Predictions of future Nearctic landbird vagrants to Europe Chandler S. Robbins

arities generate more than their share of interest among the birdwatch- Ring public. The avid field observer thrills at breaking an arrival or departure date. The 'twitcher' travels long distances to add rare species to his list. Ringers are attracted to coastal concentration points or offshore islands hoping to handle, photograph and describe some foreign waif that has had the misfortune to be carried thousands of kilometres off its normal migratory course by vagaries of the weather. An American's interest in the landfall" of vagrant Nearctic landbirds in Europe is understandably different from that of you who are on the receiving end. Our first questions are: (1) are the small numbers of American vagrants that reach European shores indicative of tremendous losses at sea of other individuals of the same species, and (2) do these losses involve such a large proportion of the population of some species that the loss could be detected by the North American Breeding Survey in the following summer? Along the western shore of the North Atlantic, one of the chief sources of vagrant is the occasional passage of a tropical hurricane north­ eastward along or parallel to the coast. Depending to some degree on the date that a hurricane affects the east coast of the United States or Canada, one may predict some of the species that will be deposited along its course. The primary concern of ornithologists who have reported on the effects of such storms, however, has been the recurrence of species such as White- tailed Tropicbirds Phaethon lepturus, Sooty Terns Sterna Juscata, Black Skimmers Rynchops nigra, or Yellow-billed * far outside their normal ranges. The. more difficult-to-assess storm-related mortality of other species has received less attention. Ringers along the Atlantic coast of the United States and Canada know that stiff westerly winds frequently cause a great accumulation of grounded migrants along the coast. On certain days, a large proportion of these birds arrives in emaciated condition, having been blown far offshore and then having struggled back, flying into the wind. The terrestrial birds that make a successful landfall, and become European vagrants, must often be only a tiny fraction of those that are lost at sea. Thus, my primary interest in North American vagrants in Europe is not so much to learn or to predict which species have successfully made the passage, but to use this information to gain a better understanding of the potential impact on Nearctic bird populations of the tremendous losses that may take place at sea.

*See tables 1 and 2 for scientific names.

448 [Bril. Birds 73: 448-457, October 1980] Future Nearciic landbird vagrants 449 Meteorological context Paramount to any understanding of storm-blown migrants is basic in­ formation on both normal and abnormal atmospheric circulation across the North Atlantic. Norman Elkins (1979) has done a masterful job of addressing this subject. Before seeing his paper, I had laboured in vain trying to relate some of the British landfalls to passage of American hurricanes. As Mr Elkins has convincingly shown, however, fast eastward- moving waves associated with certain strong warm sectors provide not only a more logical explanation than that offered by western Atlantic hurricanes, but are supported by the arrival dates of 75% of the vagrants.

Methods Accepted records, 1947-76 I have based this study on accepted records of North American and near-passerines (cuckoos, nightjars, and woodpeckers) recorded in Britain and Ireland during the same 30-year period (1947-76) summarised in Elkins's table 1. Older records, dating from before the large network of well-manned coastal and island bird observatories, may be biased by disproportionately large numbers of the bigger and more conspicuous species (eg Yellow-billed ); the records for 1977 and 1978 were not complete and were still under consideration by the Rarities Committee when this study was undertaken.

Candidate species If one were subjectively to compile a list of candidate species composed of those most likely to make a successful transatlantic crossing, it would reasonably be based on those species already proved successful. If we seek greater objectivity, however, we can employ modern statistical techniques and computer technology to examine a large array of data and make logical predictions based on information provided to this end. It will be necessary to define the 'universe' of species from which the best candidates will be selected. Many North American species have some chance of appearing in the British Isles one or more times during the next few centuries. Long-distance migrants that travel along the east coast of Canada and the United States are more likely to make transatlantic migrations than are rare, western, or non-migratory species. Thus, a logical starting point is to define the universe of interest in terms of numerical abundance along the North American east coast in autumn. The best data sources for this purpose are banding (ringing) records from coastal stations at Manomet Bird Observatory in Massachusetts, Island Beach State Park in New Jersey, and Kiptopeke Beach in Virginia. These three stations are about evenly spaced from Cape Cod Bay to the mouth of Chesapeake Bay and each has a large sample of recent banding records. The years selected for this study varied among stations, depending on consistency of coverage and availability of summarised data. The Manomet figures were from 52,081 birds banded during August to November 1969-76; the Island Beach data were from 207,253 birds banded in autumn 1956-68; the Kiptopeke data were from 450 Future Nearctic landbird vagrants 94,865 birds banded during late August to late October 1967-75. Each and near-passerine species that averaged two or more captures per year at one or more of the three stations was included in the analysis. Excluded, however, were those species that breed on both sides of the Atlantic: Sand Martin (Bank Swallow) Riparia riparia, Swallow (Barn Swallow) Hirundo rustica, Wren (Winter Wren) Troglodytes troglodytes, Treecreeper (Brown Creeper) Certhia jamiliaris, Starling (European Starling) Stumus vulgaris and House Sparrow Passer domesticus. Unfortunately, two vagrants already on the British and Irish list did not qualify for inclusion in this study. These were the Common Nighthawk, an aerial feeder that is rarely captured at the coastal banding stations, and the Summer Tanager, which nests primarily southwest of the banding stations and was too infrequently caught to qualify for inclusion. These two species were omitted from the analysis to be consistent. The analysis, therefore, is based on 31 species that occurred in Britain and Ireland during the period 1947-76, and 73 candidate species for which there were no accepted records up to and including 1976. Definition oj migration characteristics To be a likely candidate for transatlantic" migration, a passerine species must be present in good numbers at the right place at the right time, and must be physically capable of surviving an extended trip with little or no food intake along the way. Thus, an index of these characteristics, or of associated factors, should be correlated with the probability (success or failure) of the species crossing to Europe. If numerical values are assigned to various migration characteristics, a computer algorithm (calculation in a number of steps) could select those variables that are most closely cor­ related with successful crossings and derive a formula from which candi­ date species could be ranked according to their likelihood of completing the trip. The success of the exercise depends largely on the ability of the investigator to select a powerful battery of predictor variables for the computer to select from, and to assign the most appropriate numerical data to each variable. I will specify in detail the data I used so that future researchers can profit from my inadequacies and generate better predictions. ABUNDANCE ON EAST COAST OF NORTH AMERICA A glance at the list of North American vagrants that have been recorded in Britain and Ireland supports the hypothesis that abundance during autumnal migration on the western shore of the North Atlantic is a prime factor for inclusion. For each of the three banding stations, the total number of new banded birds of each species within a sampling period was divided by the total bandings during the same period to obtain the relative percentage of each species. Percentages were individually summed across the three stations to obtain a pooled index of abundance for each of the 104 species. BODY WEIGHT The North American passerines and near-passerines recorded most frequently in the British Isles tend to be medium to large birds. Even among the wood-warblers, the large species are primarily the ones with the most records. For this reason, body weight was selected as one of the input characteristics. The variable was defined as mean autumn weight of primarily east coast birds from the Operation Recovery banding program for the years 1966-68. STANDARD DEVIATION OF BODY WEIGHT This statistic, which measures variability within a sample, is high for species whose individuals differ extensively in amount of body fat, and is low for species in which all individuals weigh about the same. It is one way, therefore, of detecting Future Nearctic landbird vagrants 451 species in which some individuals are prepared for long nonstop flights. The Blackpoll Warbler, for example, stores much fat for useduring migration (SD=3.63, n= 1,635), whereas the Eastern Pewee does not (SD= 1.45, n=229); both species have a mean weight of 14g. The standard deviations were taken from the same source as the body weights and are based on samples of at least several hundred individuals for most species. For species such as the Common Grackle, in which males are much heavier than females, only male weights were used, to avoid a high standard deviation caused by sexual dimorphism rather than by amount of body fat. MIGRATORY DISTANCE All North American vagrants in Europe are migratory species, with long-distance migrants far outnumbering short-distance migrants. Only crude estimates of the usual migration distances of each species can be made until several banded North American passerine vagrants are encountered in Europe, and more is known about the relation between their normal breeding and wintering areas in the New World. Several candidate species may have a leap-frog pattern of migration, with those from the northern extremity migrating to the southern portion of the winter range. Lacking detailed information of this kind, the migratory distance was arbitrarily taken as the straight line distance in miles from the centre of the breeding range of the eastern population to the approximate centre of the winter range. WEST-TO-EAST COMPONENT OF THE AUTUMNAL MIGRATION This variable is difficult to quantify, but the concept of a long-distance migrant from the interior overshooting the coast and heading over the Atlantic is too important to ignore (Russell 1978). Consequently, I used a simple code of 0, 1 or 2 as an estimate of the west-to-east component of autumnal migration of each species. These estimates were based on: (a) the relation between breeding and wintering areas; (b) knowledge of migratory routes; (c) the abundance of transients at coastal versus inland locations; and (d) tendency for northeastward (reverse) migration in autumn. DATES OF MIGRATION Sharrock & Sharrock (1976) showed that North American vagrants are strongly clustered in the first three weeks of October, with only one or two records for the entire month of August and the first half of September. This observation does not correspond with the peak of migration within the normal range of these birds, so it appears likely that the weather patterns essential to a swift ocean crossing rarely occur before late September. Consequently, the likelihood of early migrants reaching Europe is greatly reduced. I coded migration dates on a scale of 1 to 15, using dates encompassing 90% passage at the Ocean City Operation Recovery station on the coast of Maryland (1964-68) and comparing these dates with the Sharrocks' histogram. NOCTURNAL VERSUS DIURNAL MIGRATION During marginal or poor weather conditions, one would expect nocturnal migrants to become disoriented more frequently than diurnal migrants. In an effort to determine whether this contributes to the arrival of North American vagrants in Europe, I coded each species on a scale of 1 to 4. Species I have never observed on migration during daylight were coded 4 and those for which I have no evidence of nocturnal migration were coded 1. Species known to migrate both by night and by day were coded 2 or 3. Statistical analysis A stepwise regression program (Barr et al. 1976) was used to examine the relationship between the number of records in Britain and Ireland (dependent variable) of each of the 104 species with relation to the following eight (independent) variables: (a) abundance at Manomet, (b) abundance at the three coastal stations combined, (c) body weight, (d) standard deviation of body weight, (e) migratory distance, (f) west-to-east com­ ponent of migration, (g) dates of migration, and (h) nocturnal versus diurnal migration. The variable most closely correlated with the number of records in the British Isles was selected first; each of the other variables was then examined in connection with the first, and the one that improved the correlation the most was selected. This process continued until none of the remaining variables significantly improved the correlation between in­ dependent and dependent variables. The selected variables were then used to calculate predicted values for each species. 452 Future Nearciic landbird vagrants Results The stepwise regression program sequentially selected the following four variables: 1. (f) West-to-east component (F=3.35, probability of greater F=0.0701) 2. (d) Standard deviation of body weight (14.10,0.0003) 3. (e) Migratory distance (11.45, 0.0010) 4. (b) Abundance, three stations combined (9,61,0.0025) The amount of variance in the number of records explained by the regression model (R2) is not of direct interest because it is strongly affected by the number of species included in the model. Likewise, the predicted Table 1. Predicted relative likelihood of autumn transatlantic vagrancy (see text) and actual numbers of Nearctic passerines and near-passerines observed in Britain and Ireland in spring and in autumn winter during 1947-76 The predicted values (P) in tables 1 and 2 were calculated with the following formula: P=-2.169 +0.412 (SD) +0.164 (ABUND) +0.0823 (DIST) +0.795 (VV-E) where SD is the standard deviation of body weight in grams; ABUND is the sum of the percentage abundances at Manomet, Island Beach and Kiptopeke; DIST is the distance in hundreds of miles between centres of breeding and winter ranges; and VV-E is the coded estimate of west-to-east component of autumn migration SPRING AUTUMN AND WINTER Species Observed Observed Predicted

Yellow-rumped Warbler Dendroica comnata 0 7 8.18 Yellow-billed Cuckoo Coccyzus americanus 0 18 6.26 Bobolink Dolichonyx oryzivorus 0 6 6.11 Blackpoll Warbler Dendroica striata 0 14 6.02 American Redstart Setophaga ruticilla 0 2 5.77 Grey-cheeked Thrush Catharus minimus 0 13 5.27 Black-billed Cuckoo Coccyzus erythrophthalmus 0 5 5.21 Rose-breasted Grosbeak Pheucticus ludovicianus 0 7 4.33 Red-eyed Vireo Vireo olivaceus 0 8 4.32 Swainson's Thrush Catharus ustulatus 1 3 3.96 Veery Catharusjuscescens 0 1 3.95 White-throated Sparrow Zonotrichia albicollis 6 4 3.56 Slate-coloured Juncoy«nco hyemalis 6 0 3.36 Northern Waterthrush Seiurus noveboracensis 0 2 3.22 Tennessee Warbler Vermivora peregrina 0 2 3.04 Yellowthroat Geothlypis trichas 0 1 3.04 American Robin Turdus migratorius 2 13 2.85 Northern Oriole Icterus galbula 2 10 2.77 Yellow Warbler Dendroica petechia 0 1 2.43 Scarlet Tanager Piranga olivacea 0 2 2.24 Ovenbird Seiurus aurocapillus 0 1 2.02 Black-and-white Warbler Mniotilta varia 0 1 1.96 Parula Warbler Panda americana 0 3 1.33 Yellow-bellied Sapsucker Sphyrapicus varius 0 1 1.25 Hermit Thrush Catharusguttatus 1 0 1.06 Fox Sparrow Zonotrichia iliaca 1 0 1.04 Evening Grosbeak Hesperiphona vespertina 1 0 0.95 Song Sparrow Zonotrichia melodia 4 0 0.83 Brown Thrasher Toxostoma rujum 0 1 0.50 Rufous-sided Towhee Pipilo erythrophthalmus 1 0 0.24 Hooded Warbler Wilsonia citrina 0 1 0.15 Common Nighthawk Chordeiles minor 0 6 — Summer Tanager Piranga rubra 0 1 — Future Nearciic landbird vagrants 453 number of records for each species varies inversely with the number of candidate species in the list; thus, the results cannot be used to predict probability of occurrence, but only to indicate relative position in the list of candidates. In table 1, for example, the predicted values for five of the top ten species are considerably below die observed number of British and Irish records. If only the 31 species with accepted records had been included in the analysis, the predicted values would have been much closer to the observed values; with the addition of 73 unrecorded species to the model, however, the slope Table 2. Most likely candidates for autumn transatlantic flight: predicted relative likelihood of autumn vagrancy (see text) of Nearctic passerines and near-passerines not observed in Britain or Ireland during 1947-76 Species Predicted number GROUP I Catbird Dumetella carolinensis 4.86 Connecticut Warbler Oporomis agilis 4.73 Belted Kingfisher Ceryle alcyon 4.45 Eastern Kingbird Tyrannus tyrannm 4.36 Yellow-shafted Flicker Colaptes auratus 4.29 Common Grackle Quiscalus quiscula 4.22 GROUP 2 Dickcissel Spiza americana 2.86 Alder Flycatcher Empidonax alnorum 2.86 Blue Jay Cyanocitta cristata 2.45 Wood Thrush Hylocichla mustelina 2.42 Black-throated Blue Warbler Dendroica caerulescens 2.29 Red-winged Blackbird Agelaius phoeniceus 2.15 Canada Warbler Wilsonia canadensis 2.06 Orange-crowned Warbler Vermivora celata 1.93 Black-capped Chickadee Pants atricapillus 1.83 Palm Warbler Dendroica palmarum 1.75 GROUP 3 Yellow-bellied Flycatcher Empidonaxflaviventris 1.55 Whip-poor-will Caprimulgus vocijerus 1.41 Mourning Warbler Oporomis Philadelphia 1.36 Magnolia Warbler Dendroica magnolia 1.33 Cape May Warbler Dendroica tigrina 1.33 Cedar Waxwing Bombycilla cedrorum 1.20 Yellow-breasted Chat Icteria virens 1.18 Great Crested Flycatcher Myiarchus crinitus 1.17 Savannah Sparrow Ammodramus sandwichensis 1.13 Ruby-crowned Kinglet Regulus calendula 1.09 Golden-crowned Kinglet Regulus satrapa 1.06 Eastern Pewee Contopus virens 1.03 Black-throated Green Warbler Dendroica virens 0.98 Least Flycatcher Empidonax minimus 0.96 Blue-winged Warbler Vermivora pinus 0.91 Philadelphia Vireo Vireo philadelphicus 0.91 Prairie Warbler Dendroica discolor 0.90 GROUP 4 Wilson's Warbler Wilsoniapusilla 0.83 Bay-breasted Warbler Dendroica castanea 0.79 Swamp Sparrow Zonotrichia georgiana 0.65 Northern Mockingbird Mimuspolyglottos 0.63 White-crowned Sparrow Zonotrichia leucophrys 0.52 454 Future Nearctic landbird vagrants of the regression line was decreased, resulting in a poorer fit for those species with multiple occurrences. The 38 most likely candidates for successful transatlantic crossing, based on the four variables selected by the stepwise procedure, are shown in table 2, arranged in decreasing order of likelihood. These species are divided into four groups for discussion purposes. The breaking point for inclusion in table 2 is arbitrary and does not correspond to any particular probability of occurrence, but is a point at which the predicted values in table 1 fall off sharply. Of the species that have occurred in Britain, only Rufous-sided Towhee and Hooded Warbler had lower predicted values. The remaining 35 candidate species are listed in taxonomic sequence (Voous 1973, 1977) rather than by predicted values (probabilities are given only for those species on The 'British Birds' List oj Birds oj the Western Palearctic, 1978): Mourning Dove Zenaida macroura, Downy Woodpecker Dendrocopos pubescens, Hairy Wood­ pecker D. villosus, Eastern Phoebe Sayomis phoebe, Acadian Flycatcher Empidonax virescens (0.04), Tree Swallow Tachycineta bicolor, Long-billed Marsh Wren Cistotfwmspalustris, Carolina Wren Thryothorus ludovicianus, House Wren Troglodytes aedon, Blue-grey Gnatcatcher Polioptila caerulea, Carolina Chickadee Parus carolinensis, Boreal Chickadee P. hudsonicus, Tufted Titmouse P. bicolor, Red-breasted Nuthatch Sitta canadensis (—0.14), White-breasted Nuthatch S. carolinensis, White-eyed Vireo Vireo griseus, Solitary Vireo V. solilarius, Warbling Vireo V. gilvus, American Goldfinch Carduelis tristis, Purple Finch Carpodacus purpureus, House Finch C. mexicanus, Nashville Warbler Vermivora ruficapilla, Chestnut-sided Warbler Dendroica pensyl- vanica, Pine Warbler D. pinus, Blackburnian Warbler D. jusca, Worm-eating Warbler Helmitheros vermivorus, American Tree Sparrow Spizella arborea, Chipping Sparrow S. passerina, Field Sparrow S. pusilla, Grasshopper Sparrow Ammodramus savannarum, Seaside Sparrow A. maritimus, Lincoln's Sparrow Zonotrichia lincolnii, Northern Cardinal Cardinalis cardinalis, Blue Grosbeak Guiraca caerulea (0.19), and Passerina cyanea (0.12). The Acadian Flycatcher and Red-breasted Nuthatch are on the Western Palearctic list, but have not yet been recorded in the British Isles; the Blue Grosbeak and Indigo Bunting are Category D species not yet admitted to the British and Irish list since occurrences seem more likely to have been the result of escapes from captivity than due to transatlantic vagrancy.

Discussion The six Group 1 birds in table 2 are relatively large species compared with others in the table. The smallest species in this group, the Connecticut Warbler, tends to become excessively fat in autumn and would seem to be a prime candidate for transatlantic passage. It is extremely secretive, however, and is much more apt to be trapped or netted rather than sighted. The other five species are conspicuous and easily indentified. Three of the six species appear on the Western Palearctic list. The Belted Kingfisher is an accepted Category A bird on the British and Irish list, with occurrences in 1908, 1978/79 and 1979/80; the Yellow-shafted Flicker is a Category D species known only from ship-assisted passage; and the Catbird is accepted for the Western Palearctic but not for Britain and Ireland. Neither the predictions nor the species sequence given in table 1 carry the reported degree of accuracy. Populations are constantly changing, as demonstrated by the North American Breeding Bird Survey (Bystrak & Robbins 1977; unpublished data); for example, the Eastern Kingbird is currently Future Nearctic landbird vagrants 455 decreasing while the Common Grackle is increasing, and these changes in population status will affect their relative positions in Group 1. The ten Group 2 birds include several which present identification problems. Even experienced North American banders are unable to distinguish Willow Empidonax traillii from Alder Flycatchers in the hand. Most banded individuals are still reported under the former combined name of'Traill's Flycatcher', and some are identified only to genus because of problems in separating them from Least, Acadian or Yellow-bellied Flycatchers. The female Dickcissel resembles a female House Sparrow. While adult male and immature male Black-throated Blue Warblers are distinctive, females could easily be overlooked. The Orange-crowned Warbler is rather nondescript and, in a brief view, could be mistaken for a Phylloscopus warbler; similarly, the identification of a Palm Warbler could prove problematic, especially in the hand, where its distinctive tail- wagging habit is not observed. A Black-capped Chickadee, however, should be easily distinguished from a Willow Parus montanus or a Marsh Tit P. palustris by plumage and voice. Two of the ten species in Group 2 are on the Western Palearctic list: the Wood Thrush has not been accepted for the British and Irish list; the Palm Warbler is in Category D, on the basis of a tide-line corpse. Group 3, which contains 17 species, is not clearly separable from either Group 2 or Group 4 and perhaps should have been combined with them. Group 3 contains the Whip-poor-will, Great Crested Flycatcher, Cedar Waxwing, Yellow-breasted Chat, and 13 small to very small passerines. Had there not been British records of Parula and Tennessee Warblers, I should have questioned whether birds with a normal body weight of lOg or less were capable of a transatlantic flight. Among Group 3 species, a silent, autumn Whip-poor-will might prove difficult to detect. Cedar Waxwings normally travel in flocks, but a lone transatlantic survivor might be over­ looked as a drab Waxwing Bombycilla garrulus. As with Group 2, there are several major problems of identification. An immature or female Mourning Warbler would require careful observation to separate it from a Connecti­ cut Warbler, a process unlikely to be helped by its typically skulking behaviour; it might also even be confused with a Yellowthroat. Field separation of Eastern and Western Pewees Contopus sordidulus is almost impossible. An out-of-range Myiarchus flycatcher could not safely be assumed to be a Great Crested. A Golden-crowned Kinglet could easily be passed over as a Firecrest Regulus ignicapillus, although it lacks the latter's rather bright green upperparts and bronzed breast-sides. The only Group 3 species on the British and Irish list is the Cape May Warbler, recorded in spring 1977 {Brit. Birds 73: 2-5). Among the Group 4 species, an immature Bay-breasted Warbler would require careful study to separate it from an immature Blackpoll Warbler, since both plumage and call-notes are similar: no North American field guide depicts them adequately. Swamp and Lincoln's Sparrows can be confused. Discussion so far has been limited to the results of the statistical analysis summarised in tables 1 and 2. A few comments are in order relative to the 456 Future Nearciic landbird vagrants effects of other independent variables, and to species that, because of habits or habitat requirements, are captured in very small, numbers at coastal banding stations. Carrying the stepwise regression one step farther resulted in the selection of migration date as the fifth variable (F=2.10, probability of greater F=0.15, R2=0.35, other F values and probabilities not changed appreciably). Forcing the nocturnal versus diurnal migration variable into the analysis at the sixth step (F=0.48, probability of greater F=0.49, R2=0.36) also had little effect on the other variables. There is no statistical justification for including these two additional variables, though biologi­ cally either or both could have an important influence on transatlantic vagrants. This part of the analysis, therefore, explored the effects that these additional variables would have on the predictions. Except for minor changes in sequence, there was no change in Group 1. Inclusion of date raised Whip-poor-will, Savannah Sparrow and Ruby-crowned Kinglet from Group 3 to Group 2, and elevated Swamp Sparrow from Group 4 to Group 3. Addition of nocturnal versus diurnal migration caused no further change in the placement of these species. Inclusion of date dropped Great Crested Flycatcher from Group 3 to 4, and dropped Least Flycatcher, Blue-winged Warbler and Wilson's Warbler from consideration. Inclusion of nocturnal versus diurnal migration reinstated Least Flycatcher, but only in Group 4. Inclusion of migration date also brought into consideration five species that had been excluded from table 2 because of the low predicted number: Eastern Phoebe, Lincoln's Sparrow and Nashville Warbler were assigned to Group 3, and Indigo Bunting and Solitary Vireo to Group 4. The only additional effect of including nocturnal versus diurnal migration was to raise Blue Grosbeak to Group 4. It is hoped that time and further research will cast some light on the importance of these last two variables. Species present but undersampled at banding stations were not analysed, but some of the more common ones should be mentioned. Common Nighthawk is already represented by seven British records up to 1976 (six during 1947-76). Another common aerial feeder, the Chimney Swift Chaeturapelagica, is a long-distance migrant and should be considered. Unexpectedly, the Brown-headed Cowbird Molothrus ater, Rusty Blackbird Euphagus carolinensis and Eastern Meadowlark Stumella magna were captured so infrequently at all stations during the autumn that they were excluded from the analysis. All are short-distance diurnal migrants and probably not prime candidates for transatlantic passage. Only two of the more highly migratory members of this family, the Bobolink and the Northern Oriole, have reached Britain so far. Note, however, that Larkin etal. (1979) found two species of icterids, the Brown-headed Cowbird and the Red-winged Blackbird Agelaius phoeniceus, more than 110 km offshore from New England in mid October 1971, and Durand (1963) recorded a Rusty Blackbird aboard RMS Mauretania one day out from New York City. I hoped to find correlations between occurrence of American vagrants in Europe and populations of these same species in America. Breeding Bird Survey data were examined for those species with multiple occurrences in Future Nearciic landbird vagrants 457 Britain in a single year to see if their transatlantic crossings were associated with an especially high breeding population in the year of occurrence or a decrease the following year. No consistent correlation was found with the Breeding Bird Survey data for the region east of the Mississippi River. If future meteorological analyses can define the probable area of origin of a large-scale incident of American vagrants in Europe, it may be informative to examine Breeding Bird Survey data in greater detail for those particular species.

Acknowledgments This paper would not have been possible without the East Coast banding data to which hundreds of volunteers contributed. I am especially indebted to T. Lloyd-Evans, K. S. Anderson, S. Fitch and R. Repka for compiling the autumn data from Manomet Bird Observatory; to B. Adams for a computer summary from Island Beach State Park; and to F. R. Scott for yearly summaries from the Kiptopeke Beach Operation Recovery Station. J. O'Sullivan, Dr J. T. R. Sharrock and K. Preston provided up-to-date information on North American vagrants in Britain and Ireland. Dr P. H. Geissler and Dr B. K. Williams offered many helpful suggestions regarding analysis procedures, and L. M. Moyer provided the expertise for accessing and operating the computer programs. I thank the following persons for critically reading the manuscript: M. K. Klimkiewicz, Dr B. Noon, and Dr B. K. Williams.

Summary Migratory habits of 31 species of vagrants from North America and 73 candidate species not recorded in either Britain or Ireland during the 30 years 1947-76 were analysed by multiple regression. There was a significant correlation between the number of British and Irish records and the west-to-east component of normal migration, standard deviation of body weight, migratory distance, and abundance on the west coast of the North Atlantic. The candidate species were ranked in order of likelihood of successful transatlantic flight based on the four input variables(table 2).

References BARR, A. J., GOODNIGHT, J. H., SALL, J. P., & HELWIG, J. T. 1976. A User's Guide to SAS 76. SAS Institute, Raleigh, N. C. BYSTRAK, D., & ROBBINS, C. S. 1977. Bird population trends detected by the North American Breeding Bird Survey. Pol. Ecot. Stud. 3: 131-143. DURAND, A. L. 1963. A remarkable fall of American land-birds on the 'Mauretania', New York to Southampton, October 1962. Brit. Birds 56: 157-164. ELKINS, N. 1979. Nearctic landbirds in Britain and Ireland: a meteorological analysis. Brit. Birds 72: 417-433. LARKIN, R. P., GRIFFIN, D. R., TORRE-BUENO, J. R., & TEAL, J. 1979. Radar observations of over the western North Atlantic Ocean. Behavioral Ecology and Sociobiology 4: 225-264. RUSSELL, W. 1978. Grey-cheeked Thrushes in Europe. Brit. Birds 71: 421-422. SHARROCK, J. T. R., & SHARROCK, E. M. 1976. Rare Birds inBritain and Ireland. Berkhamsted. Voous,K.H. 1973, 1977. List of recent Holarctic bird species. Ibis 115: 612-638; 119:223-250, 376-406.

Chandler S. Robbins, 7902 Brooklyn Bridge Road, Laurel, Maryland 20810, USA