1474919x, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1474-919x.2005.00400.x by National University Of Singapo, Wiley Online Library on [18/10/2022].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Ibis (2005), 147, 283– 292

The scaling of primary flight feather length and mass

µBlackwell Publishing, Ltd.

in relation to wing shape, function and habitat

ALISTAIR DAWSON*
Centre for Ecology and Hydrology, Monks Wood, Abbots Ripton, Huntingdon, Cambridgeshire PE28 2LS, UK

The mass and length of each primary flight feather was measured in 120 species of birds
(347 individuals) representing 37 families and 15 orders. The scaling relationship between
mass and length was determined using the mass of each primary as a proportion of total
primary feather mass for that individual and, similarly, length as a proportion of total length.
This eliminated errors due to intra- and interspecific differences in absolute size. In every
species there was a highly significant constant scaling relationship (log mass/log length) for
all of the primary feathers proximal to the feather that formed the wing tip. This relationship
was allometric and varied between 1.80 in Rooks Corvus frugilegus and 4.87 in Winter Wrens
Troglodytes troglodytes. The mean scaling relationship for 120 species was 2.41 ± 0.42 sd,
which was significantly less (P < 0.0001) than isometry (i.e. 3.00). In most species (117 of
120) the primary feather forming the wing tip and all feathers distal to it had a different
scaling relationship, and had a greater mass than expected from their length. The greater
relative mass of the outer primaries may reflect a protective role against physical abrasion,
or an aerodynamic role in that each of these feathers provides a leading edge to the wing.
Thus, there were two scaling relationships that pivoted about the feather forming the wing
tip, resulting in a characteristic ‘signature’ for each species. Scaling relationships can be
related to flight characteristics and habitat, rather than to phylogeny. Closely related species
often had widely varying scaling relationships. In general, species exploiting dense vegetation
had greater scaling relationships than more aerial species. However, species with a high scal-
ing relationship did not have a greater mean feather mass, so the increased relative mass of
the distal primaries was at the expense of proximal primary feather mass.

lengths. Moreover, in pointed high-aspect-ratio wings,

INTRODUCTION
the wing tip is formed by the most distal primary. In
Wing shape varies between bird species to suit the such cases, that feather alone provides a leading edge
flight characteristics and the nature of habitats to the outer section of wing. In more rounded wings,
exploited by each species (see Rayner 1988, Penny- the wing tip may be formed by an intermediate
cuick 1989, for full descriptions of flight dynamics primary, in which case both it and the more distal
and wing shape). Fast aerial species tend to have primaries provide independent leading edges. This is
longer, pointed wings with a high aspect ratio (wing- particularly true of soaring species, in which the
span/mean chord length; Pennycuick 1989) whereas outer primaries separate to form slots.
low-speed soaring species or those inhabiting dense Relative primary feather mass varies even more
vegetation tend to have rounded wings with a low than length (Underhill & Joubert 1995). During
aspect ratio (Hamilton 1961, Rayner 1988, Lockwood a study of moult in Common Starlings Sturnus
et al. 1998). Wing shape is largely derived from the vulgaris, one aim was to determine the rate at which
relative lengths of individual primary feathers. Pointed the total mass of new primary feathers increased as
wings have primary feathers that increase markedly moult progressed (Dawson 2003). In order routinely
in length from the most proximal (P1) to the most to determine the mass of a primary feather by
distal. Round wings have primaries of more even measuring its length, it was necessary to determine
the relationship between length and mass. To this
*Email: [email protected]. end, the primary feathers were plucked from some

© 2005 British Ornithologists’ Union

 1474919x, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1474-919x.2005.00400.x by National University Of Singapo, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
284 A. Dawson

individuals and their length and mass were measured. Figure 1 for data for the Common Starling. The
For all of the primaries, except the most distal, range of the y-axis is 0.5–1.3 and of the x-axis 0.8–
length and mass appeared to have a constant allo- 1.15. These parameters were chosen to accommo-
metric scaling relationship (Alexander et al. 1979, date the species with the greatest ranges. In Figure 2,
Alexander 1985, Rayner 1985). However, the most the axes of the individual graphs are unlabelled
distal primary had a greater mass than expected from but are identical to those in Figure 1. The solid line
its length. This may reflect a distinct role of the on each graph links each of the primary feathers in
outermost primary in protecting the more proximal sequence, P1 being the most proximal (innermost)
primaries from damage, or an aerodynamic role in and P9, P10 or P11 the most distal (outermost),
that it, uniquely in the case of Starlings, forms a lead- depending on species. In some species the most dis-
ing edge to the wing. The aim of this study was to tal primary is reduced or minute (Cramp 1998), in
discover whether this pattern is true for birds in which case it was omitted from the analysis. The data
general. It became apparent that it was not only the showed that it was not just the most distal primary
most distal primary that was more massive than pre- that had a greater mass than expected from its
dicted from its length but all of the primaries distal length, but that the primary that forms the wing tip
to that forming the wing tip. An additional benefit of (as defined for each species in Cramp 1998) and any
these data on relative primary feather mass is that they primaries distal to that did also. Therefore, the open
are useful when estimating moult scores (Underhill symbols represent the primary feathers proximal
& Joubert 1995, Dawson & Newton 2004). to that forming the wing tip (Cramp 1998) and the
solid symbols represent the primary feather that
forms the wing tip and any primaries distal to it. In
METHODS
the case of the Common Starling, the wing tip is
Dead birds were collected from a variety of sources, formed equally by P8 and P9. The wing tip is often
e.g. road casualties and specimens collected for post- formed not by the longest primary but by the
mortem analyses. In total, 120 species were exam- primary distal to the longest because the distance
ined (347 individuals) representing 37 families and between the insertion points on the wing exceeds
15 orders (Cramp 1998). Only birds that were not the difference in feather length. The broken line on
moulting and whose feathers were not significantly each graph is the linear regression for all of the
damaged or worn were used. The primary feathers primaries proximal to the wing tip, i.e. P1–P7 in the
were plucked and allowed to equilibrate to room Common Starling case.
temperature and humidity. The length and mass of
each feather was recorded. Length was measured
RESULTS
with the rachis held along a straight edge. Measure-
ments were made to three significant figures, i.e. mass In the Common Starling (Fig. 1), the primary feathers
to the nearest 0.1 mg for feathers less than 100 mg, P1–P7 showed a highly significant linear correlation
and length to the nearest 0.1 mm (using Vernier between log (percentage total mass) and log (percent-
callipers) for feathers less than 100 mm. In the case age total length), with r2 = 0.998. The slope of the
of small birds (< 100 g body mass) the means of linear regression (scaling relationship) was 2.218 ±
the primaries from both wings were used where 0.045 sd (Table 1). Thus,
possible.
For each primary feather, length was converted to mass ∝ length2.218.
the percentage that it contributed to the total length
of all primaries from that wing, and, similarly, mass The relationship was therefore allometric (Rayner
was converted to the percentage of total primary 1985) as opposed to isometric; in the latter the
mass. This gives the same relationship as absolute scaling relationship would equal 3.0. The two most
length and mass, but eliminates differences due to distal primary feathers (P8 and P9), which in Starlings
intra- and interspecies variation in absolute size. This equally form the wing tip, did not follow this rela-
also means that for each species the data can be tionship. They lay above the linear regression line,
plotted on graphs with identical axes, making com- i.e. log (mass)/log (length) > 2.218. Mass exceeds
parisons between species easier. The results were that predicted by their length.
plotted as log (percentage total mass) on the y-axis In each of the other 119 species examined, there
and log (percentage total length) on the x-axis. See was also a constant allometric relationship for the

© 2005 British Ornithologists’ Union, Ibis, 147, 283–292

 1474919x, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1474-919x.2005.00400.x by National University Of Singapo, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 1. Scaling relationship (slope), 95% confidence limits and r 2 for primary feathers proximal to the wing tip, for 120 species arranged according to scaling relationship.

Species Slope ± r2 Species Slope ± r2 Species Slope ± r2

Rook 1.795 0.018 1.000 Hen Harrier 2.210 0.043 0.999 Red-legged Partridge 2.446 0.074 0.996
Eurasian Hobby 1.891 0.013 1.000 Black-tailed Godwit 2.214 0.032 0.999 Mute Swan 2.455 0.030 0.999
Atlantic Puffin 1.914 0.019 0.999 Common Starling 2.218 0.045 0.998 Stone-curlew 2.456 0.057 0.997
Peregrine Falcon 1.918 0.017 1.000 Brent Goose 2.220 0.073 0.994 House Sparrow 2.469 0.067 0.996
Red-throated Diver 1.931 0.024 0.999 Eurasian Curlew 2.221 0.020 0.999 Blackcap 2.474 0.110 0.990
Manx Shearwater 1.934 0.016 1.000 Great Cormorant 2.235 0.079 0.994 Chaffinch 2.511 0.057 0.998
Tree Pipit 1.950 0.087 0.990 Common Redshank 2.249 0.037 0.998 Common Blackbird 2.518 0.096 0.994
Jackdaw 1.961 0.017 1.000 Black-headed Gull 2.256 0.024 0.999 Rock Pigeon 2.521 0.048 0.998
House Martin 1.966 0.032 0.999 Northern Goshawk 2.267 0.060 0.997 Pied Avocet 2.526 0.048 0.998
Barn Swallow 1.972 0.047 0.997 Eurasian Siskin 2.268 0.026 0.999 Great Crested Grebe 2.533 0.064 0.998
Tufted Duck 1.978 0.038 0.998 Common Linnet 2.268 0.083 0.993 Northern Lapwing 2.533 0.029 0.999
Razorbill 1.994 0.058 0.994 European Shag 2.278 0.049 0.998 Eurasian Sparrowhawk 2.547 0.155 0.989
Common Pochard 2.001 0.028 0.999 Arctic Skua 2.279 0.048 0.997 Grey Heron 2.572 0.053 0.997
Red-breasted Merganser 2.010 0.022 0.999 Common Tern 2.279 0.032 0.999 Yellowhammer 2.575 0.028 1.000
Lesser Whitethroat 2.012 0.095 0.989 Lesser Black-backed Gull 2.282 0.037 0.998 Tawny Owl 2.582 0.090 0.996
Common Guillemot 2.070 0.058 0.995 Herring Gull 2.283 0.042 0.998 Willow Ptarmigan 2.617 0.055 0.998
Common Coot 2.074 0.009 1.000 Northern Pintail 2.289 0.034 0.999 Grey Partridge 2.659 0.064 0.998
Carrion Crow 2.084 0.035 0.999 Black-legged Kittiwake 2.289 0.025 0.999 Great Tit 2.686 0.146 0.991
Merlin 2.085 0.042 0.998 Mew Gull 2.293 0.026 0.999 Corn Crake 2.691 0.161 0.979
Common Kestrel 2.087 0.019 1.000 Mallard 2.306 0.032 0.999 Common Kingfisher 2.695 0.192 0.975
European Goldfinch 2.087 0.033 0.999 Common Snipe 2.317 0.050 0.997 Black-billed Magpie 2.711 0.071 0.998
Sky Lark 2.089 0.159 0.977 European Greenfinch 2.320 0.051 0.998 Green Woodpecker 2.755 0.084 0.997
Northern Gannet 2.090 0.027 0.999 Common Goldeneye 2.324 0.028 0.999 Stonechat 2.802 0.129 0.992
Northern Shoveler 2.100 0.025 0.999 Greylag Goose 2.331 0.041 0.998 Great Spotted Woodpecker 2.817 0.084 0.997
Barn Owl 2.100 0.048 0.997 Whooper Swan 2.336 0.055 0.997 Meadow Pipit 2.821 0.215 0.978
© 2005 British Ornithologists’ Union, Ibis, 147, 283– 292

Black-browed Albatross 2.132 0.027 0.999 Bar-tailed Godwit 2.341 0.026 0.991 Little Owl 2.853 0.105 0.996

Primary flight feather length and mass

Red Knot 2.134 0.028 0.999 Dunlin 2.337 0.038 0.998 Willow Warbler 2.938 0.039 0.999
Pied Wagtail 2.136 0.061 0.996 Eurasian Woodcock 2.349 0.030 0.999 Western Capercaillie 2.967 0.037 0.999
Northern Fulmar 2.153 0.031 0.998 Common Bullfinch 2.361 0.059 0.998 European Robin 2.979 0.098 0.997
Short-eared Owl 2.154 0.069 0.993 Pink-footed Goose 2.373 0.045 0.998 Blue Tit 3.047 0.176 0.991
Whimbrel 2.159 0.053 0.996 Common Buzzard 2.382 0.040 0.999 Coal Tit 3.048 0.069 0.998
Marsh Harrier 2.170 0.020 0.997 Eurasian Collared Dove 2.399 0.059 0.996 Goldcrest 3.134 0.072 0.997
Eurasian Teal 2.187 0.050 0.997 Turtle Dove 2.399 0.059 0.996 Common Pheasant 3.150 0.029 1.000
Common Shelduck 2.191 0.033 0.999 Fieldfare 2.405 0.058 0.997 Common Chiffchaff 3.160 0.292 0.975
Eurasian Wigeon 2.198 0.037 0.998 Black Grouse 2.411 0.050 0.998 Hedge Accentor 3.249 0.122 0.994
Goosander 2.198 0.029 0.999 Common Moorhen 2.411 0.048 0.998 Marsh Tit 3.277 0.166 0.992
Long-eared Owl 2.200 0.029 0.999 Reed Bunting 2.414 0.058 0.998 Little Grebe 3.305 0.209 0.973
Common Wood Pigeon 2.203 0.074 0.994 Song Thrush 2.443 0.054 0.998 Water Rail 3.460 0.090 0.997
Common Eider 2.209 0.050 0.997 Great Bittern 2.445 0.116 0.989 Common Whitethroat 3.579 0.215 0.982
Northern Wheatear 2.209 0.071 0.995 Redwing 2.445 0.073 0.996 Winter Wren 4.874 0.228 0.994

285
 1474919x, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1474-919x.2005.00400.x by National University Of Singapo, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
286 A. Dawson

for the distal primaries was nearly always greater

than for the proximal primaries, there was no evi-
dence in any species of a gradual increase in the
mass–length relationship along the wing. Rather,
there was always a constant relationship up to the
primary proximal to the wing tip, and a different
relationship beyond. In other words, there are two
distinct scaling relationships, and these pivot about
the wing tip.
The mean scaling relationship for primary feathers
proximal to that forming the wing tip (2.41) was
significantly less than isometry, i.e. 3.00 (t = 15.24,
Figure 1. The scaling relationship of the primary flight feathers df = 119, P < 0.0001). However, this figure refers
in the Common Starling. Each point represents the log of the
percentage contribution to the total length and mass of each
only to primary feathers proximal to the wing tip
of the nine primary feathers (P1–P9), n = 5. The standard and does not take into account the more massive
deviations for length and mass of each point are smaller than the outer primaries. When the regressions were recalcu-
size of the symbol. The solid line links each of the nine feathers lated for all 120 species to take into account all of
in sequence. The open symbols represent the primary feathers the primary feathers, the mean scaling relationship
proximal to that forming the wing tip (P1–P7) and the solid
symbols represent P8 and P9. Note that P8 and P9 are
increased from 2.41 to 2.50 ± 0.49 sd. This is still
approximately the same length, but P9 has greater mass. The significantly less than isometry (t = 13.76, df = 119,
broken line represents the linear regression of the feathers P < 0.0001). Is this a characteristic of feathers
proximal to the wing tip, i.e. P1–P7. The slope = 2.218, r 2 = generally? In other words, do large feathers (feathers
0.998, P < 0.0001. from large birds) have proportionately less mass than
small feathers (feathers from small birds)? To test
this, the mean feather mass for each of the 120 spe-
primaries proximal to, but not including, the primary cies was plotted against mean feather length (Fig. 3).
that forms the wing tip (Fig. 2). For all 120 species, The scaling relationship for this was 3.00 ± 0.04 sd.
mean r2 = 0.996. The mean value for the scaling So for feathers generally, the relationship between
relationship was 2.41 ± 0.42 sd, but this varied mass and length is isometric. The relationship between
from 1.80 for Rooks Corvus frugilegus to 4.87 for mean feather mass and mean feather length is the
Winter Wrens Troglodytes troglodytes. In only 11 spe- same in Goldcrests Regulus regulus and Winter
cies did the scaling relationship exceed 3.00. The Wrens as it is in Black-browed Albatrosses Diomedea
scaling relationship varied with the habits and habi- melanophris and Mute Swans Cygnus olor. This
tats of the species. Lowest values were found in more suggests that the construction of feathers is highly
aerial species (Rook, Eurasian Hobby Falco subbuteo, conserved.
Peregrine Falcon F. peregrinus, Barn Swallow Hirundo However, although on average the scaling relation-
rustica) and highest values in species inhabiting dense ship between mean feather mass and mean feather
vegetation (Winter Wren, Common Whitethroat length was 3.00, it is possible that deviations from
Sylvia communis, Little Grebe Tachybaptus ruficollis, this mean may be related to a bird’s habits or habi-
Marsh Tit Parus palustris, Hedge Accentor Prunella tats. For example, birds inhabiting dense vegetation,
modularis). where stronger and more massive feathers may be an
In the majority of species the primary feather that advantage, could have a greater mean mass − mean
formed the wing tip and all primaries distal to it had length relationship than more aerial species. If this
a greater mass–length relationship than the proximal were true, such species would tend to lie above the
primaries. The difference between the actual mass regression line shown in Figure 3, and other species
of the distal primaries of each species and the mass would tend to lie below. To test for this, the devi-
predicted by their lengths and regression for the ation from this regression line was calculated for each
proximal primaries was highly significant (t = 18.96, species and plotted against the slope of the regres-
df = 119, P < 0.0001). Notable exceptions to this were sion (scaling relationship) for each species derived
for some of the corvids (Rook, Eurasian Jackdaw from Figure 2, because birds inhabiting dense vege-
Corvus monedula and Carrion Crow Corvus corone). tation tended to have a greater scaling relationship
Furthermore, although the mass–length relationship than more aerial species. The results are shown in

© 2005 British Ornithologists’ Union, Ibis, 147, 283–292

 1474919x, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1474-919x.2005.00400.x by National University Of Singapo, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Primary flight feather length and mass 287

Figure 2. The scaling relationship of the primary flight feathers in 120 species of birds arranged phylogenetically (see following pages).
The format and axes for each graph are the same as for Figure 1. The solid line on each graph links each of the primary feathers in
sequence, P1 being the most proximal and P9, P10 or P11 the most distal, depending on species. The open symbols represent the
primary feathers proximal to that forming the wing tip. The solid symbols represent the primary feather that forms the wing tip and any
primaries distal to it. The broken line represents the linear regression of the feathers proximal to the wing tip. When more than one
individual of each species was assessed, the number of replicates is shown in parentheses, and in these cases each point represents
the mean ± sd. If the standard deviations are not shown they are smaller than the size of the symbol. *Data for Great Crested Grebe
Podiceps cristatus (n = 10) are taken from Piersma (1988).

Figure 4. Clearly there is no relationship – species the case in Starlings. This was indeed found to be
with a greater scaling relationship, which tend to be true for nearly all species. However, it was apparent
those inhabiting denser vegetation, did not in general that it was not the most distal primary alone that was
have a greater mean mass − mean length relationship. more massive, but the primary that formed the wing
tip and all, if any, primaries distal to that. In species
such as waders and gulls, which have pointed wings
DISCUSSION
and in which the wing tip is formed by the most dis-
The original aim of this study was to determine tal primary, that primary alone had a greater mass in
whether, for birds in general, the most distal primary relation to length. In species with round wings, such
feather had a greater mass, in relation to its length, as hawks, Galliformes and some owls (Little Owl
than the more proximal primaries, as was apparently Athene noctua and Tawny Owl Strix aluco), several

© 2005 British Ornithologists’ Union, Ibis, 147, 283– 292

 1474919x, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1474-919x.2005.00400.x by National University Of Singapo, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
288 A. Dawson

Figure 2. Continued.

primaries are more distal to that forming the wing was nearly always greater than for the proximal
tip, and in each case all of these were more massive primaries, there was no evidence in any species of a
in relation to length than primaries proximal to that gradual increase in mass–length relationship along
forming the wing tip. Among the passerines, there the wing. Rather, there was always a constant scaling
are species representing both extremes (e.g. Barn relationship up to the primary proximal to the wing
Swallow and Winter Wren), and the same relation- tip, and a different relationship beyond. In other
ship normally applied. The only exceptions to this words, there were two distinct scaling relationships,
were three corvid species, Jackdaw, Rook and which pivoted around the primary forming the wing
Carrion Crow. All three have round wings with three tip. The consequence was that each species had a
primary feathers distal to that forming the wing tip. highly characteristic pattern of scaling relationships.
The mass–length relationship for these primaries Although there were often considerable differences
was not greater than for the proximal primaries. in size between individuals of the same species (par-
An additional and unexpected discovery was that ticularly among raptors in which females tend to be
for each of the 120 species examined, there was a larger than males), and hence differences in length
constant allometric relationship for the primaries up and mass of feathers, the pattern of scaling relation-
to, but not including, the primary that forms the ships was highly consistent. Standard deviations of
wing tip. The most striking feature was that although each point were remarkably small. At the same time,
the mass–length relationship for the distal primaries there were sometimes marked differences between

© 2005 British Ornithologists’ Union, Ibis, 147, 283–292

 1474919x, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1474-919x.2005.00400.x by National University Of Singapo, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Primary flight feather length and mass 289

Figure 2. Continued.

closely related species, e.g. Lesser Sylvia curruca and all of the primary feathers distal to the wing tip con-
Common Whitethroats. tribute individual leading edges. In species inhabit-
This highly conserved pattern of a constant allo- ing dense vegetation, the greater mass of distal
metric relationship up to the wing tip, and a different primaries may reflect an additional protective role
relationship beyond, suggests a functional signific- for these feathers. Such species (e.g. Winter Wren)
ance. That the distal primaries are more massive may tend to have round wings. Feathers distal to the wing
represent an aerodynamic role for these feathers. tip are likely to come into physical contact with
All of the primaries proximal to the wing tip are objects in the bird’s environment. They may need to
‘trailing’ and provide wing surface area. The primary be stronger to resist abrasion themselves, and to
that forms the wing tip, and any distal primaries, protect more proximal primaries. However, birds
each provide a leading edge as well as surface area. inhabiting dense vegetation also tend to have a higher
Presumably therefore they need to be stronger and angle of ascent following take-off. Rounded wings
more rigid for aerodynamic reasons. In species in may be advantageous for this (Swaddle & Lockwood
which the most distal primary forms the wing tip, 2003) and this flight characteristic may require more
often fast aerial species, this feather alone forms the massive (rigid) outer primaries. Thus, it is unclear
leading edge to the outer section of the wing. In spe- whether the greater mass of distal primaries in birds
cies that utilize a low-speed soaring flight and have inhabiting dense vegetation reflects an aerodynamic
rounded wings (e.g. Common Buzzard Buteo buteo), role, a protective role or both.

© 2005 British Ornithologists’ Union, Ibis, 147, 283– 292

 1474919x, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1474-919x.2005.00400.x by National University Of Singapo, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
290 A. Dawson

Figure 2. Continued.

For the vast majority of species examined, the

scaling relationship for proximal primaries was less
than isometry (i.e. 3.00). The mean scaling relation-
ship was 2.41, and in only 11 of 120 species did it
exceed 3.00. This means that in most species as one
progresses from the most proximal primary (P1) to
the primary that forms the wing tip, feather mass
becomes progressively less than expected from a
simple geometric relationship (i.e. mass ∝ length3).
Although primaries distal to the wing tip had greater
mass than proximal primaries of the same length,
they were still less massive than predicted from
length3 (mean scaling relationship of all primaries
Figure 3. Each point represents the log mean mass and log mean was 2.50). The significance of this is unclear. It may
length of all of the primary flight feathers from 120 species be that primary feathers nearer the wing tip need to
ranging in size from Winter Wren to Mute Swan. The solid line
be more flexible, for aerodynamic reasons, than
represents the linear regression, r 2 = 0.979. The scaling relationship
for mean feather mass and length is isometric. more proximal primaries. In this context, it may be
significant that the distribution of mass along each

© 2005 British Ornithologists’ Union, Ibis, 147, 283–292

 1474919x, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1474-919x.2005.00400.x by National University Of Singapo, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Primary flight feather length and mass 291

Corn Crake Crex crex < Water Rail Rallus aquaticus)

and owls (Short-eared Asio flammeus < Long-eared
Asio otus < Little < Tawny). However, Lesser and
Common Whitethroats had markedly different
scaling relationships even though there is no obvious
marked difference in their habitats (Cramp 1998).
Although species inhabiting denser vegetation
tended to have a greater scaling relationship, the
mean mass of primary feathers in these species was
not greater than that of other species (Fig. 4). So the
increased mass of the outer primaries must be at the
expense of mass of the inner primaries. Mean mass
is the same, but it is distributed more asymmetrically
Figure 4. Deviation from the regression line in Figure 3 for each (per unit length) toward the outer primaries. One
species (i.e. log(actual mean mass) − log(expected mean mass)) group of birds did have primary feathers that were
plotted against the scaling relationship for proximal primaries particularly massive relative to length – the Alcids
derived from Figure 2. The slope does not differ significantly from
(Common Guillemot Uria aalge, Razorbill Alca torda
zero (r 2 = 0.0075, P = 0.346). Thus, there was no relationship
between the scaling relationship for the proximal primaries and and Atlantic Puffin Fratercula arctica). This probably
whether actual mean feather mass was greater or less than the reflects the fact that they use their wings to swim
expected mean feather mass. The vertical broken line represents underwater (Spring 1971, Johansson & Wetterholm
a scaling relationship of 3.00, i.e. isometry. Only 11 of 120 species Aldrin 2002) and hence need massive rigid feathers.
had a relationship greater than 3.00. The Alcids (Guillemot,
Razorbill and Puffin) are notable in that their mean feather mass I would like to thank the large number of friends and
was considerably greater than expected from mean length.
colleagues who found, collected and provided the birds for
this study.
individual feather changes as one progresses from
the most proximal to the most distal primary. Mass REFERENCES
is not evenly spread along the length of a feather –
Alexander, R.M. 1985. Body support, scaling, and allometry. In
the base of a feather is more massive per unit length
Hildebrand, D., Branmble, D.M., Liem, K.F. & Wake, D.B.
than the tip. This asymmetric distribution of mass is (eds) Functional Vertebrate Morphology : 26 – 37. Cambridge,
more exaggerated in more distal primaries, at least MA: Belknap Press.
in the case of Starlings (Dawson 2003). So the base of Alexander, R.M., Jayes, A.S., Maloiy, G.M.O. & Wathuta, E.M.
a distal primary is relatively more rigid than that of 1979. Allometry of the limb bones of mammals from shrews
(Sorex) to elephant (Loxodonta). J. Zool. (Lond.) 189: 305 –314.
a proximal primary, but the tip is relatively more
Cramp, S. (ed.) 1998. The Complete Birds of the Western Pal-
flexible. earctic on CD-ROM. Oxford: Oxford University Press.
Differences in scaling relationships tended to Dawson, A. 2003. A detailed analysis of primary feather moult
relate to habits and habitats rather than to phylogeny in European Starlings Sturnus vulgaris – new feather mass
(Table 1). Eleven species had a scaling relationship increases at a constant rate. Ibis 145: E69 – E76.
Dawson, A. & Newton, I. 2004. Use and validation of a molt
greater than 3.00 and these are all species that inhabit
score index corrected for primary-feather mass. Auk 121:
fairly dense vegetation (Cramp 1998). At the other 372– 379.
extreme, 12 species had scaling relationships less Hamilton, T.H. 1961. The adaptive significances of intraspecific
than 2.00 (Table 1), most of which are aerial, aquatic trends of variation in wing length and body size among bird
or marine species. However, one species, Tree Pipit species. Evolution 15: 180 –195.
Johansson, L.C. & Wetterholm Aldrin, B.S. 2002. Kinematics
Anthus trivialis, occupies habitats not obviously dif-
of diving Atlantic Puffins (Fratercula arctica, L.): evidence for
ferent from some species with a scaling factor greater an active upstroke. J. Exp. Biol. 205: 371– 378.
than 3.00. In general, species with long wings tended Lockwood, R., Swaddle, J.P. & Rayner, J.M.V. 1998. Avian
to have a lower scaling relationship (exceptions were wingtip shape reconsidered: wingtip shape indices and
Rook and Eurasian Jackdaw), and those with short morphological adaptations to migration. J. Avian Biol. 29:
273 –292.
wings had a greater scaling relationship. Within closely
Pennycuick, C.J. 1989. Bird Flight Performance. A Practical
related species, the scaling relationship varied with Calculation Manual. Oxford: Oxford University Press.
the density of habitat, e.g. rails (Common Coot Fulica Piersma, T. 1988. The annual moult cycle of Great Crested
atra < Common Moorhen Gallinula chloropus < Grebes. Ardea 76: 82– 95.

© 2005 British Ornithologists’ Union, Ibis, 147, 283– 292

 1474919x, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1474-919x.2005.00400.x by National University Of Singapo, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
292 A. Dawson

Rayner, J.M.V. 1985. Linear relations in biomechanics: the Swaddle, J.P. & Lockwood, R. 2003. Wingtip shape and flight
statistics of scaling functions. J. Zool. (Lond.) 206: 415 – 439. performance in the European Starling Sturnus vulgaris.
Rayner, J.M.V. 1988. Form and function in avian flight. In Ibis 145: 457– 464.
Johnston, R.F. (ed.) Current Ornithology : 1– 66. New York: Underhill, L.G. & Joubert, A. 1995. Relative masses of primary
Plenum Press. feathers. Ring. Migr. 16: 109 –116.
Spring, L. 1971. A comparison of functional and morphological
adaptations in the Common Murre (Uria aalge) and Thick-
Billed Murre (Uria lomvia). Condor 73: 1–27. Received 19 May 2004; revision accepted 18 November 2004.

© 2005 British Ornithologists’ Union, Ibis, 147, 283–292