CLIMATE RESEARCH Clim Res Vol. 54: 95-112, 2012 doi: 10.3354/cr01099 Published online September 12 Drivers of population variability in phenological responses to climate change in Japanese birds Osear Gordo!, Hideyuki Doi 2 ,3,' lDepartment oi Zoology and Physical Anthropology, Complutense University oi Madrid, José Antonio Novais 2, 28040 Madrid, Spain 2Institute ior Chemistry and Biology oi the Marine Environment, Carl von Ossietzky University Oldenburg, Schleusenstrasse 1, 26382 Wilhelmshaven, Germany 3Presenl address: Institute ior Sustainable Sciences and Development, Hiroshima University, 1-3-1 Kagamiyama, Higashi-Hiroshima, 739-8530 Japan ABSTRACT: The impact of climate change on the bird communities of Asia is poorly understood. Sinee lhe 19505, lhe Japanese Meleorology Ageney has recorded !irsl arrival (Le. !irsl sighting) and first singing for a selection of resident ,:A1auda arvensis, Cettia diphone, and Lanius bucephalus) and migratory species (Hirundo rustica and Cuculus canorus) in more than 300 bird populations. Firsl records (Le. !irsl sighting or singing) showa delay 015.4 d sinee lhe end 01 lhe 1 Q70s. Nevertheless, there is él mélrkeo heterogeneity in the temporél 1 trenos élmong populéltions in each species. Most populations of A. arvensis, C. diphone and H. rustica show a negative relationship with local temperature (Le. first records were earlier in warmer years) and this sensitivity to temperatures has increased in recent decades. Exploration of the possible causes of variability in phenological trends among populations demonstrated that greater delays were observed in those populations subjected to smaHer increases in local temperature and a greater increase in human population (a surrogate for the conservation status of bird populations). Therefore, declining bird populations are the most probable cause of the observed delay in the phenology of first individuals. Migratory species were affected by clima te in theirwintering and passage areas. Overall, first sightings of H. rustica were earlier, while the onset of singing by C. canorus was delayed in response to warmer temperatures in southeastern Asia. However, there was a noteworthy variability among populations, with no discernable regionalization or spatial organization. This suggests that there is no clear connectivity between breeding and wintering populations. KEY WORDS: Arrival dale· Heal island effeel . Long-lerm sludy . Migralory birds . Phenology . Singing onset . Temperature . Warming - - - - - - - - - - - R e s a l e or republication not permitted viÍthout lNTitten consent 01 the pubJisher - - - - - - - - - - - It is widely aeeepled lhal global clima le is ehanging, and organisms are responding to this change in a variely 01 ways (Parmesan 2006). Mosl 01 lhe evidenee has be en provided by long-term studies, which usuaHy focus on single or a few populations of one or a few species. Consequently, there are just one or a few replica tes for many studies addressing the potential effect of climate change on organisms. Phenological studies are an exception. Phenological data are plentiful, and thousands of time series have be en studied lo dale (Lehikoinen el al. 2004, Menzel el al. 2006, Rubolini el al. 2007). These sludies demonslrale a common signal in the life cycles of many organisms in response to recent changes in climate, especially temperature increase (Root et al. 2005, Menzel et al. 2006). Nevertheless, responses to climate change have an important species-specific component, and physiology, ecology, or phylogeny may conslrain lhe *Corresponding author. Email: doih@hiroshima-u.ac.jp © Inter-Research 2012 . www.int-res.com 1. INTRODUCTION 96 Clim Res 54: 95-112, 2012 observed responses. For instance, Willis et al. (2008) demonstrated that evolutionary history is essential for an understanding of flowering time responses of species in plant communities. Similarly, Gordo & Sanz (2009, 2010) lound that pollination system is an important characteristic that influences phenological trends of flowering dates in Mediterranean ecosystems. Among migratory birds, migratory distance affects phenological response both in arrival (Tryjanowski et al. 2005, Lehikoinen et al. 2004, T0ttrup et al. 2006a, Hubálek & Capek 2008, but see Hüppop & Hüppop 2003) and departure dates (Jenni & Kéry 2003, T0ttrup et al. 2006b, Van Buskirk et al. 2009). Other traits, such as size, diet, moult or the number of broods per season, have also been linked to the extent of bird responses to climate change (Jenni & Kéry 2003, Végvári et al. 2010). Buller (2003) also lound that the advance in arrival dates over the last 100 yr was related to the habitat used by each species. For instance, the arrivals of grassland birds were more advanced than those of forest birds, probably beca use of the uneven effects of clima te change on the trophic level on which birds 01 each habitat depend (plants and insects, respectively). Such species-specific phenological responses are more than a mere difference in the number of days that each species shifts its phenology. Different rates of advance are the most plausible underlying cause for the uneven trends in population numbers observed among migratory bird species during recent decades (M011er et al. 2008). Those species that are less responsive, such as long-distance migrants breeding in strongly seasonal habitats, suffer the greatest declines due to increased mismatch with their environment (Both et al. 2006a, 2010). While there is now sorne understanding of why species differ in their responses to climate change, and of the biological consequences of these differences, almost nothing is known at the population level (what are the reasons for different responses of populations belonging to the same species?) or at the individual level (why are sorne individuals able to adjust to the new conditions while others are not?). Studies at intra-specific levels are probably limited by the scarcity of long-term monitoring data that are compatible for different populations of the same species. The few that exist include sorne studies of laying dates of species such as tits Parus spp. (Visser et al. 2003), pied flycatcher Ficedula hypoleuca (Both et al. 2004, 2006b, Both & te Marvelde 2007), and tree swallow Tachycineta bicolor (Dunn & Winkler 1999). These studies demonstrated that those populations subjected to larger increases in local temperatures show greater advances in their laying phenology. In the case 01 migratory phenology, the differences in the passage periods and areas for each population have been suggested as the origin for the observed differences in the temporal trends of mean passage time of birds among ringing stations (Sparks et al. 2005a, Hüppop & Winkel 2006, Jonzén et al. 2006, T0ttrup et al. 2008). While the number 01 sites with long-term ringing schemes is extremely reduced, other measures of migratory phenology, such as the first arrival date, are plentiful and would allow a more detailed exploration of variability patterns with a higher resolution and over larger areas. However, there are very few studies explaining how and why arrivals at the breeding grounds by populations of the same species are advancing at different rates (but see Forchhammer et al. 2002, Ahas & Aasa 2006). The heterogeneous nature of monitoring schemes from each region or country, as well as differences in the study periods, make comparisons difficult (Gordo 2007a, Miller-Rushing et al. 2008c, but see Rubolini et al. 2007). Nonetheless, it is 01 paramount importance to disentangle the causes of the observed variability in phenological trends, to understand the extent to which it is due to spatial variability of changes in climate, to the different abilities of populations to respond to these changes, or to other factors also operating in the long-term at local scales and affecting the responses of each particular population. We propose 3 possible explanations for differences in phenological trends among populations: (1) Differences in the magnitude 01 climate change experienced. The larger the change in clima te over time, the larger the phenological response expected (e.g. Both et al. 2004). (2) Differences in the sensitivity or responsiveness to climate. Populations more sensitive to clima te show larger responses in their phenology (e.g. Gordo & Sanz 2010). (3) Differences in the population status. Changes in population abundance affect our ability to detect the earliest individuals, and consequently long-term changes in population size may confound trends in first annual records of any phenologcial event (Tryjanowski & Sparks 2001, Tryjanowski et al. 2005, Miller-Rushing et al. 2008a,b). To examine these hypotheses, we studied an exceptional dataset with phenological information on more than 300 bird populations across Japan. Japan provides a broad latitudinal range. Bird populations of widely distributed species in Japan occur from subtropical clima tes in the south (25 0 N) to hemiboreal ones in the north (45° N). Thus, we can compare phenological responses to the same event of a certain Gordo & Doi: Climate change and bird phenology species across a wide range of environmental conditions (Doi & Takahashi 2008). Our study also aimed to provide evidence about phenological responses of birds to clima te change in the far eastern part of the Asian continent, an area in which lhis issue has been barely explored (Koike & Higuchi 2002, Gordo 2007b, Walanuki 2011, Kobori el al. 2012). In a rece nI sludy 01 planl and animal phenology in Korea and Japan, Primack el al. (2009b) explored how phenological matching between trophic levels may be disrupled by geographical and species-specific variation of phenological responses to climate. However, the community-Ievel approach carried out in that study does not allow a comprehensive scrutiny of the phenological responses of the 3 bird species studied. Moreover, as the authors recognized, lhe sludy period (November lo March) was prabably nol lhe mosl suitable lor lesting lhe effecl 01 temperatures on the phenology of animal species (Primack el al. 2009b). Periods 01 climale sensitivity are species-specitic (Doi el al. 2008), consequently, Primack el al. (2009b) may have underestimaled lhe aclual effecl 01 lemperalure on lhe phenology 01 birds from Japan and Korea. Finally, long-distance bird migrants from Japan provide an opportunity to study the effects of climate change in a migratory system that has rarely been studied, the eastern Palearctic. Because western and eastern populations of Palearctic bird species do not share passage and wintering areas, they are subjected to different environmental pressures, and it is important to ascertain the extent to which these promote different phenological responses. 2. DATA AND METHODS 2.1. Phenological dala The Japanese Meleorological Agency (JMA) monitored the phenology of 5 common bird species belween 1953 and 2005 al 102 sites spread lhraughoul Japan (Iatilude: 24.33 lo 45.41 'N, longitude: 123.01 lo 145.59' E, altilude: O lo 610 m). Observations 01 plant and animal phenology were carried out daily by prolessional slaff al each 01 lhe JMA meleorological stations according to standardized protocols (JMA 1985), which assure high reliabilily and comparability of records among observers. Furthermore, any change in the monitoring effort, which may affect species detectability, was improbable, because sampling prolocols did nol change during lhe sludy periodo 97 The 'first singing day' for each species was defined as the first date of the year when the species was heard singing around the respective JMA station. This date was examined for the Japanese skylark Alauda arvensis (subsp. japonica Temminck & Schlegel, 1848), lhe Japanese bush warbler Cettia diphone (subsp. cantans Temminck & Schlegel, 1847), the common cuckoo Cuculus canorus (subsp. subtelephonus Zarudny, 1914), and lhe bul!-headed shrike Lanius bucephalus (subsp. bucephalus Temminck & Schlegel, 1845). In lhe case 01 lhe tirsl singing day refers to the higher-tone song used to defend wintering territories in autumn. The date when the first barn swallow Hirundo rustica (subsp. gutturalis Scopoli, 1786) was seen (tirsl arrival dale) was also recorded. A. arvensis, C. diphone, and L. bucephalus are resident birds, while H. rustica and C. canorus are long-distance migrants. Japanese populations of H. rustica and C. canorus overwinter in Indochina, Philippines, the Indonesian Archipelago, New Guinea, and even in locations as distant as Auslralia (Brazil 1991, Fig. Si in lhe supplemenl al www.inl-res.comlarticles/suppl/c054p095.pdl).Al! species are common and familiar enough to avoid misidentification. Dates were transformed into a day ofyear (DOY) scale (1 = 1 January). In leap years, one day was added alter February 28. The validity 01 dala on lhe phenology 01 tirsl individuals has be en strongly criticized due to a number of potential biases that may be present in these records (Sparks el al. 2001, Tryjanowski & Sparks 2001, Tryjanowski el al. 2005, Miller-Rushing el al. 2008b). The main drawback 01 recording tirsl individuals is that they may show migratory behaviour that is unrepresentative of the whole population. However, each species studied had records for many Japanese sites per year¡ thus errors or vagueness in sorne records (an unavoidable problem when working with large dala bases) are offsel by the large number of samples covering an extensive geographical area (Gordo & Sanz 2006, Barretl 2011). Ralher lhan a single arrival dale per year, lhe data shows the distribution of first arrival dates for each species from southern to northern Japan each year. 2.2. Climale dala JMA provided monthly average temperatures lor each sludy site lor lhe period 1953 lo 2005. We also gathered monthly temperatures for this period lram lhe Global Hislorical Climalological Nel- Clim Res 54: 95-112, 2012 98 work (GHCN) databas e Version 2 (Peterson & Vose 1997, Peterson et al. 1995) lor lOS weather stations in southeastern Asia (latitude: 10 to 31 e N, longitude: 92 to 125°E), the main wintering and passage area for the Japanese populations of migratory birds (see Fig. Si). 2.3. Temporallrends We divided the study period into 2 sub-periods: 1953-1979 and 1979-2005. We adopted this proeedure beca use temperature did not change steadily throughout the entire period (Fig. 1). Between 1953 and 1979, temperature did not show any significant trend, while during the period 1979 to 2005, temperatufe increased markedly, in accordance with global warming (IPCC 2007). We established the break point for the phenological time series in 1979 to obtain 2 equallength sub-periods 0127 yr. To study the temporal trends 01 bird phenology we carried out 10 multiple regression models (5 species x 2 periods). Each model included site as a categorical factor and year as a discrete explanatory variable. The model was: Yij = a +ai x Sitei+ ｾ＠ x Yearj+ ｾｩ＠ x Sitei xYearj+ Eij (1) where Yij is the phenological date of a species recorded in Site i during Year j, ai is the intercept for r-----------------------------------,28 -0- Spring 27 -----.- Summer 26 ｾ＠ 25 ::J "E ID Q. ｾ＠ E 16 9 24 o .:'\0 15 o 14 O:, O 13 1950 1960 1970 1980 1990 2000 2010 Fig. 1. Temporal trends oí spring and summer temperatures in Japan during 1953 to 2005. Spring: average between February and April¡ summer: average oí August and September. These were the most important pericx:ls íor the phenology oí the bird species. Lines: fitted linear regression in each subpericx:l (solid: p < 0.05, dotted: non-significant). Temperature increased at a similar rate in both seasons during 1979 to 2005 (spring +0.049°C yr- 1, summer +0.045°C yr-l) Site i, ｾ＠ is the overall slope with Year, ｾﾡ＠ is the slope 01 the fitted Hne to each Site i, and 8ij is the error for eaeh observed phenologieal date. Therelore, the ｾ＠ parameter determines whether phenological dates show a significant or non-significant temporal trend (Le. slope) lor ea eh sub-period and species, while the interaction between Site and Year determines whether or not trends differ significantly among populations. We only analyzed time series with records for at least 20 years during each sub-period, to reduce potential bias es caused by differences in the length of the time series. It is important to note that the average number of records per time series was 25.0 (±2.0 SD) in both sub-periods, Le. most 01 the time series are complete since 1953. 2.4. Ellecls 01 local lemperalures To determine the effect of temperature at the study sites on bird phenology, we built 10 new multiple regression models including local temperature as a continuous explanatory variable. Models were: Yij = a + ai x Sitei + ｾ＠ x Yearj + ｾｩ＠ x Sitei x Yearj + y x Tempij + Yi x Sitei x Tempij + 8ij (2) where the y parameter estima tes the overall effect of temperature in bird phenology, and the Yi parameters estimate the particular sensitivity of each population to temperature. Therefore, the interaction term between Site and Temperature tests for differences in sensitivity to local temperatures among populations (Le. heterogeneity 01 the slopes y¡). We included Year as a covariate in this model to account for its potential confounding effect on the true relationship between climate and phenology. Phenology 01 !irst individuals may show long-term shifts unrelated to climate, e.g. due to ehanges in population size (Sparks et al. 2001, Tryjanowski & Sparks 2001, Tryjanowski et al. 2005, Miller-Rushing et al. 200Sb). Henee, this may give rise to spurious relationships with climate due to temporal collinearity. Temperature values were calculated as the average during the month when a phenological event occurred and the preceding month at the same JMA site. We used average temperature values calculated for a 2 mo period beca use the average standard deviation of phenological dates was 10.1 d, and consequently 95 % of all phenological records of a certain time series were included in an average range of 40 d (Le. ±2 SD). Once again, only those time series with at least 20 records during each sub-period were used for these models. Gordo & Doi: Climate change and bird phenology 2.5. Variabilily among populalions Causes of variability in temporal trends among populations during 1979-2005 were studied according to the 3 hypotheses proposed in the lntroduction. We focused on this period, as most of the climate warming recorded in Japan occurred during this time (see Fig. 1). Differences in the magnitude of climate change experienced at a local scale were quantified as the increase in temperatures (slope against year) from 1979-2005 during the 2 mo period adjusted to each population phenology. Differences in the sensitivity to climate were quantified by the parameterYi (Eq. 2). To quantify the status of bird populations, we used the percentage increase in the number of inhabitants between the 19S0 and 2005 censuses. This information was provided by the Statistical Bureau of the Japanese Ministry of Internal Affairs and Communications (www.stat.go.jp/index.htm) in the form of indo km- 2 for each Japanese municipality (average area of the 102 municipalities covered by this study = 412 km 2 , range = 2S to 21S7 km 2 ). Observed changes in the number of people in Japanese municipalities accurately minor landscape transformations at a small scale of a few km 2 in urban ce nters and their peripheries. One may assume that the environment suffers the greatest impacts in those areas with the highest numbers of humans (McKinney 2002). Therefore, one expects scarcer availability and poorer quality of natural habitats at those more populated sites and consequently smaller populations of wild organisms, including birds (Beissinger & Osborne 19S2, Mason 2006, Shaw et al. 200S). The land areas of Japanese cities and towns showed extraordinary growth during the second half of the 20th century as urban populations increased (Hanis 19S2, Yamada & Tukuoka 1996). Urban sprawl has caused the deterioration and even disappearance of natural and traditional Japanese landscapes, dominated by agriculturalland use, in the areas surrounding many cities and towns (Nakamura & Short 2001). These areas are the main habitats for local populations of birds and where the JMA monitors bird phenology. In spite of the fact that sorne bird species tolera te human presence (e.g. Hirundo rustica). We expect those JMA sites with a grea ter increase in human population and urban sprawl since 1980 to have a more impacted environment. Therefore, human population increase can be used as a proxy for environmental degradation and consequently for long-term declines in local bird populations. We carried out a multiple regression model for each species with phenological trends (values of ｾｩ＠ 99 from Eq. 1) as a dependent variable and temporal trends of local temperatures from 1979 to 2005 (regression slopes of temperatures from each study site against year), sensitivity to temperature (values of Yi from Eq. 2), and human population change between 19S0 and 2005 as explanatory variables. Residuals from these models were used as response variables in new multiple regression models, which included latitude, longitude, and altitude of the study sites as predictors. The aim of the latter models was to explore any remaining spatial structure in the residuals of the temporal trends observed in bird phenology (Fig. S2 and Table Si). Quadratic terms of latitude, longitude, and altitude were also included to account for potential nonlinear spatial gradients (Legendre & Legendre 1995). We also included the interaction between latitude and longitude. All spatial predictors were standardized (mean = O, SD = 1) to avoid scale measurement effects on matrix calculations. The best model was selected according to Mallow's Cp. Similar to other information criteria, Mallow's Cp chooses the best subset of predictors that optimizes the balance between model fit and the number of variables. 2.6. Migratory species In the case of Hirundo rustica and Cuculus canorus, we also as ses sed the potential effect of clima te in wintering and passage areas, because the departure date from wintering grounds and progression rate through passage areas is affected by weather and ecological conditions there (Gordo 2007b, Gordo & Sanz 200S). All phenological time series with at least 20 records during 1953-2005 for Hirundo rustica and Cuculus canorus (S6 and 25 JMA-sites, respectively) were regressed against the temperature time series of all the weather stations from southeastern Asia selected from the GHCN (lOS sites). We used temperature records for February and March for H. rustica, and data for March and April for C. canorus. Time series of both bird phenology and temperatures were detrended prior to implementing the regressions. We removed temporal trends (Le. year effect was set to O) to avoid potential spurious relationships due to temporal collinearity between phenological and climatic time series. For this purpose, all the time series were regressed against the year and the resulting residuals from these regressions were used in the analyses. Regression slopes between climate and phenology were organized in 2 matrices (one for each species), 100 Clim Res 54: 95-112, 2012 where columns were the Japanese populations of Hirundo rustica or Cuculus canorus and rows the 108 weather stations trom southeastern Asia. Therefore, each colunm was the sensitivity pattern of one population to the spring temperatures in southeastern Asia. A Principal Componenl Analysis (PCA) was applied to each one of the matrices with the Japanese bird populations (Le. columns) as dependent variables. The aim of these peAs was to summarize aH sensitivity patterns into a few principal components (PCs), which could help in idenlifying common palterns of sensitivity among the Japanese populations of H. rustica and C. canorus. We selected only those pes with an eigenvalue higher than 5. pe scores were mapped and visually inspected to detect the most influential areas in southeastern Asia for bird phenology. PC loadings provided lhe relalionship 01 each Japanese populalion 10 lhe pallerns depicled by those PC scores. Therefore, loadings allowed us to define groups of populations with similar sensitivity pa tterns to tempera tures in wintering and passage areas. We modelled the spatial structure of these PC loadings across Japan using multiple regression models wilh lalilude, longilude, and allilude 01 phenological sites as predictors and following the same procedure as previously described. The best model was selected using Mallow's Cp. The aim of these models was to determine more reliably whether Japanese populations with similar sensitivity to temperature in southeastern Asia have a regional distribution pattern in Japan or whether they are randomly distributed across the country. All statistical analyses were carried out with STATISTICA soflware ISlalSofl, Inc. (2004), version 7. (www.slalsofl.com)l· 3. RESULTS 3.1. Temporallrends All species showed different trends between periods (Fig. 2). During lhe sub-period 1953-1979, !irsl records of singing males of Alauda arvensis and Cettia diphone were markedly delayed, while !irsl sighlings 01 Hirundo rustica were advanced. The onset of singing by Cuculus canorus and Lanius bucephalus tended to advance, but not significantly. With the exception of C. canorus, all species showed significant differences in temporal trends among populations (Site x Year inleraclion: p < 0.00001 in al! cases, Fig. 3). During lhe more recenl sub-period 1979-2005, al! species showed a positive relationship with Year, Le. lhere was an overal! delay in lhe phenology 01 lhe !irsl individuals (Fig. 2). Neverlheless, lhis delay was only significant for Hirundo rustica, Cuculus canorus and Lanius bucephalus. We found strongly significant differences among populations in all species (Site x year inleraclion: p < 0.00001 in al! cases, Fig. 3). In the case of Alauda arvensis and Cettia diphone, this heterogeneity among populations was especially marked, Le. there were both significant delays and advances (Fig. 3), which counlerbalanced each olher and precluded any statistically significant temporal lrend lor lhe whole sel 01 popula lions (Fig. 2). 3.2. Ellecl 01 local lemperalures First individuals of Alauda arvensis, Cettia diphone and Hirundo rustica were detected earlier in response to warmer springs, while Cuculus canorus and Lanius bucephalus did not show any relationship wilh locallemperalures (Table 1). The effecl 01 lemperalure was slronger during 1979-2005 lhan during 1953-1979 in lhose species lhal responded significantly to local temperature, (repeated mea sures ANOVA: Fl,146 = 6.491, P = 0.012, Fig. 4). There were significant differences in the sensitivity to temperatures among populations in sorne species and sub-periods (Fig. 4). Alauda arvensis, Cettia diphone and Hirundo rustica showed differences among populations (Site x Temperature interaction: p < 0.01 in al! cases) during 1953-1979, whereas during 1979-2005 only C. diphone slill showed such heterogeneity (Site x Temperature interaction: F6s.154s = 1.523, P = 0.004). Differences among populations were due to differences in the steepness of slopes, Le. the phenology of first individuals was more advanced in sorne populations than in others in response to a given temperature increase. Cuculus canorus and Lanius bucephalus did not show significant differences among populations (Site x Temperature interaction: p > 0.42 in all cases). Therefore, in these species, the absence of an overall temperature effect was real and not a result of countervailing patterns among populations. 3.3. Variabilily among populalions Overall, there were few statistically significant effects among the selected explanatory variables, and the explanatory capacity of models was low (Table 2). Neverlheless, lhe resul1s supporl our predictions. With the exception of Alauda arvensis, tem- 101 Gordo & Doi: Climate change and bird phenology 84 Alauda arvensis 1 ｾ＠ 80 76 = 0.208 ¡ I 1 11' 1I 11 11 i! 11 1¡ 1[[' ¡,III,I' Ｎｬﾡｉｾ Ｍ ｲｈｮｪｦＡﾡ｛ Ｍ ,¡ 111[111 1 [I¡j¡l¡ 1 I¡ ¡ 1 n = 48 I 72 68 64 I 1 I 60 88 1 ｾ＠ 1 = 0032 n = 53 Cettia diphone 84 ｾ＠ al ｾ＠ 80 <D -E 76 O ｾ＠ O 72 68 n = 44 64 276 1953 Lanius bucephalus 1969 1985 Year 2001 272 268 264 260 256 1953 1969 1985 2001 Fig. 2. Day oí year (1 = 1 January) oí bird phenology during 1953-2005. Error bars show the standard error oí annual means. The estimate íor the year effect írom the multiple regression models (13 in Eq. 1) and the number oí time series used (n) during each sub-period are shown. A solid (p < 0.05) or dotted (non-signiíicant) line represents fue estimated 13 Year paral trends in local temperature during 1979-2005 had a negative effect. This variable was non-significant in A. arvensis, while in the other species p-values were low (Table 2). Therelare, the larger the recorded warming during 1979-2005, the larger the advance in population phenology. Sensitivity to temperature did not show any significant effect. This result is not surprising if one takes into account that there were no differences in Yi among populations except in the case of Cettia diphone. In agreement with our hypothesis, those bird populations from sites with the largest increases in human population showed the greatest delays in their first singing or arrival dates. Changes in human population numbers that oeeurred sinee 1980 at the study siles had a significant positive effect on the phenology of A. arvensis and C. diphone (Table 2). This variable also had a marginal positive effect in Hirundo rustica. The amount of variability in temporal trends accounted for in significant models ranged from 13 to 22 % (Table 2). Those species wilh the weakest models (Hirundo rusticaa, Cuculus canorus and Lanius bucephalus) showed significant spatial gradients in their residuals (see Fig. S2 and Table Si in the supplement). This suggests that part 01 the variability among populations is due to the location of the study site. For instance, delays in first arrival dates were especially marked in southern populations of H. rustica, while trends in singing onset dates of L. bucephaluswere increasingly delayed along a southwestern to northeastern axis (see Fig. 3). In the case of C. canorus, a second order polynomial of longitude accounted for more than 60 % of variability in the residuals of first singing dates. Populations located in the eastern and western extremes of the distribution in Japan showed the largest delays in the singing Clim Res 54: 95-112,2012 102 A. arvensis C. diphone C. canorus H. rustica L. bucephalus • , 4' 4' ) ;j ·· ® N ｾ＠ ., .. 4' ·· 4' ) ) • • ｾ＠ . ., 4' ｾ＠ ) ｾ＠ .. " Trend (d yr- 1) '. '. 4' ·, '. '. '. 4' . ., ' 4' · · ., ) ) , , .. , • . '. 4' '{ .. ｾ＠ ｾ＠ " • ｾ＠ <-1.00 • • - 0.75 - 0.50 - 0.25 0.00 0.25 0.50 0.75 >1.00 j •· ) • . ｾ＠ ••• • Fig. 3. Temporal !rends 01 bird phenology during 1953-1979 and 1979-2005. Each do! is a site. FuI! species na mes in Fig. 2 Gordo & Doi: Climate change and bird phenology A. arvensis c. diphone 103 C. canorus H. rustica L. bucephalus • • ,· }· ® N .,1 , , .·· , .- ·· } Ｎｾ＠ } ｾ＠ .,1 • .-· ,-. } · .,1 .,1 • Sensitivity (d °C-1) .\ , .-. .· Ｎ ｾ＠ ...-· '. '. .,1 ...-· '. ［ｾ＠ } .,1 , ｾ＠ •ｾ＠ • .". 0 , '. '. • . .-. · , . '. 4' ) } J .,1 • <- 7.6 • • - 5.7 - 3.8 -1 .9 0.0 1.9 3.8 5.7 >7.6 • ••• • Fig. 4. Sensitivity 01 bird phenology to local temperatures during 1953-1979 and 1979-2005. Each dot is a site. FuI! species names in Fig. 2 Clim Res 54: 95-112, 2012 104 Table 1. Results for multiple regression models oí fue phenology oí 5 cornmonly occurring migrant and resident bird species in Japan, with local temperature and year as continuous explanatory variables and the Japan Meteorological Agency (JMA) site as a categorical factor (see Eq. 2 in 'Data and methods'). Parameter estimates for temperature (r) and year (13) Species Temperature r p 1953-1919 Alauda arvensis -2.317 <0.001 -2.856 <0.001 Cettia diphone -1.964 <0.001 Hirundo rustica -0.138 Cuculus canOHlS 0.692 Lanius bucephalus 0.373 0.383 1919-2005 Alauda arvensis -2.756 <0.001 -3.906 <0.001 -2.193 <0.001 Hirundo rustica Cuculus canOHlS 0.228 0.319 Lanius bucephalus 0.511 0.134 Cettia diphone Table 3. Hirundo rustica and Cuculus canorus. Results oí the principal component analysis (PCA) oí the sensitivity oí populations to spring temperatures in southeastern Asia. Eigenvalue (Eigenv.) and cumulative percentage oí variance accollllted íor (Cum %) shown íor selected principal components (PC) pe Year ｾ＠ p 0.174 0.143 -0.116 -0.014 -0.071 0.013 0.038 0.001 0.701 0.219 0.175 0.023 0.244 <0.001 0.298 <0.001 0.136 0.005 0.190 0.012 0.854 0.735 0.668 0.399 0.559 0.847 0.843 0.715 0.473 0.502 onset date. However, note the small sample size for this species and the weak temporal trends in its populalions (Figs. 3 & S2, Table Si). In sharl, lhose spedes weakly affected by variables sueh as local ehanges in tempera ture, sensitivity to tempera ture, and human population ehange showed sorne geographical regionalization in the temporal trends in lhe phenology 01 lheir !irsl individuals. 3.4. Migratory species We lound 5 PCs wilh an eigenvalue > 5 in lhe PCA for Hirundo rustica. The first 5 PCs aeeounted for almost 66 % of variability in patterns of sensitivity to temperatures in southeastern Asia both in February and March (Table 3). PCi depicls 2 areas wilh opposite effeets of temperature on H. rustica arrival dates. Temperatures on the eoast of China had negative seores, while in the remaining areas from Thailand to Indonesia and the Philippines, temperatures had positive seores. This pattern was important for many 1 2 3 4 5 Hirundo rustica Eigenv. Cum% 19.82 14.24 8.69 7.61 6.23 Cuculus canorus Eigenv. Cum% 8.20 32.81 23.04 39.60 49.70 58.55 65.79 populalions, as lhe large magnilude 01 lhe loadings demonslraled (Fig. 5). However, loadings lar populations breeding in eastern prefeetures of Honshu (Japan's largesl island) were mainly negalive, while in western prefeetures loadings were positive. In eastern populations (negative loadings), H. rustica advaneed its arrival in response to high tempera tures in the area from Thailand to Indonesia and the Philippines, while warmer temperatures in China delayed arrivals. However, Japanese populations of H. rustica breeding in western regions oí the country (posilive loadings) had lhe opposile response 10 climatie eonditions in southeastern Asia. Their arrival was advaneed in response to high temperatures in China during February and March, bul delayed in response to warm springs in the rest of southeastern Asia. This was con!irmed by a spalial model (Table 4). Its explanatory eapacity was low, but it had a markedly significant negative gradient from the southwest to the northeast of Japan. Loadings were more negative in northeastern populations. PC2 in Hirundo rustica showed a marked effeet of climate in the Philippines, northern Borneo, eastern Java and Timor. Other regions of southeastern Asia also showed notable seores, but there were no obvious spatial patterns. The overwhelming majority of Japanese populalions had negalive loadings, al1hough Table 2. Results oí multiple regression models oí inter-site variability oí temporal trends in bird phenology in Japan¡ p: significance of the model Species Alauda anrensis Cettia diphone Hirundo rustica Cuculus canorus Lanius bucephalus Temperature trend p ｾ＠ 0.260 -1.024 -0.444 -0.613 -0.975 0.472 0.009 0.026 0.111 0.092 Sensitivity to temperature p ｾ＠ -0.052 -0.016 0.041 0.016 0.054 0.112 0.491 0.080 0.799 0.189 Change in human population p ｾ＠ 0.0146 Om05 0.0039 -0.0013 0.0031 0.002 0.004 0.085 0.767 0.615 R2 p 0.223 0.212 0.128 0.185 0.091 0.007 0.003 0.029 0.398 0.301 Gordo & Doi: Climate change and bird phenology PC 1 105 PC4 • :( ., .) ® ® N N PC 5 , '. f • ., • Ｎｾ＠ .• ｾ＠ Scores <-2 .0 - 1.0 0.0 1.0 >2.0 Loadings <-0.7 -0.35 0.0 0.35 >0.7 •••••••••• ••••••• 'f .' • • Fig. 5. Hirundo rustica. Principal Component Analysis (PCA) of sensitivity (d OC- 1 ) of migratory phenology to temperatures in southeastern Asia during February-March. PC scores for the 108 weather stations from the Global Historical Climatological Network (GHCN) used in the present study are represented by dots on the maps of southeastern Asia. PC loadings for the Japanese bird populations are shown by dots on the maps of Japan Clim Res 54: 95-112, 2012 106 Table 4. Hirundo rustica and Cuculus canorus. Results of spatial models for principal component (PC) loadings. Parameter estimates (b) and their significan ce (p) are shown only for those variables included in the best model selected according to the lowest value of the Mallow's Cp. In the lower part of the table, the explanatory capacity (R 2 ), F-statistic (F), degrees of freedom (df) and significan ce (p) of the model are shown Variable PCl b Latitude Longitude Latitude 2 Longitude 2 Lat x Long Altitude Altitude 2 R2 F df P 1.298 -1.320 0.135 6.481 2, 83 0.002 PC2 P b P H. rustica PC3 b P C. canorus PC4 b PC5 P b PCl P 0.001 0.001 1.391 0.596 -1.946 -0.149 0.189 0.186 3.644 5, 80 0.005 0.004 0.001 0.002 0.118 0.045 -0.152 0.204 0.071 0.016 -0.09 -0.045 0.079 3.570 2, 83 0.033 they were of great magnitude only in sorne populations from central regions of Honshu. This PC accounted for only 14 % of variance. Therefore, this PC summarized correlation patterns for few J apanese populations. These populations were characterized by delayed arrivals under warmer temperatures in the Philippines, Java or Timor, and advanced arrivals under warmer temperatures in Borneo or northern China. Loadings for this PC showed several significant spatial gradients (Table 4). Significant quadratic latitudinal and longitudinal gradients suggested that largest loadings occur at mid-Iatitudes and mid-Iongitudes (Le. in central Japan). Similar to PC1, there was also a marked negative gradient from southwest to northeast. Altitude was of minor importance. PCs 3, 4 and 5 together accounted for just 26 % of variance (Table 3). Each of these PCs summarized sensitivity patterns representative for a few populations (se e small magnitude of loadings in Fig. 5). In all cases, climate in China had very low importance (low scores), while climate in other areas in southeastern Asia was the most important factor. Nevertheless, the effect of climate in these areas was variable. In PC3, there were markedly negative scores in the Malay Peninsula. In PC4, the Philippines and northern stations in Myanmar had positive scores, while most of the other stations had negative values. Finally, in PC5, regions from the Malay Peninsula to northern Laos and Myanmar had again positive scores, while the Philippines showed negative score. Spatial modeIs for PC loadings were weak or even nonexistent (Table 4). This fact can be interpreted as further proof that the sensitivity patterns depicted by these PCs are only representative of a few Japanese populations. 0.002 b P 8.045 0.072 1.658 -9.541 0.109 0.077 0.165 0.023 1.958 1, 84 0.165 0.110 10.43 1, 84 0.002 0.237 2.170 3,21 0.122 PCA for Cuculus canorus provided only one PC with an eigenvalue >5. However, this PC accounted for almost 33 % of variance (Table 3). Scores showed 2 regions in southeastern Asia with an opposite influence on the onset of singing dates of C. canorus (Fig. 6). The northern coast of China (near Shanghai) and the Philippines had positive scores, while the ® N ｾ＠ Scores <-2 .0 ® N -1 .0 0.0 1.0 >2.0 •••••••• ••••••••• Loadings <-0.7 -0.350 .0 0.35 >0.7 Fig. 6. Cuculus canorus. Principal Component Analysis (PCA) of sensitivity (d OC- 1) of migratory phenology to temperatures in southeastern Asia during March-April. PC scores for the 108 weather stations from the Global Historical Climatological Network (GHCN) used in the present study are represented by dots on the map of southeastern Asia. PC loadings for the Japanese bird populations are shown by dots on the map of Japan Gordo & Doi: Climate change and bird phenology region from the Malay Peninsula to Timor had negative scores. Loadings were negative and high in all populalions. Thus, al! populalions had similar palterns of sensitivity to climate in southeastern Asia. Taking the negative sign of loadings into account, we found that warm springs in the Philippines and China are related to earlier singing onset of C. canorus in Japan, while warm springs in the Malay Peninsula and equatorial regions of Indonesia were related to a delay in lhe onsel 01 singing. Loadings lor lhe C. canorus did not show any statisticaHy significant spalial slruclure (Table 4). In summary, temperature from southeastern Asia had effects on Hirundo rustica and Cuculus canorus phenology, bul such e!fecls varied regional!y. Palterns of sensitivity were much more diverse in H. rustica than in C. canorus, as the larger number of relevant PCs and heterogeneity of loading signs among populations demonstrated. 4. DISCUSSION 4.1. Trends in bird phenology Recorded dates of the first singing and arrival of migrant individuals during the spring in Japanese bird populations have not shown an advance in recent decades. This suggests that other factors operating on a long-term temporal scale are controlling phenological responses. For instance, Alauda arvensis, Cettia diphone and Hirundo rustica had earlier records in warm years, but none of these species showed a consistent tendency of advanced arrivals in recent decades, in spite of the notable increase in lemperalures in Japan (see Fig. 1). This result is interesting, as the spring phenology of populations of A. arvensis and H. rustica from the western Palearctic is indeed advancing in response to similar increases in temperatures (Tryjanowski et al. 2002, Buller 2003, Collon 2003, Peinlinger & Schusler 2005, Ahas & Aasa 2006, Croxlon el al. 2006, Gordo & Sanz 2006, Zalakevicius el al. 2006, Askeyev el al. 2009, Barrell 2011, Eddowes 2011, Sepp el al. 2011). One may obtain evidence for a hidden driver of bird phenology by comparing lhe year e!fecl ＨｾＩ＠ in models with (Eq. 2) and withoul (Eq. 1) lemperalure. AH species showed marked and significant delays (see ｾ＠ in Table 1) when lhe e!fecl 01 lemperalure was isolated. First records of Alauda arvensis, Cettia diphone and Hirundo rustica are actually advancing in response to warmer local temperatures: the difference belween lhe observed (see Fig. 2) and lhe lalenl 107 lemporal lrend (see Table 1) was -0.143, -0.191, and -0.109 d yc 1, respectively. However, such an advance promoted by warmer temperatures in Japan was not enough lo lul!y counlerbalance lhe lalenl delay and, as a result, A. arvensis and C. diphone did not show any temporal trend and first arrivals of H. rustica were even delayed. In the case of Cuculus canorus and Lanius bucephalus, observed and expected (Le. lalenl) changes in phenology were lhe same. These species were not affected by local temperature and consequently, this did not counterbalance their latent delay. Neverlheless, L. bucephalus phenology was studied during autumn, and its singing activity during this sea son cannot be interpreted in the same way as the the onset of singing in spring. L. bucephaJus phenology was posilively (bul non -significanlly) related to temperatures during August and September; therefore a delay in the onset of its singing is expected as a response to warmer temperatures. Moreover, L. bucephalus is a double-brooded species (Lelranc & Worlolk 1997) and consequenlly its breeding sea son finishes at the end of the summer. The beginning of the defence of winter territories is probably driven by lhe end 01 lhe breeding sea son, which is in turn affected by weather conditions in la te summer. Whal is lhis hidden driver lhal is able lo modily lhe expected advance of first singing and arrival dates in response to warmer temperatures in Japan? Extreme dates of any phenological distribution are sensitive lo changes in populalion size (Miller-Rushing el al. 2008). For instance, earliest migrants arrive later in cases 01 bird populalion decline (Sparks el al. 2001, Tryjanowski & Sparks 2001, 2005, Miller-Rushing el al. 2008b, Eddowes 2011). We did nol have sufficienl information regarding population dynamics of the bird species at our study sites to determine how population changes have affected the detection of first individuals. However, we had indirect evidence that population declines were the most plausible origin lor lhe lalenl delay 01 phenology. In lhe case 01 Alauda arvensis, Cettia diphone and Hirundo rustica, greater delays were detected in those bird populations from sites that have experienced a greater increase in the number of human inhabitants; A. arvensis and C. diphone, the species most affected by changes in human populalion, had lhe grealesl delay belween lhe observed and lhe lalenl phenology. Furthermore, both species are resident and as a consequence they must endure anthropogenic habitat degradation resulting year round. Therefore, it is possible that those bird populations located at more heavily human-populated sites have also suffered 108 Clim Res 54: 95-112, 2012 lhe grealesl declines in lheir numbers (Doi 2008). Our hypothesis agrees with empirical evidence of declines in bird communities, especially in long-distance migrants, reported for sorne regions of Japan (Brazil 1991, Hirano 1996, Higuchi & Morishila 1999, Ozaki el al. 2002, Tamada 2006). In Korea, H. rustica has also shown marked delays in the arrival of their first individuals and there is indirect evidence of a severe decline in their populations in recent decades (Lee el al. 2011). 4.2. Variabilily among populalion Temporal lrends differed markedly among populations during recent decades. Good examples of this are Alauda arvensis and Cettia diphone, which did not show any temporal trend in their phenology over lhe period since 1979 as a whole (see Fig. 2). However, many of their populations are indeed shifting lheir phenology, bul in opposile direclions (see Fig. 3). Spalial differences in lhe phenological responses of the Japanese populations of plants and animals have aIready be en demonstrated (Matsumolo el al. 2003, Doi 2008, Doi & Takahashi 2008, Primack et al. 200gb, Ibáñez et al. 2010), but the environmental/biological causes for such variability among sites remains poorly understood (but see Doi el al. 2010). We praposed 3 polenlial causes lor lhe observed heterogeneity among populations: climate change at a local scale, sensitivity to clima te, and population size trends (see Introduction). Sensitivity to clima te was the only hypothesis with no empirical suppor1. In lacl, populalions did nol differ significantly in their sensitivity to local temperatures in most cases. Nevertheless, we detected an increase in sensitivity during recent decades in populations of A. arvensis, C. diphone and Hirundo rustica, Le. the phenology of these species shows greater sensitivity in response to a given increase in temperature than it did several decades ago. This phenomenon has also been reported in a few other populations of birds (Sparks & Tryjanowski 2007, Askeyev el al. 2009) and planls (Sparks el al. 2009) and may help species to deal with climate change challenges by reacting more strongly to increases in temperature. In agreement with our prediction, those populations subjected to the greatest increase in local temperatures showed the greatest advance in phenology. Temperature increased faster at those sites with a grealer human popula!ion (see Fig. SS) due 10 lhe urban he al island effecl (Karl el al. 1988, Hua el al. 2008). Therelore, lhe lemporal lrends 01 lempera- tures in Japan shown in Fig. 1 are probably overestimated, as we used a sample ofweather stations from the JMA tha t are loca ted only in towns and cities, where the heat island effect is more pronounced (Primack el al. 2009a). Neverlheless, lemperalure has markedly increased on a local scale at our study sites, whatever the cause, and the local scale is the most relevant one for bird biology. Therefore, bird populalions may be doubly jeopardized by development of human society in Japan. Gn the one hand, human population growth and urban sprawl are reducing the quality and availability of natural habilals (Nakamura & Shorl 2001, Noda & Yamaguchi 2008), and as a consequence bird communities are declining (Sugimura el al. 2003, Tamada 2006). As we showed, latent delays observed in all species are probably evidence of declining bird populations acrass lhe counlry (Doi 2008, Lee el al. 2011). On lhe other hand, urban sprawl is enhancing temperature increases at a local scale via the heat island effect, and birds are facing especially large increases in temperatures. These are causing steep advances in spring phenology at lower trophic levels, such as in plants or insects, at the same localities (Matsumoto et al. 2003, Doi 2007, MilIer-Rushing el al. 2007, Doi & Kalano 2008, Doi el al. 2008, Primack el al. 2009a,b). Therefore, if the observed shifts in first singing or arrival dates are representative of the entire popula!ion phenology, Japanese bird populalions may be suffering from an increasing phenological mismatch with their environment during the breeding period (Primack el al. 2009b). In lhe parlicular case 01 Cuculus canorus, such phenological mismatching could be especially hazardous beca use it could affect interactions with host species. The magnitude of temperature and population size effects was small and accounted for only a small part of the variability among populations. Japan provides an enormous geographical range from south to north, bul il may sliII be insufficienllo delecl marked differences among populations. For instance, the warming rate detected in our study sites during the period 1979 10 2005 ranged lram +0.01 10 +0.07 ce ye ! (Fig. S3), while in a sludy 01 laying phenology in Eurapean populalions 01 lhe pied fIycalcher during a similar period temperature shifts ranged between -0.11 and +0.17 ce ye ! (Bolh el al. 2004). Similarly, changes in human population abundance and their expected consequences for bird populations may only have deleclable effecls beyond sorne lhreshold 01 severily (Lee el al. 2011). Abundanl and 10 sorne extent human-tolerant populations, such as the species studied here, may buffer small and modera te Gordo & Doi: Climate change and bird phenology population declines without showing an appreciable e!fecl on lheir phenology. Finally, one may argue lhal observable variability among sites is simply random noise as a result of the inaccurate nature of phenological mea sures, such as first arrivals or the onset of singing. However, the latter hypothesis is not supported by our results. Residuals showed notable spatial gradients, especially in those species without a significant model (Le. Cuculus canorus and Lanius bucephalus, see Fig. S2 and Table Si). Therelore, in sorne regions of Japan, the phenology of populations advanced (or delayed) more lhan would be expecled by chanceo Another important difference between the species was the effect of clima te itself. While arrivals of many populations of Hirundo rustica were advanced in response to warmer temperatures in southeastern Asia, arrivals of Cuculus canorus tended to be delayed. The response of H. rustica concurs with the advancement of arrivals observed in Western Palearctic populations under warmer temperatures en TOute (Huin & Sparks 1998, Zalakevicius el al. 2006, Sparks & Tryjanowski 2007, Gordo & Sanz 2008) and suggesls lhal clima le a!fecls populations in both migratory systems of H. rustica in a similar way. H. rustica is an airborne feeder, and warm years are probably related to improved foraging conditions during migration. However, despite a temperature increase of about 0.5°C across southeastern Asia since lhe 19705 (IPCC 2007), !irsl individuals arrived later in Japan. Hence, the hypothesis of a decline in population size is also supported by these results. Such a decline has be en reported for sorne longdistance migrants, probably also as a result of degradation 01 lhe winlering habitals (Higuchi & Morishita 1999, Lee el al. 2011). For inslance, C. canorus has recently disappeared in sorne areas of Japan (Tamada 2006). In Britain, sorne populations 01 lhis species also show delayed phenology, and population declines have been suggested as the most probable cause (Sparks el al. 2005b, Croxlon el al. 2006). However, lhe phenology 01 Japanese populations 01 C. canorus was positively related to temperatures in sorne areas of southeastern Asia; thus the arrival of this species could be delayed in response to warmer temperatures in its wintering area. 4.3. Asian populalions 01 migratory birds Hirundo rustica and Cuculus canorus offered a unique opportunity to examine bird phenology responses to climate in wintering and passage areas in the Asian migratory system. In agreement with previous studies from the western Palearctic (Cotton 2003, Gordo el al. 2005, Rodriguez-Teijeiro el al. 2005, Hüppop & Winkel 2006, Jonzén el al. 2006, Gordo & Sanz 2006, Zalakevicius el al. 2006, Saino et al. 2007), climate in wintering and passage areas affected arrival dates. Passage areas on the coast of China showed generally smaller effects than the rest 01 lhe regions in soulheaslern Asia (Figs. 5 & 6). This suggests lesser sensitivity to climate en route than to climate from departure areas in Indonesia, Malaysia or lhe Philippines. An alternative hypolhesis is lhal Japanese populations make a direct flight from wintering areas to Japan over the sea, since recoveries on the Chinese coast are uncommon (see Fig. S1). Furthermore, the Pacific Ocean is covered by many islands and archipelagos from Australia to Japan and thus, migratory species have ample opportunity to land and stop-over. Cuculus canorus showed a much more homogeneous response to climate in southeastern Asia among populations than Hirundo rusticaa. We suggesl 2 hypolheses. (1) Di!ferences could be jusI a matter of sample size. The number of populations for H. rustica was larger than for C. canorus (86 vs. 25 sites) and lhus, lhere was more variability. (2) Differences between species indica te 2 different strategies during the winter. In the case of H. rustica, individuals would spread throughout all of southeastern Asia, while Japanese populations of C. canorus would concentra te on the Malay Peninsula. Our results do not unambiguously support one or lhe olher. 109 Acknowledgements. We fuank fue Japan Meteorological Agency for collecting phenological and climate data over so many decades and 4 anonymous reviewers for fueir comments. O.G. received a Juan de la Cierva grant (JCI-2oo905274). H.D. thanks the Aquatic Restoration Research Center, Japan, for helping him while writing fue paper. Fllllding was provided by fue Japan Society for fue Promotion of Science to H.D. LITERATURE CITED )o- )o- )o- Ahas R, Aasa A (2006) The effects of climate change on the phenology of selected Estonian plant, bird and fish populations. Int J BiometeoroI51:17-26 Askeyev OV, Sparks TH, Askeyev IV (2009) Earliest recorded Tatarstan skylark in 2008: non-linear response to temperature suggests advances in arrival dates may accelerate. Clim Res 38:189-192 Barrett RT (2011) Recent response to climate change among migrant birds in northern Norway. Ring Migr 26:83-93 Beissinger SR, Osborne DR (1982) Effects of urbanization on avian cornmurrity organization. Condor 84:75-83 110 Clim Res 54: 95-112, 2012 >- Both C, te Marvelde L (2007) Spatial and temporal variatian >- Gordo O, >>- >- >>- >>>>>>>- >- in climate change and their effects on timing oí avian breeding and migratian throughout Europe. Clim Res 35: 93-105 Both C, Artemyev AA, Blaauw B, Cowie RJ, and others (2004) Large-scale geographical variatian confirms that climate change causes birds to lay earlier. Proc Biol Sci 27U657-1662 Both C, Bouwhuis Sr Lessells CM, Visser ME (2006a) Climate change and population declines in a long-distance migratory bird. Nature 441:81-83 Both C, Sanz JJ, Artemyev AA, Blaauw B, and others (2006b) Pied Flycatchers Ficedula hypoleuca travelling from Africa to breed in Europe: differential effects oí winter and migratian conditions on breeding date. Ardea 94,511-525 Both C, Van Turnhout CAM, Bijlsma RG, Siepel H, Van Strien AJ, Foppen RPB (2010) Avian population consequences of climate change are most severe for longdistance migrants in seasonal habitats. Proc Biol Sci 277: 1259-1266 Brazil MA (1991) The birds of Japan. Christopher Helm, London Butler CJ (2003) The disproportionate effect of global warming on fue arrival dates of short-distance migratory birds in Norfu America. Ibis 145:484-495 CoUon PA (2003) Avian migration phenology and global climate change. Proc Natl Acad Sci USA 100:12219-12222 Croxton PJ, Sparks TH, Cade M, Loxton RG (2006) Trends and temperature effects in the arrival of spring migrants in Portland (United Kingdom) 1959-2005. Acta Ornifuol 4U03-111 Doi H (2007) Winter flowering phenology of Japanese apricot Prunus mume reflects climate change across Japan. Clim Res 34:99-104 Doi H (2008) Delayed phenological timing of dragonfly emergence in Japan over five decades. Biol LeU 4: 388-391 Doi H, Katano I (2008) Phenological timings of leaf budburst wifu climate change in Japan. Agric For Meteorol 148: 512-516 Doi H, Takahashi M (2008) Latitudinal patterns in the phenological responses of leaf colouring and leaf fall to climate change in Japan. Glob Ecol Biogeogr 17:556-561 Doi H, Gordo O, Katano 1 (2008) Heterogeneous intraammal climatic changes drive different phenological responses at two trophic levels. Clim Res 36: 181-190 Doi H, Takahashi M, Katano I (2010) Genetic diversity increases regional variation in phenological dates in response to climate change. Glob Change Biol 16:373-379 Dwm PO, Winkler DW (1999) Climate change has affected fue breeding date of tree swallows throughout North America. Proc Biol Sci 266:2487-2490 Eddowes MJ (2011) Longer term trends in arrival timing of long distance migrants: the influence of ablllldance and population change. Ring Migr 26:56-63 Forchhammer MC, Post E, Stenseth NC (2002) NorfuAtlantic Oscillation timing of long- and short-distance migration. J Anim Eco17U002-1014 Gordo O (2oo7a) Does climate change affect cornmon swift Apus apus migratory phenology? APUSlist 3237. www. cornmonswift.org/3237Gordo.htrnl Gordo O (2007b) Why are bird migration dates shifting? A review of weather and climate effects on avian migratory phenology. Clim Res 35:37-58 >>>>>- >- >>>- >- >>- Sanz JJ (2006) Climate change and bird phenology: a long-term study in the Iberian Peninsula. Glob Change BioI12:1993-2004 Gordo O, Sanz JJ (2008) The relative importance of conditions in wintering and passage areas on spring arrival dates: fue case of long-distance Iberian migrants. J OrnithoI149,199-210 Gordo O, Sanz JJ (2009) Long-term temporal changes of plant phenology in fue Western Mediterranean. Glob Change BioI15:1930-1948 Gordo O, Sanz JJ (2010) Impact of climate change on plant phenology in Mediterranean ecosystems. Glob Change BioI16,1082-1106 Gordo O, Brotons L, Ferrer X, Comas P (2005) Do changes in climate paUerns in wintering areas affect the timing of the spring arrival of trans-Saharan migrant birds? Glob Change Biol11:12-21 Harris CD (1982) The urban and industrial transformation of Japan. Geogr Rev 72:50-89 Higuchi H, Morishita E (1999) Population declines of tropical migratory birds in Japan. Actinia 12:51-59 Hirano T (1996) Changes in breeding avifalllla during the past 25 years at Tomatsuriyama in Utsllllomiya City, central Japan. Strix 14:25-31 [in Japanese 1 Hua LJ, Ma ZG, Guo WD (2008) The impact of urbanization on air temperature across China. Theor Appl Climatol 93,179-194 Hubálek Z, Capek M (2008) Migration distance and the effect of North Atlantic Oscillation on the spring arrival of birds in central Europe. Folia Zool (Brno) 57:212-220 Huin N, Sparks TH (1998) Arrival and progression of the swallow Hirundo rustica through Britain. Bird Study 45: 361-370 Hüppop O, Hüppop K (2003) Norfu Atlantic Oscillation and timing of spring migration in birds. Proc Biol Sci 270: 233-240 Hüppop O, Winkel W (2006) Climate change and timing of spring migration in the long-distance migrant Ficedula hypoleuca in central Europe: fue role of spatially different temperature changes along migration routes. J Ornifuol 147,344-353 Kobori H, Kamamoto T, Nomura H, Oka K. Primack RB (2012) The effects of climate change on the phenology of winter birds in Yokohama, Japan. Ecol Res 27:173-180 Ibánez 1, Primack RB, Miller-Rushing AJ, Ellwooo E and others (2010) Forecasting phenology llllder global warming. Phil Trans R Soc B 365:3247-3260 IPCC (Intergovernmental Panel on Climate Change ) (2007) Climate Change 2007: fue Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of fue Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge Jenni L, Kéry M (2003) Timing of autumn bird migration llllder climate change: advancesinlong-distance migrants, delays in short-distance migrants. Proc R Soc Lond B 270: 1467-1471 JMA (Japan Meteorological Agency) (1985) Guidelines for the observation of phenology, 3rd edn. Japan Meteorological Agency, Tokyo [in Japanese 1 Jonzén N, lindén A, Ergon T, Knudsen E, and others (2006) Rapid advance of spring arrival dates in long-distance migratory birds. Science 312: 1959-1961 Karl TR, Diaz HF, Kukla G (1988) Urbanization: its detection and effect in the United States climate record. J Clim UQ99-1123 Gordo & Doi: Climate change and bird phenology >- Koike >- >>>>>>>>>>>- >- >>- S, Higuchi H (2002) Long-term trends in the egg laying date and clutch size of red-cheeked starlings Sturma plillippensis. Ibis 144: 150-152 Lee SD, Ellwood ER, Park SY, Primack RB (2011) Latearriving barn swallows linked to population declines. Biol Conserv 144:2182-2187 Lefranc N, Worfolk T (1997) Shrikes. A guide to the shrikes of the world. Pica Press, Robertsbridge, UK Legendre P, Legendre L (1998) Numerical ecology, 2nd edn. EIsevier, Amsterdam Lehikoinen E, Sparks TH, Zalakevicius M (2004) Arrival and departure dates. Adv Ecol Res 35:1-31 Mason CF (2006) Avian species richness and numbers in the built environment: can new housing developments be good for birds? Biol Conserv 15:2365-2378 Matsumoto K. Ohta T, Irasawa M, Nakamura T (2003) Climate change and extension of the Ginkgo biloba L. growing season in Japan. Glob Change Biol 9: 1634-1642 McKinney ML (2002) Urbanization, biodiversity, and conservation. Bioscience 52:883-890 Menzel A, Sparks TH, Estrella N, Koch E, and others (2006) European phenological response to climate change matches the warming pattern. Glob Change Biol 12: 1969-1976 Miller-Rushing AJ, Katsuki T, Primack RB, Ishii Y, Lee SD, Higuchi H (2007) Impact of global warming on a group of related species and their hybrids: cherry tree (Rosaceae) flowering at Mi. Takao, Japan. AmJ Bot 94:1470-1478 Miller-Rushing AJ, Inouye DW, Primack RB (2oo8a) How well do first flowering dates measure plant responses to climate change? The effects of population size and sampling frequency. J Ecol 96: 1289-1296 Miller-Rushing AJ, Lloyd-Evans TL, Primack RB, Satzinger p (2008b) Bird migration times, climate change, and changing population sizes. Glob Change Biol 14: 1959-1972 Miller-Rushing AJ, Primack RB, Stymeist R (2008c) Interpreting variation in bird migration times as observed by vohmteers. Auk 125:565-573 Moller AP, Rubolini D, Lehikoinen E (2008) Populations of migratory bird species that did not show a phenological response to climate change are declining. Proc Natl Acad Sei USA 105,16195-16200 Nakamura T, Short K (2001) Land-use planning and distribution of threatened wildlife in a city of Japan. Landsc Urban Plan 53:1-15 Noda A, Yamaguchi Y (2008) Characterizing urban sprawl using remote sensing and a simple spatial metric for a medium-sized city in Japan. Int J Geoinformatics 4: 43-50 Ozaki K. Baba T, Komeda S, Kinjyo M, Toguchi Y, Harato T (2002) The declining distribution of the Okinawa Rail Gallira11us okinawae. J Yamashina Inst Ornithol 34: 136-144 (in Japanese) Parmesan C (2006) Ecological and evolutionary responses to recent climate change. Annu Rev Ecol Evol Syst 37: 637-669 Peintinger M, Schuster S (2005) Changes in first arrival dates of cornmon migratory bird species in southwestern Germany. Vogelwarte 43:161-169 Peterson TC, Vose RS (1997) An overview of the Global Historical Climatology Network temperature database. Bull Am Meteorol Soc 78:2837-2849 Peterson TC, Vose RS, Schmoyer R, Razuvaev V (1998) Global Historical Climatology Network (GHCN) quality >>- >>>- >>- >>>- >- 111 control of monthly temperature data. Int J Climatol 18: 1169-1179 Primack RB, Higuchi H, Miller-Rushing AJ (2009a) The impact of climate change on cherry trees and other species in Japan. Biol Conserv 142:1943-1949 Primack RB, Ibánez 1, Higuchi H, Lee SD, Miller-Rushing AJ, Wilson AM, Silander JA (2009b) Spatial and interspecific variability in phenological responses to warming temperatures. Biol Conserv 142:2569-2577 Rooríguez- Teijeiro JD, Gordo O, Puigcerver M, Gallego S, Vinyoles D, Ferrer X (2005) African climate warming advances spring arrival of the cornmon quail Coturmx coturmx. Ardeola 52:159-162 Root TL, MacMynowski DP, Mastrandrea MD, Schneider SH (2005) Human-modified temperatures induce species changes: joint attribution. Proc Natl Acad Sci USA 102: 7465-7469 Rubolini D, Moller AP, Rainio K. Lehikoinen E (2007) Intraspecific consistency and geographic variability in temporal trends of spring migration phenology among European bird species. Clim Res 35:135-146 Saino N, Rubolini D, Jonzén N, Ergon T, Montemaggiori A, Stenseth NC, Spina F (2007) Temperature and rainfall anomalies in Africa predict timing of spring migration in trans-Saharan migratory birds. Clim Res 35: 123-134 Saino N, Rubolini D, Lehikoinen E, Sokolov LV and others (2009) Climate change effects on migration phenology may mismatch brooo parasitic cuckoos and their hosts. Biol Lett 5:539-541 Sepp M, Palm V, Leito A, Piiiidam K. Truu J (2011) Effect of atmospheric circulation types on spring arrival of migratory birds and long-term trends in the first arrival dates in Estonia. Estonian J Ecol 60: 111-131 Shaw LM, Chamberlain D, Evans M (2008) The house sparrow Passer domesticus in urban are as: reviewing a possible link between post-decline distribution and human socioeconomic status. J Ornithol 149:293-299 Sparks TH, Tryjanowski P (2007) Rapid (but differential) advance of spring arrival dates in two hinmdines. Clim Res 35,159-164 Sparks TH, Roberts DR, Crick HQP (2001) What is the value of first arrival dates of spring migrants in phenology? Avian Ecol Behav 7:75-85 Sparks TH, Bairlein F, Bojarinova JG, Hüppop O and others (2005a) Examining the total arrival distribution of migratory birds. Glob Change Biol 11:22-30 Sparks TH, Croxton PJ, Collison N, Taylor PW (2005b) Examples of phenological change, past and present, in UK farming. Ann Appl Biol 146:531-537 Sparks TH, Jaroszewicz B, Krawczyk M, Tryjanowski P (2009) Advancing phenology in Europe's last lowland primeval forest: non-linear temperature response. Clim Res 39,221-226 Sugimura K. Yamada F, Miyamoto A (2003) Population trend, habitat change and conservation of the rnrique wildlife species on Amami Island, Japan. Glob Env Res 7 :79-89 Tamada K (2006) Population change of grassland birds over ten years in Nakashibetsu, eastern Hokkaido. Ornitholog Sei 5,127-131 Tottrup AP, Thorup K. Rahbek C (2oo6a) Patterns of change in timing of spring migration in North European songbird populations. J Avian BioI37:84-92 Tottrup AP, Thorup K. Rahbek C (2006b) Changes in timing of autumn migration in North European songbird populations. Ardea 94:527-536 112 Clim Res 54: 95-112, 2012 >- Tottrup >- >>- AP, Thorup K. Rainio K. Yosef R, Lehikoinen E, Rahbek e (2008) Avian migrants adjust migratian in response to environmental conditions en raute. Biol Lett 4,685-688 Tryjanowski P, Sparks TH (2001) Is fue detection oí fue first arrival date oí migrating birds influenced by population size? A case study oí fue red-backed shrike Lamus collurio. Int J Biometeoro145:217-219 Tryjanowski P, Kuzniak Sr Sparks TH (2002) Earlier arrival oí sorne farrnland migrants in western Poland. Ibis 144: 62-68 Tryjanowski P, Kuzniak Sr Sparks TH (2005) What affects fue magnitude oí change in first arrival dates oí migrant birds? J Ornithol 146:200-205 Van Buskirk J, Mulhivill RS, Leberman Re (2009) Variable shifts in spring and autumn migratian phenology in NorthAmericansongbirds associated with climate change. Glob Change Biol 15:760-'7'71 Végvári Z, Bókony V, Barta Z, Kovács G (2010) Life history Editorial responsibility: Tim Sparks, Cambridge, UK >- >>>- predicts advancement of avian spring migration in response to climate change. Glob Change Biol16:1-11 Visser ME, Adriaensen F, Van Balen JH, Blondel J and others (2003) Variable responses to large-scale climate change in European Parus populations. Proc Biol Sci 270: 367-372 Watanuki y (2011) Climate variability and the timing of breeding and reproduction in seabirds. Mem Fac Fish Sci Hokkaido Univ 53:19-26 (in Japanese) Willis CG, Rubiel B, Primack RB, Miller-Rushing AJ, Davis CC (2008) Phylogenetic patterns of species loss in Thoreau's wooos are driven by climate change. Proc Natl Aead Sei USA 105,17029-17033 Yamada H, Tokuoka K (1996) Population and urbanization trends in post-war Japan. Asian Econ J 10:187-206 Zalakevicius M, Bartkeviciene G, Raudonikis L, Janulaitis J (2006) Spring arrival response to climate change in birds: a case study from eastern Europe. J Ornithol 147: 326-343 Subrrritted: September 27, 2010¡ Accepted: Febllla¡y 26,2012 Praofs received fram author(s): July 26,2012