Digit Ratio in Birds

Authors


Abstract

The Homeobox (Hox) genes direct the development of tetrapod digits. The expression of Hox genes may be influenced by endogenous sex steroids during development. Manning (Digit ratio. New Brunswick, NJ: Rutgers University Press, 2002) predicted that the ratio between the lengths of digits 2 (2D) and 4 (4D) should be sexually dimorphic because prenatal exposure to estrogens and androgens positively influence the lengths of 2D and 4D, respectively. We measured digits and other morphological traits of birds from three orders (Passeriformes, house sparrow, Passer domesticus; tree swallow, Tachycineta bicolor; Pscittaciformes, budgerigar, Melopsittacus undulates; Galliformes, chicken, Gallus domesticus) to test this prediction. None were sexually dimorphic for 2D:4D and there were no associations between 2D:4D and other sexually dimorphic traits. When we pooled data from all four species after we averaged right and left side digits from each individual and z-transformed the resulting digit ratios, we found that males had significantly larger 2D:4D than did females. Tetrapods appear to be sexually dimorphic for 2D:4D with 2D:4D larger in males as in some birds and reptiles and 2D:4D smaller in males as in some mammals. The differences between the reptile and mammal lineages in the directionality of 2D:4D may be related to the differences between them in chromosomal sex determination. We suggest that (a) natural selection for a perching foot in the first birds may have overridden the effects of hormones on the development of digit ratio in this group of vertebrates and (b) caution be used in making inferences about prenatal exposure to hormones and digit ratio in birds. Anat Rec, 2008. © 2008 Wiley-Liss, Inc.

The Homeobox (Hox) genes direct the development of the digits, limbs, and urogenital systems in vertebrates (Herault et al., 1997; Peichel et al., 1997). These genes are dispersed over four gene clusters (HoxA-HoxD) (Lemons and McGinnis, 2006). Limb development is governed by genes within the HoxA and HoxD clusters; genes in groups 11–13 affect digit development (Zákány and Duboule, 1999). The expression of these genes is influenced by endogenous sex steroids (Block et al., 2000), so that prenatal exposure to estrogens and androgens positively influence the length of the digit 2 (2D) and digit 4 (4D), respectively (Manning, 2002 and references therein). It is hypothesized that because of these effects, the 2D:4D ratio of human males is, on average, smaller than that of females (Manning, 2002; Lutchmaya et al., 2004). The observation that human females with congenital adrenal hyperplasia, which is characterized by excess production of androgens, have male-like 2D:4D (Brown et al., 2002c) is evidence of these effects. The effects of hormones on digit ratio occur early in development and there is little plasticity thereafter (Garn et al., 1975; Brown et al., 2002c). However, more recent work has revealed that digit ratios may change over time in human children until puberty (Manning et al., 2004; McIntyre et al., 2005; Trivers et al., 2006).

The hypothesis that digit ratio can function as an easily observed indicator of prenatal exposure to androgens stimulated a plethora of studies correlating digit ratio with developmental (e.g., Manning et al., 2003; Burley and Foster, 2004; Lutchmaya et al., 2004; Forstmeier, 2005), anatomical (e.g., McFadden and Shubel, 2002; Chang et al., 2006; Rubolini et al., 2006; Garamszegi et al., 2007), physiological (e.g., Manning et al., 1998, 2004; Manning and Taylor, 2001), and behavioral (e.g., Manning et al., 2000; Williams et al., 2000; Csathó et al., 2003; Bailey and Hurd, 2004; Bailey et al., 2004; Fink et al., 2004; Romano et al., 2005a; van Anders and Hampson, 2005; Millet and Dewitte, 2006; Saino et al., 2006a, b) measures implying that digit ratio may be a reliable predictor of evolutionary fitness (Manning, 2002).

To date, most researchers have focused on correlates of human digit ratio. However, there is the potential for digit ratio to function as a predictor of evolutionary fitness in other vertebrates because Hox genes are highly conserved in vertebrates (Krumlauf, 1994; Lemons and McGinnis, 2006). Manning (2002) predicted that tetrapods descended from the common ancestor with the primitive pentadactyl (five-digit) limb will be sexually dimorphic for 2D:4D. In nonhuman mammals, smaller 2D:4D ratios have been found in laboratory mice (Mus musculus) (Brown et al., 2002a; Manning et al., 2003: Bailey et al., 2004), wood mice (Apodemus sylvaticus) (Leoni et al., 2005), and laboratory rats (Rattus norvegicus) (McMechan et al., 2004). In contrast, Guinea baboon (Papio papio) males have larger 2D:4D than do females, but this outcome may be affected by small sample sizes (Roney et al., 2004). Together these results suggest that mammals are sexually dimorphic for 2D:4D with males usually having lower 2D:4D than females.

Reptiles show inconsistent patterns of sexual dimorphism in 2D:4D. Common wall lizard (Podarcis muralis) males had larger 2D:4D on both of their front limbs than did females, whereas tree skinks (Mabuya planifrons) were sexually monomorphic for 2D:4D on both front limbs (Rubolini et al., 2006). Chang et al. (2006) found that anolis lizard (Anolis carolensis) males had larger 2D:4D on the hind right leg than did females, but males and females were monomorphic for 2D:4D on their other limbs. In contrast, Lombardo and Thorpe (2008) failed to find sexual dimorphism in 2D:4D on any limb of anolis lizards. To date, there are no reports of male reptiles with lower 2D:4D than females. When present in reptiles, sexual dimorphism in 2D:4D appears to be in the opposite direction to that in mammals.

Like other reptiles, birds show inconsistent patterns of sexual dimorphism in 2D:4D. Male zebra finches (Taenopygia guttata) had larger right foot 2D:4D than did females (Burley and Foster, 2004). In constrast, Forstmeier (2005), using a larger sample (n = 500) than Burley and Foster (2004) (n = 103), did not detect sexual dimorphism in 2D:4D on the right feet of zebra finches. Differences between these two studies may be due to population differences (Forstmeier, 2005). Romano et al. (2005b) did not detect sexual dimorphism for 2D:4D in adult ring-necked pheasants (Phasianus colchis), but adult females produced from eggs injected with testosterone during incubation had larger 2D:3D on their left than right feet. In contrast, adult males produced from testosterone injected eggs showed neither a significant differences between left and right digit ratios nor sexual dimorphism in 2D:4D (Romano et al., 2005b). More recently, Saino et al. (2007) showed that eggs with experimentally increased levels of estradiol produced male pheasants with smaller right foot 2D:4D indicating that prenatal exposure to estrogens can affect bird digit ratios. Close inspection of the data presented in Navarro et al. (2007) indicates no sexual dimorphism in 2D:4D in house sparrows (Passer domesticus). Last, collared flycatchers (Ficedula albicollis) did not show significant sexual dimorphism in 2D:4D (Garamszegi et al., 2007). To date, there are no reports of unmanipulated male birds with lower 2D:4D than females.

In summary, studies of (a) five species of mammals from two orders, (b) three species of oviparous reptiles from three families, and (c) four species of birds from two orders reveal that patterns of 2D:4D vary between taxa, species within taxa, and populations within species of tetrapods. These results suggest that steroid hormones may not consistently affect the expression of the Hox genes that direct digit development. However, in these studies the effect sizes (Table 1 in Lombardo and Thorpe, 2008) and repeatability estimates are often low making it difficult to detect sexual dimorphism in 2D:4D. Nevertheless, it is still important to test Manning's prediction on a multitude of species from a variety of taxa and populations within taxa to help reveal digit ratio patterns as a first step in elucidating the role of prenatal exposure of sex steroids on Hox gene expression.

Table 1. Sex differences in left and right 2D:4D in house sparrows, tree swallows, budgerigars, and chickens
MeasureMaleFemaletdfPd
  1. Mean ± SD are indicated for digit ratios. t = Student's t-test statistic. df = degrees of freedom. P = probability of observing t due to chance alone. d = Cohen's d, a measure of effect size.

House sparrows      
 Right-2D:4D0.924 ± 0.012 (n = 58)0.897 ± 0.089 (n = 62)1.421180.160.26
 Left-2D:4D0.919 ± 0.133 (n = 57)0.895 ± 0.098 (n = 62)1.221170.230.24
Tree swallows      
 Right-2D:4D0.944 ± 0.105 (n = 33)0.902 ± 0.100 (n = 38)1.70690.090.42
 Left-2D:4D0.965 ± 0.091 (n = 32)0.961 ± 0.119 (n = 38)0.17680.870.04
Budgerigars      
 Right-2D:4D0.705 ± 0.081 (n = 10)0.710 ± 0.141 (n = 19)0.10270.920.04
 Left-2D:4D0.761 ± 0.116 (n = 12)0.685 ± 0.125 (n = 15)1.62250.120.63
Chickens      
 Right-2D:4D0.881 ± 0.094 (n = 12)0.874 ± 0.056 (n = 12)0.31220.760.13
 Left-2D:4D0.869 ± 0.034 (n = 12)0.853 ± 0.053 (n = 12)0.85220.400.34

Our goal was to test in birds Manning's (2002) prediction of sexual dimorphism in 2D:4D. Manning's hypothesis is relevant to birds because birds evolved from an ancestral pentadactyl tetrapod (Galis et al., 2003). We were interested in testing this hypothesis because if confirmed it could provide field biologists with easily measured marker of prenatal exposure to steroid hormones (Garamszegi et al., 2007). To that end, we examined digit ratio in four species of birds from three different orders: Passeriformes; house sparrow, tree swallow (Tachycineta bicolor); Psittaciformes; budgerigar (Melopsittacus undulates); Galliformes; chicken (Gallus domesticus). We chose these species because it was easy for us to locally obtain samples. House sparrows and tree swallows are common wild birds, budgerigars are common cage birds, and chickens are common domesticated birds. In addition, we were able to examine the relationship between digit ratios and sexually dimorphic characters in house sparrows, tree swallows, and chickens (see Methods).

The birds in our study have different kinds of feet although three of the four species have the same anatomical pattern of digits. Examining feet with different arrangements of digits and primary functions allowed us to also ask whether natural selection for function may have overridden the effects of prenatal exposure of testosterone in digit ratio in birds. The feet of house sparrows and tree swallows have an anisodactyl arrangement of digits (i.e., 2D, 3D, and 4D point forward, while 1D, the hallux, points backward) (Proctor and Lynch, 1993) and use their feet for perching. Chickens have anisodactyl feet and, although capable of flight, are mostly terrestrial and primarily use their strong feet to scratch the ground while searching for food. Budgerigars are arboreal and primarily use their zygodactyl feet (2D and 3D point forward, while 1D and 4D point backward; Proctor and Lynch, 1993) to perch.

METHODS

House Sparrows

We measured the lengths of digits (i.e., toes) and tarsi of male and female house sparrows. Sparrows were captured from 6 May to 17 July, 2003, using nest-box traps, walk-in traps, and mist nets at River Ridge Farm and the Grand Valley State University (GVSU) campus in Ottawa County, John Ball Zoo, and the Thorpe House in Kent County, MI. We determined the sex of sparrows by noting their sexually dichromatic plumages. Males with their black bib, gray crown and rump, and chestnut-colored wings are more brightly colored than the mostly dull grayish brown and bib-less females (Lowther and Cink, 1992). We also measured the bib areas of adult male house sparrows (see later). Recently, fledged male sparrows lack the distinctive black bib but were also sexed by plumage characteristics (Lowther and Cink, 1992). We banded sparrows to avoid remeasuring them.

Tree Swallows

We measured the lengths of digits, the right tarsus, the right wing, and the right tail fork of male and female tree swallows. Swallows were captured using nest-box traps in wooden nest boxes arranged in grids in an old field on the GVSU campus (Lombardo et al., 2004) during May and June in 2003. We distinguished males from females by the presence of a cloacal protuberance and the lack of a brood patch in males and by plumage characteristics (Dwight, 1900; Lombardo, 2001). Each swallow was banded with a US Fish and Wildlife Service numbered aluminum band so that birds were not remeasured.

Budgerigars

We measured the lengths of digits and right wings of male and female budgerigars in the breeding colonies of two cage-bird enthusiasts in Kent Co., MI, in April 2005. The birds were sexed by noting the color of the cere; royal blue in males and white to pale-brown (nonbreeding) or brown (breeding) in females.

Chickens

We measured the lengths of the digits, tarsi, and combs of male and female partial chicken carcasses (heads and feet) obtained from a local poultry farm in March 2002. Comb size is a sexually dimorphic trait in chickens and is an indicator of testosterone levels (Allee et al., 1939; Zeller, 1971; Ligon et al., 1990; Zuk et al., 1995).

Measuring Toe and Tarsus Lengths

We took digital photographs of the ventral surface of the feet of house sparrows, tree swallows, and budgerigars with a Nikon Coolpix® 5 megapixel digital camera. Photographs of the digits were obtained by holding the birds on their backs so that they splayed out and fully extended their toes. The digital images were downloaded into Scion Image, a digital image analysis program (http://www.scioncorp.com). The analysis tool was calibrated to each individual image by measuring the ruler that appeared in each image. Once calibrated, we measured each digit to 0.01 mm. Digit 1 of house sparrows and tree swallows was measured from the distal edge of the metatarsal fold to the proximal edge of the claw (Lucas and Stettenheim, 1972). Digits 2–4 on house sparrows and tree swallows were measured from the distal edge of the metatarsal pad to the proximal edge of the claw on each digit (Lucas and Stettenheim, 1972). Digits 1–4 on budgerigars were measured from the distal edge of the metatarsal pad to the proximal edge of the claw (Lucas and Stettenheim, 1972). The complete lengths of some digits were obscured from view by other digits and so were not measured resulting in unbalanced sample sizes in some categories.

Tarsus lengths on sparrows and swallows were measured as the length of the diagonal from behind the middle of the ankle joint to the middle of the joint between the tarsus and 3D in the front. Tarsus lengths were measured to the nearest 0.01 mm using the digital image analysis program.

We measured chicken tarsus and digit lengths with electronic digital calipers to the nearest 0.01 m. Chicken digits were measured using the same landmarks as were the toes of sparrows and swallows.

B.B. measured all sparrow and swallow digit and tarsus lengths and K.S. measured all budgerigar digits. We did not measure tarsus lengths in budgerigars. Lombardo measured all chicken digit and tarsus lengths. Each person measured all the digits on each individual of their assigned species to avoid interobserver differences in measurements within species.

Measuring House Sparrow Bibs

We measured the bib areas of adult male house sparrows to determine if there was a relationship between digit ratio, which is associated with testosterone levels during embryonic development (Garn et al., 1975; Brown et al., 2002c; Trivers et al., 2006), and bib area because bib area is positively correlated with testosterone level during molt (Møller, 1988) and male house sparrow reproductive success (Møller and Erritzoe, 1988; Møller, 1990). We used digital photographs of the bibs of sparrows to calculate bib area. After calibrating the software as described earlier, sparrow bibs were measured by first outlining a digital image of a bib using the software and then allowing the program to calculate its area to 0.01 mm2. B.B. measured house sparrow bibs.

Wing and Tail Fork Measurements in Tree Swallows

To determine if sexual dimorphism in digit ratio was associated with other sexually dimorphic traits, we also measured wing and tail fork length in tree swallows. Tree swallows have slightly forked tails (Turner and Rose, 1989). Wing and tail fork length are sexually dimorphic traits in tree swallows; males have longer wings (Robertson et al., 1992) and tail forks (Lombardo, unpublished data). We measured each swallow's unflattened right wing chord length to 1 mm with a ruler with a stop fixed to one end. Tail fork length was measured as the length of tail feathers from the notch in the center of the slightly forked tail to the tip of the outer tail feathers on the right side of the tail to the nearest 1 mm with a ruler while the tail was held so that the outer edges of each outer tail feather were parallel to each other. M.L. measured all swallow wings and tail fork lengths.

Measuring Chicken Combs

We measured the combs of chickens to determine if there was a relationship between digit ratio and comb dimensions because comb size is strongly influenced by testosterone levels (Allee et al., 1939; Zeller, 1971; Ligon et al., 1990; Zuk et al., 1995). Again, our goal was to determine if digit ratio, which is influenced by testosterone levels during development (Garn et al., 1975; Brown et al., 2002c; Trivers et al., 2006), was correlated with adult testosterone levels. Comb length was measured with electronic digital calipers to the nearest 0.1 mm. Comb length is an indicator of head ornament size and highly correlated with comb height (Ligon et al., 1990). M.L. measured all combs.

Statistical Analyses

We examined the data for normality and, where appropriate, used parametric and nonparametric statistical tests to analyze data using SPSS 10.0 for Windows (SPSS, 2001). We used Student's t-tests to compare the digit lengths and digit length ratios of males and females. In all cases, Levene's test for equality of variances showed that there was no significant differences between the sexes in sample variances (all P > 0.05). Following Lessells and Boag (1987), we estimated the reliability of our calculations of left and right side 2D:4D by determining the intraclass correlation coefficients (ICC) of 2D:4D. Our calculations of 2D:4D for house sparrow (ICC = 0.34, F = 2.04, df = 117, P < 0.001) and tree swallow (ICC = 0.38, F = 2.24, df = 64, P = 0.001) were more reliable than were those for budgerigars (ICC = 0.20, F = 1.50, df = 18, P = 0.20) or chickens (ICC = 0.27, F = 1.74, df = 23, P = 0.10), but in all species there were significant correlations between left- and right-side digit lengths for 2D and 4D (all P < 0.05) and with the exception of tree swallows (paired-t = 2.71, df = 64, P = 0.01), there were no significant differences between left and right side 2D:4D (all P > 0.05; Table 1) suggesting that measurement errors did not obscure sexual dimorphism in digit ratios. Nevertheless, caution in interpretation of these results is warranted; ICCs were small making it potentially difficult to detect statistically significant sexual dimorphism in 2D:4D because the differences in 2D:4D between individuals were not much greater than the measurement error of 2D:4D.

After we separately analyzed data from each species, we pooled all of our data after calculating the average digit length of each digit (i.e., 1D, 2D, 3D, 4D) for each individual as (right digit + left digit)/2 and z-transforming the resulting digit length ratios to increase our chances of detecting sexual dimorphism in digit ratio. Using transformed data, we used ANOVA to examine the effects of species, sex, and the species–sex interaction on digit ratio.

Following McFadden and Shubel (2002), Bailey et al. (2004), and Rubolini et al. (2006), we examined sex differences in 2D:4D by calculating effect size (Cohen's d); by convention, effect size of d = 0.8 is considered large, d = 0.5 medium, and d = 0.2 small (Cohen, 1992).

Throughout we focus our discussion on morphological data from the right sides of bodies because (a) right-digit ratios appear to be more sensitive to exposure to sex steroids (Manning, 2002), (b) previous studies have reported that 2D:4D sexual dimorphism is more pronounced on the right side (Williams et al., 2000; Brown et al., 2002b; McFadden and Shubel, 2002; Rahman and Wilson, 2003), and (c) preliminary analyses indicated that right- and left-digit lengths and ratios were significantly correlated in the birds in this study (all P < 0.05).

Data are reported as mean ± SD. All tests were two-tailed and differences were considered statistically significant at P < 0.05. The Institutional Animal Care and Use Committee at GVSU approved this study.

RESULTS

House Sparrows

We pooled data from recently fledged males (n = 8) and females (n = 12) sparrows with those from of adult males (n = 50) and females (n = 51) because preliminary analyses revealed no significant differences between fledgling and adult sparrows in any digit length or digit length ratio (all P > 0.05). There were no significant differences in 4D length or 2D:4D among the four locations where sparrows were captured for either males or females (all P > 0.05) so we pooled data from the different locations. House sparrow males and females did not differ in right tarsus length (males: 18.27 ± 2.13 mm, n = 58; females: 18.87 ± 1.76 mm, n = 63; t = −1.71, df = 119, P = 0.09) allowing us to directly compare 4D length between the sexes.

Male and female sparrows did not differ in either 4D length (male: 8.69 ± 1.11 mm, n = 58; female: 8.84 ± 1.07 mm, n = 63: t = −0.76; df = 119, P = 0.45) or 2D:4D (Table 1). Effect size was small for between sex comparisons of both right and left 2D:4D (Table 1) suggesting that the differences between the sexes were of little biological significance. In addition, when male and female digit ratios were pooled, there was no laterality in 2D:4D (left side, 0.91 ± 0.11, n = 118; right side, 0.91 ± 0.11, n 118; paired-t, t = 0.17, df = 117, P = 0.87). Mean male bib area differed significantly among the four locations (ANOVA, F = 15.72, df = 3,43, P < 0.001). However, neither 4D length nor 2D:4D were significantly correlated with male bib area at any of the four locations (all P > 0.05).

Tree Swallows

Male and female tree swallows did not differ in right tarsus length (males: 14.63 ± 0.94 mm, n = 39; females: 14.51 ± 1.06 mm, n = 38; t = 0.51, df = 75, P = 0.61) allowing us to directly compare 4D length between the sexes. Male and female swallows did not differ in 4D length (male: 6.65 ± 0.95 mm, n = 34; female: 6.94 ± 0.72 mm, n = 38; t = −1.42, df = 70, P = 0.16) or 2D:4D (Table 1). Effect size was small for between sex comparisons of both right and left 2D:4D (Table 1) suggesting that the differences between the sexes were of little biological significance. When female and male digit ratios were pooled, left side 2D:4D (0.95 ± 0.10, n = 65) was significantly larger than right side 2D:4D (0.92 ± 0.10, n = 65) (paired-t, t = 2.71, df = 64, P = 0.009).

For males and females, 4D length and 2D:4D were not significantly correlated with either right wing or right tail fork length (all P > 0.05).

Budgerigars

Male and female budgerigars did not differ in either right 4D length (male: 11.00 ± 1.43 mm, n = 16; female: 10.47 ± 1.54 mm, n = 25; t = 1.12, df = 39, P = 0.27) or 2D:4D (Table 1). Effect size was medium for between sex comparisons for right 2D:4D and very small for left 2D:4D (Table 1). When female- and male-digit ratios were pooled, there was no laterality in 2D:4D (left side, 0.72 ± 0.13, n 19; right side, 0.75 ± 0.11, n = 19, paired-t, t = 0.99, df = 18, P = 0.034).

Male and female budgerigars did not differ in right wing length (male: 101.12 ± 4.12 mm, n = 17; female: 100.89 ± 3.99 mm, n = 27; t = 0.18, df = 42, P = 0.86). Right wing length was not significantly correlated with 2D:4D for either males or females (both P > 0.05).

Chickens

Chickens were sexually dimorphic for right tarsus length (male: 103.10 ± 5.63 mm, n = 12; female 87.68 ± 3.84, n = 12; t = 7.84, df = 22, P < 0.001). Therefore, we used the ratio of 4D/right tarsus length to control for sexual dimorphism in body size to compare 4D length between males and females. There was no significant difference between males (0.59 ± 0.04, n = 12) and females (0.60 ± 0.03, n = 12) in this ratio (t = -0.62, df = 22, P = 0.53). We did not detect sexual dimorphism in 2D:4D and effect size was small for between sex comparisons of both right and left 2D:4D (Table 1). When female and male digit ratios were pooled, there was no laterality in 2D:4D (left side, 0.87 ± 0.04, n = 24; right side, 0.87± 0.05, n = 24, paired-t, t = 0.64, df = 64, P = 0.53).

Chickens were sexually dimorphic for comb length (male: 54.83 ± 7.46, n = 12; female = 32.17 ± 5.49, n = 12; t = 8.48, df = 22, P < 0.001) but there was no relationship between comb length and 4D length or 2D:4D for either sex (all P > 0.05).

All Birds Together

In seven of eight (87.5%) comparisons in Table 1, males had a larger 2D:4D than did females suggesting a sexually dimorphic pattern in digit ratio.

After calculating average digit lengths and z-transforming the resulting digit length ratios, males had larger 2D:4D (0.143 ± 1.017, n = 105) than did females (−0.124 ± 0.972, n = 121) (t = 2.01, df = 224, P = 0.045) but the effect size was small (d = 0.26). Males also had significantly larger 2D:3D (0.147 ± 1.057, n = 111) than did females (−0.133 ± 0.930, n = 122) (t = 2.15, df = 231, P = 0.03); and again, the effect size was small (d = 0.28). ANOVA using transformed data showed that there were no significant species–sex interactions for 2D:4D (F = 0.07, df = 3, 218, P = 0.98)

Using data transformations, there were no significant differences between the sexes in 2D:4D within species (all P > 0.05).

DISCUSSION

We did not detect sexual dimorphism in 2D:4D when we separately analyzed digit ratio data from four species of birds from three taxonomic orders (Passeriformes, Psittaciformes, Galliformes). Effect size analyses showed that sex had little influence on 2D:4D in the birds in this study. However, repeatabilities of calculations of 2D:4D were low, therefore we may have failed to detect sexual dimorphism in 2D:4D.

Nevertheless, separately analyzing species produced results consistent with Forstmeier's (2005) and Romano et al.'s (2005b) findings of no sexual dimorphism in 2D:4D in zebra finches and ring-necked pheasants, respectively. In contrast, Burley and Foster (2004) found that 2D:4D was correlated with hatching order so that eggs laid earliest in clutches produced birds with smaller 2D:4D; female zebra finches decrease their allocation of androgens to eggs across the laying order (Gil et al., 1999). This result is consistent with Manning's (2002) prediction that prenatal exposure to testosterone reduces 2D:4D and suggests that androgens of maternal origin have the potential to affect digit ratio in birds. However, Forstmeier (2005) did not find a laying order effect on digit ratio in his much larger sample of zebra finches. Romano et al. (2005b) found that injecting testosterone into ring-necked pheasant eggs increased 2D:3D in females but did not affect 2D:4D in males or females. Furthermore, injecting eggs with estradiol resulted in males with lower 2D:4D ratios on their right feet (Saino et al., 2007). In our pooled sample, males had larger 2D:4D and 2D:3D than did females.

When we used transformed digit ratios to pool data from all four species, we found that males had larger 2D:4D and 2D:3D. In each case, effect sizes were small suggesting that sex had little influence on the digit ratio. Furthermore, there were significant differences among the species in all digit ratios, but no species–sex interactions suggesting that sex had little influence on digit ratio both within and among species.

Taken together, the directionality of digit ratio, when present, in birds is opposite to that found in some mammals (e.g., Brown et al., 2002a; Manning, 2002; Manning et al., 2003; Roney et al., 2004), but similar to that found in some reptiles (Chang et al., 2006; Rubolini et al., 2006; but see Lombardo and Thorpe, 2008).

The variations in digit ratio detected among species studied to date are not completely consistent with the phylogenetic constraint hypothesis (Chang et al., 2006) which predicts that birds should have digit ratios more similar to reptiles than to mammals because birds and reptiles share a more recent common ancestor (Feduccia, 1996). For example, larger male 2D:4D has been detected in some mammals (e.g., Roney et al., 2004), reptiles, (Rubolini et al., 2006; Chang et al., 2006), and birds (Burley and Foster, 2004; our pooled results).

There was no evidence of an association between 2D:4D and other sexually dimorphic characters in the three sexually dimorphic species (house sparrow, tree swallow, chicken) in our sample. Likewise, Garamszegi et al. (2007) failed to find any associations between 2D:4D and sexually dimorphic traits in collared flycatchers. In contrast, small 2D:4D has been found to be negatively correlated with the expression of some male secondary sexual characteristics in humans (Manning and Taylor, 2001; Manning, 2002, and references therein) suggesting that the effects of testosterone on sexual dimorphic traits may differ in birds and mammals (cf. Owens and Short, 1995).

Why are the patterns of digit ratio in birds different from those found in mammals? We propose several related First, between taxa variation in digit ratios may arise in at least two nonmutually exclusive ways (a) sex-specific patterns of steroid secretion during development may affect growth hormone secretion in different ways in different taxa (Rubolini et al. 2006) and (b) differences may arise from differential patterns of maternal deposition of steroids in eggs (e.g., Schwabl, 1993, 1996; Gil et al., 1999; Lovern et al., 2001; Lovern and Wade, 2001, 2003; Whittingham and Schwabl, 2002) which would result in differential exposure of male and female embryos to hormones (Rubolini et al., 2006). Either of these proximate mechanisms could result in the production of variation among taxa in digit ratios, but the second may be especially important.

Manning (2002) predicted that male mammals have smaller 2D:4D than do females. In most mammals, embryos are exposed to maternal estrogens during development and subsequent differentiation into male and female phenotypes is influenced by the presence or absence, respectively, of testosterone producing testes. Consequently, male embryos are exposed to higher levels of testosterone of embryonic origin than are female embryos possibly resulting in sexually dimorphic digit ratios (Manning, 2002). In contrast, bird embryos develop into males unless they are exposed to estrogens before hatching (Mittwoch, 1998; Smith and Sinclair, 2004; Adkins-Regan, 2007). Thus, in birds, the ovary is the dominant gonad secreting hormones that affect sexual differentiation into male or female phenotypes (Mittwoch, 1998; Adkins-Regan, 2007). In addition, the effects of hormones of embryonic origin on digit ratio may be attenuated by the effects of steroid hormones of maternal origin from the egg yolk (cf. Schwabl, 1993). This may obscure the sexually dimorphic patterns predicted by Manning (2002). Thus, the effects of hormone exposure on digit ratio in birds may be complicated by the patterns of development and exposure to hormones of both embryonic and maternal origin. In chickens, for example, the onsets of digit formation (Lillie, 1952) and the secretion of gonadal steroids (Eising et al., 2003) occur on about the fifth day of development. Thus digit formation would be influenced by steroids of both embryonic and maternal origin. However, the role of steroids of maternal origin on digit growth is likely to be complex because there are sometimes no differences at laying between the levels of androgens of maternal origin in eggs that produce either males or females (e.g., Schwabl, 1993; Eising et al., 2005; Pilz et al., 2005). Therefore, we strongly encourage more experimental studies like those performed by Romano et al. (2005b) and Saino et al. (2007) to elucidate the effects of manipulating hormone exposure on digit development.

Second, patterns of sexual dimorphism in the reptile lineage, which includes birds, may differ from those in mammals because of differences between these lineages in chromosomal sex determination. Females are the heterogametic sex in the reptile lineage, whereas males are the heterogametic sex in mammals. Genes on the sex chromosomes influence the development of the gonads (e.g., Sinclair et al., 1990) and thus ultimately the secretion of fetal sex hormones which, in turn, may then influence the expression of Hox genes during development (Block et al., 2000). Furthermore, alleles (e.g., SRY or others) on the sex chromosomes may directly affect hormone levels (Arnold, 2004). Consequently, the differences between the reptile and mammal lineages in the directionality of the sexual dimorphism in 2D:4D may be related to the differences between lineages in chromosomal sex determination.

Third, the difficulty in detecting sexual dimorphism in 2D:4D in birds may be because sex differences in digit ratios are generally smaller for toes than for fingers in primates (e.g., McFadden and Shubel, 2002). Studies of digit ratio in birds have focused, for practical reasons, on toe length. It may be possible to obtain data on digit length in wings from skeletal material or X-rays. However, the identity of digits in bird forelimbs is still not resolved (Burke and Feduccia, 1997; Burke et al., 1998; Chatterjee, 1998; Garner and Thomas, 1998; Wagner and Gauthier, 1999). Moreover, it is likely that extensive modifications to forelimb digits in birds related to adaptations for flight may obscure patterns related to prenatal exposure to androgens.

Fourth, strong natural selection for a perching foot in the first birds may have counteracted the effects of androgens on the development of digit ratio. Archaeopteryx lithographica, the oldest recognized bird in the fossil record, had anisodactyl feet with characteristics that suggest that they were primarily used for perching (Feduccia, 1993). Anisodactyly appears to be the primitive condition from which other toe arrangements in birds have evolved (Feduccia, 1996).

Fifth, it is unlikely that sexual dimorphism in 2D:4D in birds is associated with egg laying per se because Chang et al. (2006) and Rubolini et al. (2006) detected sexual dimorphic digit ratios in three species of egg-laying lizards. More importantly, experiments with ring-necked pheasants demonstrate that injecting steroid hormones into eggs affect digit ratios (Romano et al., 2005b; Saino et al., 2007).

Last, it is unlikely that the lack of clear patterns of 2D:4D in birds is due to the nonresponsiveness of digit development to steroid hormones. Observations (Burley and Foster, 2004) and experiments (Romano et al., 2005b; Saino et al., 2007) suggest that prenatal exposure to steroid hormones can affect digit ratio in birds (but see Forstmeier, 2005). However, the evolutionary significance of this effect is obscure especially because correlations between digit ratios and behavioral characteristics varied between successive generations in zebra finches (Forstmeier, 2005).

We think that natural selection for digits adapted for perching in the ancestors of birds is the most likely explanation for the inconsistent patterns of sexual dimorphism in digit ratio in birds. We suggest that data from several tetrapod species which cover a wide evolutionary distance could be obtained to help resolve the relationships between selection, prenatal exposure to androgens, and digit ratio. For example, adequate samples sizes could be easily obtained from tetrapods common in museum collections. Therefore, we encourage more studies of digit ratios in a wide variety of tetrapod taxa to more thoroughly test Manning's (2002) predictions and to detect instances in which natural selection for digit function may have “overridden” conserved developmental pathways.

Acknowledgements

We thank E. Adkins-Regan, W. Forstmeier, N. Saino, and anonymous reviewers for comments on previous versions of the manuscript. N. Rogness of the Statistical Consulting Center at GVSU provided statistical advice. J. Klomp and D. D'Amore assisted us in the field and laboratory. We also thank A. Anderson and K. Ellen for allowing us to measure their budgies, River Ridge Farm and John Ball Zoo, Grand Rapids, MI, for allowing us to trap sparrows on their properties, and Townline Poultry, Zeeland, MI, for supplying us with chicken carcasses.

Ancillary