Patterns of natal and breeding dispersal in birds - Wiley Online Library

56 downloads 3487 Views 457KB Size Report
2[ Using data on 64 terrestrial bird species\ we tested whether natal and breeding dispersal .... covered by birds between ringing and recovery sites are stored in ...
Journal of Animal Ecology 0887\ 56\ 407Ð425

Patterns of natal and breeding dispersal in birds EMMANUEL PARADIS$\ STEPHEN R[ BAILLIE\ WILLIAM J[ SUTHERLAND$ and RICHARD D[ GREGORY British Trust for Ornithology\ The Nunnery\ Thetford\ Norfolk\ IP13 1PU\ UK^ and $School of Biological Sciences\ University of East Anglia\ Norwich\ NR3 6TJ\ UK

Summary 0[ Dispersal is of critical ecological and evolutionary importance for several issues of population biology\ particularly population synchrony\ colonization and range expansion\ metapopulation and sourceÐsink dynamics\ and population genetic struc! ture\ but it has not previously been possible to compare dispersal patterns across a wide range of species or to study movement outside the con_nes of local study areas[ 1[ Using resampling methods\ we veri_ed that statistically unbiased estimates of average dispersal distance and of intraspeci_c variance in dispersal distance could be extracted from the bird ringing data of the British Trust for Ornithology[ 2[ Using data on 64 terrestrial bird species\ we tested whether natal and breeding dispersal were in~uenced by a species| habitat requirements\ diet\ geographical range\ abundance\ morphology\ social system\ life history or migratory status[ We used allometric techniques to ascertain whether these relationships were independent of body size\ and used the method of phylogenetically independent contrasts to ascertain whether they were independent of phylogeny[ 3[ Both natal and breeding dispersal distances were lower among abundant species and among species with large geographical ranges[ Dispersal distances and life!history variables were correlated independent of phylogeny\ but these relationships did not persist after controlling for body size[ All morphometrical variables "wing length\ tarsus length and bill length# were not signi_cantly correlated with dispersal distances after correcting for body size or phylogenetic relatedness[ 4[ Migrant species disperse further than resident ones\ this relation was independent of body size but not of phylogeny[ A signi_cant positive relation was observed between breeding dispersal distance and long!term population decline among migrants\ but not among residents[ 5[ The species living in wet habitats disperse further than those living in dry habitats\ which could be explained by the greater patchiness of wet habitats in space and:or time[ This relationship was observed only for breeding dispersal\ suggesting that this habitat variable does not impose the same constraint on natal dispersal[ Key!words] body size\ comparative analyses\ dispersal distances\ phylogeny\ ring recoveries[ Journal of Animal Ecology "0887# 56\ 407Ð425

Introduction

Þ 0887 British Ecological Society

407

The distance moved by animals during dispersal is fundamental to ecology and evolutionary biology[ All studies of animal that disperse actively show some individual variation in dispersal distances "Swingland 0872#[ Many studies of these intraspeci_c di}erences have been conducted "Murray 0856^ Greenwood + Harvey 0865\ 0871^ Greenwood\ Harvey + Perrins

0868^ Gaines + McClenaghan 0879^ Waser 0874^ Boonstra et al[ 0876^ Trewhella\ Harris + McAllister 0877^ Berven + Grudzien 0889^ Payne 0880^ Bengts! son\ Hedlund + Rundgren 0883^ Stein et al[ 0883^ Dwyer + Elkinton 0884#[ On the other hand\ studies on interspeci_c di}erences in dispersal have mainly dealt with sex! or age!speci_c philopatry "Greenwood 0879^ Dobson 0871#[ Two other issues of importance in the study of dispersal are the rates of movements

408 E[ Paradis et al[

Þ 0887 British Ecological Society Journal of Animal Ecology\ 56\ 407Ð425

between populations\ which has been recently reviewed by Nichols "0885#\ and the distribution of dispersal distances\ which is the subject of the present paper[ The distribution of dispersal distances is of fun! damental importance for several issues in population biology[ It is an essential component of meta! population and sourceÐsink dynamics "Brawn + Rob! inson 0885#\ invasion and colonization "Harrison 0878^ Hengeveld 0883^ Shigesada\ Kawasaki + Takeda 0884#\ and gene ~ow and genetic structure "Neigel + Avise 0882#[ Also\ interspeci_c di}erences in distribution of dispersal distances have been shown to be critical in species interactions "McLaughlin + Roughgarden 0881^ Neubert\ Kot + Lewis 0884#[ It has been observed repeatedly that long!distance dispersal occurs regularly but at a relatively low fre! quency "Grinnell 0811^ Mayr 0852^ Murray 0856#[ Long!distance dispersal has been shown\ on theor! etical grounds\ to be important in invasion processes "Shaw 0884# and the genetic structuring of popu! lations "Ibrahim\ Nichols + Hewitt 0885#[ Long!distance dispersal in vertebrates is still poorly understood in spite of its importance "Clobert + Leb! reton 0880^ Hansson 0880#[ Detailed _eld studies are appropriate to measure small!scale dispersal "e[g[ Greenwood et al[ 0868#\ but they are usually unable to follow individuals that leave the study area "Baker\ Nur + Geupel 0884#[ For instance\ _eld studies on microtine rodents hardly cover more than one hectare while there is reliable evidence that these animals can move several kilometres in a few days "Kirkland 0877^ Saucy 0883^ Steen 0883#[ Avian study sites are usually smaller than the distance that a bird can move within a day "Payne 0880^ Matthysen\ Adriaensen + Dhondt 0884#[ Such small!scale studies on dispersal are valu! able in evaluating the proximate behavioural causes of dispersal "Greenwood et al[ 0868^ Wiggett + Boag 0882#\ but dispersal outside such local areas can only be studied with a marking scheme operating over a much larger!scale and no such investigation has pre! viously been undertaken[ Field studies are usually focused on one or two species resulting in di.culties in identifying general ecological or evolutionary con! straints[ These issues can be studied by comparing a wide range of species encompassing a diversity of body size\ habitats or other ecological variables[ Studies on dispersal distances of a wide range of birds have shown two ubiquitous patterns] the fre! quency distribution of dispersal distances decreases with increasing distance\ and natal dispersal distances are greater than breeding dispersal ones "Newton + Marquiss 0871^ Drilling + Thompson 0877^ Baillie + Milne 0878^ Moore + Dolbeer 0878^ Newton\ Davis + Davis 0878^ Verner\ Gutierrez + Gould 0881^ Bel! ant + Dolbeer 0882^ Ferrer 0882^ Lindberg\ Sedinger + Rexstad 0884^ Matthysen et al[ 0884^ Wagner et al[ 0885#[ We report here an analysis of interspeci_c vari!

ations in dispersal distances using data on marked British breeding birds from the ringing scheme of the British Trust for Ornithology "BTO#[ Birds ringed in Britain and Ireland are potentially found anywhere\ resulting in no spatial limit to the present study[ We did not _nd any theoretical framework that would have allowed us to formulate speci_c hypotheses for the present study[ This lack of theory on the causal factors that a}ect the distribution of dispersal dis! tances may be a consequence of the di.culties en! countered in the study of dispersal distances men! tioned above[ Nevertheless\ it is possible to formulate a priori some predictions on the variations in dispersal distances expected among bird species[ So we tested whether dispersal varies between species occupying di}ering habitats\ diet\ social system\ population size\ geographical range\ local abundance\ body measure! ments\ life history variables and migratory status[ As body size is correlated with a wide range of biological\ ecological or physiological variables "Peters 0872#\ we expected body size to be signi_cantly correlated with dispersal distance\ and anticipated the need to correct for body size in our analyses[ We used allometric techniques to examine whether correlations persist independently of body size and phylogenetic tech! niques to test whether correlations persist inde! pendently of phylogeny[

Methods DATA

Raw data on dispersal distances came from the BTO ringing scheme[ Distances "to the nearest kilometre# covered by birds between ringing and recovery sites are stored in the data bases[ Birds have been ringed by volunteers in Britain and Ireland since 0898[ The data on ring recoveries are stored according to the EURING standards "EURING 0868#[ The data ana! lysed for the present study span the period 0898Ð83 inclusive[ During this period\ 13 804 534 birds were ringed\ and 381 161 of them "belonging to 394 species# have been recovered "Clark et al[ 0885#[ Here we con! sider only species breeding inland^ marine species are mostly colonial and their breeding habitats are lin! early distributed making comparisons with inland species problematic\ and many of the recoveries of shorebirds are from the tideline where they may have been carried by the sea[ We included in our analyses three species of gulls "Larus spp[# that breed com! monly inland in Britain and Ireland "Gibbons\ Reid + Chapman 0882#[ For species breeding in terrestrial habitats\ it has been shown that the probability that a dead ringed bird is found varies spatially "van Noordwijk 0873^ Baillie + Green 0876#[ This point has been raised as a limitation for inferences on dis! persal with ringing recovery data "van Noordwijk 0884#[ However\ two points must be considered with respect to the present study[ First\ ringing recoveries

419 Avian dispersal

Þ 0887 British Ecological Society Journal of Animal Ecology\ 56\ 407Ð425

were made in the same area for all species\ so we can expect that spatial variations in recovery rates would be similar for all species\ hence conserving any pos! sible interspeci_c patterns[ Second\ the bias demon! strated by van Noordwijk "0884# may particularly a}ect studies with a limited number of sites where birds can be recovered or recaptured[ On the other hand\ in large!scale studies with a very large number of recovery sites\ the bias in the estimates of dispersal distance may be lower than in small!scale studies[ This last point requires further evaluation\ but in the context of the present study where we focus on inter! speci_c patterns\ we do not expect our results to be a}ected by spatial variations in recovery rates[ The data were screened in the following ways in order to remove spurious e}ects[ First\ only birds ringed and recovered during the breeding season were selected^ recoveries in the same breeding season as they were ringed were excluded[ The breeding season of each species was de_ned as the period between the beginning of the time when most birds have eggs to the end of the time when most birds have young\ and taken from Cramp + Simmons "0866Ð83#[ Second\ data were _ltered to remove heterogeneity as far as possible[ Only birds found dead were considered in the analyses\ as the spatial distribution of birds controlled "captured\ usually in ringers| nets\ and released\ or identi_ed in the _eld without being captured# is likely to be biased as _eld ornithologists tend to catch birds in particular study areas[ The following data have been discarded from the analyses] birds that were in poor condition or had an accident when ringed^ birds that have been kept more than 13 h during ringing^ birds that have been moved or held extensively during ringing^ birds that have been intentionally killed by man other than shot^ birds that have not been found freshly dead^ birds for which the dates and places of ringing and:or recovery have not been recorded accurately "to the nearest 1 weeks for the ringing date\ to the nearest 1 months for the _nding date\ to the nearest 3| for the ringing and _nding place#^ birds whose age was unknown when ringed^ birds that had ~edged when ringed but whose age was otherwise completely unknown "age code  1\ EURING 0868#[ The analyses of migrant and partial migrant species were complicated by the di}erent timing of migration of di}erent individuals resulting in the recoveries of birds on migration routes late in the breeding season[ We chose to be conservative with respect to this prob! lem and discarded all birds which could be suspected to have been ringed and:or recovered on their migration route[ The most e.cient way to apply this correction was to de_ne a cut!o} distance above which the data were discarded "because these birds are most likely to be migrants#^ this cut!o} distance was de_ned separately for each species[ The direction of dispersal was graphically examined against the dispersal dis! tance] this showed that the largest dispersal distances were preferentially directed towards the northÐsouth

axis[ The largest dispersal distances were removed as long as their corresponding directions were the closest ones to the northÐsouth axis[ This resulted in dispersal movements that were not preferentially orientated in any direction[ This method allowed us to de_ne the cut!o} distance in a repeatable manner\ and to detect the presence of very late migrants "for instance\ some birds breeding in northern Europe have been ringed or recovered in late June in Britain\ which is their wintering ground#^ such individuals were a very small proportion of the birds recovered during the breeding season "³ 2)#[ We examined all species with this procedure as\ even for strongly resident species "e[g[ tits\ see Sellers 0873#\ movements in winter may result in positive bias in the estimation of dispersal distances[ Particular care was taken to avoid bias from birds ringed as juveniles in their year of birth "age code  2\ EURING 0868# which are known to be particularly mobile at the end of the breeding season "e[g[ da Prato + da Prato 0872^ Baker 0882#[ This was done by examining whether juveniles have signi_cantly larger dispersal distances than birds ringed as nestlings "t! test at 9=94 level#[ If this was the case\ then they were excluded from the analyses[ This potential source of bias was particularly important for migrants "Baker 0882^ see present Discussion#[ We found that in some species only juveniles ringed early in the breeding sea! son have greater movements than birds ringed as nest! lings] this could be explained by the fact that these individuals have more time to explore for future breeding sites\ and may have on average a greater dispersal distance than birds born later during the breeding season[ In all cases\ a category of individuals "ringed either early or late in the breeding season# was excluded if its mean dispersal distance was sig! ni_cantly di}erent from the remaining birds[ Finally\ we checked in species which are known to have also migratory movements between Britain and Ireland "e[g[ blackbird Turdus merula L[\ song thrush T[ philomelos L[# that the largest dispersal distances were not preferentially directed towards the eastÐwest axis after removing the northÐsouth migratory move! ments[ We found that no correction was necessary with respect to these longitudinal migratory move! ments[ After application of the selection criteria\ we selec! ted those species with su.cient data "minimum sample size of 09# to estimate the parameters of natal and breeding dispersal separately[ Seventy!_ve species remained "listed in Appendix 0#[ Natal dispersal was estimated using birds ringed in their year of birth "either in the nest or as ~edglings# and recovered at breeding age^ breeding dispersal was estimated using birds ringed at breeding age[ Breeding age was determined as the minimum age at _rst breed! ing as given by Cramp + Simmons "0866Ð83#[ Some species in our sample have a variable age at _rst breed! ing "particularly gulls#^ however\ there is evidence from _eld studies that birds at the minimum age at

410 E[ Paradis et al[

Þ 0887 British Ecological Society Journal of Animal Ecology\ 56\ 407Ð425

_rst breeding go back to their natal sites though they may not yet breed "e[g[ Lack 0855#[ Most birds were not sexed when ringed "especially those ringed in the nest# making it almost impossible to de_ne sex categories in this study[ In many birds there is a sex di}erence in dispersal pattern "Green! wood 0879^ Greenwood + Harvey 0871#^ however\ here we expected to observe larger interspeci_c di}er! ences than could be measured at the intraspeci_c level[ Another limitation of our approach is that we have no data on the reproductive status of a bird when it is recovered\ so we cannot be sure that we really estimate dispersal between a birth site and a breeding site or between two breeding sites[ This may be a serious problem for birds that are at breeding age but are not e}ectively breeding[ However\ there is some evidence from _eld studies that such individuals wander around breeding sites during the breeding season\ usually waiting for an opportunity to acquire a vacant ter! ritory "e[g[ Lack 0855#[ It is therefore likely that such individuals are recovered near their potential breeding sites[ Ecological and life history variables were extracted from Cramp + Simmons "0866Ð83#[ The quantitative variables were] mean clutch size\ incubation time "in days#\ age at ~edging "in days#\ age at independence "in days#\ age at _rst breeding "in years#\ modal number of broods per year\ adult body mass "in grams#\ body mass at hatching "in grams#\ egg mass "in grams#\ wing length "in millimetres#\ bill length "in millimetres# and tarsus length "in millimetres#[ The categorical vari! ables chosen were "see Appendix 0 for details#] main food in winter and during the breeding season "ani! mals or plants#\ social system in winter "solitary or ~ocking# and during the breeding season "territorial\ neighbourhood or colonial#\ preferred habitat in win! ter and during breeding "each split into two dichot! omous variables] open vs[ closed\ and dry vs[ wet#\ and migration "treating trans!saharan migrants as migrants and all other species as residents#[ The extracted variables were from studies conducted in the British Isles when available in Cramp + Simmons "0866Ð83#^ otherwise they were from studies con! ducted on mainland Europe on the same subspecies as occurs in the British Isles[ Additionally\ data on juvenile and adult survival rates were taken from S%ther "0878#[ This latter study compiled estimates of survival rates from the literature that mostly were made with enumeration methods\ which could result in substantial biases compared to methods based on probabilistic models "Lebreton et al[ 0881#[ However\ a recent review of the literature showed that when di}erent estimates of survival rates based on di}erent methods were available for the same species\ the di}erences were less than 9=0 "and often less than 9=94#\ and did not confound interspeci_c variations "Balmer + Peach 0886#[ Variables on population status were taken from some studies conducted at the BTO[ Four variables

were considered] "i# population size\ measured as the number of pairs breeding in Britain and Ireland "Gib! bons et al[ 0882#\ "ii# geographical range\ measured as the number of 09!km squares occupied during the breeding season in Britain "Gibbons et al[ 0882#\ "iii# local abundance\ measured as the estimated densities on Common Birds Census "CBC# plots "Marchant et al[ 0889# averaged from 0880 to 0884\ and "iv# popu! lation trend\ as evaluated by the percentage change in estimated density on CBC plots between 0858 and 0883[

ESTIMATION OF PARAMETERS

Dispersal is a complex phenomenon making the devel! opment of statistical models di.cult "North 0877#[ Several models are available but their application requires restrictive assumptions "Taylor 0879^ North 0874^ Tonkyn + Plissner 0880^ Porter + Dooley 0882#[ Our approach was to use univariate parameters to measure the distribution of dispersal distances for each species[ One potential problem was the hetero! geneity in sample size "n# among species "see Appendix 0#[ The expected value of the chosen estimator should be\ as far as possible\ insensitive to sample size\ though the variance is expected to increase with decreasing sample size[ We performed a resampling analysis with four species for which we had relatively large sample sizes] k birds were randomly selected "k ³ n# and vari! ous univariate parameters were then estimated for these k individuals^ this was replicated 09 times for each value of k[ Performance of the di}erent esti! mators was then assessed by plotting the estimates against k[ Seven parameters of dispersal distances were estimated for each replicate] arithmetic mean\ standard deviation "SD#\ coe.cient of variation "CV#\ minimum\ median\ _rst and third quartiles[ Distances were loge!transformed before analysis[ Note that the arithmetic mean of loge!transformed data is equi! valent to the loge!transformed geometric mean of the original data[ Analyses were made with SAS "0889#[

INTERSPECIFIC COMPARISONS

Three sets of analyses were performed with the esti! mates of dispersal distance across all species[ First\ we computed linear correlation coe.cients with loge! transformed continuous variables\ and one!way ANOVAs with the categorical variables[ Second\ we removed the in~uence of body size on dispersal par! ameters with linear regressions[ The residuals were then analysed in the same way as for the _rst set of analyses[ The two sets of analyses were performed with SAS "0889#[ Third\ to cope with the problem of dependence among species\ we computed phylo! genetically independent contrasts "PICs# for each vari! able according to the method described by Felsenstein "0874#[ Each PIC is associated with a node in the phylogeny and is a measure of the evolutionary diver!

411 Avian dispersal

gence between both taxa originating from this node[ The phylogeny for the 64 species was built following Sibley + Ahlquist "0889#[ Sibley + Ahlquist|s phylogeny\ based on DNA:DNA hybridization\ has been repeatedly criticized "Mindell 0881^ Sheldon + Bledsoe 0882# but is the only one published that covers all avian families and gives estimates of branch lengths\ and it is relatively uncontentious for the birds covered in this paper[ Furthermore\ some con! troversial conclusions reached by Sibley + Ahlquist "0889# have been con_rmed by several recent phylo! genetic studies using alternative methods "review in Avise 0885#[ Felsenstein|s method can cope with the problem of uncertainty in phylogenetic reconstruction by ana! lysing several phylogenies for the considered taxa but these phylogenies must all contain only dichotomous bifurcations and known branch lengths "Felsenstein 0874#[ Sibley + Ahlquist "0889# provide a phylogeny of living birds with dichotomous bifurcations and known branch lengths\ but several species in our study are not present in their phylogeny[ We resolved the branching orders of these species with data from the taxonomy which allowed us to infer the relationships between species\ and estimated branch lengths for these inferred dichotomies using information in Sibley + Ahlquist "0889#[ We used two di}erent hypotheses for the latter step] "i# we took the smallest branch length within the considered genus if available\ or the smallest branch length possible "9=0#^ or "ii# we took the largest branch length within the considered genus if available\ or the largest branch length possible "length of the branch to the nearest genus Ð 9=0#[ This procedure allowed us to evaluate more rigorously the phylogenetic e}ects\ as the phylogeny used in com! parative analyses has a critical in~uence on the results "Felsenstein 0877^ Pagel + Harvey 0878#[ The PICs for dispersal parameters and the continuous life his! tory and ecological variables were computed with PHYLIP "Felsenstein 0882#[ The categorical variables were mapped along the phylogeny to identify the tran! sitions that occurred along the tree using the criterion of parsimony[ Linear correlations were made between PICs to examine the relationships between dispersal parameters and the continuous variables[ One!way ANOVAs were computed considering the nodes where a transition in the considered categorical variable vs[ the nodes with no transition "Burt 0878#[ All analyses were made considering successively the two recon! structed phylogenies[ We considered as statistically signi_cant only those tests which were signi_cant with both phylogenies[

Results WITHIN!SPECIES DISTRIBUTION OF DISPERSAL

Þ 0887 British Ecological Society Journal of Animal Ecology\ 56\ 407Ð425

DISTANCES

Most individual birds were recovered on or close to their ringing sites "Figs 0 and 1#[ Only ringing recover!

ies between breeding seasons were used as a measure of breeding or natal dispersal\ depending on the age at which the bird was ringed[ This could generate a bias if birds tend to move signi_cantly each year[ If so\ we should expect a positive relationship between the distance recorded between ringing and recovery and the time elapsed[ However\ plots of these two variables shows no relationship "Figs 2 and 3#\ sug! gesting that once birds have dispersed they tend to stay at the same breeding site[ Such site tenacity has been described extensively in birds "Kellner\ Brawn + Karr 0881^ Peach\ du Feu + McMeeking 0884#[ We selected four species with large sample sizes to perform the resampling analyses] blackbird "n  2261#\ blue tit Parus caeruleus L[ "n  673#\ green_nch Carduelis chloris L[ "n  291#\ and starling Sturnus vulgaris L[ "n  1265#[ Estimates of mean and SD of dispersal distance were unbiased with respect to k\ the variance being greater for smaller values of k "Fig[ 4#[ Estimates of other parameters varied substantially with respect to k "CV#\ or were constant across species "minimum\ median\ _rst and third quar! tiles#\ and so were unlikely to measure di}erences among species[ We chose to use the mean and SD of natal and breeding dispersal as dispersal parameters[ Greenwood + Harvey "0871# suggested that the median should be the appropriate parameter to com! pare dispersal between species\ but this was not the case in our study[ This is due to the spatial resolution of the ringing data] in the four species examined above\ more than 49) of the individuals were re! covered near their ringing site\ and as distances are stored in the data bases to the nearest kilometre they were assigned a zero dispersal distance[ It may seem surprising that CV "the ratio of SD to mean# varies to a greater extent than mean or SD[ However\ the variance of the estimate of CV is a function of the variances and covariance of the estimates of mean and SD\ and so its variability is increased compared to mean and SD[ All species showed the same pattern of distribution of dispersal distances with the proportion of recover! ies declining with distance from the ringing site[ How! ever\ di}erences among species were noticeable\ par! ticularly the spatial extent of the peak of the distribution\ and the importance of the tail relative to the peak[ The parameters estimated for each species "mean and SD# represent these variations[ We were able to estimate natal dispersal for all 64 species\ and had su.cient data to estimate breeding dispersal for 58 of these species[ Among the latter\ 50 species had a greater mean natal dispersal distance than mean breeding dispersal distance "see Appendix 0#[ For the eight other species\ the mean estimates were very similar and the di}erence was certainly due to sampling variations[ Breeding and natal dispersal distances covary to a great extent\ so the interspeci_c variation is much more important than the intra! speci_c one that we were able to measure[ Nonethe!

412 E[ Paradis et al[

Fig[ 0[ Frequency distribution of breeding dispersal distances of four species of birds in Britain[

Þ 0887 British Ecological Society Journal of Animal Ecology\ 56\ 407Ð425

Fig[ 1[ Frequency distribution of natal dispersal distances of four species of birds in Britain[

Fig[ 2[ Distance moved "breeding dispersal# against time elapsed between ringing and recovery of four species of birds in Britain[

Fig[ 3[ Distance moved "natal dispersal# against time elapsed between ringing and recovery of four species of birds in Britain[

414 E[ Paradis et al[

Fig[ 4[ Resampling analysis of dispersal distances of four species "see text for details#[ Plot of estimated means left vertical axis# and estimated standard deviations "open circles\ right vertical axis# against sample size "k#[ Dispe were loge!transformed prior to analyses[

less\ the distinction between natal and breeding dis! persal is relevant for the present study as the di}erence is consistent across species and the data analysed allowed us to estimate the parameters separately[

INTERSPECIFIC CORRELATIONS

Þ 0887 British Ecological Society Journal of Animal Ecology\ 56\ 407Ð425

Cross!species distributions of the mean and SD of log! transformed dispersal distances were normally dis! tributed\ so use of cross!species linear correlations and ANOVA was justi_ed[ The PICs generated by PHYLIP were also normally distributed and correctly stan! dardized with respect to branch lengths[ In the follow! ing\ we detail the relationships between dispersal dis! tance and the di}erent types of variables\ examining successively for each type the simple correlations\ cor! relations after correcting for body size\ and cor! relations after correcting for phylogeny[ The former correction was justi_ed by the positive correlation between body size and mean breeding and natal dis! persal "r  9=187\ n  56\ P  9=904\ and r  9=200\ n  64\ P  9=995\ respectively#[ No signi_cant cor! relation was observed between body size and SD of breeding and natal dispersal "r  9=003\ n  56\ P  9=257\ and r  Ð 9=041\ n  64\ P  9=082\ respec! tively#[ Plots of mean dispersal distances vs[ body size reveal an important scatter with a greater variance for

small body sizes than large ones "Fig[ 5 mated slopes from these regressions "SD  9=968# and 9=103 "SD  9=966#\ f and natal dispersal\ respectively[ As thes not signi_cantly di}erent from 9=14\ this s the geometric mean of dispersal distance body size with the 0:3 exponent\ which is for several ecological variables "Peters 0 following\ we focus on correlations tha account the e}ect of body size or phylog not present results that correct simulta body size and phylogeny as the correlat dispersal distance and body size after re e}ect of phylogeny is not signi_cant[ After correcting for the in~uence of b the dispersal parameters\ the correlations life history variables and the dispersal were not signi_cant\ except for the numb per year with mean breeding and nat "Table 0#[ After correcting for phylogen ness\ three life history variables gave co ni_cant correlations with dispersal param bation time with SD of natal disper independence with both mean and SD of sal\ and age at _rst breeding with mean na "Table 1#[ For mean natal dispersal\ the were positive\ i[e[ an increase in age at in

415 Avian dispersal

Fig[ 5[ Plots of body size vs[ "a# mean breeding dispersal "n  56#\ "b# mean natal dispersal "n  64# for 64 species of British birds[ Dispersal distances "km# have been loge!transformed before computing the arithmetic mean[

Table 0[ Correlations between life history and ecological variables\ and parameters of breeding and natal dispersal of British birds after removing the e}ect of body size Breeding dispersal Mean

Clutch size Incubation time Age at ~edging Age at independence Age at _rst breeding Juvenile survival Adult survival Broods per year Population size Geographical range Local abundance SD "local abundance# Population trend

Natal dispersal SD

Mean

SD

r

n

r

n

r

n

r

n

−9=924 9=030 −9=031 −9=924 9=121 9=131 −9=908 −9=157 −9=255 −9=310 −9=033 −9=124 9=010

56 56 56 56 56 16 30 56 56 56 48 48 42

−9=947 9=909 −9=159 9=989 9=997 9=084 9=002 −9=965 −9=015 −9=199 9=911 −9=965 9=965

56 56 56 56 56 16 30 56 56 56 48 48 42

−9=919 9=139 9=001 9=096 9=027 9=960 9=996 −9=323 −9=445 −9=273 9=943 −9=951 9=049

64 64 64 64 64 21 34 64 64 64 53 55 45

−9=053 −9=985 −9=196 −9=044 9=930 9=969 9=141 9=955 9=037 −9=945 −9=081 −9=120 −9=916

64 64 64 64 64 21 34 64 64 64 53 55 45

] P ³ 9=94^ ] P ³ 9=90^ ] P ³ 9=990[ r] correlation coe.cient\ n] sample size

Þ 0887 British Ecological Society Journal of Animal Ecology\ 56\ 407Ð425

or age at _rst breeding during evolution is accompanied by an increase in mean natal dispersal[ The correlations with SD of natal dispersal were nega! tive[

Population size and geographical range were stron! gly negatively correlated with mean natal and breed! ing dispersal] plots of these correlations reveal di}er! ent relationships[ The relation between population

416 E[ Paradis et al[

Table 1[ Correlations between life history and ecological variables\ and parameters of breeding and natal dispersal of British birds using the method of phylogenetically independent contrasts[ t0 and t1 refer to the two phylogenies used to compute the PICs Breeding dispersal Mean

Body size Clutch size Incubation time Age at ~edging Age at independence Age at _rst breeding Juvenile survival Adult survival Broods per year Population size Geographical range Local abundance SD "local abundance# Population trend

t0 t1 t0 t1 t0 t1 t0 t1 t0 t1 t0 t1 t0 t1 t0 t1 t0 t1 t0 t1 t0 t1 t0 t1 t0 t1 t0 t1

Natal dispersal SD

Mean

SD

r

n

r

n

r

n

r

n

9=901 −9=126 9=978 9=370 −9=950 −9=196 −9=123 −9=091 −9=038 9=998 9=185 9=196 −9=087 −9=981 −9=076 −9=070 −9=191 −9=966 −9=216 −9=363 −9=218 −9=477 −9=217 −9=118 −9=085 9=006 9=960 −9=920

55 55 55 55 55 55 55 55 55 55 55 55 15 15 39 39 55 55 55 55 55 55 51 51 51 51 41 41

−9=159 −9=189 9=908 −9=922 9=960 9=929 −9=197 −9=947 −9=998 9=062 −9=044 −9=031 9=272 9=257 −9=060 −9=119 −9=954 −9=001 −9=122 −9=106 −9=990 9=234 −9=144 −9=495 −9=189 −9=404 −9=073 −9=138

55 55 55 55 55 55 55 55 55 55 55 55 15 15 39 39 55 55 55 55 55 55 51 51 51 51 41 41

9=391 9=119 9=061 9=045 9=184 9=059 −9=913 9=010 9=531 9=260 9=304 9=295 9=203 9=171 −9=143 9=147 −9=070 9=098 −9=683 −9=528 −9=590 −9=343 −9=707 −9=483 −9=596 −9=182 9=990 9=194

63 63 63 63 63 63 63 63 63 63 63 63 20 20 33 33 63 63 63 63 63 63 55 55 55 55 45 45

−9=356 −9=033 −9=014 −9=214 −9=270 −9=401 −9=121 −9=058 −9=584 −9=163 −9=263 −9=900 −9=205 −9=114 9=101 9=486 −9=937 −9=313 9=473 9=053 9=398 9=916 9=483 9=941 9=315 9=932 −9=990 −9=315

63 63 63 63 63 63 63 63 63 63 63 63 20 20 33 33 63 63 63 63 63 63 55 55 55 55 45 45

P ³ 9=94^ P ³ 9=90^ P ³ 9=990[ r] correlation coe.cient\ n] sample size "number of independent contrasts#

Þ 0887 British Ecological Society Journal of Animal Ecology\ 56\ 407Ð425

size and mean dispersal is curvilinear^ species that have a large population size have restricted dispersal "Fig[ 6#[ The relation between geographical range and mean dispersal is linear with scatter "Fig[ 7#[ These correlations were still signi_cant after correcting for body size "Table 0#\ or after the phylogenetic cor! rection "Table 1#[ Geographic range was weakly nega! tively correlated with SD of breeding dispersal while population size was weakly positively correlated with SD of natal dispersal[ The correlations with local abundance or its standard deviation were not sig! ni_cant after correcting for body size\ except for the standard deviation of local abundance with mean of breeding dispersal "Table 0#[ After correcting for phylogenetic relatedness\ local abundance was stron! gly negatively correlated with mean natal dispersal\ while the spatial heterogeneity in local abundance was negatively correlated with SD of breeding dispersal and with mean natal dispersal "Table 1#[ No cor! relation was signi_cant with the measure of popu! lation trend "Tables 0 and 1#[ To investigate the potential relationship between

dispersal and population trend\ we made further analyses including population trend taking into account those variables that signi_cantly in~uenced dispersal parameters "see below#[ We estimated sep! arately the correlation coe.cients between population trend and mean of breeding dispersal for the migrant and the resident species "Fig[ 8#] the correlation was signi_cant for the migrants "r  9=602\ n  00\ P  9=903# but not for the residents "r  9=129\ n  31\ P  9=032#[ The correlation coe.cient for migrants was also signi_cant after removing the e}ect of body size "r  9=613\ n  00\ P  9=901#\ or after correcting for phylogenetic relatedness among the migrant species "r  9=611\ P  9=907\ and r  9=810\ P ³ 9=990\ with both phylogenies\ respectively\ n  09 in both cases#[ The correlation between population trend and mean of natal dispersal was not signi_cant\ either for migrants "r  9=278\ n  01\ P  9=101#\ or for residents "r  9=153\ n  33\ P  9=972#[ The correlation between population trend and mean of breeding dispersal was close to signi_cance for species living in wet habitats during summer

417 Avian dispersal

Fig[ 6[ Plot of mean natal dispersal vs[ population size for 64 species of British birds[ A similar relationship was observed with breeding dispersal[

Fig[ 7[ Plot of mean natal dispersal vs[ geographical range for 64 species of British birds[ A similar relationship was observed with breeding dispersal[

Þ 0887 British Ecological Society Journal of Animal Ecology\ 56\ 407Ð425

"r  9=480\ n  09\ P  9=961#\ but migrant species living in wet habitats "2 out of 09# critically in~uenced this correlation[ Taking into account wet winter habi! tats gave no signi_cant results[ Including population trend along with population size or geographical range in a multiple regression did not yield signi_cant partial correlation coe.cients for population trend "P × 9=94#[ All morphometric variables were signi_cantly posi! tively correlated with means of natal and breeding dispersal "P ³ 9=94# but not after correction for body size "P × 9=0#[ After correcting for phylogenetic relatedness\ two morphometrical variables were sig! ni_cantly and consistently correlated with mean natal

dispersal] egg mass "r  9=219\ P ³ 9=90\ and r  9=142\ P ³ 9=94\ for both phylogenies\ respec! tively\ n  63#\ and wing length "r  9=331\ P ³ 9=990\ and r  9=132\ P ³ 9=94\ for both phylogenies\ respec! tively\ n  63#[ Migration had a signi_cant e}ect on all parameters of dispersal "Table 2#[ The migrant species disperse further than the resident species "class averages in kilometres] migrants\ 03=19\ n  04\ and residents\ 6=78\ n  41 for mean breeding dispersal^ 14=94\ n  04\ and 05=85\ n  41\ respectively\ for SD of breeding dispersal^ 11=65\ n  05\ and 04=47\ n  48\ respectively\ for mean natal dispersal^ 17=12\ n  05\ and 10=86\ n  48\ respectively\ for SD of natal dis!

418 E[ Paradis et al[

Fig[ 8[ Plot of mean breeding dispersal against long!term population trend measured by the changes in density on Common Birds Census plots during 0858Ð83\ "a# migrant species\ "b# resident species "see results for details on the analyses#[

Table 2[ E}ects of categorical variables on breedinga and natal dispersal of British birds after removing the e}ect of body size Breeding dispersal Mean

Migration Summer habitat] Open vs[ closed Dry vs[ wet Winter habitat] Open vs[ closed Dry vs[ wet Summer food Winter food Breeding social system Winter social system

Þ 0887 British Ecological Society Journal of Animal Ecology\ 56\ 407Ð425

Natal dispersal SD

Mean

SD

F

d[f[

P

F

d[f[

P

F

d[f[

P

F

d[f[

P

03=93

54\ 0

9=9993

7=86

54\ 0

9=993

02=72

62\ 0

9=9993

0=06

62\ 0

9=172

9=17 5=49

54\ 0 54\ 0

9=599 9=902

1=35 1=70

54\ 0 54\ 0

9=011 9=087

0=47 0=55

62\ 0 62\ 0

9=102 9=308

0=13 3=42

62\ 0 62\ 0

9=169 9=926

9=96 6=45 9=38 4=36 1=87 9=08

54\ 0 54\ 0 54\ 0 54\ 0 53\ 1 54\ 0

9=685 9=997 9=377 9=911 9=947 9=552

1=14 0=78 0=96 9=27 0=85 9=61

54\ 0 54\ 0 54\ 0 54\ 0 53\ 1 54\ 0

9=006 9=063 9=294 9=428 9=038 9=288

9=61 9=54 09=92 15=91 1=04 1=66

62\ 0 62\ 0 62\ 0 62\ 0 61\ 1 62\ 0

9=288 9=311 9=991 9=9990 9=013 9=090

1=57 4=97 0=65 1=51 3=77 4=10

64\ 0 62\ 0 62\ 0 62\ 0 61\ 1 62\ 0

9=095 9=916 9=078 9=009 9=909 9=914

persal#[ The dichotomous habitat variables opposing dry vs[ wet habitats had a strong e}ect on mean of breeding dispersal "Table 2#] the species living in wet habitats\ in summer and:or in winter\ disperse further than those in dry habitats "average] 05=93\ n  06\ and 6=90\ n  49\ respectively\ for summer habitat^ 06=72\ n  03\ and 6=94\ n  42\ respectively\ for winter habi! tat#[ The same variables had marginal e}ects on SD

of breeding dispersal and mean of natal dispersal[ Food in winter was also important "Table 2#] the spec! ies eating mainly animal prey in winter disperse fur! ther than those eating mainly plants "average] 09=86\ n  30\ and 5=56\ n  15\ respectively\ for mean breed! ing dispersal^ 10=02\ n  38\ and 8=43\ n  15\ respec! tively\ for mean natal dispersal#[ Food in summer had a signi_cant e}ect on mean of natal dispersal

429 Avian dispersal

"Table 2#[ Social system during the breeding season had a signi_cant e}ect on SD of natal dispersal] aver! age] colonial\ 19=70\ n  09\ neighbourhood\ 19=84\ n  7\ and territorial 04=82\ n  46[ All ANOVAs com! puted to test for a relation between evolutionary change in categorical variables and dispersal par! ameters gave non!signi_cant results "P × 9=98#[

Discussion

Þ 0887 British Ecological Society Journal of Animal Ecology\ 56\ 407Ð425

The distribution of animal dispersal distances and the factors a}ecting it are still poorly known "van Balen + Hage 0878#[ This study shows that British bird species di}er in their breeding and natal dispersal dis! tance distributions and that this variability can be related to ecological and biological variables[ The most important variables were] abundance and dis! tribution\ some life history traits\ whether the habitat is wet or dry\ winter food\ breeding social system\ migration and body size[ The phylogenetic analyses revealed that relatedness between species is critical for the evaluation of the importance of these variables on dispersal[ Some variables were no longer signi_cantly correlated after the phylogenetic correction\ sug! gesting that the simple correlations were the result of a common heritage through time "Harvey + Purvis 0880#[ On the other hand\ the correlation between age at _rst breeding and mean of natal dispersal was only revealed after phylogenetic correction[ This indicates further the importance of this kind of comparative analysis "Felsenstein 0874#[ The present study set out to decipher the inter! speci_c variations in dispersal patterns for British breeding birds[ In the following we discuss the results we obtained relative to each studied variable^ we emphasize the relationships between dispersal\ migration\ distribution and abundance in the light of the current ideas on these subjects and previous stud! ies[ Finally\ we underline the potential contribution of the present study[ The pervasive in~uence of body size on species| life history and ecology is evidenced by the positive correlations with both measures of dispersal across species[ Evolutionary trends towards larger body size have\ however\ been independent of dispersal which suggests that much of this variation is due to taxo! nomic di}erences[ Few other studies have explicitly considered the relation between dispersal and body size[ Peters "0872# reports that migration distances are greater for larger species but migration and dispersal are distinct processes[ Other morphometric variables were correlated with means of breeding and natal dispersal but these disappeared after the body size correction[ The fact that egg mass and wing length were correlated with mean natal dispersal after cor! recting for phylogenetic relatedness is intriguing[ It is interesting that Monkkonen "0884# showed that there is a relationship between wing shape and migration distance but found no relationship between wing

length and migration distance[ We have no biological explanation for a link between natal dispersal and egg mass or wing length\ and cannot exclude the possi! bility that these relations arose because of type I stat! istical errors due to the number of tests made[ Some life history traits were strongly correlated with several parameters of dispersal but an allometric relation seems to account for this as these correlations lost signi_cance after the body size correction\ except for number of broods per year[ Interestingly\ Bennett + Harvey "0877# showed there is a negative correlation between the number of broods per year and adult survival among European birds\ even after correcting for body size[ This is evidence that the number of broods per year is involved in life history trade!o}s[ Dispersal may also be involved in such trade!o}s\ which could explain the negative correlation between natal dispersal and number of broods per year once the in~uence of body size was removed[ Correlations with other life history variables may be di.cult to characterize because of the small number of species we have here compared to that in Bennett + Harvey "0877#[ An important result from the phylogenetic analyses is that the correlations between age at inde! pendence or age at _rst breeding and mean natal dis! persal were stronger than the cross!species corre! lations[ A close examination of the PICs indicates that this result is mainly due to the contrast between the raven\ Corvus corax L[\ and its closest relative in our reconstructed phylogenies\ the crow\ C[ corone L[ The raven has considerably higher age at independence and mean of natal dispersal than the crow\ and as both species are closely related this resulted in a high PIC for this pair of species[ PICs for the Corvidae are generally high as there is an important variation in age at independence and mean dispersal distance within this family[ If we removed the Corvidae from the phylogenetic analyses\ then the correlations were still positive but not signi_cant[ Correlations between population size\ geographical range and local abundance on the one hand\ and dis! persal parameters on the other\ are not surprising as dispersal is usually thought to have a critical in~uence on these variables "Tilman 0883#[ The relationships here were negative\ which we interpret to mean that the species exploiting the most abundant habitats in the British Isles do not need to disperse a long distance to _nd suitable sites for reproduction[ Brown\ Mehl! man + Stevens "0884# suggested that variations in abundance of North American birds are primarily determined by variation in the distribution of suitable habitats[ If the same explanation holds for British birds\ then this would corroborate our interpretation of the negative relationship between dispersal dis! tances and abundance[ There was no correlation between long!term population trend and dispersal parameters considering all species together\ but the correlation was positive and signi_cant when con! sidering the migrant species alone[ This correlation

420 E[ Paradis et al[

Þ 0887 British Ecological Society Journal of Animal Ecology\ 56\ 407Ð425

was not in~uenced by body size or phylogenetic relatedness[ Lack of su.cient dispersal\ and hence colonization abilities\ has been proposed as a causal factor of population decline "Harrison 0880#[ If so\ a positive correlation between dispersal and population trend is expected[ This suggests that poor dispersal in some migrant species could have a role in their long! term decline[ However\ recent investigations showed that mortality outside the breeding season could be the major cause of population decline for several British birds "Baillie + Peach 0881^ Peach\ Crick + Marchant 0884#\ indicating that further evaluation of the role of dispersal is necessary[ The negative correlation between dispersal and both species| geographical range size and species| abundance among British birds is entirely new^ the non!independence of abundance and range has been widely documented "e[g[ Gaston\ Blackburn + Law! ton 0886#[ Two studies have also shown life history traits associated with fecundity to be correlated with abundance and range size "Blackburn\ Lawton + Gre! gory 0885^ Gaston + Blackburn 0885#[ Higher rates of o}spring production tend to be found in species with higher abundances and larger range sizes\ although these relationships often disappear after cor! recting for phylogeny[ Intriguingly\ we _nd that fec! undity "measured as broods per year# declines with breeding and natal dispersal across species\ and that later development "measured as age at independence and age at _rst breeding# is associated with greater natal dispersal within taxa[ Thus species with limited dispersal tend to be the most fecund and develop earlier^ furthermore\ they attain higher abundances and larger geographical range sizes[ The interrelation between these variables and a suite of other macro! ecological variables is likely to be complex^ the role of dispersal in such considerations has until now largely been neglected[ It is unclear why some of these pat! terns are robust to phylogenetic analysis whereas others are not[ This implies that some of these patterns may be taxonomic artefacts but the results from di}erent studies are mixed "Blackburn et al[ 0885^ Gaston + Blackburn 0885# suggesting that the power to detect these patterns is low[ The signi_cant cor! relations between dispersal parameters and popu! lation size or geographical range after the phylo! genetic correction are somewhat intriguing[ It is hard to predict that these variables should diverge simul! taneously during evolutionary times[ This result is in agreement with previous reports of an in~uence of phylogenetic relatedness between species on the relations between body size and abundance in British birds "Nee et al[ 0880# or between abundance and dis! tribution "Gregory 0884#[ Competition within guilds of closely related species may have a critical in~uence on these phylogenetic relationships "Nee et al[ 0880^ Gregory 0884#[ Migration had a strong e}ect on dispersal parameters\ but this needs to be interpreted

cautiously[ Migration movements could have con! founded the present study^ Baker "0882# showed that the redstart Phoenicurus phoenicurus L[\ and the wheatear Oenanthe oenanthe L[ "two migrant birds# in Britain tend to move signi_cantly along the northÐ south axis and to a greater extent than the robin Eri! thacus rubecula L[\ a closely related resident species[ Baker "0882# suggested that such post~edging move! ments have a function in locating a future breeding site\ an overwintering site\ or in creating a navi! gational target^ some of these explanations could be non!exclusive[ We tested the possibility that young migrant birds could be ringed far away from their birth sites by analysing separately the birds ringed in the nest "EURING age code 0# and those ringed in their year of birth but as freely ~ying "EURING age code 2# for each migrant species[ Our expectation\ according to Baker|s results\ was that the mean\ and especially the SD\ should be greater in the latter cate! gory[ This was not veri_ed for 04 species out of 05\ the exception being the swallow Hirundo rustica L[ where the {age 2| birds ringed in August were re! covered at a signi_cantly greater distance than all the other birds ringed as juveniles "these birds were excluded for the estimation of the natal dispersal par! ameters#[ The other 04 species did not verify our expectation because birds ringed as {age 2| were a small part of the sample or were ringed mostly early in the breeding season\ thus introducing no bias in our analyses[ We were quite restrictive with respect to the distances of migrant birds and imposed a threshold so that all distances below it were not preferentially directed in any direction^ this threshold varied from 29 km for the turtle dove Streptopelia turtur L[ to 299 km for the reed warbler Acrocephalus scirpaceus L[ The distributions of dispersal distances were di}erent between migrant and resident species "see Figs 0 and 1#[ We consider that our results reveal a real pattern and that migrants disperse further than resident birds\ although\ as we cannot be absolutely certain that our analytical methods eliminate all migratory move! ments\ this conclusion should be regarded as provisional[ British migrant birds di}er from resident species with respect to several ecological characters "O|Connor 0870#\ and this could explain our results[ O|Connor "0874# hypothesized that migrant birds in Britain use more intermediate successional habitats compared to residents[ Hence\ migrants may have an opportunistic settling strategy "Johnson + Grier 0877^ Spendelow et al[ 0884#[ However\ there was variation within the migrants and some species have relatively short dispersal distances similar to most residents "e[g[ swallow or house martin Delichon urbica L[#[ Habitat was an important factor for breeding dis! persal with respect to the {wet vs[ dry| feature[ Possible confounding interactions with migration seem improbable as only four migrant species use wet habi! tats in summer\ and only three in winter[ Furthermore\ the e}ect of {wet vs[ dry| habitat was limited to breed!

421 Avian dispersal

ing dispersal[ The most sensible explanation of this result is that as wet habitats are more discontinuously distributed in space "patchy distribution of wet areas# or in time "wet areas are less stable# than dry ones\ the species requiring a wet habitat to breed need to disperse further[ This is supported by the absence of e}ect of this habitat variable on natal dispersal[ The average of mean natal dispersal over the species living in dry habitats in summer "03=89\ n  46# or in winter "03=79\ n  59# is close to the average of mean breed! ing dispersal over the species living in wet habitats in summer "05=93\ n  06# or in winter "06=72\ n  03#[ This suggests that wet habitats impose a constraint on breeding dispersal but not on natal dispersal[ Food in winter had a signi_cant e}ect on dispersal parameters\ but there is little doubt that this was caused by an indirect e}ect of migration as all but one migrant species "the turtle dove# eat animal prey in winter[ The present paper reports the _rst comparative study of large spatial scale dispersal patterns for a taxonomic group of animals[ Such studies are critical as parameters on dispersal are usually lacking in popu! lation studies "Brawn + Robinson 0885^ Kot\ Lewis + van den Driessche 0885#[ We were not able to inves! tigate here the variation in dispersal distances at the intraspeci_c level\ but we characterized interspeci_c di}erences[ Our _ndings suggest that several factors critically in~uence the distribution of dispersal distances\ and that it may be relevant to consider these factors in future studies on the costs and bene_ts of dispersal[ Our results should be valuable for other comparative studies in order to evaluate the role of dispersal in several issues of population biology\ for instance population synchrony\ colonization and range expansion\ metapopulation and sourceÐsink dynamics\ and population genetic structure[

Acknowledgements

Þ 0887 British Ecological Society Journal of Animal Ecology\ 56\ 407Ð425

This study was possible with the constant e}ort of volunteer ringers in Britain and Ireland over many years[ The authors thank the Ringing Unit of the BTO for their permanent work with the ringing data and for particular advice for the present study[ Special thanks to Rob Williams and Will Peach for stimu! lating discussions and to Jean Clobert\ Will Peach\ John Reynolds and David Thomson for helpful com! ments on a previous version of this paper[ We are grateful to two anonymous referees who made com! ments that considerably helped us in clarifying our manuscript[ John Marchant computed the CBC esti! mates for the present analyses\ and Graham Austin helped with the resampling analysis[ Financial sup! port for the present study was provided by NERC grant GST:91:0086 as part of the NERC:SOAFED special topic on Large Scale Processes in Ecology and Hydrology[ The posts held by Stephen Baillie and Richard Gregory are supported by a contract from

the Joint Nature Conservation Committee on behalf of English Nature\ the Countryside Council for Wales\ Scottish Natural Heritage and the Department of the Environment for Northern Ireland[

References Avise\ J[C[ "0885# Three fundamental contributions of molecular genetics to avian ecology and evolution[ Ibis\ 027\ 05Ð14[ Baillie\ S[R[ + Green\ R[E[ "0876# The importance of vari! ation in recovery rates when estimating survival rates from ringing recoveries[ Acta Ornithologica\ 12\ 30Ð59[ Baillie\ S[R[ + Milne\ H[ "0878# Movements of eiders Som! ateria mollissima on the east coast of Britain[ Ibis\ 020\ 210Ð224[ Baillie\ S[R[ + Peach\ W[J[ "0881# Population limitation in Palaearctic!African migrant passerines[ Ibis\ 023 "Suppl[ 0#\ 019Ð021[ Baker\ R[R[ "0882# The function of post!~edging explo! ration] a pilot study of three species of passerines ringed in Britain[ Ornis Scandinavica\ 13\ 60Ð68[ Baker\ M[\ Nur\ N[ + Geupel\ G[R[ "0884# Correcting biased estimates of dispersal and survival due to limited study area] theory and an application using wrentits[ Condor\ 86\ 552Ð563[ van Balen\ J[H[ + Hage\ F[ "0878# The e}ects of environ! mental factors on tit movements[ Ornis Scandinavica\ 19\ 88Ð093[ Balmer\ D[E[ + Peach\ W[J[ "0886# Review of Natural Avian Mortality Rates[ BTO Research Report no[ 064[ British Trust for Ornithology\ Thetford[ Belant\ J[L[ + Dolbeer\ R[A[ "0882# Migration and dispersal of laughing gulls in the United States[ Journal of Field Ornithology\ 53\ 446Ð454[ Bengtsson\ G[\ Hedlund\ K[ + Rundgren\ S[ "0883# Food! dependent and density!dependent dispersal] evidence from a soil collembolan[ Journal of Animal Ecology\ 52\ 402Ð19[ Bennett\ P[M[ + Harvey\ P[H[ "0877# How fecundity bal! ances mortality in birds[ Nature\ 222\ 105[ Berven\ K[A[ + Grudzien\ T[A[ "0889# Dispersal in the wood frog "Rana sylvatica#] implications for genetic population structure[ Evolution\ 33\ 1936Ð1945[ Blackburn\ T[M[\ Lawton\ J[H[ + Gregory\ R[D[ "0885# Relationships between abundances and life histories of British birds[ Journal of Animal Ecology\ 54\ 41Ð51[ Boonstra\ R[\ Krebs\ C[J[\ Gaines\ M[S[\ Johnson\ M[L[ + Craine\ I[T[M[ "0876# Natal philopatry and breeding sys! tems in voles "Microtus spp[#[ Journal of Animal Ecology\ 45\ 544Ð562[ Brawn\ J[D[ + Robinson\ S[K[ "0885# SourceÐsink popu! lation dynamics may complicate the interpretation of long! term census!data[ Ecology\ 66\ 2Ð01[ Brown\ J[H[\ Mehlman\ D[W[ + Stevens\ G[C[ "0884# Spatial variation in abundance[ Ecology\ 65\ 1917Ð1932[ Burt\ A[ "0878# Comparative methods using phylogenetically independent contrasts[ Oxford Surveys in Evolutionary Biology\ 5\ 22Ð42[ Clark\ J[A[\ Adams\ S[Y[\ Peach\ W[J[ + Simons\ J[R[ "0885# Report on bird ringing in Britain and Ireland for 0883[ Ringing and Migration\ 06\ 25Ð68[ Clobert\ J[ + Lebreton\ J[!D[ "0880# Estimation of demo! graphic parameters in bird populations[ Bird Population Studies\ Relevance to Conservation and Management "eds C[M[ Perrins\ J[!D[ Lebreton + G[ Hirons#\ pp[ 64Ð093[ Oxford University Press\ Oxford[ Cramp\ S[ + Simmons\ K[E[L[ "0866Ð83# The Birds of the Western Palearctic\ Vols 0Ð7[ Oxford University Press\ Oxford[ Dobson\ F[S[ "0871# Competition for mates and pre!

422 E[ Paradis et al[

Þ 0887 British Ecological Society Journal of Animal Ecology\ 56\ 407Ð425

dominant juvenile male dispersal in mammals[ Animal Behaviour\ 29\ 0072Ð0081[ Drilling\ N[E[ + Thompson\ C[F[ "0877# Natal and breeding dispersal in house wrens "Troglodytes aedon#[ Auk\ 094\ 379Ð380[ Dwyer\ G[ + Elkinton\ J[S[ "0884# Host dispersal and the spatial spread of insect pathogens[ Ecology\ 65\ 0151Ð0164[ EURING "0868# Code manual] new EURING[ British Trust for Ornithology\ Tring[ Felsenstein\ J[ "0874# Phylogenies and the comparative method[ American Naturalist\ 014\ 0Ð04[ Felsenstein\ J[ "0877# Phylogenies and quantitative charac! ters[ Annual Review of Ecology and Systematics\ 08\ 334Ð360[ Felsenstein\ J[ "0882# PHYLIP "Phylogeny Inference Pack! age# version 2[4c[ Distributed by the author[ Department of Genetics\ University of Washington\ Seattle[ Ferrer\ M[ "0882# Juvenile dispersal behaviour and natal philopatry of a long!lived raptor\ the Spanish Imperial Eagle Aquila adalberti[ Ibis\ 024\ 021Ð027[ Gaines\ M[S[ + McClenaghan\ L[R[ Jr "0879# Dispersal in small mammals[ Annual Review of Ecology and System! atics\ 00\ 052Ð085[ Gaston\ K[J[ + Blackburn\ T[M[ "0885# Global scale mac! roecology] interactions between population size\ geo! graphic range size and body size in Anseriformes[ Journal of Animal Ecology\ 54\ 690Ð603[ Gaston\ K[J[\ Blackburn\ T[M[ + Lawton\ J[H[ "0886# Inter! speci_c abundanceÐrange size relationships] an appraisal of mechanisms[ Journal of Animal Ecology\ 55\ 468Ð480[ Gibbons\ D[W[\ Reid\ J[B[ + Chapman\ R[A[ "eds# "0882# The New Atlas of Breeding Birds in Britain and Ireland] 0877Ð80[ T[ + A[D[ Poyser\ London[ Greenwood\ P[J[ "0879# Mating systems\ philopatry and dis! persal in birds and mammals[ Animal Behaviour\ 17\ 0039Ð 0051[ Greenwood\ P[J[ + Harvey\ P[H[ "0865# The adaptive sig! ni_cance of variation in breeding area _delity of the black! bird "Turdus merula L[#[ Journal of Animal Ecology\ 34\ 776Ð787[ Greenwood\ P[J[ + Harvey\ P[H[ "0871# The natal and breed! ing dispersal of birds[ Annual Review of Ecology and Sys! tematics\ 02\ 0Ð10[ Greenwood\ P[J[\ Harvey\ P[H[ + Perrins\ C[M[ "0868# The role of dispersal in the great tit "Parus major#] the causes\ consequences and heritability of natal dispersal[ Journal of Animal Ecology\ 37\ 012Ð031[ Gregory\ R[D[ "0884# Phylogeny and relations among abun! dance\ geographical range and body size of British breed! ing birds[ Philosophical Transactions of the Royal Society of London[ Series B[ Biological Sciences\ 238\ 234Ð240[ Grinnell\ J[ "0811# The role of the {accidental|[ Auk\ 28\ 262Ð 279[ Hansson\ L[ "0880# Dispersal and connectivity in meta! populations[ Metapopulation Dynamics] Empirical and Theoretical Investigations "eds M[ Gilpin + I[ Hanski#\ pp[ 78Ð092[ Academic Press\ London[ Harrison\ S[ "0878# Long!distance dispersal and colonization in the bay checkerspot butter~y\ Euphydryas editha bayen! sis[ Ecology\ 69\ 0125Ð0132[ Harrison\ S[ "0880# Local extinction in a metapopulation context] an empirical evaluation[ Metapopulation Dynam! ics] Empirical and Theoretical Investigations "eds M[ Gilpin + I[ Hanski#\ pp[ 62Ð77[ Academic Press\ London[ Harvey\ P[H[ + Purvis\ A[ "0880# Comparative methods for explaining adaptations[ Nature\ 240\ 508Ð513[ Hengeveld\ R[ "0883# Small step invasion research[ Trends in Ecology and Evolution\ 8\ 228Ð231[ Ibrahim\ K[M[\ Nichols\ R[A[ + Hewitt\ G[M[ "0885# Spatial patterns of genetic variation generated by di}erent forms of dispersal during range expansion[ Heredity\ 66\ 171Ð180[

Johnson\ D[H[ + Grier\ J[W[ "0877# Determinants of breed! ing distributions of ducks[ Wildlife Monographs\ 099\ 0Ð26[ Kellner\ C[J[\ Brawn\ J[D[ + Karr\ J[R[ "0881# What is habi! tat suitability and how should it be measured< Wildlife 1990] Populations "eds D[R[ McCullough + R[H[ Barrett#\ pp[ 365Ð377[ Elsevier Applied Science Ltd\ London and New York[ Kirkland\ G[L[ Jr "0877# Meadow voles "Microtus pennsyl! vanicus# on forest clearcuts] the role of long!distance dis! persal[ Journal of the Pennsylvania Academy of Sciences\ 51\ 72Ð74[ Kot\ M[\ Lewis\ M[A[ + van den Driessche\ P[ "0885# Dis! persal data and the spread of invading organisms[ Ecology\ 66\ 1916Ð1931[ Lack\ D[ "0855# Population studies of birds[ Clarendon Press\ Oxford[ Lebreton\ J[D[\ Burnham\ K[P[\ Clobert\ J[ + Anderson\ D[R[ "0881# Modeling survival and testing biological hypotheses using marked animals] a uni_ed approach with case studies[ Ecological Monographs\ 51\ 56Ð007[ Lindberg\ M[S[\ Sedinger\ J[S[ + Rexstad\ E[A[ "0884# Esti! mating nest site _delity of adult female black brant with multi!state modeling and geographic information systems[ Journal of Applied Statistics\ 11\ 614Ð624[ Marchant\ J[H[\ Hudson\ R[\ Carter\ S[P[ + Whittington\ P[ "0889# Population Trends in British birds[ British Trust for Ornithology\ Tring[ Matthysen\ E[\ Adriaensen\ F[ + Dhondt\ A[A[ "0884# Dis! persal distances of nuthatches\ Sitta europaea\ in a highly fragmented forest habitat[ Oikos\ 61\ 264Ð270[ Mayr\ E[ "0852# Animal Species and Evolution[ Belknap Press of Harvard University Press\ Cambridge\ Massachusetts[ McLaughlin\ J[F[ + Roughgarden\ J[ "0881# Predation across spatial scales in heterogeneous environments[ Theoretical Population Biology\ 30\ 166Ð188[ Mindell\ D[P[ "0881# DNA!DNA hybridization and avian phylogeny "review of Phylogeny and Classi_cation of Birds] a Study in Molecular Evolution by C[G[ Sibley and J[E[ Ahlquist and Distribution and Taxonomy of Birds of the World by C[G[ Sibley and B[L[ Monroe Jr#[ Systematic Biology\ 30\ 015Ð023[ Monkkonen\ M[ "0884# Do migrant birds have more pointed wings