The Goldilocks effect: female geladas in mid-sized groups have higher fitness.

Elizabeth Tinsley Johnson,Amy Lu,Jacinta C Beehner,Noah Snyder-Mackler,Jacob A Feder,Thore J Bergman

doi:10.1098/rspb.2021.0820

Abstract

The cost–benefit ratio of group living is thought to vary with group size: individuals in ‘optimally sized’ groups should have higher fitness than individuals in groups that are either too large or too small. However, the relationship between group size and individual fitness has been difficult to establish for long-lived species where the number of groups studied is typically quite low. Here, we present evidence for optimal group size that maximizes female fitness in a population of geladas (Theropithecus gelada). Drawing on 14 years of demographic data, we found that females in small groups experienced the highest death rates, while females in mid-sized groups exhibited the highest reproductive performance. This group size effect on female reproductive performance was largely explained by variation in infant mortality (and, in particular, by infanticide from immigrant males) but not by variation in reproductive rates. Taken together, females in mid-sized groups are projected to attain optimal fitness due to conspecific infanticide and, potentially, predation. Our findings provide insight into how and why group size shapes fitness in long-lived species.

Highlights

Variability in group size within species reflects a balance between the costs and benefits of group living [1,2]
Our results show that females in mid-sized units have the highest fitness
Male takeovers are more frequent in larger units and are more likely to lead to infanticide when they do occur in smaller units

Summary

Introduction

Variability in group size within species reflects a balance between the costs and benefits of group living [1,2]. The dependent binary variable indicated whether a female conceived a surviving offspring during each female-month (273 surviving offspring, 996.5 female-years total, 188 females) We modelled this outcome variable as a function of the following predictors: female age (both the linear and quadratic term, to control for the known effects of female age on reproductive output, reviewed in [61]), number of females and the number of males. We modelled whether the infant died of each respective mortality cause (i.e. maternal death, infanticide and unknown) as a binary variable, using unit size (both linear and quadratic) as a fixed effect and unit as a random effect For these three analyses, we used only infants whose survival outcomes were known (n = 377; i.e. no right-censored offspring). All figures were constructed using ggplot2 [62], and prediction intervals were extracted from mixed-models using the effects package [63]

Results

Discussion

Findings

Accelerated reproduction is not an adaptive