Abstract # 13128 Event # 36:

Scheduled for Thursday, August 9, 2018 11:00 AM-11:15 AM: (Chula Vista ) Oral Presentation


USE OF REGRESSION TREE MODELING TO PREDICT PAIR SUCCESS IN RHESUS MACAQUES (MACACA MULATTA)

A. Heagerty1,2, C. Johnson3 and K. Coleman1,2
1Oregon National Primate Research Center, 505 NW 185th Ave, Beaverton, OR 97006, USA, 2Oregon Health & Science University, 3Portland State University
line
     

Social housing is integral to the psychological well-being of research macaques, but not all pairs of individuals are socially compatible. Pair introductions inherently risk stress and injury for animals involved, particularly when partners are not compatible. Refining the process of selecting partners reduces these risks. We applied regression tree modeling to determine demographic (age, sex) and social history variables (time since group housing, pairing history) that predicted introduction success (co-housed for ?2 weeks post-introduction) in 843 rhesus macaque pairs. The regression tree method generates a decision-tree, revealing a hierarchy of variable importance. In our data, the use of protected contact during the introduction was associated with lower rates of success (18.6% versus 44% success). Among pairs with protected contact, prior successful introductions was associated with current success (62.8% success for animals with > 2.5 prior successes versus 32.8% for animals with < 2.5 prior successes). Finally, among animals with prior pair success, older animals were less likely to form compatible pairs than younger animals (40% success for animals >15 years versus 66.2% success for animals <15 years). Interestingly, prior pair success and age were not predictive for pair introductions in which protected contact was not used. The regression tree method is a useful tool for incorporating multiple factors into the pair selection process, and can be applied to other aspects of behavioral management decision making.