Spring bird migration phenology in Eilat, Israel

Analysis of the mean date of fi rst captures and median arrival dates of spring migration for 34 species of birds at Eilat, Israel, revealed that the earlier a species migrates through Eilat, the greater is the inter-annual variation in the total time of its passage. Birds arrive during spring migration in Eilat in four structured and independent waves. Th e annual fl uctuation in the initial arrival dates (initial capture dates) and median dates (median date of all captures), not including recaptures, did not depend on the length of the mi gratory route. Th is implies that migrants crossing the Sahara desert depart from their winter quarters on diff erent Julian days in diff erent years. We suggest that negative correlations between the median date of the spring migration of early and late migrants depends upon the easterly (Hamsin) wind period. Moreover, we believe that the phenology of all birds during spring migration in Eilat is possibly also determined by external factors such as weather conditions on the African continent or global climatic processes in the Northern hemisphere. Orphean Warblers (Sylvia hortensis) show a strong negative correlation (rs = -0.502) of initial capture date with calendar years, whereas other species such as Barred Warbler (S. nisoria; rs = -0.391) and Spotted Flycatcher (Muscicapa striata; rs = -0.398) display an insignifi cant trend. Th e Dead Sea Sparrow (Passer moabiticus) and Red-Backed Shrike (Lanius collurio) are positively correlated regarding initial arrival date and medians of spring migration.


Introduction
During the last two decades in many European countries, spring arrival for numerous bird species has been recorded earlier than during the 1970s. Th is includes both short-distance migrants and species wintering in Africa (Mason 1995;Sokolov et al. 1998;Sparks 1999;Barrett 2002;Hüppop and Hüppop 2003). Th e considerable shift in timing of spring migration towards earlier Julian dates has been associated with global warming and climatic change (Bairlein and Winkel 2001;Sokolov 2001;Sparks et al. 2001Sparks et al. , 2003Hüppop and Hüppop 2003). However, the way in which timing of migration actually aff ects the early stages of spring migration has yet to be defi ned.
For many migrating species, the Eilat oasis is the fi rst stopover site encountered after crossing the combined geophysical barrier of the Sahel, Sahara and Sinai deserts (Yosef and Tryjanovski 2002a). In some passerines species, e.g., Blackcap Sylvia atricapilla (Izhaki and Maitiav 1998), Red-backed and Masked Shrikes (Lanius collurio, nubicus resp.; Yosef and Tryjanovski 2002b, c), and Ortolan and Cretzschmar's Bunting (Emberiza hortulana, caesia resp.; Yosef and Tryjanovski 2002d, e), a signifi cant interannual variation of the timing of spring migration has been documented.
Here, we present an analysis of the initial arrival (capture) date and median date of migration of 34 avian species in order to better understand if any changes have occurred in the phenology during spring migration for the past 20 years at Eilat.

Materials and methods
From 1984-2003, the International Birding and Research Center in Eilat (IBRCE), Israel (29 º 33΄N, 34 º 57΄E) has trapped and ringed passerines in autumn and in spring. Th is has resulted in dataset containing information on more than 180.000 individual birds from 272 avian species. In most species, the numbers trapped in spring were higher than in the autumn. Th e IBRCE ringing program is aimed at three major groups of birds: raptors, waders, and passerines. Because the ringing program during spring began at diff erent times of the month (between 1-15 February), we analyzed data from 15 February only.
In order to analyze the phenology of spring migration, we selected species based on the following criteria: (1) all species were caught in mist nets; (2) no special eff ort was made to capture individual species, e.g., swallows (Hirundo spp.) and wagtails (Motacilla spp.) at roost; (3) the species were captured in a minimum of 17 seasons; (4) in suffi cient numbers (>20); and (5) the species was not sedentary. Based on these criteria, we included in our fi nal analyses a total of 34 avian species comprising 32 passerine species, Quail (Coturnix coturnix) and Wryneck (Jynx torquilla) (Table 1).
For all individual birds from all species captured, we calculated the date of initial capture, and the average and median of all captures during the season. Th e average and median were calculated if the number of birds caught exceeded 20, and excluded all recaptures.
Because of Eilat's unique geographical location on the migratory fl yway, our analysis focused on the date of the initial capture and the median date of all captures. It was assumed that the date of initial capture refl ected the date of fi rst arrival of the species in that season (Sokolov et al. 1998). Th e species' median arrival date was considered to be of major importance because the duration of spring migration in Eilat for many species exceeds 3.5 months (Morgan and Shirihai 1997). Further, in many species there are several geographically distinct subspecies, many of which have not been identifi ed to date, which constitute the species population for the season (Morgan and Shirihai 1997).
To corroborate our conjectures, we conducted a correlation analysis between initial and median captures of the 34 species included in our study from 1984-2003. We also compared dates of initial capture of these species for the two decades from 1984-1993 and from 1994-2003. Standard statistical methods were used to characterize and analyze the data (Sokal and Rohlf 1995). Calculations were made using the STATISTICA v.5 package.

Results
Analysis of the phenology in 34 species at Eilat showed a large inter-annual variation ( Table 1). More specifi cally, early migrants showed large inter-annual variation regarding initial capture date resulting in no statistically signifi cant trends overall (r s = -0.2547, n = 34, p = 0.146).
Correlation analyses of initial capture date during the past 20 years revealed that in the studied species, change in initial captures ranged from a strong positive infl uence to results that were statistically insignifi cant (Table 2). Correlation analyses of median capture dates of the individuals during the past 20 years highlighted groups of species that had similar spring migration timing and phenology (Table 3).
Initial captures of the 34 species confi rmed that the fi rst of the spring arrivals are the long-distance migrants, and that they are already present in Eilat by mid-February. Owing to its geophysical position in relation to African winter quarters, one of the important characteristics for Eilat is the simultaneous occurrence of groupings of species (Fig. 1). Combining the results of the correlations analysis of fi rst arrival data and median data of spring migration in Eilat for 20 years showed that there are at least four such waves of arrival.   -1984-1993and 1994-2003.
Th e fi rst species to arrive in the spring are Quail, Reed Warbler (A. scirpaceus), Savi's Warbler (Locustella lusciniodes), Blackcap (S. atricapilla), Common Whitethroat (S. communis), and Lesser Whitethroat (S. curruca). Th ese species usually appear in Eilat in the second half of February (Table 1); their initial captures are positively and signifi cantly correlated among themselves (Table 2). Th e second wave of migrants arrives in the second 10 days of March. Th e initial capture of Wryneck, Black-eared Wheatear (Oenanthe hispanica), Northern Wheatear (0. oenanthe), Bonelli's Warbler (P. bonelli), Olivaceous Warbler (H. pallida), Woodchat Shrike (L. senator), Cretzschmar's Bunting (E.ceasia) occurs almost simultaneously (Table 1) By correlation analyses, we found a strong and positive correlation among the medians of all species that arrive together (Table 3). However, there are signifi cant negative correlations between the median date of spring migration among early migrants and those that migrate late (e.g., between the medians of spring migration of Garden Warbler and six early migrants).
In several species we found a signifi cant correlation between initial capture dates in the spring and in the calendar years ( A comparison of mean data regarding initial capture dates during the two consecutive decades indicated that all 34 species arrived on average three days earlier from 1984-1993 than from 1994-2003 (17 vs 14 of March). However, the diff erence is not statistically signifi cant (Mann-Whitney U Test, U= 563, Z= 0.184, p = 0.854).

Discussions
Th e beginning of spring migration for birds is controlled by the photoperiod or by means of special endogenous programs that are initiated in the breeding areas, and are specifi c for species that winter in the equatorial regions or further south (Gwinner 1996;Berthold 2001). Following this concept, many researchers believe that spring migration in Africa begins approximately at the same time, and that this is specifi c for each species, subspecies or separate population. Alerstam (1990) suggested that weather conditions determine the speed of spring migration, but not the timing of its initiation. Sparks et al. (2001) and Hüppop and Hüppop (2003) found that weather conditions on the European continent determined strong interannual fl uctuations in the timing of arrival of long-distance migrants in breeding areas.
To date, Eilat is the southernmost site for which there exists a long-term monitoring data set of spring migration phenology, and is closest to the wintering grounds. Of the 34 species chosen for analyses, the majority (24) were long-distance migrants from Africa to Europe and Asia, some (9) breed in the Mediterranean basin, and one, the Rufus Bush-Robin, is a summer visitor that breeds in the Eilat region and the Negev desert. Th e initial capture dates and median for many species varied considerably between years. Th e median date of spring migration (median of all captures) for some species also varied greatly between years (Table 1). Moreover, initial arrival dates and the median date of spring migration were very positively connected with each other, though the median dates of early migrants were very negatively associated with median dates of late migrants (Tables 2, 3). Assuming that birds do not stop over for longer periods in the desert, we infer that crossing the Sahara occurs on variable dates. It also suggests that the spring migration phenology at Eilat is strongly connected to extrinsic factors such as weather conditions in Africa or global climatic processes in the Northern hemisphere. Th e concomitant arrival of the same groups of species also suggests that weather infl uences the initiation of spring migration of those birds arriving at Eilat. A major weather pattern observed to have an eff ect on the medians of migration is known as the Hamsin. A Hamsin weather pattern consists of days (sometimes up to a week) of very hot, dry weather due to strong easterly winds occurring mostly in the spring. Th e Hamsin winds delay migration in late-migrating species such as the Garden and Barred Warbler and the Spotted Flycatcher (pers. obs.).
For several species of birds we found a connection between dates of initial capture and the calendar year. Some of these can be attributed to the infl uence of global warming on avian spring migration, as shown by several studies (Bairlein and Winkel 2001;Sokolov 2001;Sparks et al. 2001;Hüppop and Hüppop 2003;Butler 2003). Only the Orphean Warbler showed a strong positive correlation of initial capture with the calendar year, whereas the Barred Warbler and Grey Flycatcher displayed insignifi cant trends. Th e Dead Sea Sparrow and Red-Backed Shrike showed positive correlations of initial arrival date with the medians of spring migration. Similar to long-term studies by Sokolov (2001;43 years) and Butler (2003;90 years), we concluded that it is imperative that the long-term ringing program at Eilat continues in such a way that the infl uence of environmental pressures on phenological trends of migratory birds will be evident in future decades.
L. Mitchell, Mrs. B. Bausch, Swarovski Optics, and Earthwatch for their support of the ringing program in Eilat. Ms. Susan Craig improved an earlier draft of the manuscript.