Abstract
The questions asked were (1) whether claws and toes of birds of prey are actually different from those of other perching birds, and, if so, (2) what parameter can describe such a difference. The structure of toes and claws of the first and third toe was then evaluated in three groups of birds: Falconiformes, Strigiformes and non‐raptorial species. One adult male per species was considered, and, among non‐raptorial birds, only species belonging to typically or partially perching families. Only one specimen per species was chosen, from museum skin bird collections. All species examined occur in the western Palearctic, according to Cramp & Simmons (Citation1977–1994). Discriminant Function Analysis showed clear separation of the groups. In particular, the first canonical function segregated Falconiformes from Strigiformes, whereas the second separated Strigiformes from non‐raptorials. However, Falconiformes and non‐raptorials partially overlapped and were not separated. The characteristics segregating Falconiformes from Strigiformes mainly concerned claw curvature, claw length in relation to toe length, and the last phalanx shape, thin or rounded. Characters contributing to segregate Strigiformes from non‐raptorials were claw curvature in relation to their radius and the shape, thin or rounded, of both claws and last phalanxes. Results indicate that Strigiformes toes and claws only superficially resemble those of Falconiformes and the shape of claws and toes of Falconiformes are much more similar to those of non‐raptorial species than they are to those of Strigiformes.
Introduction
That birds of prey (Falconiformes and Strigiformes) possess claws which differ from those of other bird species is a long‐held concept. That is likely the reason why those features are usually named talons instead of claws. This idea is reflected today by many authors, when reporting the morphology of talons with very limited variations. Claw morphology seems not to appeal to researchers, as there are almost no primary bibliographic sources dealing with the topic. A few papers have dealt with the claw geometry of Archaeopteryx lithographica, studied mainly in comparison to modern birds to ascertain whether it was a cursorial or perching bird (Yalden Citation1985; Peters & Görgner Citation1992; Feduccia Citation1993); the only other available source dealing with modern birds is Richardson (Citation1942, quoted by Feduccia Citation1993).
In addition, there are several, although generic, descriptions in books about birds, where only slightly varying feet and claws of birds of prey are described as long toes ending in long sharp talons (Brown Citation1976; Johnsgard Citation1990; Feduccia Citation1996). Birds of prey claws (talons) are reported to be different in shape from those of non‐raptorial birds and typically in their ability to seize and kill prey (Brown & Amadon Citation1968; Brown Citation1976; Johnsgard Citation1990; Feduccia Citation1996). When considered in detail, these descriptions do not rely on objective data, but rather offer generic descriptions of both structure and use. We all subjectively feel that claws of birds of prey appear somewhat different from those of a perching Passeriform bird, but if we observe them, when reported to the same size, they look similar. For instance, let us just enlarge the claw of, say, a spotted flycatcher (Muscicapa striata) and it becomes surprisingly similar to a raptor’s talon.
To our knowledge, nobody has systematically studied the structure of claws of extant birds and the present paper therefore aims to investigate claw geometry quantitatively. Considering that adaptations and/or convergence could alter the interpretation, we aimed to compare those structures in birds of prey and in non‐raptor birds. We worked on three groups: Falconiformes, Strigiformes, as representatives of raptorial birds, and arboreal non‐raptorial species as a whole, as representatives of birds with well‐developed claws, but not specialized as predators.
Materials and methods
All specimens but one (Sturnus vulgaris, which was found freshly dead in perfect condition in the wild) used in this study were skin preparations housed in museum collections. The species used were chosen among those occurring in the western Palearctic, according to Cramp & Simmons (1980), with the only exception being those recorded as very accidental. Only species described as typically perching or partially perching were considered for non‐raptorials, in order not to introduce distortions in claw morphology. One adult male per species was considered in the study, in order not to introduce possible confounding within‐species variability. The variability between species within a certain group, described below, can be considered as the obvious within‐sample variability, as each species was represented by a single specimen. Specimens were chosen randomly among those present in the collections, provided they had undamaged claws.
The species were assigned to three groups: (A) Falconiformes (n = 44), (B) Strigiformes (n = 17), (C) non‐raptorials (n = 80). Groups (A) and (B) comprised all species covered by Cramp & Simmons (1980). Conversely, as group (C) was too large to be represented by all species, we considered representatives of each Family containing species with at least partial perching habits; these families were represented in our study by a variable number of species, but each one with one single specimen, as done for both groups (A) and (B). Families having no species with those habits, e.g. ground‐dwelling species, or represented by species with only accidental recordings were not considered. The full list of species considered as well as the relative specimen identification code are listed in Appendix I.
We dealt with the right foot only and, particularly, with measurements of toe and claw of toe‐1 and toe‐3, as they are the shortest and the longest toe, respectively. Such a decision was justified by considering that toe‐1 and toe‐3 are the most important toes to grip a branch or, in the case of raptorial species, to grip prey (Goslow Citation1971; Csermely et al. Citation1998, Citation2002), as they are in opposition to each other. The measurements were made using a digital caliper (Borletti, Switzerland) measuring to the nearest 0.01 mm. Sixteen morphological measurements were originally obtained from the first and the third toe of each specimen. Ten of them, five for each toe considered, were direct measurements expressed in mm: (1) length of toe (LT); (2) height (HP) and (3) width (WP) of the last phalanx; (4) height (HC) and (5) width (WC) of claw (figure ).
Figure 1 Sketch showing the measurements considered for toes and claws. A, width of last phalanx or of claw; B, length of toe; C, height of last phalanx or of claw.
Toe length (1) was measured along its dorsal margin, from the joint between tarsus and toe to the beginning of claw horn (Baker Citation1993). The height of the last phalanx (2) was measured just before the toe tip, which is very conspicuous in raptors and owls, on the vertical plane. The width of the same phalanx (3) was the measurement at the same location, but in cross section, i.e. on the horizontal plane. Similarly, the height of claw (4) was measured on the vertical plane at its base, i.e. where the claw meets the phalanx skin, whereas the width was measured at the same location, but in cross section, i.e. on the horizontal plane.
In addition, the outline of the external margin of both claws was traced onto graph paper by leaning the claw on the paper and running a sharp 0.5 mm pencil along its external margin for its whole length, from where the claw meets the phalanx skin to the claw very tip. Such an outline was used later to obtain three other measurements for each toe claw: (6) length (LC), (7) curvature (CC), and (8) radius (R) (figure ).
Figure 2 Last phalanx and claw outline showing some details of the measurements considered for the claws. D‐E, section showing where the height of claw was measured; D‐D1 (arc), length of claw; O‐D = O‐D1, claw radius; α, claw curvature angle.
The latter measurements were obtained with mathematical formulae using the MATLAB software (version 4.0 for Windows) (The MathWorks Citation1994). The first step was to record the Cartesian co‐ordinates of the claw outline extremities on the graph paper, as well as of a third point about mid‐way between the others along the outline. These co‐ordinates were recorded visually, reading the outline with a 14× stereo‐microscope to the nearest 0.1 mm. Assuming that such an outline was, or could be considered, an arc of a circle, the software firstly calculated the co‐ordinates of the centre of such a circle and then the three above‐mentioned measurements.
We considered claw length (6) to be the length of the arc of a circle (figure ). The curvature (7) of the same claw was the angle (degrees) at the centre subtended by claw length, and the radius length (8) was the distance between the calculated centre and any of the three points whose co‐ordinates were previously recorded visually (figure ). Our method of calculating the curvature differed from that used by Peters & Görgner (Citation1992). In fact, because of different dorsal (external) and ventral (inner) curvatures, as a measurement of curvature they considered the angle constituted by the two lines tangential to the middle (ideal) line of curvature at the fulcrum point of the joint and at the very tip of the claw, respectively. Instead, we considered that the dorsal curvature alone was responsible for claw curvature as a whole. On the other hand, Feduccia (Citation1993) used a method similar to our own, but considered the inner claw curvature.
Because of the large size variability between the species, 14 additional morphometric parameters were calculated as ratios of the above measurements. They were the descriptive variables used for this study’s statistical evaluation, in order not to distort the data distribution. The parameters used were the following:
LT/LC 1: the ratio between Length of Toe‐1 and Length of Claw‐1; | |||||
LT/LC 3: the ratio between Length of Toe‐3 and Length of Claw‐3; | |||||
LT/CC 1: the ratio between Length of Toe‐1 and Curvature of Claw‐1; | |||||
LT/CC 3: the ratio between Length of Toe‐3 and Curvature of Claw‐3; | |||||
HP/HC 1: the ratio between Height of toe‐1 Phalanx and Height of Claw‐1; | |||||
HP/HC 3: the ratio between Height of toe‐3 Phalanx and Height of Claw‐3; | |||||
WP/WC 1: the ratio between Width of toe‐1 Phalanx and Width of Claw‐1; | |||||
WP/WC 3: the ratio between Width of toe‐3 Phalanx and Width of Claw‐3; | |||||
H/W P1: the ratio between Height and Width of toe‐1 Phalanx; | |||||
H/W P3: the ratio between Height and Width of toe‐3 Phalanx; | |||||
H/W C1: the ratio between Height and Width of toe‐1 Claw; | |||||
H/W C3: the ratio between Height and Width of toe‐3 Claw; | |||||
R/C C1: the ratio between Radius and Curvature of toe‐1 Claw; | |||||
R/C C3: the ratio between Radius and Curvature of toe‐3 Claw. | |||||
In addition to these ratios, two other absolute measurements were used for the analysis:
C C1: the Curvature, in degrees, of toe‐1 Claw; | |||||
C C3: the Curvature, in degrees, of toe‐3 Claw. | |||||
Statistical analyses
The statistical analysis was performed using the software SPSS 9.0 for Windows (SPSS Citation1999). All the variables were analysed using the t‐test and the Discriminant Function Analysis, whereas body mass comparisons were performed using the Mann–Whitney U test. Means are reported±SE unless differently specified.
Results
The seven pairs of ratio values used for the analysis were rather similar between the groups. The height of the last phalanx (HP) was moderately greater than the height of its claw (HC) in both toe‐1 and toe‐3 (table ); this was more evident in the first toe of Strigiformes and non‐raptorial species, where the ratio reached mean values of 1.453±0.075 mm and 1.556±0.020 mm, respectively.
Table I. Sample size, mean value±SE and SD of every ratio value between the morphological measurements considered for the analysis in each group of birds used in the study.
While Falconiformes and Strigiformes showed no difference (t‐test, P>0.05) between the values recorded for the first and third toe, Falconiformes displayed a ratio of toe‐1 greater (t = 2.096, df = 43, P<0.05) than that of toe‐3, whereas Strigiformes did not. The similar ratios between the width of phalanx (WP) and of claw (WC) revealed in each group that the phalanx was expectedly slightly larger than its claw. In addition, the comparison of the ratios concerning the first and third toe was significant in all groups (P<0.01), showing that the claw is proportionally thicker in toe‐3.
The last phalanx was almost “rounded” (i.e. as high as it was large) in Falconiformes, since the ratio between height (HP) and width (WP) was very close to the mean value 1 in both toes (P>0.1). In contrast, Strigiformes had a similar value for toe‐3, but a greater one for toe‐1 (t = 3.382, df = 16, P<0.01), whereas non‐raptorials showed the opposite trend, the ratio value of toe‐1 being smaller (t = 9.806, df = 79, P<0.0001) than that of toe‐3. Concerning the claw itself, that of toe‐1 was constantly higher than wide, particularly in non‐raptorial species, where the height (H)/width (W) ratio value was 1.704±0.029 mm.
The two parameters were more similar concerning toe‐3 in Falconiformes and, particularly, in Strigiformes, where the claw was larger than high (mean ratio value 0.974±0.037 mm), whereas in non‐raptorials that ratio value was similar to that of toe‐1 (1.666±0.044 mm). The values of both toes differed greatly in Falconiformes (t = 8.096, df = 43, P<0.0001) and in Strigiformes (t = 9.841, df = 16, P<0.0001), but not in non‐raptorials. Finally, the ratio between the claw radius (R) and its curvature (C), in degrees of an arc, showed similar mean values for Falconiformes and Strigiformes, but a lower value for non‐raptorial species. This occurred for toe‐1, where the mean value was almost half of the Falconiformes’ one (0.108±0.006 and 0.050±0.003, respectively) and, more evidently, in toe‐3 (0.116±0.006 and 0.063±0.004, respectively). The comparison between the values of toe‐1 and toe‐3 in each group revealed significant differences in each group (Falconiformes: t = −2.082, df = 43, P<0.05; Strigiformes: t = −3.653, df = 16, P<0.01; non‐raptorials: t = −3.774, df = 79, P<0.001).
The Discriminant Function Analysis revealed distinct clusters among the three groups considered (figure ).
Figure 3 Plots of the two canonical variables from the discriminant analysis showing segregation of the three groups considered. The black squares show the centroid location of each group.
Table II. Eigenvectors for both canonical functions and the relative canonical correlation values.
Table III. The Wilks’ Lambda value and its significance for both canonical functions.
Most data (92.9%) belonged to the expected group (table ); in particular, Falconiformes comprised 40 (90.9%) of 44, Strigiformes 15 (88.2%) of 17, and non‐raptorial species 76 (95.0%) of 80.
Table IV. Results of classification. Among the original grouped cases for the three groups, 92.9% were correctly classified.
The first canonical discriminant function separated clearly non‐raptorial species from Falconiformes and Strigiformes together (figure ). The first coefficient (H/W C3) related to the first canonical function obtained from the structure matrix accounted for 54% of variance (table ). The second ranking coefficient (R/C C1) was negative and explained 43.3% of variance, whereas the following two in ranking (H/W P3 and H/W C1, respectively) accounted for 40.2% and 36.9% of variance.
Table V. Values of the four most extreme positive and negative characters obtained from the structure matrix.
The second canonical discriminant function clearly separated Falconiformes from Strigiformes, but failed to separate non‐raptorials (figure ). The coefficients related to the second canonical function obtained from the structure matrix had values similar to those of the first one (table ). Three of four higher rank correlation coefficients had very low rank for the first canonical function. The first ranking correlation coefficient, H/W P1, was negative and explained 61.1% of variance. The other three coefficients along the ranking order were all positive and accounted for 43.5% (LT/CC 1), 37.6% (RC/C1) and 31.8% (LT/CC 3) of variance, respectively. In particular, R/C C1, ranking 3 with the second canonical, was rank 2 with the first canonical function and, instead, negative.
The three groups considered can then be separated on the basis of several morphometric ratios of claws. Not only were non‐raptorial species separated from birds of prey as a whole, but even among these the Falconiformes differed from Strigiformes. The first canonical function discriminated between non‐raptorial species and Strigiformes, mostly on the basis of characteristics of the claws and, to a lesser extent, of the last phalanx of toes. In particular, Strigiformes had both claw‐1 and claw‐3 longer in relation to the radius of curvature (R/C C1, R/C C3) and more rounded at the base (H/W C1, H/W C3) than non‐raptorial birds. In addition, Strigiformes had claw‐1 less high at the base compared to the relative phalanx (HP/HC 1), whereas claw‐3 was longer in relation to its toe (LT/LC 3) than non‐raptorial species. Strigiformes had also phalanx‐1 more high than wide (H/W P1), whereas phalanx‐3 was, instead, more rounded (H/W P1) than that of non‐raptorial species. The same canonical function showed that Falconiformes and non‐raptorial birds have widely overlapping measurements and it is not possible to discriminate between them.
Similar differences were apparent between Falconiformes and Strigiformes on the basis of the second canonical function. The major differences segregating the two groups concerned claw‐1, which in Strigiformes proved thinner (H/W C1), lower in relation to phalanx height (HP/HC 1), less curved (C C1), but with the curvature more similar to the amount of its radius (R/C C1). Moreover, the curvature of both claw‐1 and claw‐3 of Strigiformes was smaller in relation to toe length (LT/CC 1, LT/CC 3), Falconiformes having longer curvature relative to the toe. Strigiformes showed also both phalanx‐1 and phalanx‐3 more high than wide, and thinner (H/W P1, H/W P3) than those of Falconiformes.
There is long‐held concern about the consequences of using ratios in Multivariate Analysis (cf. Green Citation1979, p. 105). Therefore, we repeated the same statistical analysis using the ranked data for each variable. We carried out the Discriminant Analysis on those transformed data and the results substantially matched those obtained using the ratios directly (Tables and ), as can otherwise be noted comparing Figures and .
Table VI. Wilks’ Lambda values and their significance for both canonical functions after data ranking.
Table VII. Values of the four most extreme positive and negative characters obtained from the structure matrix after data ranking.
Figure 4 Plots of the two canonical variables from the discriminant analysis showing segregation of the three groups considered after data ranking. The black squares show the centroid location of each group.
In addition, to obtain a more reliable fit of results, we also checked the relationship between the considered ratios’ numerator and denominator. Linearity is always observed and the intercept was found crossing the axes very close to the origin (figure ). Even the Discriminant Analysis carried out on the linear regression residuals showed results substantially similar to those obtained from the analysis carried out on ranks.
Figure 5 Three diagrams used as example showing the linearity of the residuals in the first three ranking coefficients: (A) H/W C3, (B) H/W P3, (C) H/W C1. The measurements are in millimetres and the line shows the regression line y = ax+b, which crosses the axes very close to the origin in each diagram.
Coming back to the overlapping area between Falconiformes and non‐raptorials, not clearly separated by the second canonical discriminant function, we can add one final consideration. Twenty‐six species fell within that area: 19 (43.18% of total species considered) were Falconiformes and only 7 (8.75% of total species considered) were non‐raptorials. Among the former, 10 belonged to the Accipitridae (Pernis apivorus, Elanus caeruleus, Milvus milvus, Circaetus gallicus, Circus aeruginosus, Circus macrourus, Melierax metabates mechowi, Accipiter nisus, Accipiter brevipes, Aquila pomarina) and 9 to the Falconidae (Falco naumanni, F. sparverius, F. vespertinus, F. columbarius aesalon, F. subbuteo, F. eleonorae, F. concolor, F. biarmicus feldeggi, F. peregrinus). Falcons were thus highly represented, corresponding to 69.23% of the species considered in that Family.
The seven non‐raptorial species were one species of Columbiformes (Streptopelia turtur) and six species of Passeriformes (Turdus philomelos, Garrulus glandarius, Nucifraga caryocatactes, Corvus frugilegus, Corvus corone cornix, Carduelis carduelis). In this case, apart form Streptopelia, those species were represented almost totally by Passeriformes and, among them, by Corvidae (4 of the 6 species), the heaviest passeriform species. However, these six species were just 9.68% of the passeriform species considered in this study.
The mean body mass of the species falling in the overlapping area was extremely variable in both groups. The body mass variance was very high, suggesting the use of the non‐parametric Mann–Whitney U test to compare body mass in species within the area with species of the same group falling outside the area. The body mass of Falconiformes within the area was much smaller than outside it (474.53±429.62 [SD] g vs 1570.60±1131.09 g; U = 4.028, n = 44, P<0.0001), but that of non‐raptorials was instead more than double that of species falling outside the area (231.71±198.50 [SD] g vs 93.29±234.96 g; U = 2.879, n = 80, P<0.01). Consequently, there was no difference in body mass between the Falconiformes and the non‐raptorial species, both falling within the overlapping area (U = 1.475, n = 26, P>0.1).
Discussion
The Discriminant Function Analysis revealed clearly that the claw morphology of the groups considered differs considerably, mostly in the shape of the section, thin or more rounded, of claw and of last phalanx and claw length in relation to radius of curvature. This applies almost equally to toe‐1 and toe‐3. The amount of claw curvature is a discriminant character only segregating Strigiformes from Falconiformes. It shows the curvature of claws is the same in Falconiformes and in non‐raptorial species and that statements reporting a stronger curvature of toes for the former (e.g. Feduccia Citation1996) are incorrect.
The curvature of toe‐1 proved a discriminant character between Falconiformes and Strigiformes, showing a more curved claw in the former, whereas we would have expected a similarity because the claws (talons) are usually considered a convergent adaptation in these groups. An explanation for such a difference could lie in the Strigiformes toes: they have a very pronounced movement of the last phalanx, which can be easily turned, as in a very closed grip. This aspect is very often observed in dead owls as well as in museum specimens, where the toes are strongly closed on themselves, as in a fist. As this ability is not so developed in Falconiformes, it is likely that Falconiformes compensate this minor ability by developing a more strongly curved claw. The above difference between the groups concerns claw‐1 only and it could show that toe‐1 is particularly important for gripping the prey, mostly small mammals as in the case of Strigiformes, using both toe and talon to encircle the prey rather than using the latter as a weapon to kill. This is also in agreement with previous observations by Csermely et al. (Citation1998, Citation2002) on the predatory behaviour of Eurasian buzzards (Buteo buteo), common kestrels (F. tinnunculus) and some owl species.
Strigiformes have claw‐3 longer in relation to its toe than do non‐raptorials, confirming its use to encircle the prey, as such a toe is opposed to toe‐1. Both, therefore, can be seen as a tool forming a ring for prey grasping, with toe‐2 and toe‐4 used to assist the grasp more laterally. On the other hand, Strigiformes have an intermediate value between the other groups considered, non‐raptorials having the thinnest section and Falconiformes the roundest one.
It thus emerges from our data that the characteristics of claws are rather different among the groups considered and that birds of prey are not particularly similar from this point of view. As the Discriminant Function Analysis failed to separate Falconiformes from non‐raptorials, both these groups must be considered very similar concerning the general design of claws, at least the first and third one. Strigiformes are clearly different from the other groups, having a more specialized shape of claws, and this feature can be considered more derived than in other birds.
One final consideration concerns the relatively few Falconiformes and non‐raptorial species falling inside the overlapping area of both groups. They were represented by species similar in body mass. In fact, many of the lightest species, principally Falconidae, among the Falconiformes and several of the heaviest species, principally corvids, among the non‐raptorials were plotted in that area. Both subgroups, instead, differed greatly from the species of the same group falling outside the overlapping area, suggesting that similarity in body mass might have led to similarity in claw shape. However, this is not true for every species of that mass range and thus that suggestion cannot be anything more than a working hypothesis. In fact, the small number of those species and their high body mass variance prevent any definitive conclusions. It could nevertheless be interesting to explore that hypothesis in future studies.
Acknowledgements
We received much help from several people in carrying out this study. Specimens were examined in the following museums through the courtesy of their respective Curators and/or Directors: Museo di Storia Naturale dell’Università, Parma, Italy (Professor Vittorio Parisi); The Natural History Museum, Tring, U.K. (Dr Robert Prys‐Jones and Dr Mark P. Adams); University Museum of Zoology, Cambridge, U.K. (Dr Ray J. Symonds); Museo Civico di Storia Naturale, Milano, Italy (Dr Giorgio Chiozzi); Istituto Nazionale per la Fauna Selvatica, Ozzano Emilia, Italy (Dr Marco Zenatello); Museo Regionale di Scienze Naturali, Torino, Italy (Dr Claudio Pulcher); Museo Civico di Zoologia, Roma, Italy (Dr Rossella Carlini). Dr Maria Groppi is thanked for her help with using the mathematical software, as well as Dr Andrea Pilastro for his sound and improving comments and suggestions as referee. We appreciate the improvements in English usage made by Felipe Chavez‐Ramirez through the Association of Field Ornithologists’ program of editorial assistance, led by Daniel M. Brooks. This study was financially supported by the Italian Ministero dell’Istruzione, dell’Università e della Ricerca.
Related Research Data
References
- Baker , K. 1993 . Identification guide to European non‐Passerines , Thetford : British Trust for Ornithology . Field Guide Number 24
- Brown , L. and Amadon , D. 1968 . Eagles, hawks and falcons of the world , Feltham : Country Life Books .
- Brown , L. 1976 . British birds of prey , London : Collins .
- Cramp , S and Simmons , K. E. L , eds. 1977–1994 . Handbook of the birds of Europe, the Middle East and North Africa. The birds of western Palearctic , Oxford : Oxford University Press .
- Csermely , D. , Bertè , L. and Camoni , R. 1998 . Prey killing by Eurasian kestrels: the role of the foot and the significance of bill and talons. . Journal of Avian Biology , 29 : 10 – 16 .
- Csermely , D. , Casagrande , S. and Sponza , S. 2002 . Some adaptive details in the comparison of predatory behaviour of four owl species. . Italian Journal of Zoology , 69 : 239 – 243 .
- Feduccia , A. 1993 . Evidence from claw geometry indicating arboreal habits of Archaeopteryx. . Science , 259 : 790 – 793 .
- Feduccia , A. 1996 . The origin and evolution of birds , New Haven : Yale University Press .
- Goslow , G. E. 1971 . The attack and strike of some North American raptors. . Auk , 88 : 815 – 827 .
- Green , R. H. 1979 . Sampling design and statistical methods for environmental biologists , New York : J. Wiley & Sons .
- Johnsgard , P. A. 1990 . Hawks, eagles & falcons of North America , Washington, DC : Smithsonian Institution Press .
- Peters , D. S. and Görgner , E. 1992 . “ A comparative study on the claws of Archaeopteryx. ” . In Papers in avian paleontology. Honoring Pierce Brodkorb , Edited by: Campbell , K. E . 29 – 37 . Los Angeles : Science Series Natural History, Museum of Los Angeles County . In
- Richardson , F. 1942 . Adaptive modifications for tree‐trunk foraging in birds. . University of California Publications in Zoology , 46 : 317 – 368 .
- SPSS Inc . 1999 . SPSS 9.0 for Windows: Base, professional statistics and advanced statistics , Chicago : SPSS Inc .
- The MathWorks Inc . 1994 . MATLAB 4.0, high‐performance numeric computation and visualization software , Natick : The MathWorks Inc .
- Yalden , D. W. Forelimb function in Archaeopteryx. Edited by: Hecht , M. K , Ostrom , J. H , Viohl , G and Wellnhofer , P . pp. 91 – 97 .