Back

Changes in Social Position Predict Survival in Bottlenose Dolphins

Rankin, R. W.; Foroughirad, V. J.; Krzyszczyk, E. B.; Frere, C. H.; Mann, J.

2022-08-26 animal behavior and cognition
10.1101/2022.08.25.505273 bioRxiv
Show abstract

1Social bonds and social structure are important features of animal systems that impact individual fitness. Few studies have examined how temporal dynamics in individual social bonds predict fitness outcomes. This is critical to understand given the high variation in types of social structures and strategies within populations of social mammals, both across time and among individuals. If individually-differentiated social bonds are important, it might be the change in bonds which has fitness consequences, not the absolute number of bonds nor their strength. We investigated how network dynamics predict survival in a wild population of bottlenose dolphins using a 35-year longitudinal study. In particular, we were interested in two sex-specific measures of the "widowhood effect", as well as a more general investigation into the predictability of mortality from changes in higher-order social network metrics. We used two inferential frameworks to provide complimentary evidence for or against hypotheses; namely, a gradient-boosting predictivist approach with relative importance measures, and an inclusion probabilities approach based on stability-selection. We found evidence of a widowhood effect among males but not females. Surprisingly, the most robust predictor of survival was closeness-centrality, whereby the loss of closeness centrality preceded death. This finding, that absolute network position may not be as relevant to survival as changes in network status, is consistent with a large social-psychological literature in humans on the impacts of loss of social capital, dissolution of close friendships, and loss through death of partners (widowhood effect). This study highlights the critical nature of social connections and how its disruption can be a matter of life and death.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.