RESULTS
Male replacement, pregnant females, unweaned infants and their fortunes
We observed 18 male replacement events involving 16 OMUs and 15 adult males between January 2006 and March 2020 (Case #1 - #18,Table S1 ). Two events occurred when no females were pregnant or caring for unweaned offspring (#7 and #18). Eight pregnant females were present during male replacement events, four of which were attacked by new males, which subsequently resulted in two abortions (#1 and #3), one gave birth but the newborn was killed by the new male (#11), and one returned to stay with ousted male and gave birth and weaned her offspring (#17). All other pregnant females were not attacked by the new males and gave birth to their offspring.
Thirty-one unweaned infants were present following male replacements. Two of them and their mothers left the unit with the ousted male and survived. Although half of all remaining unweaned infants (15/29) were attacked by new males, only three infants (#2, #8 and #11) were killed by the first set of attack, and one more infant was killed after repeated attacks (#8). Therefore, except of four infanticides, most unweaned infants (27/31) present during male replacements are known to have survived throughout the mating season either because their mothers stayed with the ousted male, transferred to another social unit after an unsuccessful attack, or were tolerated by the new male (Fig 1 and Table S1) .
The difference in mortality of fetus and infants during male replacement compared to stable periods
Two abortions were observed during a total of 117 pregnancies outside of male takeover events. Therefore, there is a significant increase of abortions following male replacement (Fig 2 , 2/8 vs 2/117, Z-Score = 3.6212, p < 0.001). Twenty-three unweaned infant deaths were observed during a total of 121 births, with two infants who were present during times of male replacement dying for unknown reasons in winter (potentially because of low temperature and disease). We compared the difference of death rates between the infants and the fetus + infants present during times of male replacement and those present at other times. There is no significant difference of death rates between the infants present during times of male replacement and those present stable periods (6/31 vs 17/90, Z-Score = 0.3408, p > 0.05), and between the fetus + infants present during times of male replacement and those present stable periods (8/33 vs 19/92, Z-Score = 0.7102, p> 0.05).