1
1
2 A POLYMORPHISM IN THE ESTROGEN RECEPTOR GENE
3 EXPLAINS COVARIANCE BETWEEN
4 DIGIT RATIO AND MATING BEHAVIOUR
5
6 APPENDIX
(a)
Chr4A: 6,416,086 6,447,982
3’ 5’
ARmicro1 ARmicro2
(b)
Chr3: 56,300,102 56,480,907
3’ 5’
ESR1micro1 ESR1micro3
ESR1micro2
7
8 Supplementary Figure 1. Exon-intron structure and location of microsatellite markers in
9 (a) the zebra finch androgen receptor (AR) gene and (b) the zebra finch estrogen receptor
10 α (ESR1) gene. Chromosomal positions are from the zebra finch July 2008 assembly.
11
2
12
13
14 Supplementary Figure 2. Illustration of the putative linear relationship between fetal
15 testosterone levels and digit ratio. Random data were generated for 1000 females and
16 1000 males according to the sex-specific means and standard deviations for fetal
17 testosterone and digit ratio, respectively. Means and SDs of ln-transformed testosterone
18 levels were calculated from raw data extracted from Abramovich (1974) and Rodeck et
19 al. (1985). The study of Reyes et al. (1974) was discarded due to many testosterone
20 values being below the detection limit. Data for 15 females (gestational weeks 11-18)
21 showed a mean of 292ng/l (after ln-transformation: mean±SD = 5.52±0.62), while data
22 for 44 males of same age showed 2,248ng/l (mean±SD = 7.50±0.67), yielding an effect
23 size of d=3.07. This may be an underestimate of the true difference, since these values
24 contain measurement error. Note that measurements of amniotic testosterone levels show
3
25 less of a sex difference (e.g. Lutchmaya et al. 2004), but these are only rough proxies for
26 the true fetal testosterone levels (see Rodeck et al. 1985). As means and SDs for digit
27 ratios we used 0.971±0.034 for females and 0.954±0.035 for males, yielding an effect
28 size for the sex difference of d=-0.482 (mean values of 73 studies with high-quality
29 measurements, taken from Hönekopp & Watson 2010). Assuming that a straight line goes
30 through these male and female means (bold line), we first drew the x-values for all
31 individuals from the respective standard normal distributions and then assigned them y-
32 values according to this regression line (y=1.018-0.0086x). In a second step we added
33 random noise to the y-value of every individual (drawn from a normal distribution) such
34 that the expected intra-sex SDs in digit ratio of 0.034 and 0.035 were reached. In the
35 depicted example, sampling error led to minor deviations from the simulated values: the
36 effect sizes for the sex differences in testosterone and digit ratio are d=2.97 and d=-0.483
37 respectively. The observed across-sex correlation is r=-0.291, the average within-sex
38 correlation is r=-0.178, values that also depend on sampling error. Analytically the
39 expected within-sex correlation can be shown to equal the ratio of the two effect sizes
40 (r=-0.483/2.97=-0.163). This is because the slope of an ordinary least squares regression
41 line equals the correlation coefficient times the ratio of the SDs in y and x, respectively:
42 b=r*(SDy/SDx). Alternatively, the slope can be expressed as the ratio of the two sex
43 differences b=∆y/∆x. Solving the two equations for r yields r=(∆y/SDy)/(∆x/SDx) which
44 is just the ratio of the two effect sizes.
45 Note that our calculation also rests on the assumption that digit ratio is affected by the
46 testosterone levels present during weeks 11-18 of embryo development. Malas et al.
47 (2006) found that fetal digit ratio showed a significant sex difference (higher in females),
4
48 but neither male nor female digit ratio showed a significant temporal change from week 9
49 to end-of-term. Thus is may be that digit ratio is determined before the peak in
50 testosterone in mid-trimester, i.e. at a time of potentially smaller sex difference in
51 testosterone levels (no measurements available). If so, the association between fetal
52 testosterone and digit ratio could potentially be stronger than our simulation suggests.
53 This, however, would not be consistent with the findings by Berenbaum et al. (2009),
54 (see §4 of main text).
55
56
5
57 Supplementary Table 1. Primer information on the five microsatellite loci studied.
58 Chromosomal positions are from the zebra finch July 2008 assembly. Ta = annealing
59 temperature
Locus Position Repeat Primer Ta
(midpoint) motif
ARmicro1 Chr4A: GGAT 5’-GAGAAACAAATTCCCCAGCA-3’ 57
6,436,964
5’-TGCCACTGGAGTTGGTACTG-3’
ARmicro2 Chr4A: GGAT 5’-CTCTGTCCACACTGCGCTAA-3’ 57
6,440,160 5’-CAGGCTGAGGTCTGACAGC-3’
ESR1micro1 Chr3: TA 5’-ATTTCACCTGCTGGGATGAG-3’ 57
56,386,640 5’-TCCAAGGCTGAAACTTGACA-3’
ESR1micro2 Chr3: TA 5’-TTTTAATCCTTGTTTCTTTCATGTC-3’ 57
56,447,177 5’-AAAAGTTTACAATCACAGGGATGC-3’
ESR1micro3 Chr3: TA 5’-GCWTTTTAAGAGTRCCAYCAG-3’ 57
56,449,371 5’-AAGTATCARGAAACWAAATGTAGCT-3’
60
61
62
6
63 Supplementary Table 2. General information on the AR and ESR1 haplotypes found in
64 our study population. Allele calls are approximate sizes of PCR products in base pairs
65 from the two AR and the three ESR1 microsatellites (same order as in Suppl. Table 1).
66 Triple zeroes refer to null alleles. The table also indicates the frequency of the haplotypes
67 in the entire sample, the number of informative full-sib families with varying numbers of
68 that haplotype, as well as the number of individuals in those informative families. The
69 effect is the estimated regression slope of digit ratio over the number of copies of the
70 respective haplotype. Standard errors of slopes are given with corresponding z and P-
71 values (before Bonferroni correction). Bold print indicates significance after Bonferroni
72 correction.
Haplotype Allele calls Frequency Families Individuals Effect SE z P
AR_1 000-382 41 13 73 -0.01270 0.00678 -1.87 0.070
AR_2 254-366 23 7 32 -0.01571 0.00956 -1.64 0.105
AR_3 286-382 30 8 42 -0.00910 0.00791 -1.15 0.240
AR_4 288-382 102 28 139 0.00490 0.00459 1.07 0.285
AR_5 289-362 34 7 28 -0.00141 0.00813 -0.17 0.869
AR_6 291-408 101 32 149 0.00020 0.00459 0.04 0.978
AR_7 293-370 77 25 110 0.00401 0.00579 0.69 0.483
AR_8 296-373 6 2 5 0.00087 0.03016 0.03 0.970
AR_9 296-378 49 12 62 -0.01188 0.00741 -1.60 0.148
AR_10 298-000 182 53 234 0.00113 0.00328 0.35 0.729
AR_11 298-378 258 55 276 -0.00006 0.00271 -0.02 0.983
AR_12 299-386 12 5 20 0.00964 0.01234 0.78 0.414
AR_13 299-390 142 27 161 -0.00717 0.00360 -1.99 0.046
AR_14 301-378 804 129 674 0.00329 0.00178 1.80 0.065
AR_15 301-382 10 4 14 -0.02655 0.01234 -2.15 0.037
AR_16 301-400 117 35 178 -0.00328 0.00383 -0.86 0.388
AR_17 305-396 330 60 308 -0.00012 0.00305 -0.04 0.968
AR_18 318-378 13 3 19 -0.00770 0.00892 -0.86 0.413
AR_19 322-378 41 7 39 0.00758 0.00875 0.87 0.363
AR_20 342-373 32 12 55 0.00024 0.00714 0.03 0.986
ESR1_1 186-000-225 175 49 218 -0.00419 0.00301 -1.39 0.164
ESR1_2 190-231-228 48 14 59 0.00303 0.00757 0.40 0.696
ESR1_3 190-233-228 91 24 137 0.00500 0.00463 1.08 0.277
ESR1_4 192-231-228 33 9 44 -0.00382 0.00752 -0.51 0.595
ESR1_5 194-000-221 33 9 48 -0.00338 0.00623 -0.54 0.592
ESR1_6 194-231-228 212 52 236 -0.00463 0.00344 -1.35 0.178
ESR1_7 198-223-221 701 120 600 -0.00786 0.00188 -4.20 0.00003
ESR1_8 200-221-220 418 104 530 -0.00069 0.00222 -0.30 0.757
ESR1_9 200-225-221 69 19 101 0.00393 0.00558 0.70 0.478
ESR1_10 202-231-220 343 66 332 0.01465 0.00273 5.37 1.4*10-7
ESR1_11 202-231-228 45 15 57 0.00690 0.00690 1.00 0.323
ESR1_12 204-247-224 230 53 274 0.00297 0.00324 0.92 0.357
ESR1_13 212-227-221 17 6 34 0.01626 0.00846 1.92 0.076
73