Longer breeding dispersal than natal dispersal in the ortolan bunting

Transcription

1 Longer breeding dispersal than natal dispersal in the ortolan bunting Behavioral Ecology doi: /beheco/arh129 Advance Access publication 28 July 2004 Svein Dale, Anne Lunde, and Øyvind Steifetten Department of Ecology and Natural Resource Management, Agricultural University of Norway, PO Box 5003, NO-1432 Ås, Norway Theoretical models dealing with dispersal patterns are currently limited by a lack of empirical data, and existing data may be biased because of small spatial scale of many previous studies. We studied the whole known population of a small passerine bird, the ortolan bunting (Emberiza hortulana), in Norway. Males conducted extraordinary long-distance breeding dispersal of up to 45 km during their first years of life but showed high territory fidelity when older. Males that failed to attract a female in their first singing territory were especially likely to disperse, and their movements regularly occurred within a breeding season or until the next year (such movements were also defined as breeding dispersal). Breeding dispersal distances of males (median ¼ 11.9 km) were more than four times as long as their natal dispersal distances (median ¼ 2.7 km). These data contradict a classical view of dispersal in birds, namely, that the longest dispersal movements occur before the first territory is established (natal dispersal) and subsequent movements (breeding dispersal) are shorter. Thus, breeding dispersal plays a larger role than does natal dispersal in gene flow and population connectivity in the ortolan bunting. We suggest that short natal dispersal and subsequent long breeding dispersal within the breeding season may be an optimal dispersal strategy in ortolan buntings owing to their patchy distribution in our study area, and we predict that this may also be the case for other species with patchy or fragmented distribution. Key words: breeding dispersal, Emberiza hortulana, natal dispersal, ortolan bunting, territory switching. [Behav Ecol] Understanding patterns of dispersal is important in biogeography (MacArthur and Wilson, 1967), population ecology (Hanski, 1999), behavioral ecology (Greenwood, 1980), and conservation biology (Baillie et al., 2000), but theoretical models are currently limited by a lack of empirical data (Clobert et al., 2001). One long-standing tenet is that among birds, natal dispersal is longer than is breeding dispersal, and that breeding dispersal is usually short or absent owing to extensive reuse of territories between years, particularly among males (Greenwood and Harvey, 1982). In their influential paper, Greenwood and Harvey (1982: 6) stated that No examples have yet been recorded where adults leave an area to a greater extent than do juveniles... To our knowledge, no studies have contradicted this clear-cut claim in the 20 years since this publication. However, many studies have been conducted on small study areas that may bias dispersal estimates (Koenig et al., 1996), and the logistical difficulties involved in large-scale studies of dispersal have effectively restricted thorough tests of even basic patterns (Clobert et al., 2001). The problem of small study areas in analyses of natal and breeding dispersal was avoided by Paradis et al. (1998). They used ringing recoveries of 75 species of birds at a large spatial scale, based on the ringing data of the British Trust for Ornithology. Their analyses apparently confirmed the pattern that natal dispersal is longer than is breeding dispersal. The dispersal pattern proposed by Greenwood and Harvey (1982) is also reflected by the prominent view that location of future breeding territories is the result of exploratory movements during the postfledging period (Adams and Brewer, 1981; Brewer and Harrison, 1975; Morton, 1991; Reed et al., 1999). Thus, when a bird has settled in a territory for the first time, subsequent movements later in life are usually shorter. However, when resightings (recoveries) are not the result of Address correspondence to S. Dale. svein.dale@ina.nlh.no. Received 29 January 2004; revised 2 April 2004; accepted 18 May detailed observations of the dispersal history of individuals (see Paradis et al., 1998), there is a possibility that data may become biased. For example, if rapid dispersal movements (territory changes) occur early in adult life, breeding dispersal could easily be classified as natal dispersal (for further details, see Discussion). Thus, to fully understand the relationship between natal and breeding dispersal, and the process of dispersal itself, it may be necessary to follow in detail the movements made by individually marked birds over years. The Norwegian population of ortolan buntings is particularly well suited for studies of dispersal at a large spatial scale. The known population currently contains about 150 singing males, with nearly all individuals distributed in about 40 welldefined habitat patches in an area of nearly 500 km 2.We identified individuals with color-rings throughout seven consecutive breeding seasons. In each year up to 77% of all males had color-rings, and, combined with the low number of individuals and breeding patches, this greatly increased our chances of documenting dispersal events. There was a malebiased sex ratio in our population (Dale, 2001a), and we defined movements made by males after they had failed to attract females to their first singing territory as breeding dispersal (see Methods and Discussion). We used our data to test whether long-distance movements in the ortolan bunting occurred before males became territorial (natal dispersal) or during adult life (breeding dispersal). In particular, we compared data on natal dispersal of birds ringed in the nest with the dispersal behavior of the same individuals as adults. METHODS The study was conducted from Most Norwegian ortolan buntings occur in easily accessible agricultural areas of central Hedmark County ( N, E, maximum extent of this area was 40 km both north south and east west), where they nest on raised peat bogs, forest clearcuts on poor sand, land being cleared for cultivation, and one forest burn (Dale, 2001a). Minor relict populations (about 10 Behavioral Ecology vol. 16 no. 1 Ó International Society for Behavioral Ecology 2005; all rights reserved.

2 Dale et al. Long breeding dispersal in a passerine bird 21 Figure 1 Frequency distributions of breeding (a) and natal dispersal distances (b) of male ortolan buntings in Norway. Breeding dispersal includes 145 dispersal events (changes in singing territory) by 104 individuals; natal dispersal is for 30 individuals. males) occur km to the south and southwest. Ortolan buntings are small (20 25 g), migratory passerine birds, and in the breeding areas in Norway, pairs raise one clutch from May July. Both sexes are necessary for successful breeding (Cramp and Perrins, 1994), so dispersal cannot take place during a breeding attempt. Adult males were captured in mist-nets with the aid of playback of song and were given a combination of one metal ring and three color-rings to permit individual identification. Resightings were confirmed during visits at 1 3-day intervals throughout the breeding season to all habitat patches which have been used by ortolan buntings. During visits to each habitat patch, the mating status (unpaired or paired) of each male was recorded. In addition, all other potentially suitable habitat patches within the study area, although hitherto unused, were visited frequently. Potentially suitable patches were determined from aerial surveys and ground checks. From , males were captured and ringed in one large subpopulation, in which local ornithologists also had ringed some birds from From 1999, surveys and ringing took place in all known breeding areas in Norway. A total of 301 males had color-rings as adults. Analyses were restricted to males because females often behave cryptically, and dispersal events can therefore be missed, in particular those movements that might occur before females become mated. Age was known with certainty only for birds ringed in the nest. All birds ringed for the first time as adults were scored as 1 year or older (2Kþ in the Euring system). Natal dispersal of a male was defined as the movement between the place of birth (ringed as nestling) and the first singing territory. Breeding dispersal was defined as movements involving a change of singing territory (both within and between years). These definitions follow the methods of Greenwood and Harvey (1982), who defined natal dispersal as dispersal from the site of birth to that of first reproduction or potential reproduction and subsequent movements between sites as breeding dispersal, and Paradis et al. (1998), who defined natal dispersal as movement between place of birth and recovery at minimum breeding age without requiring evidence that breeding took place. For ortolan buntings and most passerine birds, minimum breeding age is their first year as adults (2K). Territories of ortolan buntings are roughly 100 m in diameter (Cramp and Perrins, 1994; Dale and Olsen, 2002), which is typical for small passerine birds, and territory changes were therefore defined as movements more than 200 m. All observations concern typical defended singing territories; floaters have not been recorded in our study population. Floater is here defined as a nonterritorial male, that is, not attempting to attract a female (cf. Newton, 1998; Zack and Stutchbury, 1992). For a bird to be recorded as a nondisperser, we required that it return to the same territory the subsequent year. Because of our intense field surveys, we were sometimes able to record multiple movements of one individual within a short time period (e.g., within one breeding season). Many other studies would only have recorded the starting and finishing points of such movement series. Thus, to compare with most existing data on breeding dispersal, we used the combined breeding dispersal distances in the pairwise comparison of natal and breeding dispersal. However, in this analysis we also used a single dispersal distance value for each individual (the longest distance). In all other analyses, we used the single distance value only. Analyses of dispersal may depend on the resighting probability for dispersing individuals. For data on movements between years, we had an almost 100% resighting probability. A total of 157 males were recorded in at least 2 years (mean ¼ 2.78 years), and in 156 of these cases, the time series were continuous without holes (e.g., not recorded in year y, but recorded in years y ÿ 1 and y þ 1). Hence, there is no reason to believe that the observed distribution of dispersal distances is biased, and the use of mark-recapture modeling is not needed. For movements within a breeding season we cannot calculate a resighting probability because we have no means of detecting if there are any holes in a movement series, but for birds that were recorded moving from one territory to another, the median time to resighting was only 7 days (range ¼ 1 21 days, n ¼ 68). Nonparametric statistical tests were used because of nonnormal distributions of variables. All tests were two-tailed. The standard error of the estimates of the proportion of individuals dispersing was calculated by assuming that dispersal rate was a variable with a binomial distribution (Sokal and Rohlf, 1981). RESULTS Many males changed singing territories (Figure 1a; cases of no dispersal between years were not included because the total sample of birds was biased toward old individuals), and the distance between these territories was 1 km or more in 99 cases, and 10 km or more in 43 cases. The longest recorded breeding dispersal was 45 km. Males changed territories both

3 22 Behavioral Ecology Figure 2 Within-individual comparison of natal and breeding dispersal distances of male ortolan buntings (n ¼ 17). between breeding seasons (n ¼ 70) and within a breeding season (n ¼ 69) (Figure 1a). Dispersal within a breeding season was significantly longer than between seasons (Mann- WhitneyU test: U ¼ 1694, p ¼.0024), and males moved 43 km in less than 16 days and 22 km within 1 day. In most cases (112/142) males had failed to attract a female before they dispersed, and this was always the case for within-season breeding dispersal. Males who had been paired before moved shorter distances than did unpaired males (U ¼ 783, p,.0001), but previously paired males moved 1 km or more in 12 cases, with 36 km as the longest distance. Natal dispersal had a peak frequency between 1 5 km (Figure 1b). For 17 males, both natal and breeding dispersal was recorded. Contrary to expectations, breeding dispersal (median ¼ 11.9 km) was significantly longer than natal dispersal (median ¼ 2.7 km) (Wilcoxon matched-pairs signedrank test using combined breeding dispersal distances for each male: Z ¼ÿ2.49, p ¼.013) (Figure 2). The difference remained significant even when restricting the analysis to a single dispersal distance for each male (the longest; Z ¼ ÿ2.15, p ¼.031). Long-distance breeding dispersal was predominantly undertaken by young males, either within their first year as adult or from their first to their second year. The proportion of males dispersing declined with age among those ringed as nestlings (exact age; likelihood ratio test: G 2 ¼ 22.07, df ¼ 3, p,.0001) (Figure 3a). Distance dispersed among those that did move, declined with age as well (Spearman rank correlation: r s ¼ÿ.57, n ¼ 19, p ¼.016) (Figure 3a). The same pattern was evident among males that were ringed as adults (minimum age; proportion dispersing: G 2 ¼ 24.05, df ¼ 4, p,.0001; distance dispersed: r s ¼ÿ.23, n ¼ 99, p ¼.026) (Figure 3b). Finally, reduced dispersal with increasing age was confirmed by within-individual analyses of dispersal behavior as first-year adults compared with behavior in their second year. Adult males moved longer distances in their first year than in their second year (Wilcoxon matched-pairs signed-ranks tests; exact age: Z ¼ÿ2.02, n ¼ 7, p ¼.043; minimum age: Z ¼ÿ2.20, n ¼ 66, p ¼.028). DISCUSSION Our data indicated that a large proportion of Norwegian male ortolan buntings undertook long-distance breeding dispersal when young. The distances were far greater than those usually Figure 3 Age-related changes in the proportion of male ortolan buntings dispersing (left y-axis), and the distance moved for those that did disperse (right y-axis). Data for birds ringed as nestlings (exact age) (a) and for birds ringed as adults (minimum age) (b). Dispersal at age class x includes within-season breeding dispersal at this age and breeding dispersal occurring between ages x and x þ 1. Bars indicate standard error for proportion dispersing and percentile range for median dispersal distance. recorded for small passerine birds (Greenwood and Harvey, 1982; Hansson et al., 2002). Males often dispersed after a stay of only a few days or a few weeks in their first territory, and the short time window could mean that this phenomenon has been overlooked in other species. Other males made longdistance movements at some time between two breeding seasons. The small geographical scale of many studies would lead to long-distance dispersers simply being recorded as disappeared (Koenig et al., 1996). Long-distance breeding dispersal of passerine birds has been reported in a few other studies (Lang et al., 2002; Newton, 2000), but ours is the first to show that breeding dispersal can be far longer than natal dispersal, in contrast to the expected pattern (Greenwood and Harvey, 1982). Similarly, the observation that long-distance movements often occurred within the breeding season, and by males that had already become territorial, contrasts with the widespread view that location of future breeding

4 Dale et al. Long breeding dispersal in a passerine bird 23 territories occur during the postfledging period (Adams and Brewer, 1981; Brewer and Harrison, 1975; Morton, 1991; Reed et al., 1999). Although sample size in our main test (n ¼ 17) was not very large, the difference between natal and breeding dispersal distances was statistically significant and numerically substantial. It remains to be seen whether the dispersal pattern of ortolan buntings occurs in other birds as well (see below). Definitional issues The above conclusions were based on similar definitions of natal and breeding dispersal as used by Greenwood and Harvey (1982) and Paradis et al. (1998; see Methods). One important point to note is that with these definitions, natal dispersal is from the site of birth to the site of first reproduction or potential reproduction. The qualifier, potential reproduction, is important in the case of the ortolan bunting. Because of a male-biased sex ratio in our study population, more than 25% of all males remained unpaired every year (Dale, 2001a), and some males sang in the same or different territories year after year without attracting a female. Thus, the definitions do not take into account whether breeding actually took place in the territories that were used before and after a breeding dispersal event, only that the male tried to attract a mate for breeding in those territories. This differs from the definitions of dispersal used by some other investigators in which actual breeding both before and after breeding dispersal was required, and natal dispersal was defined as the movement from birth place to the place of first actual breeding (see Kenward et al., 2002; Lang et al., 2002). However, the latter definitions are difficult to use in populations in which some males may fail to attract a female. For example, one male ortolan bunting was present for 5 years in a row with failed attempts to attract a female every year, and we would not be able to assign natal or breeding dispersal values. Consequently, sample sizes in our tests would be drastically reduced if we used the latter definitions because many males never attracted females or did not return the year after being paired for the first time, and important information on the actual behavior of many individuals would thereby be excluded. Thus, both for the purpose of unambiguous definitions and biological meaningfulness, the former definitions, following the method of Greenwood and Harvey (1982), and Paradis et al. (1998), work better. We are therefore of the opinion that the long-distance movements undertaken by young male ortolan buntings are best regarded as cases of breeding dispersal. The fact that long-distance dispersal (up to 36 km) was also observed among males that had been paired before supports the conclusion that longdistance movements represented breeding dispersal. However, should one prefer to use the latter definition requiring actual breeding before breeding dispersal, our results would change so that natal dispersal would be longer than breeding dispersal (Z ¼ÿ1.35, n ¼ 7, p ¼.18). Despite this, our results would still differ from most previous studies in that the timing of the long-distance movements is later, either within the breeding season of second-year males or between the second and third years of age. Because the long-distance dispersal movements documented in the present study were often performed by males during a short time in their first breeding season, they would not necessarily be recorded in ringing recovery analyses of the type used by Paradis et al. (1998). In their study, recoveries were only based on ringed birds found dead. Thus, the probability of recording a male ortolan bunting in the first shortterm territory would be very low, and in many cases, natal dispersal would include the first dispersal event between singing territories, whereas cases of breeding dispersal would be biased to short ones typical of older males. This means that the data presented by Paradis et al. (1998), although unbiased with respect to spatial scale, may contain an unknown proportion of classification of long-distance movements as belonging to natal dispersal, which we would prefer to classify as breeding dispersal. Furthermore, for relatively long-lived species, one long-distance breeding dispersal at young age can be swamped by the much higher chances of recording site fidelity at older ages. Hence, gross comparisons of natal and breeding dispersal distances can give the impression that the longest movements are made during natal dispersal, whereas breeding dispersal performed at young age may actually be longer. Finally, Paradis et al. (1998) excluded from the analyses all recoveries made within the same breeding season as the birds had been ringed. If long-distance breeding dispersal often occurs within one breeding season (about half of all cases in the present study), such movements could be missed by Paradis et al. (1998). Spatial scale Compared to most previous studies of color-ringed bird populations, the spatial scale in our study was much larger (study area of nearly 500 km 2 ). Thus, we were able to document long-distance movements of up to 45 km. One could still object, however, that the distribution of dispersal distances in Figure 1 is truncated so that a righthand tail of the distribution is missed. All field studies will have a spatial limitation, but we argue that the size of our study area provided a fairly representative picture of ortolan bunting dispersal range for the following reasons: (1) we studied the whole known Norwegian population of ortolan buntings, and our population is separated from the closest known neighboring populations in Sweden by about 250 km; (2) the annual return rate of color-ringed adult males was on average 62% (Dale, 2001a), which is exceptionally high given the small body size of the ortolan bunting and the fact that it is a tropical long-distance migrant; and (3) the return rate of male nestlings was on average 26% (Dale, 2001a), which is also a very high figure. The high return rates made it unlikely that a substantial number of males left our study population neither during natal dispersal nor during breeding dispersal. Ecological implications We suggest that short natal dispersal and subsequent long breeding dispersal within the breeding season may be an optimal dispersal strategy in ortolan buntings owing to their patchy distribution in our study area, and in other areas as well (Cramp and Perrins, 1994). Note that natal dispersal distances of ortolan buntings recorded in the present study are similar to those recorded for other passerine birds (Paradis et al., 1998), and the term short natal dispersal is therefore relative to the breeding dispersal distances we recorded. During the breeding season, singing males make suitable habitat patches more easy to locate. On the other hand, exploratory movements during the postfledging period may not be profitable in the ortolan bunting because adult males do not sing any longer and start migration to Africa only 1 2 months after young have fledged (Cramp and Perrins, 1994). Thus, it may be more advantageous for males to return to their natal area or close by, and if they are not successful in establishing a territory or attracting a female there, they may search for alternative breeding sites farther away during the breeding season by using conspecifics as a cue to good habitat patches (Reed and Dobson, 1993). We predict that other species with patchy distribution might also show

5 24 Behavioral Ecology longer breeding dispersal than natal dispersal, and with longdistance movements occurring within breeding seasons. On the other hand, the high rate of unpaired males in our population, owing to a lack of females, could stimulate males to higher-than-normal emigration rates (Dale, 2001a,b), but a male-biased sex ratio is typical for small bird populations occupying patchy or fragmented habitat (Dale, 2001b). Longdistance dispersal in general may also be a consequence of low population size or density (Paradis et al., 1998), perhaps because dispersal can be a successful way of escaping local competition (McCarthy, 1997). In the case of the ortolan bunting, competition for breeding space is unlikely to have caused long-distance dispersal because there seems to be vacant breeding sites in most habitat patches (Dale, 2001a). In addition, a large proportion of the dispersal movements is toward the central areas of the population (Dale S, unpublished data). In conclusion, we suggest that longer breeding dispersal rather than natal dispersal may be a consequence of a patchy or fragmented distribution, or low population size or density, but the pattern may be opposite for more abundant and widespread species. Our findings also have implications for understanding population dynamics. Even birds in the isolated and declining Norwegian population of the endangered ortolan bunting (Dale, 2001a), have a surprising ability to move between habitat patches that are separated by at least 1 10 km of unsuitable breeding habitat. Furthermore, these movements often occurred quickly so that failure to attract a mate did not necessarily mean that breeding would be delayed until the next year. Dispersal range and behavior is also of great importance when assessing long-term persistence of patchy or fragmented populations (Hanski, 1999). Previously, natal dispersal has been assumed to be the main mechanism maintaining gene flow and population connectivity in birds (Greenwood and Harvey, 1982), but our results from the ortolan bunting indicate that long-distance breeding dispersal may be more important, and dispersal within the breeding season may be an especially effective way for birds to locate distant breeding populations. We thank V. Bunes, P. Christiansen, J. P. Cygan, T. Granerud, B. Hessel, N. Manceau, B. F. G. Olsen, T. Osiejuk, K. Ratynska, T. I. Starholm, and C. Sunding for help in the field; A. Johnsen, L. Powell, G. A. Sonerud, and J. Swenson for comments on the manuscript; and Selberg legacy/wwf Norway, Mr. and Mrs. Sørlie s Foundation, the Nansen Foundation, the Norwegian Directorate for Nature Management, and the Environmental Authorities of Hedmark and Akershus Counties for financial support. Dale S, Olsen BFG, Use of farmland by Ortolan buntings (Emberiza hortulana) nesting on a burned forest area. J Ornithol 143: Greenwood PJ, Mating systems, philopatry and dispersal in birds and mammals. Anim Behav 28: Greenwood PJ, Harvey PH, The natal and breeding dispersal of birds. Ann Rev Ecol Syst 13:1 21. Hanski I, Metapopulation ecology. Oxford: Oxford University Press. Hansson B, Bensch S, Hasselquist D, Nielsen B, Restricted dispersal in a long-distance migrant bird with patchy distribution, the great reed warbler. Oecologia 130: Kenward RE, Rushton SP, Perrins CM, MacDonald DW, South AB, From marking to modelling: dispersal study techniques for land vertebrates. In: Dispersal ecology (Bullock JM, Kenward RE, Hails RS, eds). Oxford: Blackwell; Koenig WD, Van Vuren D, Hooge PN, Detectability, philopatry, and the distribution of dispersal distances in vertebrates. Trends Ecol Evol 11: Lang JD, Powell LA, Krementz DG, Conroy MJ, Wood thrush movements and habitat use: effects of forest management for redcockaded woodpeckers. Auk 119: MacArthur RH, Wilson EO, The theory of island biogeography. Princeton, New Jersey: Princeton University Press. McCarthy MA, Competition and dispersal from multiple nests. Ecology 78: Morton ML, Wakamatsu MW, Pereyra ME, Morton GA, Postfledging dispersal, habitat imprinting, and philopatry in a montane, migratory sparrow. Ornis Scand 22: Newton I, Population limitation in birds. San Diego, California: Academic Press. Newton I, Movements of bullfinches Pyrrhula pyrrhula within the breeding season. Bird Study 47: Paradis E, Baillie SR, Sutherland WJ, Gregory RD, Patterns of natal and breeding dispersal in birds. J Anim Ecol 67: Reed JM, Boulinier T, Danchin E, Oring LW, Informed dispersal. Prospecting by birds for breeding sites. Curr Ornithol 15: Reed JM, Dobson AP, Behavioural constraints and conservation biology: conspecific attraction and recruitment. Trends Ecol Evol 8: Sokal RR, Rohlf FJ, Biometry. New York: Freeman. Zack S, Stutchbury BJ, Delayed breeding in avian social systems: the role of territory quality and floater tactics. Behaviour 123: REFERENCES Adams RJJ, Brewer R, Autumn selection of breeding location by field sparrows. Auk 98: Baillie SR, Sutherland WJ, Freeman SN, Gregory RD, Paradis E, Consequences of large-scale processes for the conservation of bird populations. J Appl Ecol 37: Brewer R, Harrison KG, The time of habitat selection by birds. Ibis 117: Clobert J, Danchin E, Dhondt AA, Nichols JD, Dispersal. Oxford: Oxford University Press. Cramp S, Perrins CM, The birds of the Western Palearctic, vol. 9: buntings and New World warblers. Oxford: Oxford University Press. Dale S, 2001a. Causes of population decline of the Ortolan bunting in Norway. In: Bunting studies in Europe (Tryjanowski P, Osiejuk TS, Kupczyk M, eds). Poznan: Bogucki Wydawnictwo Naukowe; Dale S, 2001b. Female-biased dispersal, low female recruitment, unpaired males, and the extinction of small and isolated bird populations. Oikos 92: