Data from: Adaptive benefits of group fission: evidence from blue monkeys
Data files
May 03, 2025 version files 1.69 MB
Abstract
Permanent group fissions are thought to represent the tipping point at which a group has become too large and therefore splits into two, allowing for an evaluation of the consequences of living in too large a group and if fission can alleviate those costs. We first examined how adult female activity budgets (feeding, moving, resting) differed among periods surrounding (i.e., before and after) multiple fission events, accounting for seasonal variation, and using five mixed-effects beta regression models. We then assessed how rates of agonism differed among periods surrounding these fission events using two negative binomial models, one examining all agonistic interactions and one focusing on agonistic interactions that were lost. Our third analysis used a generalized linear mixed model to investigate a female’s likelihood of conception in a given month, based on her individual characteristics, which post-fission group size she joined, and whether that month fell before vs. after fission, vs. neither. Finally, we used a mixed effects Cox proportional hazards model to evaluate the relationship between infant survival, whether the infant’s mother joined the small vs. large post-fission group, and whether the month in which the infant was born fell before vs. after fission vs. neither. Here we present the three datasets used for these analyses, thus presenting individualized records of both behavioral and life history variables in relation to group fissions.
https://doi.org/10.5061/dryad.0cfxpnwbb
Description of the data and file structure
Permanent group fissions represent the tipping point where a group has become too large and thus splits into two, allowing for an evaluation of the consequences of living in too large a group and if fission can alleviate those costs. Our study investigated how female activity budgets, rates of agonism, likelihood of conception, and offspring survival changed before vs. after group fission. Accounting for seasonal variation, we found that females spent less time feeding in general and on preferred food items right before fission than during any other period. We also found that females were less likely to conceive in the two years leading up to fission, and that offspring survival was reduced for infants born up to two years before fission compared to those born after or during stable periods, with infants born one year before fission having the lowest survival rates. Here we present three datasets used in this analysis, one related to changes in activity budgets and rates of agonism, the second related to a female’s likelihood of conception, and the last describing infant survival.
Files and variables
Dataset 1 (AdaptiveBenefitsFission_BlueMonkeys_2025_Dataset1.csv)
Dataset 1 is set up to evaluate a female’s behavior in different periods surrounding a fission event. It includes (a) how much time a female blue monkey spends on three major activity categories (feeding, moving, resting) as a deviation from mean time spent performing that activity during the same time of year, and (b) how much time a female spends consuming particular food types (fruits and young leaves, which are the top items in the blue monkey diet at this site) as a deviation from mean time spent eating those food types during the same time of year. It also includes (c) a count of agonistic interactions in which the female was involved in a given period, broken down into total number of interactions and interactions she lost. Each female can have up to four rows of data per fission, one for each period of analysis surrounding the fission event. This dataset was used to complete the first two analyses in the paper: changes to activity budgets and rates of agonism. Each row includes the following variables, listed from left to right:
female: unique ID of each adult female present during any of the four periods of analysis for each of five fission events. Each female can have up to four rows of data for a given fission event, one per analysis period.
pre_group: the unique ID of the pre-fission group of the female.
period: coded 1- 4, 1 for the analysis period that fell a year prior to the onset of the fission, 2 for the period that fell immediately before the onset of the fission, 3 for the period that fell immediately after the fission completed, and 4 for the period that fell a year after the fission completed. Each period is 60 days long and is set relative to the fission event for the group entered in pre-group.
post_group: the daughter (post-fission) group of the female. There were two post-groups per fission event (pre-group ID). NAs in this column mean that the female died before joining a post-group.
postfission_group_size: coded small/large, “small” if the post-group was the smaller of the two daughter groups and “large” if the group was the larger of the two. NAs in this column mean that the female died before joining a post-group.
total_beeps: the total number of instantaneous activity records from focal samples of the female for the given period.
pct_feeding_s: the percentage of all activity records recorded during focal samples (total_beeps) in a given period when the female was ingesting food (any type) expressed as a deviation (in percentage points) from the mean time spent feeding during the same months that fall in the period.
pct_moving_s: the percentage of all activity records recorded during focal samples (total_beeps) in a given period when the female was locomoting expressed as a deviation (in percentage points) from the mean time spent moving during the same months that fall in the period.
pct_resting_s: the percentage of all activity records recorded during focal samples (total_beeps) in a given period when the female was resting expressed as a deviation (in percentage points) from the mean time spent resting during the same months that fall in the period.
pct_feeding_fruit: the percentage of all activity records recorded during focal samples (total_beeps) in a given period when the female was ingesting fruit. This is the raw data, which is not adjusted for seasonality.
pct_feeding_fruit_s: the percentage of all activity records recorded during focal samples (total_beeps) in a given period when the female was ingesting fruit expressed as a deviation (in percentage points) from the mean time spent feeding on fruit during the same months that fall in the period.
pct_feeding_youngleaves_s: the percentage of all activity records recorded during focal samples (total_beeps) in a given period when the female was ingesting young leaves expressed as a deviation (in percentage points) from the mean time spent feeding on young leaves during the same months that fall in the period.
agonism_losses: the count of agonistic interactions that the female lost (acted submissively) during the given period.
total_agonism: the count of all agonistic interactions in which the female participated (won, lost, or draw) during the given period.
Dataset 2 (AdaptiveBenefitsFission_BlueMonkeys_2025_Dataset2.csv)
Dataset 2 describes the reproductive status of all adult females October 1997 to December 2022. This dataset is set up to evaluate how a female’s likelihood of conception differs before vs. after fission vs. all other stable months. Each row of data represents a month during a female’s reproductive life. This dataset also includes characteristics of the female and time of year data.
The dataset includes the following variables, listed from left to right:
female: unique ID code for each reproductively active female in the study population during the study period.
group: unique ID code for each study group. The value here is the female’s group during a particular month.
month: The month (falling between October 1997 and December 2022) represented in this row of data. All the following columns’ values are for the month recorded in this column. Females will have as many rows of data as months during which they were reproductively active (minus the month of their first conception and any months in which their reproductive status could not be confirmed, see Methods). Expressed as the number corresponding to each calendar month.
year: the year in which the month fell. Expressed as a four-digit year.
reproductive_status: coded pregnant/conceived/gave birth/non-reproductive, “pregnant” if the female was pregnant the entire month, “gave birth” if she gave birth during that month, “conceived” if she conceived during that month, and “non-reproductive” if no other status applied. We assigned the month of conception by back-counting one average gestation length (176 days) from an offspring’s date of birth.
exposure: the number of days in the month on which the female could conceive. For months when females gave birth, this value was the number of days remaining in the month after the birth. For months in which females were not pregnant, the exposure term value was the number of days in the month. For months during which a female was pregnant for the whole month, the exposure term was 0.
conceived: coded 0/1, 0 if the female did not conceive during the month, 1 if she did conceive during that month.
time_since_last_conception: the number of months (count) elapsed since the female last conceived and the current month.
lactation_stage: represents the intensity of lactation a female was experiencing in a given month as a function of the age of her youngest infant: coded 1-5: 1 (infant age < 5 months), 2 (infant age 5-9 months), 3 (infant age 10-15 months), 4 (infant age 15-32 months), and 5 (infant age > 32 months or most recent infant died).
age: the age of the individual entered in female during the month recorded in month.
total_group_size: the average daily number of individuals (adult males, adult females, and immatures) in the study group in a given month.
fission_category_2_years: coded before/after/none, “before” if the month fell 24 months before the study group fissioned, “after” if the month fell in the 24 months after the group fissioned, and “neither” if it fell during any other month in the study period or if the group did not fission during the study period.
fission_category_1_year: coded before/after/none, “before” if the month fell 12 months before the study group fissioned, “after” if the month fell in the 12 months after the group fissioned, and “neither” if it fell during any other month in the study period or if the group did not fission during the study period.
fission_category_6_months: coded before/after/none, “before” if the month fell 6 months before the study group fissioned, “after” if the month fell in the 6 months after the group fissioned, and “neither” if it fell during any other month in the study period or if the group did not fission during the study period.
time_of_year: coded Dec-Feb/Mar-May/Jun-Aug/Sep-Nov, according to what month was entered in month.
postfission_group_size: coded small/large, “small” if the female joined the smaller post-fission group and “large” if she joined the larger post-fission group. NAs in this column mean that the month did not fall within 24 months before or after a fission event or the female died before joining a post-fission group.
Dataset 3 (AdaptiveBenefitsFission_BlueMonkeys_2025_Dataset3.csv)
Dataset 3 describes infant survival to two years of age for all offspring born in the study population between October 1997 and December 2022. This dataset is set up to evaluate how infant survival differs before vs. after fission vs all other stable months in the study period. Each row represents one infant and includes the following variables, listed from left to right:
infant: unique ID for each full-term offspring born (whether first seen alive or dead) during the study period.
date_of_birth: date of birth for the infant. Expressed as MM/DD/YYYY.
date_of_death: date of death for the infant. Expressed as MM/DD/YYYY. NA means the individual did not die during the study period.
mom: unique ID of the infant’s mother.
mom_age: the age of the infant’s mother at the time of its birth.
group: the unique ID code for each study group. This column’s value is the group that the infant was born into.
days_survived: count of the days the infant survived, censored at 730 days.
survive_two_years: coded yes/no, “yes” if the infant lived to its second birthday and “no” if it died before age 2 years.
fission_category_2_years: coded before/after/none, “before” if the infant was born in the 24 months before the study group fissioned, “after” if it was born in the 24 months after the study group fissioned, and “neither” if it fell during any other month in the study period or if the group did not fission during the study period.
fission_category_1_year: coded before/after/none, “before” if the infant was born in the 12 months before the study group fissioned, “after” if it was born in the 12 months after the study group fissioned, and “neither” if it fell during any other month in the study period or if the group did not fission during the study period.
fission_category_6_months: coded before/after/none, “before” if the infant was born in the 6 months before the study group fissioned, “after” if it was born in the 6 months after the study group fissioned, and “neither” if it fell during any other month in the study period or if the group did not fission during the study period.
peak_birth_season: coded yes/no, “yes” if the infant was born during the population’s peak birth season (December to March) and “no” otherwise.
postfission_group_size: coded small/large, “small” if the infant’s mom joined the smaller post-fission group and “large” if she joined the larger post-fission group. NAs in this column mean that the infant was not born within 24 months before or after a fission event or the mom died before joining a post-fission group.
Code/software
We ran this code in R Studio version 2024.04.2 and used the following packages: dplyr (v 1.1.4), tidyverse (v 2.0.0), ggplot2 (v 3.5.1), sjPlot (v 2.8.17), glmmTMB (v 1.1.10), DHARMa (v 0.4.7), coxme (v 2.2-22), and survival (v 3.6-4).
The datasets relate to seven fission events that occurred between 1999 and 2019 in the blue monkey population inhabiting the Kakamega Forest, western Kenya. We used data from all seven fissions for records of female conceptions and infant survival and data from the last five fissions only (2008 to 2019) for records of female behavior, because only these last five fissions occurred while the long-term monitoring protocol included focal animal follows of adult females, which allowed systematic recording of activity.
Throughout the study period, a team of trained observers monitored the study groups for all or part of a day on a near daily basis. All group members could be identified as individuals. Observers documented which individuals were present and whether any sub-grouping occurred, meaning that group members were separated into two parties that traveled and foraged separately for at least part of the day. They also recorded all observed agonistic interactions, noting winners and losers when one and only one animal (the loser) showed submission. Beginning in September 2006, the team also conducted systematic 30-minute focal animal follows of adult females, selecting subjects to maintain even sampling across females and across the morning (until 10:30 AM), midday (10:30 AM-14:30 PM) and afternoon (14:30 and later). During focal follows, observers recorded the subject’s activity at 1-minute intervals: main activity categories included feeding (if the subject ingested food on or within 2 sec of the minute mark), moving (involving hindlimb locomotion), and resting. Observers also noted the food item if the focal subject was feeding and the identity of any social partner. Observers recorded all occurrences of agonistic interactions involving the focal subject during focal follows; agonistic interactions between the same opponents were considered separate events if there was a lull in aggressive behavior for at least 30 seconds.
We used the census data to identify periods of sub-grouping. Specifically, we identified a sub-grouping period as when the group was split into spatially distinct parties on at least five days, and consecutive sub-grouping days were less than 14 days apart. We considered a fission to be complete when the two sub-groups had their first aggressive intergroup encounter. We designated four 60-day periods representing different times relative to each sub-grouping period. The earliest period was centered on the day that fell a year before the onset of sub-grouping. The last day of the second period fell immediately (a week) before the onset of sub-grouping, and the first day of the third period fell immediately (a week) after fission was complete. The fourth and latest period was centered on the day that fell one year after the date of fission.
We aggregated activity records from focal follows for each female in each of the four periods. We calculated individuals’ activity budgets for each period by dividing the total number of instantaneous records when a female performed a given activity by the total number of instantaneous records when she was a focal subject. We accounted for seasonal variation by calculating a population-wide mean percentage for a given activity for each month using all focal follows from 2006 to 2013. We then calculated the mean during the time of year matching each 60-day analysis period as a weighted mean based on the number of days of each month that matched the analysis period. Finally, we expressed the percentage of a female’s activity budget as a deviation in percentage points from the mean time spent on that activity during the same time of year.
To investigate how agonism rates varied by period, we aggregated all agonism that a female experienced during her focal samples in each period, breaking it down into total agonism and agonism losses. Agonistic interactions included aggressive (spatial displacements, threats, chases, contact aggression) and submissive (flee, cower, gecker, trill) behavior. Females did not need to be present in all four periods to be included in either analysis. However, we excluded females that were sampled for less than 6 hours in a given period, as these females were prone to having outlying data values.
To analyze likelihood of conception, we focused on females who were adults at any time from October 1997 to December 2022. Females that were already reproductively mature (i.e., had already conceived their first offspring) in October 1997 were included in the dataset beginning that month. Females that matured after October 1997 were added to the dataset starting the month after their first confirmed conception. For females that died during the study period, the last month we included in the dataset was 7 months before their death or the month of their last birth, whichever occurred later. All other females remained in the data set through December 2022.
We excluded the month of a female’s first conception because it had missing values for certain predictors, including time since last conception. Conceptions could be confirmed only if an offspring was born, whether it was first seen alive or dead (either stillbirth or peri-natal death). Therefore, the month of a female’s first conception fell 176 days before her first birth of a full-term infant (whether living or stillborn). For one female that had a miscarriage after her first confirmed birth, we omitted all months from seven months before the miscarriage to the month after the subsequent conception (because we could not confirm a value for the time since last conception for these months).
We assigned each adult female a monthly reproductive status (pregnant, gave birth, conceived, or non-reproductive). We categorized a female as “pregnant” if she was pregnant the entire month, “gave birth” if she gave birth during that month, “conceived” if she conceived during that month, and “non-reproductive” if no other status applied. We created three categorical variables to assess the influence of fission on probability of conception at six months, one year, and two years. We calculated time since last conception and maternal age to the nearest month. We classified lactation stage as one of five categories based on the age of her most recent surviving infant: 1 (infant age < 5 months), 2 (infant age 5-9 months), 3 (infant age 10-15 months), 4 (infant age 15-32 months), and 5 (infant age > 32 months). We also created an exposure variable that equaled the number of days in each month in which a female could conceive. For months during which females gave birth, this value was the number of days remaining in the month after the birth. Pregnant females, who took a value of 0, were excluded from the model of conception probability. We added a variable identifying which post-fission group a female ended up in for months falling within 2 years before or after a fission event.
For the infant survival analysis, we created three categorical variables to assess the influence of fission on infant survival, assigning each infant as being born before vs. after fission vs. neither, and using timescales of six months, one year, and two years to assess “before” and “after”. We used the infant’s mother’s age at the time of the infant’s birth and designated whether the infant was born during the peak birth season (December-March) or not. We added a variable identifying which post-fission group an infant’s mother ended up in for infants born two years before or after fission.