After weaning, 4 sows died (1 from 6% CWG, 2 from 6% A-V blend, and 1 from 2% CWG treatments).
The remaining 387 sows were used for analysis of subsequent reproductive performance. When the response was binary (i.e., sows bred or sows not bred), the total number of successes were counted across all replicates, and a Poisson regression model (Ott and Longnecker,2001) was fitted to the number of successes, with source and level of fat, parity, and source and level of fat by parity interaction as fixed effects and the total number of sows as an offset variable. The SAS software GENMOD procedure (Littell et al., 2006) was used for the analysis.
From the 299 sows that farrowed in the next cycle, subsequent litter data were collected. Information included total pigs born, pigs born alive, stillborn pigs, and mummified pigs. Similar to feed intake and sow and litter performance, subsequent litter data were analyzed using a completely randomized design using the MIXED procedure of SAS. Degrees of freedom, least squares means, and standard errors were calculated applying