ESTRUS DETECTION IN GROUP-HOUSED SOWS BY ANALYSIS OF DATA ON VISITS TO THE BOAR

This study deals with (automated) oestrus detection in a group-housing system where several sows may come into oestrus at the same time. Time-lapse video recordings were made from five groups of ten recently weaned sows. Data on both the frequency and duration of visits by these sows to a Detection Area in front of a boar-pen were collected. Checklists can be drawn up to enable the stockman to determine which sows should be examined for oestrus. We show that a formula combining the parameters frequency and duration can be used to detect sows coming into oestrus in time for them to be served. The total number of false-positive listings for the 50 sows used in this study was 21 for the first 10 days they stayed in the Service House. Using duration of visits as the only parameter was not sufficient to detect a minimum of 95% of the sows in time for them to be served. Using only frequency, it was possible to detect at least 95% of the sows in time for them to be served, but with a total number of 100 false-positive listings for this study. The system was improved by excluding data collected during periods of generally increased activity during the morning and the afternoon. The same system can also be used to detect at least 95% of the sows correctly at the first time they show the standing response. The total number of false-positive listings found for this study was 132. With an automated system based on the guidelines described in this paper it should be possible to detect sows that come into oestrus in time for them to be served and even as early as the time of the first standing response.