Een the subject of intensive breeding programs. For instance, the Churra breed has experienced a 15?0 increase in milk production during the last 25 years (Churra Breeding Association web, http://www.anche.org). In the light of these facts, we expected to find selective sweeps MS023 biological activity related with meat vs milk production in our dataset. When we built a population tree based on SNPs mapping to the three selective sweeps, we did not observe a clustering of the Churra and Latxa dairy breeds, though they were located in close positions (Supplementary Fig. S2). Consistently, local trees based on SNPs that mapped to the Oar3 and Oar6 selective sweeps did not show a clustering of Churra and Latxa. In contrast, both breeds grouped together in the local tree based on SNPs located within the Oar13 selective sweep. Moreover, the analysis of the allele frequencies of SNPs mapping to the Oar3, Oar6 and Oar13 selective sweeps did not reveal any meaningful pattern (Supplementary Fig. S3). These inconclusive results could be due to the limited power and the stringency of our experiment. We may have missed many selective sweeps that did not reach statistical significance due to the moderate sample size employed in our study or because they were not simultaneously identified with BayeScan and hapFLK. Genetic heterogeneity amongst breeds, where distinct mutations have similar effects on milk yield or growth, could be another reason. It is also possible that the selective sweeps we have detected do not have any relationship with meat or milk production but with other traits (e.g. morphology, adaptation, reproduction, disease resistance) that we did not take into consideration in our selection analysis. A fourth factor could be that artificial selection for meat and dairy traits has mainly evolved through polygenic adaptation, shifting the allele frequencies of hundreds or thousands of loci instead of fixing novel mutations with major phenotypic effects. Finally, the methods used by us are good at detecting ongoing or recently completed selective sweeps but they have difficulties in identifying ancient sweeps that ended a long time ago40. Though we have found patterns of variation on Oar3, Oar6, and Oar13 that are compatible with the occurrence of selective sweeps, it is difficult to envisage which set of phenotypes were really targeted by selection. Indeed, intensive selection of Spanish sheep breeds, as Churra and Latxa, for milk production is relatively recent (it began 2? decades ago) and genetic exchanges between dairy and non-dairy populations may have taken place, thus obscuring the effects of selection. Importantly, several of the selective sweeps detected with BayeScan and hapFLK contained genes encoding transcriptional regulators with effects on body size (e.g. HGMA2 on Oar3 and LCORL and NCAPG on Oar6). This phenotype experienced a substantial reduction during the early times of domestication and subsequently increased as a consequence of artificial selection for growth rate. Changes in the selection pressure conferring a TAPI-2 cost higher biological efficacy to a mutation that was previously deleterious are expected to generate hard sweep signatures41. Our finding, however, is difficult to interpret because the set of breeds employed in the current work do not differ substantially in terms of body size, weight or stature. Such cryptic selective sweeps have been also observed in cattle41, and so far their biological significance remains unknown. Noteworthy, neutra.Een the subject of intensive breeding programs. For instance, the Churra breed has experienced a 15?0 increase in milk production during the last 25 years (Churra Breeding Association web, http://www.anche.org). In the light of these facts, we expected to find selective sweeps related with meat vs milk production in our dataset. When we built a population tree based on SNPs mapping to the three selective sweeps, we did not observe a clustering of the Churra and Latxa dairy breeds, though they were located in close positions (Supplementary Fig. S2). Consistently, local trees based on SNPs that mapped to the Oar3 and Oar6 selective sweeps did not show a clustering of Churra and Latxa. In contrast, both breeds grouped together in the local tree based on SNPs located within the Oar13 selective sweep. Moreover, the analysis of the allele frequencies of SNPs mapping to the Oar3, Oar6 and Oar13 selective sweeps did not reveal any meaningful pattern (Supplementary Fig. S3). These inconclusive results could be due to the limited power and the stringency of our experiment. We may have missed many selective sweeps that did not reach statistical significance due to the moderate sample size employed in our study or because they were not simultaneously identified with BayeScan and hapFLK. Genetic heterogeneity amongst breeds, where distinct mutations have similar effects on milk yield or growth, could be another reason. It is also possible that the selective sweeps we have detected do not have any relationship with meat or milk production but with other traits (e.g. morphology, adaptation, reproduction, disease resistance) that we did not take into consideration in our selection analysis. A fourth factor could be that artificial selection for meat and dairy traits has mainly evolved through polygenic adaptation, shifting the allele frequencies of hundreds or thousands of loci instead of fixing novel mutations with major phenotypic effects. Finally, the methods used by us are good at detecting ongoing or recently completed selective sweeps but they have difficulties in identifying ancient sweeps that ended a long time ago40. Though we have found patterns of variation on Oar3, Oar6, and Oar13 that are compatible with the occurrence of selective sweeps, it is difficult to envisage which set of phenotypes were really targeted by selection. Indeed, intensive selection of Spanish sheep breeds, as Churra and Latxa, for milk production is relatively recent (it began 2? decades ago) and genetic exchanges between dairy and non-dairy populations may have taken place, thus obscuring the effects of selection. Importantly, several of the selective sweeps detected with BayeScan and hapFLK contained genes encoding transcriptional regulators with effects on body size (e.g. HGMA2 on Oar3 and LCORL and NCAPG on Oar6). This phenotype experienced a substantial reduction during the early times of domestication and subsequently increased as a consequence of artificial selection for growth rate. Changes in the selection pressure conferring a higher biological efficacy to a mutation that was previously deleterious are expected to generate hard sweep signatures41. Our finding, however, is difficult to interpret because the set of breeds employed in the current work do not differ substantially in terms of body size, weight or stature. Such cryptic selective sweeps have been also observed in cattle41, and so far their biological significance remains unknown. Noteworthy, neutra.
