All the Pups We Cannot See: Cannibalism Masks Perinatal Death in Laboratory Mouse Breeding but Infanticide Is Rare

SIMPLE SUMMARY: Perinatal mortality is a large problem in laboratory mouse breeding. Dead pups are often eaten by the adult mice. Pups cannibalised between birth and first husbandry check are likely not to be accounted for, leading to an underestimation of the number of pups are born and die. This study aimed at understanding to what extent perinatal mortality is underestimated in a breeding facility. The roles of cannibalism and infanticide (killing of a live pup) in pup mortality were also investigated. The results indicated that the standard pup-counting method by daily cage checks led to an underestimation of the number of pups that are born by 35% and those that die by 102% compared to data where daily checks were combined with video observations from which cannibalistic events could be recorded. Cannibalism of dead pups before the first check explained this inaccuracy in death counts, while infanticide was rare. Beyond considerations of animal welfare and ethics, and a conflict with the 3Rs principle, high perinatal mortality means that larger colonies of breeding animals are needed to supply animals in sufficient numbers for research. Paradoxically, the common practice of not disturbing cages around parturition conceals the extent of perinatal mortality but has also constrained the determination of causes of pup death. ABSTRACT: Perinatal mortality is a major issue in laboratory mouse breeding. We compared a counting method using daily checks (DAILY_CHECK) with a method combining daily checks with detailed video analyses to detect cannibalisms (VIDEO_TRACK) for estimating the number of C57BL/6 pups that were born, that died and that were weaned in 193 litters from trios with (TRIO-OVERLAP) or without (TRIO-NO_OVERLAP) the presence of another litter. Linear mixed models were used at litter level. To understand whether cannibalism was associated with active killing (infanticide), we analysed VIDEO_TRACK recordings of 109 litters from TRIO-OVERLAP, TRIO-NO_OVERLAP or SOLO (single dams). We used Kaplan-Meier method and logistic regression at pup level. For DAILY_CHECK, the mean litter size was 35% smaller than for VIDEO_TRACK (p < 0.0001) and the number of dead pups was twice lower (p < 0.0001). The risk of pup loss was higher for TRIO-OVERLAP than TRIO-NO_OVERLAP (p < 0.0001). A high number of pup losses occurred between birth and the first cage check. Analyses of VIDEO_TRACK data indicated that pups were clearly dead at the start of most of the cannibalism events and infanticide was rare. As most pups die and disappear before the first cage check, many breeding facilities are likely to be unaware of their real rates of mouse pup mortality.