Document S2: Supplementary Material for Johnston et al..
Horn length and circumference QTL models adjusted for body weight.
INTRODUCTION
In the main paper, we present QTL models for horn measurements without adjusting for body weight. We chose to do so because some animals had horn measurements, but body weight was unknown. Therefore, fitting body weight in our models would have resulted in a smaller dataset. Below we outline why the QTL for horn traits located at Ho is not a QTL for overall body size, but is instead is a horns-specific locus.
METHODS
This supplementary material presents results using the same methods in the main text of the paper, including body weight as a covariate. Only records where both horn length, base circumference and body weight were measured simultaneouslyhave beenconsidered. These constituted 696 records in 291 normal horned male sheep, of which 209 were genotyped.Body weight is mainly recorded during the annual August catch, with occasional measurements made during the rut (November) or after death (usually winter).
RESULTS AND DISCUSSION
Significance of body weight as a fixed effect: Weight had a significant contribution to variation in horn length (Wald test, F = 147.39, d.f = 1, p < 0.001) and base circumference (Wald test, F = 107.04, d.f. = 1, p < 0.001) when fitted as a covariate in a generalized linear model. The significance of body weight was also tested when fitting the animal model in ASReml2 by using the !DDF qualifier to calculate the approximate denominator degrees of freedom in an analysis of variance (Gilmour et al. 2006). Again, weight had a significant contribution to horn length (ANOVA, F1,568.8= 38.52, p 0.001) and horn base circumference (ANOVA, F1,579.9 = 47.27, p0.001).
Variance component estimation and QTL mapping:Fitting body weight as a covariate led to slightly higher estimates of heritability of horn length/base circumference and QTL effects (on horns), anda decrease in the proportion of variance explained by permanent environment effects (Table S3). This is consistent with the correlation between body weight and horns measurements being non-genetic. It is more likely that the covariance between body weight and horns is driven by permanent environmental effects.
Could some of the additive genetic variation in horn morphology be explained by body weight?
Although there was a phenotypic correlation between body size measurements and horn morphology (horn length: rP = 0.410, s.e. = 0.047; horn base circumference: rP = 0.349, s.e. = 0.048), the heritability of body weight within the normal horned males was negligible (h2 < 0.001), with variation in body weight mainly due to the permanent environment effect (c2 = 0.321, s.e. 0.045) and the residual effect (ε2 = 0.547, s.e. 0.047).
Could there be a body size QTL on at the position of Horns(Ho)?
If the horn length and base circumference variation at Ho was due to a QTL for overall body size (rather than specific to horns), then the horn morphology QTL should not be significant when body weight is fitted as a covariate. However, QTL were still detected at Ho(LOD = 2.60 and 1.43 for horn length and base circumference respectively), which is consistent with the horn morphology QTL being independent of body size QTL.Furthermore, a previous mapping experiment failed to identify QTL for body weight or leg length on chromosome 10 (Beraldi et al. 2007b) and conducting a scan of the Ho confidence interval in the current study found no evidence of a body weight QTL within normal-horned males (LRT < 0.02, d.f. = 1, p > 0.89).
Trait / Model / Mean (SD) / VOBS / VP / Additive Genetic(Heritability) / QTL / Permanent Environment / Year of Birth / Year of Growth / Residual
VA / h2 / VQ / q2 / VC / c2 / VB / b2 / VY / y2 / VΕ / ε2
HORN LENGTH / Polygenic / 319.07 / 20766 / 3011 / 1005 / 0.334 / NF / NF / 754.6 / 0.251 / 80.80 / 0.027 / 515.4 / 0.171 / 655.0 / 0.218
(144.1) / (309.1) / (383.4) / (0.118) / NF / NF / (312.3) / (0.109) / (45.54) / (0.015) / (238.9) / (0.067) / (47.74) / (0.027)
Polygenic (inc. Weight) / 2644 / 951.2 / 0.360 / NF / NF / 596.8 / 0.226 / 70.29 / 0.027 / 370.9 / 0.140 / 654.5 / 0.248
(255.0) / (344.0) / (0.119) / NF / NF / (275.4) / (0.109) / (41.44) / (0.016) / (180.5) / (0.060) / (47.97) / (0.029)
QTL / 3077 / 0.0001 / 0 / 1017 / 0.331 / 755.8 / 0.246 / 83.55 / 0.027 / 563.8 / 0.183 / 656.4 / 0.213
(330.4) / (0) / (0) / (289.2) / (0.083) / (204.5) / (0.073) / (47.10) / (0.015) / (256.0) / (0.070) / (47.87) / (0.027)
QTL (inc. Weight) / 2705 / 0.0005 / 0 / 1019 / 0.377 / 545.2 / 0.202 / 71.42 / 0.026 / 411.4 / 0.152 / 658.2 / 0.243
(274.5) / (0) / (0) / (262.0) / (0.081) / (173.6) / (0.069) / (42.30) / (0.016) / (193.8) / (0.062) / (48.28) / (0.030)
HORN BASE CIRCUMFERENCE / Polygenic / 150.75 / 762.78 / 324 / 130.9 / 0.404 / NF / NF / 73.25 / 0.226 / 11.49 / 0.036 / 24.55 / 0.076 / 83.81 / 0.259
(27.62) / (28.17) / (45.20) / (0.124) / NF / NF / (35.61) / (0.114) / (6.035) / (0.018) / (14.96) / (0.044) / (6.162) / (0.028)
Polygenic (inc. Weight) / 289.8 / 138.7 / 0.479 / NF / NF / 42.74 / 0.148 / 11.68 / 0.040 / 14.13 / 0.049 / 82.59 / 0.285
(24.62) / (41.05) / (0.120) / NF / NF / (30.40) / (0.108) / (6.108) / (0.021) / (10.06) / (0.034) / (6.099) / (0.031)
QTL / 321.1 / 34.24 / 0.107 / 74.68 / 0.233 / 89.29 / 0.278 / 11.27 / 0.035 / 27.40 / 0.085 / 84.20 / 0.262
(28.31) / (56.63) / (0.175) / (40.18) / (0.121) / (35.66) / (0.1158) / (5.959) / (0.018) / (16.04) / (0.047) / (6.201) / (0.029)
QTL (inc. Weight) / 286.2 / 31.85 / 0.111 / 85.18 / 0.300 / 57.75 / 0.202 / 11.27 / 0.039 / 16.84 / 0.059 / 83.32 / 0.291
(24.71) / (50.87) / (0.177) / (38.44) / (0.127) / (30.30) / (0.110) / (5.970) / (0.020) / (11.02) / (0.037) / (6.171) / (0.032)
TABLE S3. —Estimated variance components for the study traits for the polygenic and QTL model, with and without body weight as a fixed effect in an animal model. VOBS is the observed phenotypic variance and VP is the phenotypic variance defined as the sum of the variance components. Each random effect is reported as the variance component (VI) and effect size (i2) respectively, as described in the Materials and Methods. Numbers in parenthesis are the standard error unless otherwise stated.NF = random effect not fitted.
REFERENCES
Beraldi D, McRae AF, Gratten J, Slate J, Visscher PM, Pemberton JM (2007). Mapping quantitative trait loci underlying fitness-related traits in a free-living sheep population. Evolution61(6): 1403-1416.
Gilmour AR, Gogel BJ, Cullis BR, Thompson R (2002). ASReml User Guide Release 2.0. VSN International Ltd, Hemel Hempstead, HP1 1ES, UK.