No evidence for epistasis
Failure to detect significant epistasis for the evaluated traits was unexpected given that other studies working on similar traits detected epistatic QTL pairs for body weight at early growth stages and for Gompertz curve parameters [4, 7, 18, 19]. Most approaches to detect epistasis are subject to a high level of false positive results  and the detection of only one suggestive epistatic QTL pair could be due to the stricter thresholds enforced in this analysis compared to those reported by other studies, the low density of marker coverage or the relatively small size of the population.
Bodyweight QTL at Specific Ages
Body weight QTL at 3 weeks on chromosomes 1, 4, 13 and Z (Table 1) were similar to those detected earlier in a similar population raised as broilers for maximum growth . In a previous report, chromosomes 1, 2, 3, 6 and 11 were shown to harbour QTL for body weight at 46 days  and also appear in the list of significant and suggestive QTL for weight at 6 weeks of age in this study (Table 1). The positions of the QTL for 6 week weight on chromosome 3 at 235 cM and 161 cM on chromosome 4 are similar to the QTL at 252 and 149 cM respectively on chromosomes 3 and 4 reported by Jacobsson et al. . Most of the QTL for this trait are similar to those identified on chromosomes, 1, 2, 4, 6, 8, 13 and Z in a parallel broiler study .
QTL detected in the latter study for body weight at 9 weeks on chromosomes 1, 2, 4, 6, 8, and 13 are similar to those at 12 weeks of age. Positions for the QTL on chromosomes 4 and 13 respectively were at 177 and 15 cM in the earlier report and at 177 and 7 cM in this study. These are probably the same QTL because of the high correlations expected between body weight at 9 and 12 weeks of age. The significant 12 week body weight QTL on chromosomes 1, 4 and 27 are at similar locations as carcass weight QTL in a related broiler experiment .
QTL for body weight at 6 weeks on chromosomes 2, 3 (two locations), 4 and 6 were also identified at 48 weeks of age. Whereas body weight at two ages includes a part-whole component and the phenotypic correlation was only 0.52 (Additional file 2: Table S2), the results emphasise the importance of early growth in determining adult body weight.
Growth rate QTL
QTL for Gr3-6 on chromosomes 1, 2, 3, 4 and 13 in this study were similar to those reported for growth from 2 to 4 weeks on chromosomes 1 and 2 [4, 23]; and the QTL for Gr6-12 on chromosomes 1 and 3 are similar to those previously published for growth from 6 to 12 weeks .
QTL for Gompertz parameters
The Laird form of the Gompertz equation is a function of initial growth rate, relative growth rate at the point of inflection (the rate of exponential decay of the relative growth rate) and body weight at time t0 compared to the original Gompertz model which is a function of mature body weight . It was chosen to maximise the number of data points available for analysis (requiring the estimation of only 3 parameters) and provided an excellent fit to the data (results not shown).
The Gompertz parameter estimates (Additional file 1: Table S1) fall within the range reported in the literature: WA: 2483 – 5698 g, L: 0.0908 – 0.141 g/d, K: 0.0224 – 0.031, Ti: 42.2 – 63 d, W0: 39.8 – 64 g [14, 25–27]. The similar locations of significant and suggestive QTL for body weight and Gompertz parameters (Table 4) confirm assertions made earlier that these parameters are genetically determined and can be exploited to improve traits through selection [1, 8].
Whereas sexual maturity occurs earlier in the male-line broilers than the White Leghorn layers used in this experiment , there was evidence of only one significant QTL for the point of inflection of the growth curve. The broiler allele of this QTL decreased the point of inflection by 1.8 d or 3.6 d in the homozygous state and if the three significant and suggestive effects from Table 3 are summed the effect is substantial (almost 7 d or 14 d in the homozygous state).
We chose to use a constant 14 h light (L): 10 h dark (D) photoschedule rather than use a typical 8 L:16D during rearing gradually increasing to, for example, 16 L:8D during lay starting when the birds are between 16 and 22 weeks of age. This was done in order to avoid compromising the growth of the birds by artificially photostimulating them at the same age when individuals would be at different physiological states caused at least in part by genetic differences. This practice therefore ensured that all the birds received the same photostimulation and that growth per se was not affected by differences in the reproductive responses to increasing photoperiods. It also ensured that growth was not constrained by limiting opportunity for ad libitum feed consumption to only 8 h per day. Significant QTL for mature body weight (WA) were detected on chromosomes 2, 4, 8 and 27 while suggestive QTL were detected on chromosomes 3, 7, 9 and 15. Asymptotic or mature body weight QTL have been reported by Le Rouzic et al. (2008) on chromosomes 2 and 27 and additional QTL that were not detected in our study on chromosome 1, 6 and 11 . A similar QTL for age at the point of inflection was detected on chromosome 11 in both studies whereas suggestive QTL for this trait on chromosomes 2 and 4 contrast with significant QTL on chromosomes 1, 12 and 20 . No genome wide significant QTL were detected for W0, K or L. W0 may be determined more by egg size and therefore maternal QTL than by QTL inherited by the chick. Genetic variation for K and L may exist but be explained more by the other parameters, particularly mature weight. Changing growth rate while maintaining acceptable hatch and mature weights may be a desirable goal in commercial meat production systems and growth curves have been modified by differential selection on early growth in chickens and quail [28, 29]. Selection based on growth curve parameters is therefore likely to be effective but difficult to implement because of the length of time required to obtain body weights at older ages on which to base parameter estimates. The results suggest that relatively few areas of the genome may contain the genes largely responsible for controlling growth curves suggesting that whole genome selection based on high density SNP chips could ameliorate this problem. However, in contrast to the successful selection experiments noted above, Ibanez and Blasco  suggest that genomic selection to change the growth curve will be difficult and require constant re-evaluation of the associations between the SNPs and the genes determining curve parameters.
The architecture of growth QTL
Examination of Table 4 suggests that relatively few chromosomal locations affect growth to an extent that they can be detected in this population. Taking the large confidence intervals into consideration (Tables 1, 2, 3), the results suggest that there were at least 15 such locations: 2 on each of chromosomes 1, 2, 3 and 4, one each on chromosomes 6, 8, 9, 11 and Z with an additional 2 on microchromosomes 11 and 27. These are generally consistent with other experiments and provide a smaller list of QTL in the search for causative mutations compared with the list of QTL for body weight identified at different ages such as in Table 1.
Evidence of a QTL at 166 cM on chromosome 4 for mature body weight had the largest additive effect and explained the highest proportion (9.5%) of the phenotypic variation. The results generally confirm earlier observations about the critical role of chromosome 4 in controlling growth and other traits of economic importance [3, 31, 32].
QTL detected for growth rate intervals (Table 2) were generally similar to those for body weight at the corresponding ages (Table 1). Both the body weight and growth rate approaches identified more significant QTL than analysis of the Gompertz curve parameters but all methods identified a significant QTL for adult body weight (WA) on chromosomes 2, 4, and 8.
More QTL were detected for growth before sexual maturity than for later growth and suggest that genetic variation for growth is more important during early life (Tables 1 and 2). This may in part be associated with differential development of the reproductive organs and fat deposition as the birds approach sexual maturity and by changes in reproductive status with increasing age. These effects are likely to decrease the apparent importance of growth QTL whereas those QTL associated with reproductive senescence or fatness may become more important and overshadow the growth QTL identified during the rearing of the birds. The latter are more likely to be associated with growth of the lean tissue mass (muscle, bone and nervous tissues) and after peak rates of lay by QTL for fatness. Alternatively the environmental variance may become larger with age and time in the cages. These observations are consistent with the lack of correlation between WA and growth from 48–72 weeks (Additional file 2: Table S2). Further research will be necessary to determine the relative role of non-growth QTL on variation in adult weight including differences in fatness.
Age specific body weight QTL were detected for each growth stage which supports earlier observations that there are different genes and gene actions involved in growth at different developmental stages [1, 4, 8, 9, 33]. Chromosomes 3, 4 and 8 had QTL involved with body weight throughout the lifetime of the birds. Chromosome 11 harboured QTL involved mainly in very early growth and those on chromosome 1, 13 and Z from 3 to 24 weeks age. QTL on chromosomes 15, 27, and 28 mainly affected growth from 24 weeks. The few QTL that were detected at older ages (48 and 72 weeks) compared to earlier periods may reflect large environmental effects, particularly with respect to individual differences in the age that egg laying started to decline or cease altogether, and the accompanying changes in fat deposition, as discussed above.
Moderately high negative correlations between early body weight and age at inflection of the growth curve are consistent with previous research on the age at the onset of puberty which showed that large body weight QTL were associated with early onset of egg laying . Thus large, faster growing birds will tend to mature earlier and therefore the age at inflection of the growth curve will also be earlier than in slower growing, smaller birds.