 1–5, 6–10, …, For spring captures, median capture dates were and 26–30/31) during the study (1991–2017). We from early April to early June in 28 species investi- used the mean wind speed and the most frequent gated (Table 1). The Eurasian Wren Troglodytes trog- wind direction for each five-day interval from 26 lodytes, a resident, had the earliest median capture March to 10 June and 26 August to 15 November in date (8 April). Long-distance summer visitors to the spring and autumn, respectively. nearby mountains showed later median capture dates than other migratory types, mostly arriving from late 4) Data analyses April to early May. The latest median capture date A generalized linear model (GLM) was used to (3 June) was observed in the Japanese Leaf War- investigate the factors influencing migration timing bler Phylloscopus xanthodryas, a long-distance sum- (median capture date) in spring and autumn. The mer visitor to the nearby mountains. In contrast, the median capture date was adjusted so that 1 March Brambling Fringilla montifringilla, a long-distance was day one in spring and 1 August was day one winter visitor, had the third-earliest median capture in autumn. We assumed that the response variable, date (13 April) of the 28 species investigated. median capture date, was normally distributed. The best GLM models for spring, based on meteo- Explanatory variables included meteorological fac- rological factors and year, were selected for 20 spe- tors (mean temperature and TWA) and year. No sig- cies (Table 2). The mean spring temperature was significant correlations between these three explanatory nificantly negatively associated with median capture variables were identified for either spring or autumn date for ten species (P<0.05). Of these, five were (P>0.05; data not shown). Correlation analyses were long-distance summer visitors to the nearby moun- investigated using Spearman’s rank correlation coef- tains, and two were resident and wandering birds. ficient. The statistical analyses were performed using The other migratory types except residents were each R version 3.5.3 (R Core Team 2019). represented by one species. Mean spring temperature To explain the significant annual advance of the was also negatively associated with median capture median capture date in autumn, we calculated the date for three other species, but not significantly so percentage of individuals recaptured among total (P>0.05). The average rate of change in median cap- number captured in each spring for the Japanese ture date of the five long-distance summer visitors to White-eye and the Black-faced Bunting (this will the nearby mountains was −2.599 d/°C (range: −1.751 be discussed in detail below). Furthermore, to con- to −3.175). The rate of change of the Brambling, a firm the breeding of these species, we calculated the long-distance winter visitor, was −4.515 d/°C; the percentage of juvenile individuals, assumed to be largest change of the ten species. The average rate of fledglings, captured during the first month (25 August change was −3.775 d/°C (range: −3.326 to −4.249) to 24 September) among total number captured in for the other four species representing the other the autumn banding season. To explain the signifi- migratory types: resident and wandering, wander- cant annual delay of median capture date in autumn, ing, and short-distance winter visitor. TWA was sig- we calculated the percentage of individuals recap- nificantly positively associated with median capture

45 A. DORZHIEVA et al.

Table 2. Results of the selection of best GLM, including meteorological factors and year as explanatory variables, for spring.

Coefficient Species Explanatory variables t statistics P value 1) Estimate Std. error Residents Intercept 42.5 3.4 12.5 *** Japanese Tit TWA 6.432 3.072 2.094 * Intercept −297.7 249.5 −1.2 0.272 Temperature −3.280 1.536 −2.135 0.070 Varied Tit TWA 2.053 1.512 1.357 0.217 Year 0.201 0.127 1.585 0.157 Intercept 1522.1 916.1 1.7 0.112 Eurasian Tree Sparrow TWA 10.093 6.137 1.645 0.116 Year −0.728 0.458 −1.588 0.128 Intercept 1084.3 539.8 2.0 0.068 White-cheeked Starling Year −0.497 0.269 −1.844 0.090 Resident and wandering birds Intercept −199.4 146.0 −1.4 0.185 Japanese White-eye Temperature −4.249 0.726 −5.851 *** Year 0.157 0.073 2.137 * Intercept 120.7 18.1 6.7 *** Brown-eared Bulbul Temperature −4.054 1.229 −3.298 ** Intercept 102.8 24.5 4.2 *** Oriental Greenfinch Temperature −2.605 1.664 −1.566 0.130 Wandering birds Intercept 287.1 151.8 1.9 0.071 Black-faced Bunting TWA 1.822 1.034 1.762 0.091 Year −0.116 0.076 −1.532 0.139 Intercept −97.4 120.7 −0.8 0.429 Brown-headed Thrush Temperature −3.472 0.558 −6.227 *** Year 0.107 0.061 1.763 0.093 Short-distance migrants (winter visitors) Intercept 51.3 1.3 39.0 *** Pale Thrush TWA 3.158 1.181 2.673 * Intercept 395.5 189.7 2.1 * Grey Bunting Temperature −3.326 0.904 −3.679 ** Year −1.445 0.095 −1.524 0.141 Long-distance migrants (summer visitors) to the nearby mountains Intercept 103.1 14.0 7.4 *** Asian Stubtail Temperature −3.175 0.949 −3.345 ** Intercept 81.7 12.8 6.4 *** Japanese Robin Temperature −1.739 0.858 −2.026 0.068 TWA 2.029 1.246 1.628 0.132 Intercept 563.7 214.7 2.6 * Japanese Thrush Temperature −2.616 1.068 −2.449 * Year −0.233 0.108 −2.154 * Intercept 103.2 12.3 8.4 *** Sakhalin Leaf Warbler Temperature −2.430 0.834 −2.912 ** Intercept 91.0 12.8 7.1 *** Blue-and-white Flycatcher Year −1.597 0.871 −1.835 0.081 Intercept 113.2 14.3 7.9 *** Eastern Crowned Leaf Warbler Temperature −3.024 0.971 −3.115 ** Intercept 745.1 162.3 4.6 *** Narcissus Flycatcher Year −0.337 0.081 −4.161 *** Intercept 97.3 11.9 8.1 *** Siberian Blue Robin Temperature −1.751 0.814 −2.152 * Long-distance migrant (winter visitor) Intercept 111.3 26.1 4.3 *** Brambling Temperature −4.515 1.767 −2.555 * 1) ***, P<0.001; **, P<0.01; *, P<0.05

46 Migration timing based on bird banding date for two species (P<0.05), the resident Japanese summer visitors to the nearby mountains showed ear- Tit Parus minor and the short-distance winter visitor lier median capture dates (mostly in September) than Pale Thrush Turdus pallidus. Year was significantly other migratory types. The Eastern Crowned Leaf associated with median capture date for three species Warbler, a long-distance summer visitor to the nearby (P<0.05). The Japanese White-eye, a resident and mountains, had the earliest median capture date (1 wandering bird, tended to be captured later (0.157 d/ September). In contrast, the Eurasian Siskin Cardue- year). In contrast, the Japanese Thrush and the Nar- lis spinus and the Brambling, both long-distance win- cissus Flycatcher, both long-distance summer visitors ter visitors, had the second- and third-latest median to the nearby mountains, tended to be captured earlier capture dates (30 and 29 October, respectively) of the (−0.233 and −0.337 d/year, respectively). 26 species. The latest median capture date (5 Novem- ber) was recorded for the Red-flanked Bluetail Tar- 2) Changes in the timing of autumn migration siger cyanurus, a short-distance winter visitor. The median capture dates for the autumn captures The best GLM models for autumn, based on mete- were from early September to early November for orological factors and year, were selected for 13 spe- the 26 species investigated (Table 1). Long-distance cies (Table 3). Mean autumn temperature was signifi-

Table 3. Results of the selection of best GLM, including meteorological factors and year as explanatory variables, for autumn.

Coefficient Species Explanatory variables t statistics P value 1) Estimate Std. error Resident Intercept 1060.8 599.5 1.8 0.090 Japanese Tit Temperature −5.826 3.205 −1.818 0.082 Year −0.435 0.302 −1.442 0.162 Resident and wandering birds Intercept 1012.6 254.2 4.0 *** Japanese White-eye Year −0.462 0.127 −3.639 ** Intercept 54.6 14.6 3.7 *** Japanese Bush Warbler Temperature 1.686 0.782 2.156 * Wandering bird Intercept 997.9 359.3 2.8 * Black-faced Bunting Year −0.457 0.179 −2.551 * Short-distance migrants (winter visitors) Intercept 95.4 1.3 72.9 *** Red-flanked Bluetail TWA −2.862 1.590 −1.800 0.084 Intercept 45.1 18.8 2.4 * Pale Thrush Temperature 1.907 1.007 1.893 0.070 Intercept 86.9 0.9 95.8 *** Grey Bunting TWA −2.104 1.082 −1.945 0.064 Long-distance migrants (summer visitors) to the nearby mountains Intercept −335.3 194.7 −1.7 0.101 Japanese Robin Year 0.210 0.097 2.165 * Intercept 102.4 13.3 7.7 *** Japanese Thrush Temperature −1.261 0.718 −1.755 0.092 TWA −1.469 0.953 −1.542 0.136 Intercept −521.1 251.7 −2.1 0.059 Eastern Crowned Leaf Warbler Year 0.276 0.126 2.199 * ** Narcissus Flycatcher Intercept 106.3 32.5 3.3 Temperature −2.560 1.747 −1.466 0.167 Intercept 58.5 1.0 58.3 *** Japanese Leaf Warbler TWA 1.905 1.220 1.561 0.131 Passage visitor Intercept −902.2 381.0 −2.4 * Eyebrowed Thrush Temperature −3.632 2.072 −1.753 0.098 Year 0.516 0.190 2.711 * 1) ***, P<0.001; **, P<0.01; *, P<0.05

47 A. DORZHIEVA et al. cantly associated with median capture date only for 3) Percentage of recaptured and juvenile birds, the Japanese Bush Warbler Cettia diphone, a resident and length of stay and wandering bird (1.686 d/°C; P<0.05). TWA did The Japanese White-eye and the Black-faced Bun- not significantly affect any species’ median capture ting showed significant positive correlations between date (P>0.05). Year was significantly associated with year and the percentage of individuals recaptured median capture date for five species (P<0.05). Of among the total number captured of spring banding these, the Japanese White-eye, a resident and wan- (Japanese White-eye: r=0.640, P<0.001; Black-faced dering bird, and the Black-faced Bunting, a wander- Bunting: r=0.799, P<0.001; Fig. 1). There were also ing bird, tended to be captured earlier (both ca. −0.46 significant positive correlations between year and the d/year). The other three species, the Japanese Robin, percentage of juvenile individuals captured during the Eastern Crowned Leaf Warbler, and the Eye- the first month (25 August to 24 September) among browed Thrush (the former two long-distance sum- the total number captured of autumn banding for both mer visitors to the nearby mountains and the latter Japanese White-eye (r=0.858, P<0.001) and Black- a passage visitor) tended to be captured later (0.210, faced Bunting (r=0.778, P<0.001; Fig. 2). 0.276, and 0.516 d/year, respectively). There was a significant positive correlation between

Fig. 2. The relationship between year and the percentage of juveniles captured during the first month among total number Fig. 1. The relationship between year and the percentage of captured in the autumn banding season: (a) Japanese White- individuals recaptured among total number captured in spring: eye, (b) Black-faced Bunting. Recaptures were excluded from (a) Japanese White-eye, (b) Black-faced Bunting. the calculation.

48 Migration timing based on bird banding

mean spring temperature for nine species, of which seven showed similar results in our study, but two, the Japanese Robin and the Japanese Tit, did not. However, three additional species showed similar significant trends in our study: the Sakhalin Leaf Warbler Phylloscopus borealoides, the Siberian Blue Robin Luscinia cyane and the Brambling. All three species are long-distance migrants, the former two being summer visitors to the nearby mountains and the latter a winter visitor. Mean spring temperature was calculated using data from March 1 to May 31 in Nakata et al. (2011a), however this method pro- duced similar results to our study when the relationship between mean spring temperature and median capture dates was investigated (data not shown). The Siberian Blue Robin showed a significant trend only in our analyses, but this was not the case in the previous method. These differences may be attributed to the length of the study period: 20 years in Nakata et al. (2011a) and 27 years in the present study. A longer monitoring period may be needed to show a change in bird migration timing caused by changes in spring temperature. Mean spring temperature showed no relationship to year for the 27 study years, suggesting that spring migration timing was predominantly influenced by temperature. The ten species whose median capture dates were significantly negatively associated with spring temperature covered all migratory types except residents (Table 2). Of the four resident species included in this Fig. 3. The relationship between year and (a) the percentage of individuals recaptured among total number captured in analysis, the GLMs showed that median capture date autumn and (b) the average length of stay in autumn for the was not significantly associated with mean spring Japanese Robin. An outlier (of 18 days) from the 1995 data temperature. On the other hand, that of at least half of was excluded from figure 3b. the species in each of the other migratory types was significantly associated with mean spring temperature (Table 2). Many resident Japanese White-eyes year and the percentage of individuals recaptured in and Brown-eared Bulbuls Hypsipetes amaurotis are autumn for the Japanese Robin (r=0.548, P<0.01; observed around the study area, but the significant Fig. 3a), whose median capture date in autumn was effect of mean spring temperature on the median delayed yearly. Further, there was also a significant capture date of these species may be attributable to positive correlation between year and average length the migratory portions of these populations (Niigata of stay for this species (r=0.601, P<0.01; Fig. 3b). Prefectural Branch of Wild Bird Society of Japan 2010). Among the ten species with significant trends, the highest rate of change in median capture date DISCUSSION was in the Brambling, a long-distance winter visitor, 1) Changes in the timing of spring migration which departed 4.515 days earlier when mean spring Nakata et al. (2011a) analyzed the relationship temperature rose by one degree. The average rates of between median arrival date and spring temperature change of median capture date were −2.599 d/°C for in the coastal forest of Sekiya for a period of 20 the five long-distance summer visitors to the nearby years (1989 to 2008). They identified significant neg- mountains, and −3.775 d/°C for the four shorter-dis- ative correlations between median arrival date and tance migrants (including resident and wandering,

49 A. DORZHIEVA et al. wandering, and short-distance winter visitor). This TWA was significantly positively associated with is probably because the local temperature (i.e. the the median capture date of the Japanese Tit and Pale temperature in Niigata City) has a greater effect on Thrush, suggesting that stronger tailwinds delay the shorter-distance migrants than on long-distance sum- median capture or departure dates of these species. mer visitors. However, long-distance summer visi- Sinelschikova et al. (2007) demonstrated that the tors to the nearby mountains arrived at the study site advance in the timing of spring migration of the Song later than other migratory types, mostly from late Thrush Turdus philomelos is primarily due to the April to early May. This suggests that these species increased frequency of tailwinds favorable for migra- are susceptible to negative impacts of temperature tory flights over the migratory route. Therefore, the changes because their later arrival at the breeding relationship between tailwinds and migration timing areas increases their risk of missing the suitable or the behavior of these two species remains unclear. breeding period if the rate of spring warming exceeds The median capture date of the Japanese White- their acceleration of migration (Marra et al. 2005). eye was delayed by a mean of 0.157 days per year, Indeed, in the Netherlands, a mismatch between the whereas that of the Japanese Thrush and the Nar- breeding period and the peak of food availability led cissus Flycatcher shifted earlier by 0.233 and 0.337 to a population decline in the Pied Flycatcher Fice- days per year, respectively (Table 2). In the case of dula hypoleuca, a long-distance migrant (Both et al. the Narcissus Flycatcher, therefore, spring arrival 2006). at the study area became an average of nine days Rates of change in avian spring arrival have been earlier over the 27-year study period. Since both of studied in other countries. In Europe, a long-term these species are long-distance summer visitors to the survey from 1828 to 2010 showed that the average nearby mountains, Nakata et al. (2011a) suggested arrival of seven short-distance migrants was 0.65 that this trend may be due to environmental changes d/°C earlier and that of six species of long-distance in their wintering grounds or along their migration migrants was 0.96 d/°C earlier (Kolářová et al. 2017). routes, although the exact reasons remained unclear. Notably, Barn Swallows arrive two days earlier per One possible reason for the delayed median capture degree (Huin & Sparks 1998). In central Russia, a date of the Japanese White-eye, could be an increase long-term survey from 1936 to 2013 showed that in breeding around the study area. Breeding adults 25 bird species arrived 15–37 days earlier during stay around the study area for a few months, which warm springs relative to cold ones (Vengerov 2015). may increase the possibility of capturing them in Long-term observations from 1967 to 2008 in North the spring banding season. Nevertheless, the expla- America showed that the mean spring first-arrival nations for this trend remain unclear, because the dates of short-range migrants have advanced by 0.15 Black-faced Bunting did not show the same tendency days per year, whereas those of long-range migrants (discussed below). have advanced by 0.06 days per year (Deleon et al. 2011). Significant changes in the temperature 2) Changes in the timing of autumn migration regime indicate a change towards an earlier arrival The spring migration timing of birds is directly date of birds by –1.1±0.8 days per decade in the related to the timing of breeding (Sokolov 2006). Southern Hemisphere (Chambers et al. 2013). The Changes in the median autumn capture date of shifts towards earlier spring arrival or departure date birds in the coastal forest of Sekiya depended on related to increases in spring temperature observed in both annual changes in environment factors and on the present study were thus on par with, or slightly the results of breeding. The median capture date in greater than, those identified in previous studies autumn of the Japanese White-eye and the Black- from Europe, North America, and other countries. faced Bunting advanced by 0.46 days per year (on Our analysis of spring migration indicated that arrival average approximately 12 days earlier over the 27 or departure dates changed because of temperature study years), possibly due to changes in population variation. With the onset of spring, food availabil- structure in the coastal forest. The first case of Black- ity increases as temperature affects insect activity faced Bunting breeding in this forest was recorded and growth of shoots (Carey 2009). Accordingly, a in 1996, whereas the Japanese White-eye had been change in bird migration timing caused by changes in confirmed breeding there when bird-banding research spring temperature is regarded as a response to food began (Ito 2006). The yearly increase in the percent- availability during the breeding period. age of Japanese White-eye and Black-faced Bun-

50 Migration timing based on bird banding ting individuals recaptured in spring (Fig. 1) reflects The median capture date of the Eyebrowed Thrush, an increase in breeding in the study area. Because a passage visitor, was positively associated with year, breeding adults stay around the study area for a few being delayed by 0.516 days per year (on average 14 months in the spring banding season, the chances days over the 27 study years) in autumn. The species of recapturing them increase. It is possible that the prefers forest conditions and is omnivorous, foraging increase in breeding during spring influenced the ear- for both insects and fruit on the ground and in the lier median capture date in autumn. Juveniles tend canopy (Tojo 2015). Recaptured individuals were rec- to remain in their natal area to establish their pre- ognized in 14 years of the 27-year study period, but ferred territories in autumn (Dhondt 1971), and this this quantity of data was insufficient for investigating may increase the total number of individuals stay- factors driving this delay. However, the number of ing around the study area when the autumn banding individuals newly caught clearly increased annually season starts. Actually, the percentage of juveniles (r=0.781, P<0.001; data not shown). Similarly, 48 captured during the first month of autumn banding individuals were recaptured during the last 11 years, showed a significant annual increase for these two which was four times larger than the 12 individuals species (Fig. 2). The increase of breeding in both spe- recaptured during the first 16 years of the study. This cies has been attributed to forest succession (Nakata suggests that the progress of forest succession most et al. 2011b) since the growth of the sub-canopy and likely influenced Eyebrowed Thrush behavior. The shrub layers, composed mainly of evergreen broad- yearly delay in the species’ arrival or departure date leaved species, has produce suitable nesting sites for was double that of the Japanese Robin and Eastern these birds (Deguchi et al. 2015). Crowned Leaf Warbler (both long-distance summer The median capture date of the Japanese Robin visitors to the nearby mountains), but factors driving and the Eаstern Crowned Leaf Warbler, both long- this trend could not be clarified during this study. distance summer visitors to the nearby mountains, The median capture date of the Japanese Bush was delayed by 0.210–0.276 days per year (on Warbler, a resident and wandering bird in the Niigata average 6–7 days later over the 27 study years) in Prefecture, was positively associated with mean autumn. To date there are no reports of these two autumn temperature. Its median capture or arrival species breeding in the coastal forest of Sekiya. The date was delayed by 1.686 days for a 1°C increase increase in the percentage of recaptured Japanese in mean autumn temperature, possibly due to popula- Robin individuals in autumn (Fig. 3a) may also be tions moving down from the mountains. The species attributed to forest succession. The species prefers was confirmed as breeding in the coastal forest of coniferous or mixed forests with thick undergrowth Sekiya (Ito 2006), however the percentage of recap- and forages for insects or earthworms in bushes or on tured individuals did not significantly correlate with the ground (Seki et al. 2012). Vegetation succession either year or mean temperature in either spring or from a simple pine forest to a mixed forest composed autumn (data not shown). Mean autumn temperature of many evergreen and deciduous broad-leaved did not correlate with year across the 27 study years. trees with well-developed evergreen shrub layers This suggests that autumn migration timing or behav- (Yamaguchi & Nakata 2008; Nakata et al. 2011b), ior of the Japanese Bush Warbler may be directly has resulted in an increase in these food resources. controlled by autumn temperature. These conditions have enabled the Japanese Robin to stay in the forest for longer (Fig. 3b), which may CONCLUSION explain the increase in its recapture rate (Fig. 3a) and resultant delayed departure date. In contrast, only The migration timing of the birds studied in spring nine Eastern Crowned Leaf Warblers were recaptured was found to be strongly influenced by meteorologi- during the 27-year study, thus these data were not cal factors. Median capture date was significantly suitable for identifying factors driving their delay in negatively associated with mean spring temperature median capture date in autumn. Our results indicate for half of the bird species studied (including rep- that bird-banding records must be used with caution resentatives of all the migratory types investigated when analyzing autumn migration timing because except residents), suggesting an earlier response the median capture date does not always reflect the to the onset of warmer springs. However, autumn actual migration timing if forest succession or veg- migration timing was less influenced by meteoro- etation progression occurs at the study site. logical factors. Forest succession from a simple pine

51 A. DORZHIEVA et al. forest to a mixed forest with well-developed sub- (Habitat selection of the Black-faced Bunting in rela- canopy and shrub layers may strongly influence the tion to vegetation structure during the breeding season autumn median capture date for some species due to in coastal Japanese Black Pine forest). J Yamashina an increase in breeding and to longer length of stay Inst Ornithol 46: 101–107 (in Japanese). in the study area. Environmental factors thus produce Deguchi T, Yoshiyasu K & Ozaki K (2012) Hyoshiki different migration timing responses between spring chosa joho ni motodzuita 2000-nendai to 1960-nendai and autumn in birds. no tsubame no watari jiki to hanshoku jokyo no hikaku (Comparison of Barn Swallow migration and breeding based on banding records from the 1960s and ACKNOWLEDGMENTS 2000s). Jpn J Ornithol 61: 273–282 (in Japanese). Deleon RL, Deleon EE & Rising GR (2011) Influence We are grateful to all staff and co-workers of the of climate change on avian migrants’ first arrival Niigata Group, the Japanese Bird Banding Associa- dates. Condor 113: 915–923. tion, for their long-term bird-banding research. We Dhondt AA (1971) Some factors influencing territory in also thank the Yamashina Institute for Ornithology the Great Tit (Parus major L). Giervalk 61: 125–135. for allowing us the use of their bird-banding records Dunn PO & Winkler DW (2010) Effects of climate (permit No. 30–44). change on timing of breeding and reproductive suc- cess in birds. In: Dunn PO & Moller AP (eds) Effects REFERENCES of Climate Change on Birds. pp 113–128. Oxford Univ Press, New York. Acharia BK & Chettri B (2010) Effect of climate change Gordo O (2007) Why are bird migration dates shifting? on birds, herpetofauna and butterflies in Sikkim A review of weather and climate effects on avian Himalaya: a preliminary investigation. In: Arrawatia migratory phenology. Clim Res 35: 37–58. ML & Tambe S (eds) Climate Change in Sikkim Pat- Huin N & Sparks TH (1998) Arrival and progression of terns. pp 141–160. IPR Department, Government of the Swallow Hirundo rustica through Britain. Bird Sikkim, Gangtok. Study 45: 361–370. Both C, Bouwhuis S, Lessells CM & Visser ME (2006) IPCC (2013) Climate change 2013: the physical sci- Climate change and population declines in a long- ence basis. Contribution of Working Group I to the distance migratory bird. Nature 441: 81–83. fifth assessment report of the Intergovernmental Buskirk JV, Robert S, Mulvihill L & Leberman RC Panel on Climate Change. Available at http://www. (2009) Variable shifts in spring and autumn migration climatechange2013.org/report (accessed on 23 Octo- phenology in North American songbirds associated ber 2018). with climate change. Glob Change Biol 15: 760–771. Ito Y (2006) Niigata-shi Sekiya kaigan hoan-rin no Carey C (2009) The impacts of climate change on the choso (Avian composition in a coastal protection annual cycles of birds. Philos T R Soc B 364: 3321– forest of Sekiya in Niigata city). Bull Niigata Pref 3330. Branch, Jpn Wild Bird Soc 61: 2–8 (in Japanese). Chambers LE, Altwegg R, Barbraud C, Barnard P, Japan Meteorological Agency (2018a) Climate change Beaumont LJ, Crawford RJM et al. (2013) Pheno- monitoring report 2017. Available at http://www. logical changes in the Southern Hemisphere. PLoS jma.go.jp/jma/en/NMHS/ccmr/ccmr2017_low.pdf ONE 8: 1–12. (accessed on 12 November 2018). Charmantier A & Gienapp P (2013) Climate change and Japan Meteorological Agency (2018b) Kako no kisho timing of avian breeding and migration: evolutionary deta kensaku (The data retrieval system of Japan versus plastic. Evol Appl 7: 15–28. Meteorological Agency). Available at https://www. Cleland EE, Chuine I, Menzel A, Mooney HA & data.jma.go.jp/obd/stats/etrn/index.php (accessed on Schwartz MD (2007) Shifting plant phenology in 4 January 2019) (in Japanese). response to global change. Trends Ecol Evol 22: Jenni L & Kery M (2003) Timing of autumn bird migra- 357–365. tion under climate change: advances in long-distance Cotton PA (2003) Avian migration phenology and migrants, delays in short-distance migrants. P R Soc global climate change. Proc Natl Acad Sci USA 100: B 270: 1467–1471. 12219–12222. Kobori H, Kamamoto T, Nomura H, Oka K & Primack Deguchi S, Chiba A & Nakata M (2015) Kaigan R (2012) The effects of climate change on the phe- kuromatsu-rin ni okeru hanshoku-ki no aoji no nology of winter birds in Yokohama, Japan. Ecol Res shokusei kozo ni kanren shita seisoku basho sentaku 27: 173–180.

52 Migration timing based on bird banding

Koike S & Higuchi H (2002) Long-term trends in the R Core Team (2019) R: A language and environment egg-laying date and clutch size of Red-cheeked Star- for statistical computing. R Foundation for Statisti- lings Sturnia philippensis. Ibis 144: 150–152. cal Computing, Vienna, Austria. Available at https:// Kolářová E, Matiu M, Menzel A, Nekovář J, Lumpe P www.R-project.org/ (accessed on 12 April 2019). & Adamík P (2017) Changes in spring arrival dates Sapir N, Wikelski M, Avissar R & Nathan R (2011) and temperature sensitivity of migratory birds over Timing and flight mode of departure in migrating two centuries. Int J Biometeorol 61: 1279–1289. European bee-eaters in relation to multi-scale meteo- Lawler JJ, Shafer SL, White D, Kareiva P, Maurer EP, rological processes. Behav Ecol Sociobiol 65: 1353– Blaustein AR et al. (2009) Projected climate-induced 1365. faunal change in the Western Hemisphere. Ecology Seki S, Nishiumi I & Saitoh T (2012) Distribution of 90: 588–597. two distinctive mitochondrial DNA lineages of the Marra PP, Francis CM, Mulvihill RS & Moore FR Japanese Robin Luscinia akahige across its breed- (2005) The influence of climate on the timing and rate ing range around the Japanese islands. Zool Sci 29: of spring bird migration. Oecologia 142: 307–315. 681–689. Mills AM (2005) Changes in the timing of spring and Sinelschikova A, Kosarev V, Panov P & Baushev AN autumn migration in North American migrant pas- (2007) The influence of wind conditions in Europe on serines during a period of global warming. Ibis 147: the advance in timing of the spring migration of the 259–269. Song Thrush (Turdus philomelos) in the south-east Ministry of the Environment (2017) Stop global warm- Baltic region. Int J Biometeorol 51: 431–440. ing 2017. Available at https://www.env.go.jp/earth/ Sokolov LV (2006) Effect of global warming on the ondanka/knowledge/Stop2017e.pdf (accessed on 5 timing of migration and breeding of passerine birds November 2018). in the 20th Century. Entomol Rev 86: 59–81. Møller AP (2013) Long-term trends in wind speed, Sparks TH & Mason CF (2001) Dates of arrivals and insect abundance and ecology of an insectivorous departures of spring migrants taken from Essex bird bird. Ecosphere 4: 1–11. reports 1950–1998. Essex Bird Rep 1999: 154–169. Nakata M, Chino N, Chiba A, Komatsu Y, Itoh Y, The Ornithological Society of Japan (2012) Check-list Akahara K et al. (2011a). Niigata-shi no kaigan-rin of Japanese birds. 7th ed. The Ornithological Society ni okeru chōrui no shunki torai jiki no keinen henka of Japan, Sanda (in English and Japanese). to kion to no kankei (Chronological trends in the Tojo H (2015) Seasonality of the bird community in the timing of spring bird migration and its relationship Ogawa Forest Reserve, an old-growth deciduous for- with temperature in a coastal forest near the city of est in central Japan. Ornithol Sci 14: 79–88. Niigata). Jpn J Ornithol 60: 63–72 (in Japanese). Vengerov PD (2015) Сроки весеннего прилета птиц Nakata M, Minami S, Chiba A, Itoh Y, Komatsu Y, в Воронежском заповеднике на фоне длительных Akahara K et al. (2011b). Niigata-shi no kaigan- климатических изменений (Spring arrival dates of rin ni okeru chorui hokaku-su no keinen henka to birds in the Voronezh reserve against the background shinrin sen’i to no kankei (Chronological trends in of long-term climate change). Bull Belgorod State the number of birds captured and the relationship to Univ Ser Sci 3: 82–92 (in Russian). forest succession in a coastal forest near the city of Wild Bird Society of Japan (2008) Firudo gaido, Nip- Niigata). Jpn J Ornithol 60: 178–190 (in Japanese). pon no yacho (A field guide to the birds of Japan). Niigata Prefectural Branch of Wild Bird Society of Wild Bird Society of Japan, Tokyo (in Japanese). Japan (2010) Niigata-ken no tori, Niigata-ken chorui Yamaguchi Y & Nakata M (2008) Shigai-chi ni rinsetsu mokuroku (Check-list of birds of Niigata). Niigata shita kaigan kuromatsu-rin ni okeru joryoku koyoju Prefectural Branch of Wild Bird Society of Japan, no shin’nyu to seicho—Niigata-ken Niigata-shi no Niigata (in Japanese). jirei (Growth status of evergreen broad-leaved trees in Ogawa-Onishi Y & Berry PM (2013) Ecological a coastal Pinus thunbergii forest adjacent to an urban impacts of climate change in Japan: The importance area—A case study in Niigata City, Niigata Prefec- of integrating local and international publications. ture). J Jpn Soc Coast For 7: 7–15 (in Japanese). Biol Conserv 157: 361–371. Zhou LM, Tucker CJ, Kaufmann RK, Slayback D, Palm V, Leito A, Truu J & Tomingas O (2009) The Shabanov NV & Myneni RB (2001) Variations in spring timing of arrival of migratory birds: Depen- northern vegetation activity inferred from satellite dence on climate variables and migration route. Ornis data of vegetation index during 1981 to 1999. J Geo- Fennica 86: 97–108. phys Res-Atmos 106: 20069–20083.