Data from: Multiple motivations: Agonistic coalitions and interventions in blue monkeys (Cercopithecus mitis stuhlmanni)
Data files
Dec 26, 2024 version files 686.30 KB
-
Cords-Rotter_agonistic_support_data.csv
672.01 KB
-
README.md
14.30 KB
Abstract
Primates are known for forming agonistic coalitions, but most data come from species in which agonism occurs frequently and rank predicts fitness. We analyzed coalitions and interventions in wild blue monkeys (Cercopithecus mitis), in which both agonism and third-party involvement are rare, and in which rank does not predict fitness. Data came from a long-term study in the Kakamega Forest, western Kenya, spanning 12 years and 12 groups. These data revealed that intervening animals both supported winners and defended losers, and that coalition partners nearly always prevailed over their opponent. Adult females were joiners (intervening in the initial agonism) and juveniles were coalition-recipients disproportionately, while opponents were disproportionately adults, especially males. Joiners were most likely to support the original contestant who was winning (vs. unclear outcome or losing), but also favored the smaller of the original contestants (vs. same-sized or larger) and the one to whom they were more closely related. A subset of the data, in which ranks were known for the two original contestants, showed higher odds of joining the higher- vs. lower-ranking original opponent. In high risk interventions (coalition recipient was losing, joiner was smaller than opponent), the preferences for more related and smaller opponents were magnified. This Dryad contribution constitutes the data set used for these analyses.
README: Data from: Multiple Motivations: Agonistic Coalitions and Interventions in Blue Monkeys (Cercopithecus mitis stuhlmanni)
https://doi.org/10.5061/dryad.fxpnvx11q
Description of the data and file structure
In social animals, cooperation can take the form of agonistic support, when two or more group-mates join forces in an aggressive interaction against a third party. While these coalitions are often successful, intervening in a dispute requires energy and likely increases the risk of injury relative to standing by. We therefore expect animals to intervene in agonistic conflicts strategically, minimizing their risks and costs and maximizing their benefits. This data set allows one to examine predictions from this hypothesis from a naturally living population of monkeys, focusing on the choice of which contestant to support when a ‘joiner’ intervenes in their dispute. The data set also allows one to compute descriptive statistics about these coalitions and interventions.
DATASET 1 (Cords-Rotter agonistic support data.csv)
In this data set, each line of data represents one coalition (i.e., two animals (the joiner and recipient) siding with each other against a third (the opponent). The following variables, listed from left to right, are included for each case. NA indicates missing data.
sort: an integer that uniquely identifies each row of data
group: the social group in which the coalition occurred
date: the date on which the coalition occurred
clearjoiner: yes/no (y/n), ‘y’ if dyadic aggression was witnessed before any joiner became involved, and ‘n’ if two (or more) animals were jointly aggressing against a third when first detected.
joiner_ID: unique code for an individual joiner. When clear_joiner is ‘y’, then the joiner is an animal who intervened on behalf of the recipient against the opponent. When clear_joiner is ‘n’ the distinction between joiner and recipient is arbitrary. When entry starts with ‘UID’, it means that the animal was not identified as a named individual, but in some cases was identified to age sex class; for example, ‘UIDaf’ refers to an unidentified adult female, and ‘UIDsj’ to an unidentified small juvenile. ‘UIDs’ refers to multiple unidentified animals acting as joiners, but only one entry was made for these cases as the exact number could not be specified in field records. A question mark preceding a name indicates uncertainty about the joiner’s ID: in some cases, however, its age-sex class was known from field notes and assigned accordingly.
support_or_defend: when clear_joiner is coded ‘n’, this variable is always coded as ‘support’. When clear joiner is coded ‘y’, this variable is coded ‘support’ if the joiner took the side of a recipient who was already winning against the opponent, ‘defend’ if the joiner defended a recipient (by directing aggression to the opponent) when the recipient was losing at the time of the intervention, and ‘def-sup’ if the winner-loser status of the recipient was not clear at the time of the intervention.
recip_ID: unique code for individual recipient. When clear_joiner is ‘y’, the recipient is the animal who the joiner supported. When clear_joiner is ‘n’, the distinction between joiner and recipient is arbitrary. When entry starts with ‘UID’, it means that the animal was not identified as a named individual, but in some cases was identified to age sex class; for example, ‘UIDaf’ refers to an unidentified adult female, and ‘UIDsj’ to an unidentified small juvenile. ‘UIDs’ refers to multiple unidentified animals acting as joiners, but only one entry was made for these cases as the exact number could not be specified in field records. A question mark preceding a name indicates uncertainty about the joiner’s ID: in some cases, however, its age-sex class was known from field notes and assigned accordingly.
opp_ID: unique code for the opponent. When clear_joiner is ‘y,’ opponent is the animal to whom the joiner (and sometimes also the recipient) directed aggressive behavior. When clear_joiner is ‘n,’ both joiner and recipient directed aggression to the opponent. When entry starts with ‘UID’, it means that the animal was not identified as a named individual, but in some cases was identified to age sex class; for example, ‘UIDam’ refers to an unidentified adult male, and ‘UIDsj’ to an unidentified small juvenile. A question mark preceding a name indicates uncertainty about the joiner’s ID: in some cases, however, its age-sex class was known from field notes and assigned accordingly.
joiner_own_01rank: dominance rank of joiner, on 0-1 (low to high) scale. See Methods and associated paper for details of how rank was assessed. Ranks were first assigned for females in the year in which they turned 5 years old, and then annually throughout their lives. Males of all ages were not assigned a rank (NA). Animals who were not fully identified did not have an assigned rank (NA).
recip_own_01rank: dominance rank of recipient, on 0-1 (low to high) scale. See Methods and associated paper for details of how rank was assessed. Ranks were first assigned for females in the year in which they turned 5 years old, and then annually throughout their lives. Males of all ages were not assigned a rank (NA). Animals who were not fully identified did not have an assigned rank (NA).
opp_own_01rank: dominance rank of opponent, on 0-1 (low to high) scale. See Methods and associated paper for details of how rank was assessed. Ranks were first assigned for females in the year in which they turned 5 years old, and then annually throughout their lives. Males of all ages were not assigned a rank (NA). Animals who were not fully identified did not have an assigned rank (NA).
all_w_own_rank: yes/no, coded ‘y’ if joiner, recipient and opponent all had an assigned rank, and ‘n’ otherwise.
coalition_type: assigned based on relative ranks of joiner, recipient and opponent when all were known. ‘alldown’ means that joiner and recipient outranked the opponent. ‘allup’ means that the opponent outranked the joiner and recipient. In bridging coalitions, the opponent’s rank was intermediate to those of joiner and recipient. ‘Bridge w high joiner’ means that joiner outranked the other two individuals, and the recipient was the lowest-ranking, whereas ‘bridge w low joiner’ means that the recipient outranked the other two, and the joiner was the lowest-ranking. Data missing (NA) if ranks were not assigned to all three animals.
recip_winning: If clear_joiner = y, this variable takes values of ‘win’ if recipient was prevailing over opponent (i.e., opponent was showing submission) at the time of the intervention, ‘unclear’ if neither party was prevailing, and ‘lose’ if recipient was behaving submissively to opponent at time of intervention. If there was no clear joiner, coded as missing data (NA).
successful: yes/no, coded only if there was a clear joiner, else coded as missing data (NA). Coded ‘y’ if intervention was successful meaning that: (1) the joiner defended a losing recipient and stopped the aggression it was receiving, or (2) the joiner supported a winning recipient (or one for whom the win/loss status was unclear) leading to the opponent’s retreat. Coded ‘n’ if not successful, meaning that: (1) the joiner defended a losing recipient but failed to stop the aggression it was receiving, or (2) the joiner supported a winning recipient who ended up receiving aggression from its opponent.
joiner_size_rel_to_opp: joiner’s size relative to the opponent, coded as ‘bigger’, ‘same’ or ‘smaller.’ Relative size was based on the two animals’ age-sex classes. Specifically, from largest to smallest, these were adult males, adult females, large juveniles, medium juveniles, small juveniles, and infants. Missing if either joiner or opponent was not fully identified.
primary: When clear_joiner = yes, this variable is coded ‘y’ if individual in Joiner_ID was the first single animal to join in an initially dyadic dispute and ‘n’ otherwise. If two (or more) animals were the first to join but joined simultaneously, neither was assigned as “the first” and this variable was coded ‘n’. NA if clear_joiner = 'no’.
primaryplus: When clear joiner = yes, this variable is coded ‘y’ if individual in Joiner_ID was the first single animal to join in an initially dyadic dispute, OR if it was the first to join in simultaneously with another individual. In contrast to ‘primary’, this filter allows one to capture cases in which multiple individuals were the first to join in simultaneously. NA if clear_joiner = ‘no’.
kinship_joiner_recip: maternal kinship class (e.g. mother-daughter, brother-sister, first cousins) of joiner and recipient
kinship_joiner_opp: maternal kinship class (e.g. mother-daughter, brother-sister, first cousins) of joiner and opponent
kinship_recip_opp: maternal kinship class (e.g. mother-daughter, brother-sister, first cousins) of recipient and opponent
joiner_agesex: agesex class can take values of ‘admale’ (adult male), ‘adfem’ (adult female), ‘lj’ (large juvenile, 5 years or older but has not yet reproduced if female, has not yet emigrated from natal group if male), ‘mj’ (medium juvenile, 3-5 years old), ‘sj’ (small juvenile, 1-3 years old) and ‘infant’ (up to 1 year old). Data missing (NA) when joiner was not sufficiently identified.
joiner_sex: coded ‘m’ (male), ‘f’ (female) or ‘u’ (unsexed)
recip_agesex: agesex class can take values of ‘admale’ (adult male), ‘adfem’ (adult female), ‘lj’ (large juvenile, 5 years or older but has not yet reproduced if female, has not yet emigrated from natal group if male), ‘mj’ (medium juvenile, 3-5 years old), ‘sj’ (small juvenile, 1-3 years old) and ‘infant’ (up to 1 year old). Data missing (NA) when recipient was not sufficiently identified.
recip_sex: coded ‘m’ (male), ‘f’ (female) or ‘u’ (unsexed)
opp_agesex: agesex class can take values of ‘admale’ (adult male), ‘adfem’ (adult female), ‘lj’ (large juvenile, 5 years or older but has not yet reproduced if female, has not yet emigrated from natal group if male), ‘mj’ (medium juvenile, 3-5 years old), ‘sj’ (small juvenile, 1-3 years old) and ‘infant’ (up to 1 year old). Data missing (NA) when opponent was not sufficiently identified.
opp_sex: coded ‘m’ (male), ‘f’ (female) or ‘u’ (unsexed)
01kinship_joiner_recip: coded 1 if joiner is related to recipient, 0 if joiner is unrelated to recipient. Coded NA if joiner-recipient kinship was unknown.
choseA1: randomly generated binary variable used to assign recipient and opponents as Anim1 and Anim2 for joiner’s choice in models. Coded only for rows with clear joiners, else missing (NA).
Anim1: identity of individual (either recipient or opponent) assigned (randomly) as Animal1
A1rank: rank of Animal1 (0-1, low to high, scale)
A1agesex: age-sex class of Animal1. Can take values of ‘admale’ (adult male), ‘adfem’ (adult female), ‘lj’ (large juvenile, 5 years or older but has not yet reproduced if female, has not yet emigrated from natal group if male), ‘mj’ (medium juvenile, 3-5 years old), ‘sj’ (small juvenile, 1-3 years old) and ‘infant’ (up to 1 year old).
kinclass_JA1: maternal kinship class (e.g. mother-daughter, brother-sister) of joiner and Animal1
kinR_JA1: average R value for the kinship class entered in kinclass_JA1
Anim2: identity of individual (either recipient or opponent) assigned (randomly) as Animal2
A2rank: rank of Animal2 (0-1, low to high, scale)
A2agesex: age-sex class of Animal2. Can take values of ‘admale’ (adult male), ‘adfem’ (adult female), ‘lj’ (large juvenile, 5 years or older but has not yet reproduced if female, has not yet emigrated from natal group if male), ‘mj’ (medium juvenile, 3-5 years old), ‘sj’ (small juvenile, 1-3 years old) and ‘infant’ (up to 1 year old).
kinclass_JA2: maternal kinship class (e.g. mother-daughter, brother-sister) of joiner and Animal2
kinR_JA2: average R value (coefficient of relatedness) for the kinship class entered in kinclass_JA2
A1winning: ‘yes’, ‘no’ or ‘unclear’ to indicate whether A1 was behaving submissively (i.e. losing) prior to the intervention: ‘yes’ means A1 was winning, A2 behaving submissively; ‘no’ means A2 was winning, A1 behaving submissively; ‘unclear’ means neither A1 nor A2 was behaving submissively, so there was no clear winner. Coded only for rows with clear joiners, else missing (NA).
A1vsA2Rdiff: equals kinR_JA1 minus kinR_JA2. Can take negative values if Animal2 is more closely related to the joiner than is Animal1. Coded only for rows with clear joiners, and non-missing values for both kinR_JA1 and kinR_JA2, else missing (NA).
A1sA2size: Animal1’s size relative to Animal2, could be coded ‘bigger’, ‘same’ or ‘smaller.’ Relative size was based on the two animals’ age-sex classes. Specifically, from largest to smallest we included adult males, adult females, large juveniles, medium juveniles, small juveniles and infants. Coded only for rows with clear joiners, but could be a missing value (NA) if either Animal1 or Animal2 was not sufficiently identified (UID).
A1vsA2rankdiff: equals A1_rank minus A2_rank. Can take negative values if A2 ranks higher than A1. Coded only for rows with clear joiners.
recip_and_opp_own_rank_1st_joiners: filters (‘y’) for cases in which (i) both recipient and opponent had a known rank and (ii) primary = ‘yes’.
high_risk: filters (‘y’) for cases in which (i) primary = ‘yes’, (ii) the joiner was smaller than the opponent (based on comparison of their age-sex classes) and (iii) the recipient was losing its dyadic contest with the opponent when the joiner intervened. Else coded ‘n’.
Methods
The study population of blue monkeys (Cercopithecus mitis stuhlmanni) inhabited the Kakamega Forest, western Kenya. During the study period (June 2006 - December 2018), a team of trained observers monitored multiple monkey groups on a near daily basis, moving around the group to conduct a daily census of its members, and conducting focal animal follows of adult females in a systematic way. They noted agonism whenever it occurred, which could be both ad libitum and as part of the focal animal follows carried out on all adult females; they also noted all coalitions and interventions. From these field records, we compiled a list of coalitions and interventions, which both involve one animal aiding another in an agonistic context. We distinguished interventions as a type of coalition in which observers witnessed one animal intervening in an initially dyadic dispute, in contrast to other “in progress” coalitions in which two individuals might be simultaneously threatening a third when first observed. In a few interventions, the first intervention involved two or more animals, who took one side simultaneously, so it was not possible to identify a single first “joiner.” Most of the analyses we reported in our associated paper focused on primary coalitions, i.e., on the first single animal to take a side in an ongoing dyadic dispute. We dubbed this individual the “joiner”, the animal helped the “recipient”, and their common target the “opponent.”
We used data on maternal kinship, known from pedigree information back to 1979, to assess the relatedness between pairs of animals in the joiner, partner and opponent trio. We assumed that adult males resident in heterosexual groups were unrelated to the females in those groups, as males invariably disperse from natal groups before breeding in this species. We assigned age-sex classes to all individuals sufficiently well identified. We assessed dominance rank based on winner:loser matrices compiled for each group over a calendar year. Specifically, we included all observed dyadic agonism in which one and only one animal showed submission. A few matrices were compiled from somewhat shorter or longer periods if a group split in two during the year. We ran these matrices through the program Domicalc (Schmid & DeVries 2013) to extract a unique rank ordering in a given group on an annual basis.
For primary interventions, we ran several models to investigate which of the original two contestants the joiner sided with. We first randomly assigned the recipient and opponent, the original opponents in the dyadic agonistic encounter, as Animal1 and Animal2. We then structured our analysis to investigate whether the first single joiner sided with Animal1 as a function of (1) whether Animal1 had been winning the original dyadic contest, (2) how big Animal1 was relative to Animal2, (3) whether the joiner was more closely related to Animal1 than to Animal2, and (4+5) the age-sex classes of Animal1 and Animal2, respectively. The data set includes columns expressing these predictor variables.
*Reference: Schmid, V.S., & deVries, H. (2013). Finding a dominance order most consistent with a linear hierarchy: an improved algorithm for the I&SI method. Animal Behaviour, 86, 1097-1105. http://dx.doi.org/10.1016/j.anbehav.2013.08.019