M. Marquiss (correspondence), Centre for Ecology & Hydrology, Banchory Research Station, Hill of Brathens, Banchory, Aberdeenshire AB31 4BW, Scotland. E-mail: email@example.com. Present address of M.M.: Saughtrees, South Bandodle, Aberdeenshire, AB51 7NN, Scotland. E-mail: firstname.lastname@example.org. – K. A. Hobson, Environment Canad, 11 Innovation Blvd., Saskatoon, Saskatchewan S7N 3H5, Canada. – I. Newton, Centre for Ecology & Hydrology, Monks Wood Research Station, Abbots Ripton, Huntingdon, Cambs PE28 2LS, England.
We used the stable isotope composition of body feathers to show the extraordinarily varying regional provenance of an irruptive conifer seed specialist, the common crossbill Loxia curvirostra. In the boreal region of the western Palearctic, this species specialises on the seeds of Norway spruce Picea abies. The patterns of deuterium in the feathers of migrant common crossbills collected in Britain suggested that irruptions in different years originated in widely separated regions of the Palearctic boreal forest. The birds from some irruptions were relatively narrow-billed compared with those from others, but bill width was not correlated with deuterium values. However, deuterium values did co-vary with the dates that the irruptions reached Britain, with the lowest deuterium levels in years with the latest arrivals. This finding was consistent with the idea that birds with low values came further, from regions far to the northeast. This pattern was very different from that found in irruptive northern bullfinches Pyrrhula p.pyrrhula which had larger variation in deuterium values within an irruption, suggestive of a much more widespread provenance. It is argued that the difference between species is the result of their differing foraging ecology. Bullfinches have a varied summer diet and, unlike common crossbills, do not concentrate regionally to breed where a single type of food is available that year.
Regular migration in birds involves annual population movements from breeding to wintering areas and return (Alerstam 1991, Newton 2007a). The marking of samples of individuals is often sufficient to document the origins, routes and destinations of migrants, together with the phenology of movement (Wernham et al. 2002). In contrast, irregular migration, involving irruptive movements or invasions (Lack 1954, Newton 2007b), is less easy to detail because of its sporadic occurrence and the variable provenance of migrants. The available data are even scarcer where the species concerned inhabit geographically remote regions, with sparse human populations, and move over thousands of kilometres. However, the stable isotopic composition of feathers can now be used to indicate the geographical region where birds lived during feather growth, or at least to separate birds from different regions, and has thus become a useful tool in migration studies (Hobson et al. 2004). Here we use the patterns of variation in deuterium ratios to show the extraordinarily varying regional provenance of an irruptive conifer seed specialist.
Common crossbills Loxia curvirostra are well-known irruptive migrants, every few years moving in large numbers outside of their regular range in the boreal forest (Newton 1972, 2006, 2007a). They are specialist finches that in the western Palaearctic feed principally on Norway spruce Picea abies, a ‘masting’ conifer that produces seed sporadically. In any one region, seed is produced in abundance only in occasional years, separated by several years with little or no seed production (Hagner 1965). The temporal patterns of seed production vary between widely separated regions. Crossbill populations track spruce seed production, settling each year in large numbers where seed is abundant, then moving as the seed is consumed or shed to another region where seed is available (Newton 1972). Norway spruce seed production is annually synchronised over large geographical areas (Koenig and Knops 1998), so common crossbills have to move hundreds of km each summer from regions where one crop is coming to an end to other regions where a new crop is forming. The birds breed over winter when spruce seed is most readily available. Thus, crossbills in Norway spruce areas make only one major movement each year, unlike regular migrant birds that move in autumn and return to breed in the following spring (Newton 2006). As Norway spruce produces little seed for one or more years after a masting event, few (if any) common crossbills use the same area in successive breeding seasons, and their population movements are thus nomadic in nature.
Whenever spruce seed is scarce or absent over much of their normal breeding range, common crossbills move far greater distances (often >1,000 km), and then reach south and west Europe in substantial numbers (Newton 1972, Marquiss 2002). Such ‘irruptions’ are irregular but not rare – at least 40 occurred in the 120-year period 1881–2000 (Newton 2006). The locations of recoveries of migrant crossbills ringed in central Europe suggest that, following irruptive movements southwestwards, the birds return northeastwards in a later year to northern Russia west of the Urals. It is possible that such ‘return’ movements take the birds back to their region of origin (discussed in Knox 1992), but few crossbills have been ringed in their boreal forest breeding areas so the precise source of irrupting migrants remains obscure.
The stable hydrogen isotopic composition (deuterium, δD) of bird feathers can indicate the geographical region where the birds lived when they grew the feathers (Hobson 2003, Hobson et al. 2004, Bowen et al. 2005, Fox et al. 2007). In practice, the northeast-southwest gradient of δD in rainfall across the Palearctic is shallow, compromising the precision of spatial estimates. Nevertheless, the pattern of variation within samples of birds can be informative: for example, by indicating differential provenance (Hobson 2005a,b).
There are alternative ideas about the provenance of common crossbill irruptions to western Europe. On the one hand, if irruptions are driven primarily by food shortage across vast expanses of boreal forest, the birds arriving in western Europe will be from widespread parts of the northern breeding range, numbers accruing as migrants proceed southwestwards. In this case, δD values for feather samples from migrants will be highly variable between birds within irruptions, but have similar average values for different irruption years. This pattern was shown in feather samples from irruptions of northern bullfinches Pyrrhula p. pyrrhula (Newton et al. 2006).
The alternative idea is that crossbill irruptions have more regionalised origins driven by a high regional abundance of birds, followed by food shortage as the conifer seed is consumed or abruptly shed, but with different regions involved in different years. This view gains circumstantial support from the well-established variation in the morphology of common crossbills from different irruptions (Davis 1964, Herremans 1988, Marquiss and Rae 1994). In this case, δD values for feather material will vary less between birds within irruptions, and more in mean values between irruptions. The present study uses samples of body feathers from common crossbills collected in different irruptions to detect patterns that will distinguish between these two ideas about provenance.
Feather material came from 66 common crossbills; 49 that had been collected and preserved as skins in the National Museums of Scotland, and 17 that had been caught (and released) in eastern Scotland. Fifty-five were migrants from five different irruptions into Britain (1909, 1927, 1990, 1997 and 2002), collected or live-caught between 4 June and 2 October, mainly at sites around the western coast of the North Sea away from breeding habitat (Table 1). These birds were likely to be recently arrived, as most came from offshore islands or other coastal localities lacking conifers, where migrants were unlikely to linger. The only specimens away from the North Sea were four collected in August 1927 in Gloucestershire. The 11 other birds were not recent migrants. They were live-caught between 2 and 6 April 1995, in coniferous woodland in Aberdeenshire, Scotland. Common crossbills in early April have moulted but not yet migrated so these latter birds had grown their body feathers in Scotland.
Table 1. Numbers and collection dates of common crossbill samples from each year.
Collection dates median (range)
Forty-nine feather samples were from museum skins. Live-caught (and released) birds were the eleven from 1995 and six from June 1997.
(21 June–2 Oct.)
(8 July–15 Sept.)
(4 June–15 Sept.)
(30 June–4 July)
(17 June–17 Sept.)
Each individual was sexed using plumage colour; wing length, bill length, depth and width were measured (Svensson 1992, Summers et al. 1996), and a small amount of vane material was clipped from flank feathers (Newton et al. 2006). Measurements of both live birds and museum skins can vary between observers (Summers et al. 1996), so one author (MM) took all wing and bill measurements. Estimates of repeatability (Lessells and Boag 1987) were derived from 32 of the museum specimens that were re-measured by this observer. For all four measurements there were highly significant ‘F’ ratios (>56.8; df=1, 31; P<0.001). Repeatability was very high for wing length (r=0.94) and bill depth (0.92), slightly less so for bill length (0.85) and bill width (0.80). Museum skin specimens were dried and shrunken so, to compare biometrics across samples, we adjusted the measurements of the 17 live birds by average shrinkage values derived from 15 common crossbill carcases measured fresh, then again as dried museum skins (M. Marquiss unpubl. data). The wing length measurements of live birds were thus reduced by 0.60 mm, bill depths by 0.37 mm, bill lengths by 1.01 mm and bill widths by 0.33 mm.
Prior to isotopic analysis, feather material from each individual was cleaned with a 2:1 chloroform: methanol solvent mixture to remove surface contaminants and oils. Cleaned feather vanes were then subsampled and analysed for δD using the methods described by Hobson et al. (2004). Stable isotope ratios were expressed in δ-notation as parts per thousand deviation from the VSMOW/SLAP international standards. Reported values correspond to non-exchangeable hydrogen using the comparative equilibration technique as described by Wassenaar and Hobson (2003).
The resulting data were grouped into six series (five from different irruption years and one ‘Scottish grown’ sample, Table 1) and analysed by General Linear Models (GLM), with δD or body dimensions (ln-transformed) as the response variable and series and sex as the explanatory variables. Sex was included as an explanatory factor because common crossbill males average larger than females (e.g. Marquiss 1980), and in a study of migrant bullfinches (Newton et al. 2006) males had slightly higher δD values than females.
Variation in δD values within and between series
A GLM with δD values as the response variable, and series and sex as explanatory factors, showed no evidence for a difference in δD values between sexes (F1.59=0.13, P=0.7). However, significant variation was found between series (F5.59=240.7, P<0.001), much of it due to the substantial difference between Scottish-moulted birds and immigrants (Tukey simultaneous tests; all T values >19.7, P<0.0001) (Table 2, Fig. 1). An analysis of immigrants alone revealed significant differences in δD values between irruptions (F4.49=34.1, P<0.001), explaining 73% of the overall variation. Amongst mean values (Table 2), those for the 2002 and 1927 irruptions were lowest, and significantly lower than those for 1909 and 1997 (all T values >3.6, P<0.005). The highest mean value for immigrants was from the 1990 irruption, significantly higher than the mean for 1909 immigrants (T>3.0, p<0.03).
Table 2. δD value (‰) of the flank feathers of common crossbills from five irruptions compared with that of birds that grew their feathers in Scotland.
SE of the mean
−71 to −44
−139 to −117
−151 to −128
−127 to −108
−140 to −113
−159 to −133
There was no suggestion of greater variation within some series compared with others (Bartlett's test for equal variances=2.73, P=0.74). In particular, the standard deviation of values for Scottish moulted birds was no less than the standard deviations of values within each of five irruptions (Table 2).
Variation in the morphology of crossbills from different series
As anticipated, males had longer wings and larger bills than females, but after accounting for variation associated with sex, there was no suggestion of differences in wing and bill measurements between series, except for bill width (F5.58=4.68, P=0.001). Birds from the irruptions in 1997 and 2002 had wider bills than birds from 1909 and 1990 (Tukey simultaneous tests; all T values >3.04, P<0.04; Table 3). This result was not an artefact of the shrinkage adjustments applied to live-caught birds in 1995 and 1997 because the significant variation remained when these specimens were excluded (F4.43=4.54, P=0.004).
Table 3. Bill width (mm) of common crossbills from five irruptions and of birds in Scotland in April 1995.
SE of the mean
SE of the mean
Birds from 1997 and 2002 irruptions had wider bills than those from 1909 and 1990.
The timing of immigration in relation to mean δD values and body shape
The distinct differences in δD values between irruption years suggested a fairly regionalised provenance of migrant crossbills, with two irruptions (1927 and 2002) originating much further northeast than the others according to the δD gradient. Under these circumstances, it might be expected that birds in irruptions characterised by very low δD values would have travelled further to reach western Europe, and might thus have arrived later in the summer. Moreover, the variation in the bill shape of crossbills might reflect their original provenance. We therefore examined data from the five irruptions comparing mean δD values (Table 2) with the median collection (or live-capture) dates and mean bill width. Median collection dates of immigrants varied by more than six weeks between years (Table 1), while mean bill widths ranged between 10.50 and 11.35 mm (Table 3).
There was no suggestion of relationships between mean ln-transformed bill width and mean δD (Pearson's correlation coefficient, r=0.489, n=5, P=0.740), or median date (r=0.351, n=5, P=0.562). However, a regression of median date (response) on mean δD values (predictor) for the five irruption years revealed a very good association, with 92% of the variation in date explained by δD (F1..3=44.3, P=0.007). As predicted, birds with lower δD values arrived in Britain later in the summer (Fig. 2). This did not apply to birds within an irruption because there was no significant regression for collection date on δD amongst migrants after taking irruption year into account (F1.49=0.96, P=0.33). These findings endorsed the idea of a restricted regional provenance for each irruption, with later arrivals in Britain indicating more distant origins.
The pattern of δD values in the feathers of common crossbills refutes the idea of their irruptions being driven by migrants accumulating in numbers as they move west and southwards across boreal forest regions with little or no food (conifer seed). Rather, the low variance in values within individual irruptions, together with significant differences in the mean δD between irruptions supports the idea of more regionalised provenance, varying from year to year. Thus common crossbill irruptions in different years originate in different parts of the boreal forest, and are probably predisposed by very high local abundance, coupled with widespread food shortage, as spruce seed is consumed and shed. The idea of regionalised and varying provenance is further endorsed by the timing of birds arriving in Britain – several weeks later for birds with very low δD values, suggesting that they had travelled longer distances.
Without suitable ‘reference’ specimens from across the boreal zone (from regions known to be where they grew their feathers), we cannot be sure of the precise provenance of irruption immigrants. Of the five crossbill irruptions represented in the present study, the 1990 immigrants had the highest δD and the earliest arrival date. From the presence of migrants around the southern Baltic in May 1990 (Marquiss 2002) and the recovery of an April ringed bird from Karelya (Newton 2006), this irruption was thought to originate in northwestern Russia. The other four irruptions, with lower δD values, would therefore have originated further east, towards the Ural mountain chain and perhaps beyond, in the far northeast.
Little information is available on year-to-year variance in average deuterium deposition in precipitation over the boreal region of Eurasia during the last century. However, the isotopic basemap used to create the depiction in Hobson et al. (2004) was based on the 40 year International Atomic Energy Association (IAEA) dataset, and the year-to-year variance in δD in precipitation for Moscow, Russia (55°′49N, 37°57′E, elevation 158 m) based on those data is only 40/00 (95% CI; http://waterisotopes.org). Hence, we believe that the substantial isotopic differences among irruptions reflect primarily differences in the geographical origins of birds rather than annual differences in δD deposited in precipitation (see also Hobson 2005b).
Only five irruptions were represented in our samples, with relatively few birds from each (and unbalanced sex ratios), so there was little statistical power to resolve morphological variation. Nevertheless, the lack of correlation between crossbill morphometrics and δD is inconsistent with the idea of a simple gradient in crossbill size or shape eastwards across the boreal zone. We therefore entertain the alternate view that discrete populations of birds differing in size or shape might be endemic to specific regions of the boreal forest, perhaps associated with localised cone morphology (Benkman 1993) or particular conifer mosaics (Marquiss and Rae 2002). Much larger samples, including more birds from more irruptions, would be required to attempt to map crossbill size and shape against δD.
The pattern in δD values documented here for common crossbills contrasts with that for irruptive migrant bullfinches (Newton et al. 2006). This difference seems logical considering the foraging ecology of the two species. Common crossbills feed primarily on Norway spruce seed and track seed production with nomadic annual movements. Norway spruce seed is prolific in any one region only at intervals of several years; so breeding (and moulting) crossbills are locally concentrated in regions of spruce ‘mast’. By comparison, bullfinches have a varied summer diet (buds, ovules, seeds and invertebrates, Newton 1967, Marquiss 2007), that is inevitably available every summer over much of their geographical breeding range. Bullfinches are thus unlikely to be concentrated in particular regions during the time they grow their body feathers. Their irruptive migrations occur as full grown and moulted birds move large distances when their main winter seed foods, such as rowan Sorbus aucuparia or birch Betula spp., are in short supply across widespread parts of the boreal forest.
Because of their narrow diet and the regional concentration of spruce seed production, irrupting common crossbills must usually originate over smaller spatial scales than those of bullfinches that have a varied summer diet. By analogy, we might expect irruptions of other species with varied summer diet (e.g. Bohemian waxwing Bombycilla garrulous) to be drawn from birds distributed over a large spatial scale.
We are grateful to Bob McGowan for help and access to crossbill skins in the National Museum of Scotland (Edinburgh), which formed an important part of this study. Len Wassenaar assisted with the isotopic measurement of samples, which were run at the National Water Research Institute in Saskatoon, Saskatchewan, Canada. Funding was provided by an operating grant to KAH from the Canadian Wildlife Service. We thank S. Bearhop, R. W. Summers and P. Edelaar for helpful comments on the manuscript.