In avian species the fertile period has been defined as the interval after sperm deposition during which a female can lay fertile eggs. The length of the fertile period is dependent on sperm storage in the tubules at the utero-vaginal junction where the spermatozoa are released for upward transport towards the infundibulum for ova fertilization . The purpose of this selection experiment was to investigate what genetic progress could be made to extend the duration of fertility in the Brown Tsaiya duck. The selection was carried out using an animal model and the BLUP of breeding values. The selection experiment was continued up to 12 generations using the same methodology of selection already discussed in Cheng et al. . In addition the durations of fertility and hatchability were determined and their correlated responses to selection on F were analyzed. The selection responses were calculated, using the common method of calculating selection responses by taking the differences between the average phenotypic values for the S and C lines across the generations of selection [25, 33]. Sorensen and Kennedy  described an alternative way of estimating response to selection based on the mixed model approach, as the phenotypic trend can be further divided into genetic and environmental trends. We therefore estimated the genetic trends by averaging the multiple-trait BLUP animal model values for each trait in each generation and determined the differences between the S and C lines.
The measured selection responses and the calculated predicted genetic responses were found to be similar. This indicated the adequacy of the data representation model with no confounding with environmental trends and the accuracy of the genetic parameter estimates in the base population. Given the large variability in selection response, especially of H, we have chosen to discuss the predicted genetic response. The genetic progress in F measured by the predicted genetic response was significant i.e. 4.40 genetic standard deviations in total or 40% of the genetic standard deviation per generation. The correlated genetic progress in Dm and H was also significant, i.e. 4.89 and 3.56 genetic standard deviations in total, or 45% and 32% of the average genetic standard deviation per generation, respectively. The frequency of embryo mortality was not increased by selection. These results are consistent with the estimated genetic parameters, thereby showing a high genetic correlation between F and Dm (0.92), H (0.91) and between Dm and H (0.82). In contrast to results obtained in the chicken hen [34, 35] and according to the genetic parameter estimates, our results showed that selection on F seemed to be more effective in increasing H than direct selection of that trait. Brun et al.  reported heritabilities of 0.25 and 0.23 for F, 0.17 and 0.13 for H, and 0.27 and 0.16 for Dm in pure breeding INRA44 duck line and intergeneric crossbreeding, respectively. Our result can be explained by the fact that the heritability of F is greater than that of H (0.26 versus 0.19) and the genetic correlation between F and H is 0.91.
This study showed that the selection of F through 11 generations had major correlative effects on parameter τ of the logistic curves, which fitted the daily variations (d2-d15) in fertility rates (F/Ie) and hatchability rates (H/Ie). The S-C differences represented selection responses to the duration of fertility and hatchability which were correlated with the selection response of F. Selection for F modified the evolution of the fertility and hatchability rates, as a function of time after a single AI of the Tsaiya duck with pooled Muscovy semen mainly by increasing the time of half maximal fertility and hatchability rates. The largest increases in the fertility rates per day after single AI were between d5 and d11. Selection for F also had correlated effects on the maximum fertility rates, but these were smaller than the effect on fertility duration. Moreover, the fertility rate in the selected line was over 90% from d2 to d5 and above 80% until d8. The same tendencies were observed for changes in the evolution of hatchability rates, showing that embryo viability was not impaired. Consequently, in accordance with Brillard et al.  it is suggested that selection on F acted by increasing the storage capacity of spermatozoa, which remained able to fertilize the ova for longer. In addition, the increased duration of fertility when selecting on F was not deleterious to embryo viability. The overall fertility (F/Ie) and hatchability (H/Ie) rates at days 2–8 after AI were higher in the S line than in the C line. The embryonic viability rates in the C line (73.1%) and S line (73.0%), measured from the hatchability of fertile eggs (H/F), were not statistically different for G12 (36–42 weeks of age), confirming the results for G8 and G11 [38, 14, 15]. The differences in hatchability of fertile eggs (H/F) between the S and C lines over the 11 generations of selection were not statistically different either.
On the basis of the results of Tai et al. , a long-term selection experiment on F, with a selected and a control line, was begun in 1992. Analysis of this experiment after 11 generations of selection revealed a selection response for F (3.83 eggs), with correlated selection responses for increasing H (1.91 ducklings) and maximum duration of the fertile period (4 days), with no increase in embryo mortality rate. The genetic progress in F measured by the selection response was 2.77 genetic standard deviations or 39.6% of the genetic standard deviation per generation in G8 and 4.07 genetic standard deviations or 37% of the genetic standard deviation per generation in G12. The correlated selection response in Dm was also increased from 2.93 to 4.14 genetic standard deviations between G8 and G12. There was no increase in H in G12 compared to G8, due to an electric cut off problem in the incubator and M was increased. However there was a large variability of selection response in H. In G11 the selection response in H (2.79) was higher than in G8 (2.02). In G12 the correlated selection response on H measured at 36, 39 and 42 weeks of age (2.59) was a more relevant value.
The realized selection response for F can be compared with the theoretically expected one if selection has been done with the conventional combined selection index although that prediction of response is valid in principle for only one generation of selection. The expected selection response on F, according to the accuracy of the combined selection index on F, would be higher than the realized one. That can be explained by variation of response due to random genetic drift and sampling errors . In addition there was a loss in selection intensity especially because some animals with a high-predicted breeding value were discarded from reproduction to avoid full sib and half sib mating.
Selection to extend the fertile period was shown to be feasible [14, 15]. The present results confirm the absence of a selection plateau in responses up to the 11th generation. Selection was effective in increasing the number of ova that could be fertilized after a single AI with pooled Muscovy semen, and consequently the number of eggs able to develop a viable embryo. These changes considerably increased the maximum duration of the fertile period, and the physiological effects now need to be investigated. Selection brought about a correlated increase in fertility and hatchability rates according to egg set, especially for days 2–8 after AI, thereby demonstrating the feasibility of selection for a single AI per week in this strain of laying duck. This did not produce a concomitant increase in the rate of embryonic death, (previously thought to occur in fowl) which would have impaired the benefits of selection. Thus fertilization of the ova would seem to be a key point in the intergeneric crossbreeding of ducks [39, 40]. Nevertheless, the total mortality rate in relation to the number of fertile eggs was high (23 to 36% (G11)). It would therefore be useful to continue this selection experiment and study the long-term effects on fertility and embryo viability. A better understanding of the consequences of selection was obtained by comparing the fertility rate curves  according to the number of days after AI in the S and C lines. The genetic variability of viability in ducks needs to be determined to evaluate the possibilities of improving mule embryo viability. The results obtained here might depend on the use of Brown Tsaiya, which is a laying duck. Nonetheless, it should be feasible to select for an extension of the fertile period in meat-type ducks such as the Peking breed, which is being used effectively as parents for commercial mule ducks. Furthermore, research can now be focused on ways to improve the viability of the hybrid mule duck embryo.