ArticlePDF Available

Kinship influences sperm whale social organization within, but generally not among, social units

  • Pacific Science Enterprise Centre

Abstract and Figures

Sperm whales have a multi-level social structure based upon long-term, cooperative social units. What role kinship plays in structuring this society is poorly understood. We combined extensive association data (518 days, during 2005–2016) and genetic data (18 microsatellites and 346 bp mitochondrial DNA (mtDNA) control region sequences) for 65 individuals from 12 social units from the Eastern Caribbean to examine patterns of kinship and social behaviour. Social units were clearly matrilineally based, evidenced by greater relatedness within social units (mean r = 0.14) than between them (mean r = 0.00) and uniform mtDNA haplotypes within social units. Additionally, most individuals (82.5%) had a first-degree relative in their social unit, while we found no first-degree relatives between social units. Generally and within social units, individuals associated more with their closer relatives (matrix correlations: 0.18–0.25). However, excepting a highly related pair of social units that merged over the study period, associations between social units were not correlated with kinship (p > 0.1). These results are the first to robustly demonstrate kinship's contribution to social unit composition and association preferences, though they also reveal variability in association preferences that is unexplained by kinship. Comparisons with other matrilineal species highlight the range of possible matrilineal societies and how they can vary between and even within species.
This content is subject to copyright.
Cite this article: Konrad CM, Gero S, Frasier T,
Whitehead H. 2018 Kinship influences sperm
whale social organization within, but generally
not among, social units. R. Soc. open sci. 5:
Received: 8 June 2018
Accepted: 6 August 2018
Subject Category:
Biology (whole organism)
Subject Areas:
kin selection, social structure, cooperation,
matrilineality, relatedness, cetaceans
Author for correspondence:
Christine M. Konrad
Electronic supplementary material is available
online at
Kinship influences sperm
whale social organization
within, but generally not
among, social units
Christine M. Konrad1, Shane Gero2, Timothy Frasier3
and Hal Whitehead1
Department of Biology, Dalhousie University, 1355 Oxford Street, Halifax, Nova Scotia,
Canada B3H 4J1
Department of Zoophysiology, Institute for Bioscience, Aarhus University, C.F. Møllers Alle
Aarhus 8000, Denmark
Department of Biology, Saint Mary’s University, 923 Robie Street, Halifax, Nova Scotia,
Canada B3H 3C3
CMK, 0000-0002-6737-4520; SG, 0000-0001-6854-044X;
TF, 0000-0002-3055-0199; HW, 0000-0001-5469-3429
Sperm whales have a multi-level social structure based upon
long-term, cooperative social units. What role kinship plays in
structuring this society is poorly understood. We combined
extensive association data (518 days, during 20052016) and
genetic data (18 microsatellites and 346 bp mitochondrial
DNA (mtDNA) control region sequences) for 65 individuals
from 12 social units from the Eastern Caribbean to examine
patterns of kinship and social behaviour. Social units were
clearly matrilineally based, evidenced by greater relatedness
within social units (mean r¼0.14) than between them (mean
r¼0.00) and uniform mtDNA haplotypes within social units.
Additionally, most individuals (82.5%) had a first-degree
relative in their social unit, while we found no first-degree
relatives between social units. Generally and within social
units, individuals associated more with their closer relatives
(matrix correlations: 0.18– 0.25). However, excepting a highly
related pair of social units that merged over the study period,
associations between social units were not correlated with
kinship ( p.0.1). These results are the first to robustly
demonstrate kinship’s contribution to social unit composition
and association preferences, though they also reveal
variability in association preferences that is unexplained by
kinship. Comparisons with other matrilineal species highlight
the range of possible matrilineal societies and how they can
vary between and even within species.
&2018 The Authors. Published by the Royal Society under the terms of the Creative
Commons Attribution License, which permits
unrestricted use, provided the original author and source are credited.
1. Introduction
Cooperative societies are pervasive in the animal kingdom [1– 3]. For these systems to evolve and persist,
the benefits to cooperating individuals must outweigh the costs [1,4]. Costly cooperative behaviours
between kin are typically explained in terms of kin selection [5,6], which predicts that individuals
maximize their ‘inclusive fitness’ by helping relatives. This hypothesis, however, cannot explain
cooperation between non-relatives, and often fails to explain observed variation in cooperation
between relatives [1]. In such cases, other mechanisms in lieu of kin selection, or in addition to it, are
required to explain seemingly altruistic behaviours. Another frequently considered mechanism is
reciprocal altruism, in which individuals exchange favours that have a fitness cost [7,8]. However,
despite much focused attention on this mechanism, relatively few examples have been firmly
demonstrated [9]. Instead, many cases of cooperation may be driven by processes involving by-
product benefits, without reciprocation of costly investments [10,11]. For example, if the very existence
of more group members is beneficial, in and of itself (e.g. through safety in numbers), and if group
members are difficult to gain or replace, individuals can benefit from helping raise and protect the
offspring of others in their group—a process referred to as group augmentation [12,13].
To disentangle potential mechanisms driving cooperative behaviours, long-term studies of social
relationships and behaviour are required, together with comprehensive genetic sampling for kinship.
These types of datasets are rare among mammals, particularly among marine mammals.
The sperm whale (Physeter macrocephalus) provides an interesting case study of social structure and
cooperation because it has a multi-level cooperative social structure [14]. Female and juvenile sperm
whales live in ‘social units’ that are stable over a time frame of years [15,16], from which males
disperse before sexual maturity to live primarily solitarily or with other males [17]. Social units
sometimes join together to form temporary ‘groups’, which can last hours to days [16]. Social units
have only been observed to form groups with other social units that are members of the same ‘clan’
[18,19], with clan being a higher level of social structure composed of social units that share socially
learned behaviours, including distinguishable vocal repertoires [20].
The evolution and persistence of cooperative social units and groups in sperm whales have not been
explicitly examined, in part owing to the difficulty of conducting the necessary behavioural studies on
this long-lived, nomadic and deep-diving species. Calf care, specifically communal defence against
predators, is hypothesized as the primary force driving and maintaining social units [17,21], but it is
unclear how these cooperative behaviours evolved. Sperm whale social units are often described as
matrilineally based [14,22,23], which makes kin selection a logical hypothesis for explaining
cooperation. Yet, the degree to which social units are matrilineal is poorly understood.
Owing to the long-term observations required to confidently delineate long-term social units, kinship
has typically been studied at the level of temporary groups, which can contain multiple social units
[2426]. Genetic data on social units have been published for only a few social units to date [26–29],
with little or no support for matrilineal social units [26,28,29], except in one well-studied social unit in
the Caribbean [27]. Most of these assessments considered matrilineality generically and as a dichotomy:
that social units are either matrilineal or not [2628]. In conventional wisdom, ‘matrilineal’ would refer
to social units where most females (or all offspring, in species without male dispersion) remain, for life,
with their mothers and other close female relatives. However, examining the degree to which social
units are composed of maternal relatives (even if not exclusively so) can deepen our understanding of
the role of kinship in shaping patterns of cooperation within sperm whale social units.
If kin selection is a driving force for cooperation, we would also expect rates of association among
individuals to vary depending on their degree of relatedness. However, in the few cases where social
behaviour has been explicitly examined in relation to kinship, results have been mixed, with kinship
and social association positively correlated in one study [27], while no clear correlation was found in
two other studies [28,30]. These past studies have been limited by small sample sizes (e.g. a single
social unit [27,29]), coarse measures of association (e.g. individuals within 8.3 km of each other [28])
or short-term observations (e.g. a single week [29]). Additionally, it is unknown whether association
preferences between social units [31] relate to kinship.
Here, through a decadal study of well-known sperm whale social units, we are able to lessen these
limitations and address questions regarding the role of kinship in sperm whale social structure with
new depth and detail. In this study, we examine patterns of kinship and social behaviour using data
for 65 individuals from 12 sperm whale social units from the Eastern Caribbean. We explicitly
consider three possible categories along a gradient of matrilineality. We address three primary R. Soc. open sci. 5: 180914
questions: (i) to what degree are social units matrilineal, (ii) do rates of association between individuals
within social units correlate with relatedness, and (iii) does kinship between social units predict
association preferences?
2. Methods
2.1. Field methods
Fieldwork was carried out in an area of approximately 2000 km
, off the leeward, western coast of
Dominica, in the Caribbean Sea (15.58N; 61.58W) from 2005 to 2016 as a part of a longitudinal
research project on sperm whale behaviour [15]. Annual field seasons ranged from two to four
months in duration, and occurred between January and June, using various research platforms (total
effort: 518 days).
Sperm whales were located and followed, visually by observers on deck during daylight hours, as
well as acoustically using hydrophones up to 24 h a day [15]. Photographs were taken of the trailing
edge of flukes of juveniles and adults [32] and of the dorsal fins of calves [33] for individual
identification. In conjunction with these identification photographs, we recorded observations of
associations of individuals in clusters [15], with ‘cluster’ being defined as in box 1, as groupings
of individuals at the surface in close proximity to each other (less than 40 m) with coordinated
behaviour [16].
We used dip nets to opportunistically collect sloughed skin found floating within the flukeprints of
individual whales or clusters of whales [37]. In 2015 and 2016, we also collected biopsy skin samples
from specific individuals, to fill known gaps in our sample set. We used a 90 lb draw weight
crossbow and bolts with 2.5 cm long tips with 0.5 cm circumferences. (See [38] for details.) Skin
samples collected from 2005 to 2010 were stored in ethanol (at a concentration of 70% or greater), and
samples collected from 2011 onwards were stored in a 20% DMSO solution saturated with salt [39].
2.2. Identifications
As in Gero et al. [31], identification photographs were assigned quality ratings, and only high-quality
photographs were used for assigning final identifications.
In some cases, well-known adults and juveniles could not be photographed when they fluked,
because multiple animals fluked synchronously. In such cases, if the flukes of these well-known
individuals were confidently observed by S.G., they were recorded as having been identified (407 out
of 6938 identifications). Past analyses have demonstrated that patterns of association do not differ
when including these identifications [31]. Likewise, well-known calves who were not photographed
Box 1. Key social structure terms.
Cluster: a grouping of individual sperm whales at the surface in close proximity to each other (less
than 40 m) with coordinated behaviour [16].
Social unit: sperm whales with long-term, stable social relationships, defined as individuals identified
within 2 h of each other in at least two different years [15].
Group: sets of sperm whales temporarily travelling together for hours or days, which may include
more than one social unit. The group’s members may aggregate in close clusters while socializing,
or spread out across kilometres while foraging [34].
Clan: a higher level of sperm whale social structure composed of social units that share socially
learned behaviours, including distinguishable vocal repertoires [20].
Definition of association: a way to designate whether two individuals are ‘together’, for the purposes of
calculating association indices. In this study, we used three different definitions of association: (i)
both photo-identified in the same day, (ii) both photo-identified within 2 h of each other, and (iii)
in the same ‘cluster’ (as defined above).
Association index: a quantification of the proportion of time that two individuals spend ‘together’
(based on definition of association). Two association indices are used in this study: (i) ‘both
identified’ [35] and (ii) half-weight index [36]. R. Soc. open sci. 5: 180914
but were readily identifiable due to distinct dorsal markings that were visible by eye or because they were
known to be the only calf in the social unit were also recorded as having been identified (521 out of 2074
2.3. Measuring association and defining social units
For our analysis, we considered three definitions of association (box 1). First, as our finest spatiotemporal
scale of association, individuals in clusters at the surface, and so likely within visual contact and often in
physical contact, were considered to be associated. Second, we defined association more loosely as
individuals identified within 2 h of each other. Individuals seen within this time frame are likely close
enough to be in acoustic contact. Third, we defined association as being identified on the same day, to
capture potential avoidances or behavioural coordination occurring on larger spatiotemporal scales.
Existence of preferences and avoidances on this scale are reasonable, considering that sperm whales
travel an average of 50 km day
[40] and individuals in a group can spread out across several
kilometres while foraging [16].
To quantify the proportion of time that pairs of individuals spent associated, based on the above
definitions of association, we used two different association indices: (i) ‘both identified’ [35] and (ii)
half-weight index [36]. See the electronic supplementary material for further details on the calculation
of these indices.
Social units were delineated as in Gero et al. [15], so that they reflect long-term, stable social
relationships (see box 1). For our analysis, we also designated social units as genetically ‘well-
sampled’ or not. Well-sampled social units were those for which all adult females were included in
the genetic analysis (determined based on the availability of genetic samples). These were the social
units that we included for analyses examining association preferences within social units relative to
2.4. DNA extraction, quality control and sexing
We extracted DNA from all skin samples using standard phenol– chloroform procedures [41]. After
extraction, DNA from all samples was quantified via spectrophotometry, using a NanoDrop 2000
(Thermo Scientific, Waltham, MA, USA), and DNA concentrations were standardized accordingly for
use in polymerase chain reactions (PCRs).
To determine the sex of individuals, we amplified a 94 bp fragment of the ZFX/ZFY gene [42]. Within
this fragment, a Taq1 restriction site is present in the ZFX but not the ZFY sequence, due to a fixed
difference between the X- and Y-chromosomes. We digested the amplicon, and we size-separated and
visualized the post-restriction enzyme PCR product using ethidium bromide and agarose gel
electrophoresis to distinguish females (37 and 57 bp fragments only) from males (37, 57 and 94 bp
fragments) [42].
Sperm whale sloughed skin samples vary greatly in the amount and quality of DNA they yield [42],
and DNA quantification via spectrophotometry can overestimate the amount of viable DNA in these
samples, because it includes fragments that are too short to be amplified in PCRs. Therefore, we also
used the results of this sexing reaction as a first stage of quality control, to screen for samples that
were degraded beyond being useful to this study. Samples that failed to amplify at the 94 bp ZFX/
ZFY gene fragment were deemed too degraded for subsequent attempts at genotyping or sequencing.
Additionally, we used this sexing assay to determine and optimize DNA amplifiability for
downstream genotyping [42]. We adjusted DNA concentrations of sloughed skin samples, in
proportion to the brightness of the sample’s amplified ZFX/ZFY gene fragments relative to those of
a biopsy sample, to maximize success of amplification across microsatellite loci [42]. For example, a
sample with ZFX/ZFY gene fragments half as bright as those of a biopsy sample would be assigned
a functional concentration that is half of the concentration which was determined via
spectrophotometry, and thus we would use twice as much template DNA in the subsequent
amplification reaction for that sample. Samples that still genotyped poorly (genotyped at less than 10
microsatellite loci; see below) were excluded from further analysis.
2.5. Microsatellite genotyping
We amplified DNA samples at 18 microsatellite loci. For PCR conditions and genotyping methods, see
the electronic supplementary material. Other loci were also screened for amplification success with R. Soc. open sci. 5: 180914
sperm whale skin samples but did not produce usable results and were excluded from our analysis
(see the electronic supplementary material, table S1).
To address issues associated with low-quality DNA, particularly allelic dropout [43], we applied a
multiple-tubes PCR approach. This allowed us to determine rates of genotyping errors and improve
confidence in genotypes. For 17 samples, selected at random with respect to DNA quality and
quantity, we performed at least two independent PCRs for apparent heterozygotes and seven
independent PCRs for apparent homozygotes. These numbers of replicate PCRs were selected based
on the conservative approach described by Taberlet et al. [44]. This process of multiple amplifications
was repeated in its entirety for seven of the samples, with the identities of these samples masked, so
that they were blind controls. We determined genotyping error rates by comparing the genotypes of
the blind controls to their counterparts and calculating the rate of discrepancies. Using these rates, we
determined the number of tubes required to reach a minimum desired level of confidence in
genotypes of 99% per locus, and we performed this number of reactions to achieve this level of
confidence. If scores from replicate reactions for an individual were inconsistent, additional reactions
were performed until one genotype score emerged as at least 100 times more likely (based on above
error rates) than the other observed scores. If this likelihood ratio was not achieved in a reasonable
number of reactions, no data were included in the analysis for that individual at that locus.
Previous work on sperm whales has demonstrated the absence of significant population
differentiation at microsatellite loci within the North Atlantic [45]. Therefore, all genetic individuals
sampled off Dominica were considered to be from a single population for the purposes of calculating
allele frequencies.
We tested for linkage disequilibrium using GENEPOP v. 4.2 [46], and tested for null alleles and
deviation from the Hardy– Weinberg equilibrium using CERVUS 3.0.7 [47].
2.6. mtDNA haplotype sequencing
To determine mitochondrial DNA (mtDNA) haplotypes, we amplified and sequenced 346 bp at the 50
end of the mtDNA control region, using the primers t-Pro and Primer 2 [48]. Mitogenomic diversity is
relatively low in sperm whales, compared to estimates for other mammalian species, but out of
partitions of the sperm whale mitogenome that have been compared, nucleotide diversity was greatest
in the control region [49]. For amplification and sequencing reaction conditions and methods, see the
electronic supplementary material.
2.7. Identification of genetic individuals
To assign whether or not samples with the same or very similar microsatellite genotypes were from the
same individual, we estimated the probability of these pairs of samples originating from the same
individual (P
), while incorporating genotyping errors (as determined above), and the probability of
the two samples being from full-siblings (P
;sensu [50,51]). We classified samples as from the same
individual if log
) was greater than 3, and we classified them as from different individuals
if log
) was greater than 3. For pairs of samples where neither criterion was met, the
sample with the less complete genotype was excluded from further analysis. We also checked
the conclusions of this analysis for consistency with mtDNA haplotypes, sex and photographic
field identifications. We also calculated a probability of identity for unrelated individuals and for
full-siblings for the entire dataset, using a custom R script.
Genetic identities were linked to photo-identities directly when a biopsy sample was collected from a
photo-identified whale or a sloughed skin sample was collected from a photo-identified whale, with no
other whales in the immediate vicinity. When sloughed skin samples were collected from clusters
containing multiple individuals, the sample was assumed to be from any of the whales in the cluster.
If all individuals in the cluster except one could be excluded as providers of the skin (based on sex or
mismatching microsatellite genotypes with other known samples) then the sample was deduced to be
from the remaining individual. If multiple samples collected from different clusters were matched as
the same genetic individual, the photo-identities of the whales that were present in all of these
clusters were used to aid deduction. For some genetic individuals, more than one photo-identified
individual remained non-excluded. These genetic individuals were not used in individual-level
analyses, but if all non-excluded photo-identified individuals were from the same social unit, the
genetic individual was assigned to this social unit and used in social unit-level analyses. Individuals
were also excluded from further analyses if they were not members of known social units or if the R. Soc. open sci. 5: 180914
photo-identity of the genetic individual could not be deduced, such as when clusters contained
unidentified individuals.
2.8. Age class
Age classification of social unit members was accomplished based on observations of size and nursing in
the field, as in Gero et al. [15], combined with inference based on sex assignment. Individuals were
classified as either adult females, juveniles or dependent calves. The category ‘juveniles’ included
individuals that were noticeably smaller than adult females, but no longer nursing (see [52] for a
description of nursing behaviour). Additionally, because mature males are notably larger than adult
females [17,53], individuals that were indistinguishable from adult females based on size but sexed as
male were also classified as juveniles. Dependent calves were small individuals that were observed
nursing. Some individuals that were initially classified as dependent calves were re-classified as
juveniles in subsequent years if they were no longer observed nursing.
2.9. Assigning maternity and determining likely genetic relationships
To infer maternity of juveniles and dependent calves, we used a full-maximum likelihood method
for polygamous diploids implemented in COLONY [54]. We based error rates on the final
genotyping error rates estimated for our multiple-tubes PCR approach (0.16% for allelic dropout rate
and 0.1% for other errors). We performed a set of three runs, to increase the chances of finding the
configuration of relationships with the maximum likelihood, and repeated these runs with two
different random seed numbers, to confirm the repeatability of the results. All adult females were
included as putative mothers, and individuals classified as juveniles or dependent calves were
included as offspring. No putative fathers were included. One juvenile female observed throughout
the 12-year study period was assumed to be mature by the end of the study period (based on
pregnancy ages reported in [53]). Therefore, the runs were performed in replicate with this individual
as a putative mother instead of an offspring, but maternity assignment results did not change. We
assigned maternity if the female had a mean probability greater than 90% across all runs. Maternity
assignments were checked for agreement with mtDNA haplotypes. Individuals were classified as
maternal half-siblings if they were assigned the same mother.
To test hypotheses about relationships between adult females, where relative age is unknown, we
used the program ML-RELATE [55]. We evaluated which relationships (out of parentoffspring, half-
sibling/grandmother– granddaughter, full-sibling and unrelated) were consistent with the genetic data
at the 0.05 level of significance, by calculating likelihood ratios and using simulations to reject unlikely
relationships. If multiple relationships were consistent with the genetic data, this method was also
used to identify the most likely relationship.
2.10. Determining pairwise relatedness
To estimate relatedness between individuals, we used the R package related [56]. Performance of different
relatedness estimators varies depending on the relatedness structure of the population, and no single
estimator performs best across all relatedness structures [57,58]. Therefore, to select the best estimator
for our dataset, we used a comparative function in related that uses our population allele frequencies
to generate virtual pairs of individuals with specified genetic relationships, and to estimate the
relatedness of these pairs using four different relatedness estimators [59– 62]. For use in subsequent
analysis, we selected the estimator with the highest correlation between observed and expected
relatedness values, which was Wang’s estimator [62]. We used this estimator to calculate pairwise
relatedness values for all pairs of individuals.
2.11. Testing relationships between haplotype sharing, pairwise relatedness and association
Across all identified individuals from known social units, we tested for matrix correlations between
measures of genetic similarity and social association. A large proportion of pairs of individuals were
never both identified in the same time period, leading to many cells with no data in the matrices of
social association, which rendered Mantel tests [63] inappropriate for obtaining reliable p-values.
Instead, we calculated standard analytical p-values based on matrix correlation values (excluding
dyads with missing data in the association matrix), which, while not strictly valid for matrix data (the R. Soc. open sci. 5: 180914
assumption of independent observations is not met), provide an approximate indication of statistical
significance. The measures of genetic similarity used were mtDNA haplotype sharing (0 or 1) and
pairwise relatedness. The measures of association used were: (i) same cluster, in 6 h sampling period,
(ii) same cluster, in a year sampling period, (iii) identified within 2 h, in a 10-day sampling period,
and (iv) same day, in a year sampling period. To remove the effect of mothers associating with their
dependent calves, we omitted the pairwise data for mothers associating with their calves from all
analyses. We repeated the analyses with only data for pairs of individuals in the same social unit
Similarly, across all genetic individuals that were assigned to a known social unit, we tested for a
matrix correlation between pairwise relatedness and shared social unit membership (0 or 1), by
performing Mantel tests [63], using SOCPROG2.7 [64]. We also examined the distributions of pairwise
relatedness values within and between social units.
2.12. Composition of well-sampled social units
For well-sampled social units, we determined the proportions of relationships classified as mother–
offspring, second-degree relatives (half-sibling or grandparent– grandoffspring) or more distantly
related. We classified individuals as a motheroffspring pair if they were assigned as such based on
maternity assignment in colony or if parentoffspring was the most probable relationship in ML-
RELATE. We classified individuals as second-degree relatives if they could be inferred as such based on
motheroffspring relationships or if second-degree was the most probable relationship in ML-RELATE.
All other pairs were classified as more distantly related, which could also include unrelated individuals.
2.13. Categories of matrilineality
To assess the degree to which social units are matrilineal, we defined three possible categories of
‘matrilineal’: strictly matrilineal, generally matrilineal and matrilineally based. Most stringently, a social
unit could be categorized as ‘strictly matrilineal’ if all members have a common maternal ancestor who
is still living in the social unit. By this definition, social units are expected to split after the death of
their common maternal ancestor, but could contain several generations (female sperm whales can live
into their 80s [16] and may first conceive at around 9 years of age [53]). At a coarser scale, a social unit
could be categorized as ‘generally matrilineal’ if members have a relatively recent common maternal
ancestor, who need not be alive. In such cases, social units should have a common mtDNA haplotype
and an average genetic relatedness that is above that of the population. A social unit that is not strictly
or generally matrilineal could still be considered ‘matrilineally based’ if it is made up of two or more
strictly or generally matrilineal families. We assessed which of these categories were consistent with the
social and genetic data for the social units in this study.
2.14. Within-social unit association
Within each well-sampled social unit, we performed Mantel tests [63], using SOCPROG2.7 [64] to test for
significant matrix correlations between pairwise relatedness and association in clusters, at two sampling
periods—2 h and a day.
Within the social unit with the most sampled members (unit A), we also examined social modularity
in relation to within-social unit genetic structure. We defined social modules such that association indices
(based on association in clusters in a daily sampling period) were generally high among individuals in
the same module and generally low among individuals in different modules. We used an eigenvector-
based method, as suggested by Newman [65], and implemented in SOCPROG2.7 [64] to test for
the presence of meaningful modularity (modularity values greater than approximately 0.3) and to
delineate the modules. To account for demographic changes, we examined modularity within
three different years that span the study period (2005, 2010 and 2015). We examined the
congruence between the social modules identified by this method and the matrilineal clusters defined
by motheroffspring relationships.
2.15. Between-social unit association
For social units for which at least three members were included in the genetic analysis, we tested for
relationships between social association and genetic similarity. If one or more members of each of two R. Soc. open sci. 5: 180914
social units were associated in a sampling period, then those individuals’ social units were considered
associated in that sampling period. We used four measures of association: (i) same cluster, in 2 h, (ii)
same cluster, in a year, (iii) identified within 2 h, in a day, and (iv) same day, in a year. For measures
of genetic similarity, we classified each pair of social units’ mtDNA haplotypes as same or different,
and calculated mean relatedness values. As a measure of recent common maternal ancestry, we
calculated mean relatedness values between social units, by averaging the pairwise relatedness values
between all pairs of individuals across each pairwise combination of social units.
We performed Mantel tests [63], using SOCPROG2.7 [64] to test for matrix correlations between each
index of association and each measure of genetic similarity. One pair of social units appeared to be
contributing strongly to correlations, and so the tests were repeated with pairwise data for that dyad
3. Results
3.1. Association dataset
On average, individuals were identified in 54 different 2 h periods (range: 5 192), 26 days (range: 3 90)
and 4.7 years (range: 1 10). On average, social units were identified in 131 different 2 h periods (range:
31401), 45 days (range: 12–117) and 6.8 years (range: 2– 10). For additional details, see the electronic
supplementary material.
3.2. Microsatellite dataset and quality control
Out of 153 samples (94.8% sloughed skin and 5.2% biopsy samples), 30 were excluded by quality control
(one biopsy and 22 sloughed skin samples failed to sex, seven sloughed skin samples failed to genotype
at a minimum of 10 microsatellite loci). After consolidating duplicates (as determined by log
values) and excluding three likely duplicate samples that did not meet the log-likelihood ratio criteria, 95
unique individuals remained, 88.4% of which were scored at all 18 microsatellites, and all of which were
scored at no fewer than 16 microsatellites. The mean allelic diversity was 9.3 (range: 3– 17) and the mean
observed heterozygosity was 0.75 (range: 0.52–0.93). See the electronic supplementary material, table S2
for locus-specific allelic diversity and heterozygosity.
We calculated the total per-allele genotyping error rate for apparent heterozygotes (E
) to be 1.1%,
incorporating contamination and spurious alleles (1.0% collectively) and manual scoring errors (less than
0.1%). For apparent homozygotes (E
), the mean error rate was 2.9%, incorporating allele dropout
(2.82%) and manual scoring errors (less than 0.1%), but dropout rate varied widely across samples
(max ¼11.6%). These rates are comparable to those reported previously for sperm whale skin samples
(e.g. per-allele microsatellite error rate of 1.27% [66]). Based on our error rates, for apparent
heterozygotes, a minimum desired level of confidence in genotypes of 99% per locus was reached
with two tubes (compound error rate ¼0.013%). For apparent homozygotes, this level was reached
with two tubes based on the average dropout rate (compound error rate ¼0.085%), but three tubes
were required based on the sample with the highest dropout rate (compound error rate ¼0.16%).
Thus, we performed a second reaction for loci at which an individual appeared heterozygous, and, to
account for low-quality samples, we performed at least three reactions for loci at which an individual
appeared homozygous.
No loci showed strong indications of null alleles (all frequencies less than 0.05), and we detected no
evidence of deviations from the HardyWeinberg equilibrium; however, it is challenging to reliably
detect null alleles [67]. Two pairs of loci had evidence of linkage disequilibrium after a Bonferroni
correction, but given that our dataset is composed of social units of related individuals, this was not
unexpected, and it would be difficult to distinguish true linkage from effects of the similarity of
genotypes of relatives. Therefore, we did not exclude any loci from the analysis. Probabilities of
identity for this dataset were 2.7 10
for unrelated individuals and 1.5 10
for full-siblings.
After exclusion of unidentified individuals and individuals that were not members of known social
units, 65 genetic individuals remained that were assigned to 12 known social units and were used in the
social unit-level analyses (table 1). Of these, 55 were linked to single photographically identified
individuals from those social units and were used in the individual-level analyses (table 1). Six social
units qualified as well sampled, with genetic data for all adult females (and at least 70% of all unit R. Soc. open sci. 5: 180914
Table 1. Composition and mitochondrial haplotype (mtDNA Hap) of 12 social units sampled off Dominica. Social units were delineated as in Gero et al. [15]. Composition includes past and present sampled members.
Well-sampled social units (those for which all adult females were genetically sampled) were used for intra-social unit analyses, and are listed in the top section of the table. The number of genetically sampled social
unit members includes only those linked to a single identified individual. The number listed in parentheses counts all sampled social unit members, including samples for which individual identity was unknown. For
well-sampled social units, the mean relatedness (mean r) was calculated according to Wang [62]. Members of these social units were also categorized as adult females (A
) or offspring (O). Mother offspring (18)
relationships were determined using COLONY and ML-RELATE. Second-degree (28) relationships were determined using ML-RELATE, or inferred-based shared 18relatives.
social unit
social unit members sex
mtDNA Hap
age class relationships (%)
known sampled F M A
O1828greater than 28
A 12 12 9 3 BB 0.137 4 8 15.2 12.1 72.7
F 10 9 5 4 A 0.232 5 4 16.7 25.0 58.3
J 6 5 5 A 0.136 3 2 20.0 0 80.0
R 10 7 6 1 A 0.106 5 2 14.3 14.3 71.4
S 4 3 3 A 0.212 3 33.3 0 66.7
U 4 4 3 1 A 0.333 2 2 33.3 33.3 33.3
C 6 1 1 A total 22 18 16.9 15.5 67.6
D 7 (4) 2 3 1 A
N 9 (8) 5 7 1 A
P 9 (3) 1 1 2 BB
T 9 (6) 4 6 A
V 12 (3) 2 3 A
total 98 (65) 55 49 16 49 A, 15 BB
Haplotypes for this social unit were obtained for seven of eight samples. R. Soc. open sci. 5: 180914
members, when calves and juveniles were included), and these social units were included in the
within-social unit analyses (table 1).
3.3. Mitochondrial haplotypes
For mtDNA haplotype assignment, no errors were detected in blind replicates (n¼14) nor any
inconsistencies for pairs of samples determined to be from the same individuals based on multi-locus
microsatellite genotypes (n¼7). Haplotypes were successfully sequenced for 61 of 65 sampled social
unit members. For samples from three calves, which failed to sequence successfully, haplotypes were
inferred based on the haplotypes of their mothers (because we found consistent agreement in the
mtDNA haplotypes of all other calves and their genetically assigned mothers, see below).
Two mtDNA haplotypes (A and BB) were identified in individuals from known social units, both of
which have been previously observed in the western North Atlantic Ocean [66]. These haplotypes differ
by a single nucleotide substitution. This low level of mitochondrial diversity is consistent with previous
observations on a global and mitogenome-wide scale [66].
Mitochondrial haplotypes were consistent within social units (which is a requisite result for the social
units to be considered matrilineal), though each haplotype was shared by multiple social units.
Haplotype A was much more common, being shared by 10 out of 12 social units (table 1).
3.4. First- and second-degree relationships
We classified 30 individuals as adult females and the remaining 25 individuals as offspring. Thirteen of
these females were assigned as the mothers of 18 offspring; in all cases, the assigned mother was from the
same social unit as the offspring. Ten females were assigned to a single offspring each, two were assigned
to two offspring each and one to four offspring. These maternity assignments were in agreement with the
mtDNA haplotypes of mothers and their offspring, when both had been successfully sequenced. Seven
offspring did not have mothers identified from the sampled females. Average pairwise relatedness
between identified motheroffspring pairs was 0.52 (range: 0.42– 0.67, n¼18) and for half-siblings
inferred based on shared maternity, average pairwise relatedness was 0.32 (range: 0.120.50, n¼8).
Out of the adult females, we identified eight pairs of individuals for which parent– offspring was the
relationship with the highest likelihood. For six of these pairs, parent– offspring was the only relationship
consistent with the genetic data at the 0.05 level of significance, but for two pairs, sibling relationships
also met this level of significance. All of these parent– offspring pairs were within social units, rather
than between them (figure 1). Average pairwise relatedness between mother– offspring pairs
identified among adults was 0.50 (range: 0.43– 0.59)
Figure 1. Genetic relationships between adult females, within and between social units. Letters indicate social unit. Shading of
social unit block indicates mitochondrial haplotype (unshaded: haplotype A; grey shading: haplotype BB). Solid edges between
individuals denote mother offspring relationships, and dashed edges indicate second-degree relationships, as determined using
ML-RELATE, including only those relationships for which ‘unrelated’ was not also a likely option. (Note: variation in edge length
is an artefact of the figure arrangement and does not convey information.) Social units with no missing adult members are
indicated by an asterisk. R. Soc. open sci. 5: 180914
Pairs of adult females for which the most likely relationship was second degree (half-siblings/
grandmother– granddaughter) were much more common (n¼43), but for the majority of these
(74.4%), the genotypes were also consistent with the individuals being unrelated. For two inter-social
unit pairs of individuals, full-siblings was the most likely relationship, but in both cases, the
genotypes were also consistent with the individuals being second-degree relatives. Of these putative
second-degree relationships, 88.9% were split across different social units. For the subset of 13 pairs
for which ‘unrelated’ was not a probable relationship, all but one were between social units (figure 1).
In two of these cases, the putative second-degree relatives had different mtDNA haplotypes,
suggesting that any kinship between these individuals is paternal. For the remaining 10 well-
supported sibling pairs between social units, it could not be readily distinguished whether they
resulted from shared paternity or from maternal relatives (half-siblings or grandmother–
granddaughter) splitting into separate social units. Average pairwise relatedness between putative
second-degree relatives for which ‘unrelated’ was not a plausible option was 0.32 (range: 0.20– 0.54).
3.5. Relatedness and haplotype sharing predicting association across all individuals
Across all known social unit members, association was significantly positively correlated with pairwise
relatedness and with mtDNA haplotype sharing for all four measures of association examined (table 2).
When the dataset was restricted to pairs of individuals in the same social unit, the correlations between
pairwise relatedness and all scales of social association were positive and significant (table 2), though
only marginally so for long-term close associations (i.e. clusters in a yearly sampling period).
Members of the same social unit were also more closely related to each other than expected by chance
(matrix correlation ¼0.273, p,0.001, n¼65). The mean relatedness between individuals in the same
social unit was 0.139 (s.d.: 0.221, n¼200), whereas between individuals in different social units it was
0.004 (s.d.: 0.132, n¼1880). Relatedness values within social units were bimodally distributed, with a
local maximum at approximately 0.5, and a global maximum just above zero (figure 2).
3.6. Relationships within social units
Within social units, parent– offspring relationships made up between 14.3 and 33.3% of relationships (16.9%
overall), and between 0 and 33.3% of relationships were defined as second-degree relationships (15.5%
overall), leaving between 33.3 and 80% of relationships as more distant than second degree, potentially
including unrelated individuals (table 1). Most individuals (82.5%) had a mother or offspring in their
social unit, and out of those who did not, the majority (57.1%) had a second-degree relative (figure 3).
The remaining 7.5% of individuals had no relatives deemed to be first- or second-degree relatives
sampled from their social unit.
Table 2. Correlation between measures of social association and pairwise relatedness (Rel) or mtDNA haplotype sharing (Hap)
across all individuals (n¼55). Pairwise values for mothercalf pairs were excluded. This relationship was also tested after
restricting to members of the same social unit. Association values were calculated using ‘both identified’ as the association index.
interval predictor
all individuals the same social unit
corr. p-value
corr. p-value
day year Rel 0.200 ,0.001 0.221 ,0.001
Hap 0.248 ,0.001 —
2 h 10 day Rel 0.217 ,0.001 0.187 0.003
Hap 0.251 ,0.001 —
cluster year Rel 0.218 ,0.001 0.130 0.035
Hap 0.244 ,0.001 —
cluster 6 h Rel 0.206 ,0.001 0.197 0.001
Hap 0.175 ,0.001 — R. Soc. open sci. 5: 180914
For no social unit could all members be connected in a single network using only parent– offspring
relationships, but for two social units, all members could be connected when second-degree relationships
were included (figure 3). The remaining four social units had one or two missing connections between
members, even when second-degree relationships were included (figure 3). In these social units, all
unsampled members were calves, whose mothers were assumed (based on social data) to be among
pairwise relatedness
Figure 2. Distributions of pairwise relatedness values within (light grey) and between (dark grey) sperm whale social units.
Relatedness values were calculated using Wang’s estimator [62].
Figure 3. Relationship networks of well-sampled social units, based on genetic data. Females are indicated by circles and males by
squares. Dark grey indicates adults and light grey indicates offspring. Solid lines denote motheroffspring relationships, as
determined using COLONY or ML-RELATE. Dotted lines indicate pairs that were most likely second-degree relatives, but for which
‘unrelated’ was also a likely option (as determined using ML-RELATE). Genetic data were unavailable for six offspring; these
individuals are not shown. (Note: variation in edge length is an artefact of the figure arrangement and does not convey information.) R. Soc. open sci. 5: 180914
the sampled individuals, and so breaks in the genetic network are not likely due to the omission of these
individuals, but could be due to deceased, unknown relatives.
3.7. Kinship predicting association within social units
For the two best-sampled large social units, association was statistically significantly and positively
correlated with pairwise relatedness, at both sampling intervals (table 3). For the remaining four well-
sampled social units, these correlations were non-significant and had mixed directions (table 3;
electronic supplementary material, figure S1).
Two social modules were identified within unit A, the composition of which remained similar across
years (though the strength of social modularity decreased across the years examined). In all years, the
delineation of these social modules corresponded to the social unit’s two matrilineal clusters of
motheroffspring pairs (table 4).
3.8. Kinship predicting association between social units
Association between social units was not significantly correlated with having a shared mtDNA
haplotype at any level tested ( p0.17 for all four measures of association; table 5). Some pairs of
social units with the same mtDNA haplotype never associated (e.g. units P and A, and units S and T),
while units A and D, with differing haplotypes, frequently associated.
Association between social units, defined as being in a cluster together, seemed to be weakly
correlated with mean relatedness for both sampling periods (table 5). For the coarser measures of
association, association and mean relatedness between social units were not significantly correlated.
The marginally non-significant correlations between relatedness and fine-scale association were
primarily driven by one pair of social units, U and F, which had the highest mean relatedness value
of any pair of social units (mean relatedness ¼0.112) and the highest association index at all levels of
association. When the data point for this pair of social units is removed, the size and significance
of all correlations dropped (table 5).
4. Discussion
This study improves upon what has been previously achieved by examining more social units, with a
higher genetic resolution (3880% more microsatellite loci), than previous studies [2629], resulting in
a uniquely detailed exploration of sperm whale kinship patterns in relation to social structure.
Table 3. Intra-unit social association preferences predicted by pairwise relatedness. Association was defined as identification in
the same cluster, using ‘both identified’ as the association index. Pairwise values for mother-dependent calf pairs were excluded.
Mantel tests were performed with 10 000 permutations.
social unit Nsampling interval matrix correlation p-value
A 12 2 h 0.26 0.010
day 0.45 ,0.001
F 9 2 h 0.17 0.006
day 0.11 0.012
J52h 20.05 0.740
day 0.12 0.529
R72h 20.12 0.836
day 20.10 0.896
S 3 2 h 0.90 0.505
day 0.99 0.164
U42h 20.75 0.882
day 20.63 0.961 R. Soc. open sci. 5: 180914
Kinship was clearly correlated with several measures of social association, particularly social unit
membership (figure 2) and intra-social unit association preferences within two well-sampled social
units (table 3), suggesting that kin-selection may indeed be a contributing driver of sperm whale
social structure. However, substantial variation in patterns of associations remained unexplained by
kinship, notably including association preferences between the majority of social units. Thus, other
drivers of cooperation, such as group augmentation and reciprocal altruism, likely interact with kin-
selection to drive the cooperative social structure we observe. Preferential cooperation between
Table 5. Correlation between measures of inter-unit social association and mean pairwise relatedness (Rel) or mtDNA haplotype
sharing (Hap). Association values were calculated using half-weight indices. The tests were repeated with the pairwise values for
units U and F omitted (no UF). Mantel tests were performed with 10 000 permutations (n¼11).
interval predictor
all social units no UF
correlation p-value correlation p-value
day year Hap 0.33 0.17 0.32 0.11
Rel 0.23 0.15 0.12 0.43
2 h day Hap 0.14 0.31 0.11 0.42
Rel 0.13 0.33 20.06 0.71
cluster year Hap 0.07 0.60 0.02 0.71
Rel 0.25 0.09 0.09 0.48
cluster 2 h Hap 0.13 0.24 0.09 0.38
Rel 0.26 0.06 0.03 0.78
Table 4. Social modules and strict matrilines in social unit A across time. Strict matrilines were defined based on mother
offspring relationships (figure 3). Social modules were based on association as clusters in a daily sampling period, using half-
weight indices. Module composition is indicated by block shade, stippled shading indicates uncertainty in module assignment
(jeigenvectorjless than 0.1), and missing blocks indicate the individual was not seen (and presumably was not alive) in that
year. Good divisions within a network are generally indicated by modularity values of roughly 0.3 or greater [68]. Per cent
agreement with matrilines (% agreement) does not include uncertain module assignments.
social module
individual matriline 2005 2010 2015
Fruit salad
Lady oracle
N (days)
% agreement
100 R. Soc. open sci. 5: 180914
particular individuals or between particular social units could also be based in culture or personality, or
some preferences may be by-products of circumstance and convenience.
We found a higher degree of relatedness and matrilineality in social units than has been reported in
other regions [26,28,29]. Even so, it is unlikely that all of the social units that we examined were strict
matrilines. The presence of a living common ancestor was not conclusively demonstrated in any social
unit, despite all adult female members of these social units being genetically sampled. Rather,
inference of one or two intermediary relatives that are dead or gone would be required in each social
unit before the presence of a living common ancestor could be assumed (figure 3). Additionally, no
social unit splits have been observed in 96 social unit-years of observation off Dominica [69]. This is
despite a mean 4.5% per year decrease in the number of adults over the study period [69], which
would predict the death of common living ancestors in roughly four social units across the study
period. The presence of second-degree relationships between social units (figure 1) could indicate past
social unit splits after the death of a common ancestor, but such relationships could also be explained
by paternal relatedness. Indeed, paternal relatedness is the only explanation in cases where the
second-degree relatives have different mitochondrial haplotypes (e.g. relationships between unit A
and either unit J or F). The genetic data for all social units examined were consistent with our less
stringent category, ‘generally matrilineal’. However, haplotype sharing does not necessitate close
matrilineal co-ancestry, especially for the very common haplotype, A. As such, we could not rule out
the possibility that the social units contained unrelated matrilines. Even so, social units composed of
multiple matrilines would still be matrilineally based.
As our resolution of social data improves, so does our ability to investigate stability and distinguish
constant companions from preferred associates. This is exemplified by unit A, which had all members
genetically sampled, and was observed in seven different years between 2005 and 2016. This social
unit was composed of two strict matrilines (figure 3), which were unrelated or separated by at least
two absent intermediary relatives. This social unit was composed of two social modules that aligned
well with the delineation of these two matrilines (table 4), indicating higher association within than
between the strict matrilines in this unit. Based on our definition of social units (box 1), these
individuals qualified as members of a single social unit, but the rate at which these two matrilines
associated varied substantially across the study, and they were often observed apart (electronic
supplementary material, table S3). Additionally, over the course of our study period, we documented
the merger of two social units, U and F, which were originally classified as separate social units, using
data as far back as 1995 [15]. From 2008 onwards, their association rate generally increased, such that
by 2012 they were scarcely seen apart (electronic supplementary material, table S4). They met our
definition (box 1) to be classified as a single social unit by 2009, even though at that point in their
gradual merger the two social units were often seen apart (electronic supplementary material, table
S4). This suggests that social unit members, as we have defined them, are not such constant
companions as previously assumed, despite our definition of social units (based on [15]) being more
stringent than or similar to what has been used in other studies [19,30,70,71]. Rather, it appears that,
in some cases, sub-social unit social structures may exist but go undetected with the types of analyses
currently used to define constant companions, which often rely on sparse data. It is not clear how
sub-social unit social structures are actually expressed in the day-to-day life of a social unit at sea,
perhaps by a separation of several kilometres between subunits, or perhaps by much greater distances.
A multi-level social structure, composed of nested layers of increasingly close kin, is not unique to
sperm whales, but appears convergently in other marine and terrestrial mammals. Killer whale
(Orcinus orca) social structure is also composed of several nested tiers, the smallest tier comprising
matrilines of two to nine individuals that associate closely and constantly [72]. These matrilines then
belong to higher-level groups. The association patterns between the within-social unit matrilines that
we observed in social unit A are concordant with this social organization among killer whales.
Sperm whales have very similar life histories and social systems to elephants [73]. Distributions of
relatedness within and between sperm whale social units (figure 2) and within and between African
elephant (Loxodonta africana) core social groups (see [74]—Fig. 1) are remarkably similar, suggesting
similar degrees of matrilineality in their social structures. Also as in sperm whale social units,
kinship predicts association within African elephant core groups [74] but kinship does not well
predict higher levels of association, among either sperm whale social units or African elephant core
groups [74].
Social and ecological contexts, such as the presence of dependent calves or limited resources, likely
influence social structure and may encourage flexibility in how widely cooperation is extended
beyond close kin. For example, changes in social context may have motivated the merger of units F R. Soc. open sci. 5: 180914
and U because this merger occurred in parallel with changes in group size and composition (see
electronic supplementary material, discussion and table S4). Social unit fusion has the potential to
decrease the degree of matrilineality, depending on the relationship between the merging social units.
In the case of units F and U, the merging social units were the most closely related pair of social units
in our study, suggesting the possibility that mergers may preferentially occur between kin, which
would minimize the breakdown of matrilineality. Similarly, among African elephants, core group
splits and mergers are predicted by kinship [74].
The degree of relatedness within social units that we observed in the Eastern Caribbean is greater
than those reported for sperm whale social units in the eastern tropical Pacific [26,29]. One potential
reason for such differences in patterns of kinship and association is the degree to which populations
were affected by modern whaling; sperm whales were much more heavily targeted in the eastern
tropical Pacific than in the Caribbean [14]. Likewise, in African elephants, the relative importance of
kinship to social structure was diminished in a more heavily poached population [75]. In African
elephants, evidence suggests that individuals form associations with non-relatives if their relatives are
poached [74,75], and the same is likely true for sperm whales [14].
Alternatively, these regional differences in the matrilineally based social structure of sperm whales
may relate to differences in characteristics of prey affecting optimal group size [14], as appears to be
the case among killer whales [76]. Two distinct ‘ecotypes’ of killer whales, known as ‘residents’ and
‘transients’, eat primarily fish and marine mammals, respectively, and have notably different
matrilineal social structures [72]. Among ‘residents’ up to four generations are found in stable
matrilineal groups, but among ‘transients’ smaller groups are found containing only one to two
generations [72], seemingly because the smaller group size is more optimal for hunting the ‘transient’
killer whales’ primary prey, the harbour seal (Phoca vitulina) [76].
Another factoraffecting the social context of the individuals in this study is the population’s current state
of critical decline, with most social units losing members [69]. Social relationships can have fitness
consequences, as in baboons, where female sociality correlates with reproductive success [77], and social
structure influences processes like the transmission of disease or information [78]. The loss of individuals
may also lead to changes to the matrilineal structure of sperm whale social units, perhaps encouraging
mergers of unrelated social units, or discouraging splits of social units that have lost their common
maternal ancestor, similar to the effect that whaling may have had on social structure in the Eastern
Tropical Pacific [14]. As such, understanding the social structure of this population, and the genetic
diversity underpinning it, is important from a conservation perspective. Understanding drivers of social
structure can aid our understanding of how it may change when individuals are lost from this population.
5. Conclusion
While this study demonstrates that kinship is clearly an important factor influencing sperm whale social
relationships, we also see that it is not the be-all and end-all. Social units were largely composed of kin
but did not appear to be rigidly delineated by matrilines. Likewise, social associations within social units
were biased towards closer relatives, but as a general trend, rather than a strict rule. Social and ecological
context likely affect the degree of matrilineality in sperm whale social structure, leading to variation both
within social units across time, and broadly across populations in different ocean basins. Overall, our
findings support sperm whale society as being matrilineally based, but not strictly so; rather, it is
nuanced and multi-faceted, resembling other complex matrilineal societies, such as among elephants
and killer whales.
Ethics. Our data and samples were collected in Dominica under scientific research permits from the Fisheries Division of
the Ministry of Agriculture and Environment: SCR 013/05-02, RP-2/12 IW-1, RP-09/014 IW-1, RP-01/079 W-2, RP-
03/059 W-4, P-122/4 W-2, P-40/2 W-7 and RP-16-04/88-FIS-9. Samples were transported through CITES permits
for the import and export of animal parts issued by Environment Canada and the Environmental Coordinating
Unit of Dominica. The field protocols for approaching, photographing and recording sperm whales were approved
by the University Committee on Laboratory Animals of Dalhousie University and the Animal Welfare and Ethics
Committee of the University of St Andrews. Biopsy sample collection procedures were also approved by the Saint
Mary’s University Animal Care Committee.
Data accessibility. Data are available from the Dryad Digital Repository: ( [79].
Authors’ contributions. C.M.K. participated in the collection of the field data, carried out all molecular laboratory work and
statistical analysis, and wrote the manuscript; S.G. coordinated and funded the field operations of the study,
completed the photo-identification, undertook all of the fieldwork to collect data and edited the manuscript; T.F.
aided sample collection efforts, supervised the molecular laboratory work, contributed funding to the laboratory R. Soc. open sci. 5: 180914
work and edited the manuscript; H.W. funded the field operations and contributed funds to the laboratory work,
participated in the collection of the field data and edited the manuscript. All authors collaborated in the conception
and design of the study and gave final approval for publication.
Competing interests. We have no competing interests.
Funding. Fieldwork was funded through a Carlsberg Foundation field expedition grant and an FNU fellowship from the
Danish Council for Independent Research supplemented by a Sapere Aude Research Talent Award to S.G., as well as
by Discovery and Equipment grants to H.W. from the Natural Sciences and Engineering Research Council of Canada
(NSERC) and by a Discovery Development Grant from NSERC to T.F. Supplementary funding was provided through
a FNU Large Frame Grant to Peter Madsen from Aarhus University. S.G. is supported by a technical and scientific
research grant from the Villum Foundation, and C.K. by a NSERC CGS, a Nova Scotia Research and Innovation
Graduate Scholarship and the Patrick F. Lett Fund.
Acknowledgements. We thank Mr Riviere Sebastien and the Dominica Fisheries Division officers, staff at the Anchorage
Hotel, Dive Dominica, Al Dive, and W.E.T. Dominica for logistical support while in Dominica; all the crews of R/
V Balaena; and Peter Madsen and the crew members from the Marine Bioacoustics Lab at Aarhus University. We
thank our two anonymous reviewers for their constructive reviews. This paper emanates from The Dominica
Sperm Whale Project— Follow @DomWhale.
1. Clutton-Brock TH. 2002 Breeding together: kin
selection and mutualism in cooperative
vertebrates. Science 296, 69– 72. (doi:10.1126/
2. Cockburn A. 2006 Prevalence of different modes
of parental care in birds. Proc. R. Soc. B 273,
1375–1383. (doi:10.1098/rspb.2005.3458)
3. Clutton-Brock TH. 2009 Cooperation between
non-kin in animal societies. Nature 462,
51– 57. (doi:10.1038/nature08366)
4. Axelrod R, Hamilton WD. 1981 The evolution of
cooperation. Science 211, 1390– 1396. (doi:10.
5. Hamilton WD. 1964 The genetical evolution of
social behaviour. I. J. Theor. Biol. 7, 1– 16.
6. Hamilton WD. 1964 The genetical evolution of
social behaviour II. J. Theor. Biol. 7, 17– 52.
7. Trivers RL. 2006 Reciprocal altruism: 30 years
later. In Cooperation in primates and humans:
mechanisms and evolution (eds P Kappeler, CP
van Schaik), pp. 67 83. Berlin, Germany:
8. Trivers RL. 1971 The evolution of reciprocal
altruism. Q. Rev. Biol. 46, 35– 37. (doi:10.1086/
9. Hammerstein P. 2003 Why is reciprocity so rare
in social animals? A protestant appeal. In
Genetic and cultural evolution of cooperation
(ed. P Hammerstein), pp. 83– 93. Cambridge,
MA: MIT Press.
10. Connor RC. 2007 Invested, extracted and
byproduct benefits: a modified scheme for the
evolution of cooperation. Behav. Processes 76,
109– 113. (doi:10.1016/j.beproc.2007.01.014)
11. Connor RC. 2010 Cooperation beyond the dyad:
on simple models and a complex society. Phil.
Trans. R. Soc. B 365, 2687– 2697. (doi:10.1098/
12. Kingma SA, Santema P, Taborsky M, Komdeur J.
2014 Group augmentation and the evolution of
cooperation. Trends Ecol. Evol. 29, 476– 484.
13. Kokko H, Johnstone RA, Clutton-Brock TH. 2001
The evolution of cooperative breeding through
group augmentation. Proc. R. Soc. Lond. B 268,
187– 196. (doi:10.1098/rspb.2000.1349)
14. Whitehead H, Antunes R, Gero S, Wong SNP,
Engelhaupt D, Rendell L. 2012 Multilevel
societies of female sperm whales (Physeter
macrocephalus) in the Atlantic and Pacific: why
are they so different? Int. J. Primatol. 33,
1142–1164. (doi:10.1007/s10764-012-9598-z)
15. Gero S et al. 2014 Behavior and social structure
of the sperm whales of Dominica, West Indies.
Mar. Mammal Sci. 30, 905–922. (doi:10.1111/
16. Whitehead H. 2003 Sperm whales: social
evolution in the ocean. Chicago, IL: University of
Chicago Press.
17. Best PB. 1979 Social organization in sperm
whales, Physeter macrocephalus.InBehavior of
marine animals (eds HE Winn, BL Olla), pp.
227– 289. Boston, MA: Springer.
18. Gero S, Whitehead H, Rendell L. 2016
Individual, unit and vocal clan level identity
cues in sperm whale codas. R. Soc. open sci. 3,
150372. (doi:10.1098/rsos.150372)
19. Rendell L, Whitehead H. 2003 Vocal clans in
sperm whales (Physeter macrocephalus).
Proc. R. Soc. Lond. B 270, 225– 231. (doi:10.
20. Cantor M, Whitehead H. 2015 How does social
behavior differ among sperm whale clans? Mar.
Mammal Sci. 31, 1275–1290. (doi:10.1111/
21. Gero S, Gordon J, Whitehead H. 2013 Calves as
social hubs: dynamics of the social network
within sperm whale units. Proc. R. Soc. B 280,
20131113. (doi:10.1098/rspb.2013.1113)
22. Borrell A, Vacca AV, Pinela AM, Kinze C, Lockyer
CH, Vighi M, Aguilar A. 2013 Stable isotopes
provide insight into population structure and
segregation in eastern North Atlantic sperm
whales. PLoS ONE 8, e82398. (doi:10.1371/
23. Cantor M, Shoemaker LG, Cabral RB, Flores CO,
Varga M, Whitehead H. 2015 Multilevel animal
societies can emerge from cultural transmission.
Nat. Commun. 6, 8091. (doi:10.1038/
24. Richard KR, Dillon MC, Whitehead H, Wright JM.
1996 Patterns of kinship in groups of free-living
sperm whales (Physeter macrocephalus)
revealed by multiple molecular genetic
analyses. Proc. Natl Acad. Sci. USA 93,
8792–8795. (doi:10.1073/pnas.93.16.8792)
25. Pinela AM, Que
´rouil S, Magalha
˜es S, Silva MA,
Prieto R, Matos JA, Santos RS. 2009 Population
genetics and social organization of the sperm
whale (Physeter macrocephalus) in the Azores
inferred by microsatellite analyses. Can. J. Zool.
87, 802– 813. (doi:10.1139/Z09-066)
26. Mesnick SL. 2001 Genetic relatedness in sperm
whales: evidence and cultural implications.
Behav. Brain Sci. 24, 346– 347. (doi:10.1017/
27. Gero S, Engelhaupt D, Whitehead H. 2008
Heterogeneous social associations within a
sperm whale, Physeter macrocephalus, unit
reflect pairwise relatedness. Behav. Ecol.
Sociobiol. 63, 143– 151. (doi:10.1007/s00265-
28. Ortega-Ortiz JG, Engelhaupt D, Winsor M, Mate
BR, Rus Hoelzel A. 2012 Kinship of long-term
associates in the highly social sperm whale.
Mol. Ecol. 21, 732–744. (doi:10.1111/j.1365-
29. Christal J. 1998 An analysis of sperm whale
social structure: patterns of association and
genetic relatedness. PhD thesis, Dalhousie
University, Halifax, Canada.
30. Christal J, Whitehead H, Lettevall E. 1998 Sperm
whale social units: variation and change.
Can. J. Zool. 76, 1431 1440. (doi:10.1139/z98-
31. Gero S, Gordon J, Whitehead H. 2015
Individualized social preferences and long-term
social fidelity between social units of sperm
whales. Anim. Behav. 102, 15– 23. (doi:10.
32. Arnbom T. 1987 Individual identification of
sperm whales. Rep. Int. Whal. Comm. 37,
201– 204.
33. Gero S, Engelhaupt D, Rendell L, Whitehead H.
2009 Who cares? Between-group variation in
alloparental caregiving in sperm whales. Behav. R. Soc. open sci. 5: 180914
Ecol. 20, 838 –843. (doi:10.1093/beheco/
34. Whitehead H, Waters S, Lyrholm T. 1991 Social
organization of female sperm whales and their
offspring: constant companions and casual
acquaintances. Behav. Ecol. Sociobiol. 29,
385– 389. (doi:10.1007/BF00165964)
35. Whitehead H. 2008 Analyzing animal societies.
Chicago, IL: The University of Chicago Press.
36. Cairns SJ, Schwager SJ. 1987 A comparison of
association indices. Anim. Behav. 35, 1454– 1469.
37. Whitehead H, Gordon J, Mathews EA, Richard
KR. 1990 Obtaining skin samples from living
sperm whales. Mar. Mammal Sci. 6, 316 326.
38. Kowarski KA, Augusto JF, Frasier TR, Whitehead
H. 2014 Effects of remote biopsy sampling on
long-finned pilot whales (Globicephala melas)in
Nova Scotia. Aquat. Mamm. 40, 117– 125.
39. Seutin G, White BN, Boag PT. 1991 Preservation
of blood and tissue samples for DNA analyses.
Can. J. Zool. 69, 82 –90. (doi:10.1139/z91-013)
40. Whitehead H, Coakes A, Jaquet N, Lusseau S.
2008 Movements of sperm whales in the
tropical Pacific. Mar. Ecol. Prog. Ser. 361,
291– 300. (doi:10.3354/meps07412)
41. Sambrook J, Russell DW. 2001 Molecular
cloning: a laboratory manual. Cold Spring
Harbor, NY: Cold Spring Harbor Laboratory
42. Konrad CM, Dupuis A, Gero S, Frasier T. 2017 A
sexing technique for highly degraded cetacean
DNA. Aquat. Mamm. 43, 653– 658. (doi:10.
43. Gagneux P, Boesch C, Woodruff DS. 1997
Microsatellite scoring errors associated with
noninvasive genotyping based on nuclear DNA
amplified from shed hair. Mol. Ecol. 6,
861– 868. (doi:10.1111/j.1365-294X.1997.
44. Taberlet P, Griffin S, Goossens B, Questiau S,
Manceau V, Escaravage N, Waits LP, Bouvet J.
1996 Reliable genotyping of samples with very
low DNA quantities using PCR. Nucleic Acids Res.
24, 3189–3194. (doi:10.1093/nar/24.16.3189)
45. Engelhaupt D et al. 2009 Female philopatry in
coastal basins and male dispersion across the
North Atlantic in a highly mobile marine
species, the sperm whale (Physeter
macrocephalus). Mol. Ecol. 18, 4193– 4205.
46. Raymond M, Rousset F. 1995 GENEPOP (version
1.2): population genetics software for exact tests
and ecumenicism. J. Hered. 86, 248– 249.
47. Kalinowski ST, Taper ML, Marshall TC. 2007
Revising how the computer program cervus
accommodates genotyping error increases
success in paternity assignment. Mol. Ecol. 16,
1099–1106. (doi:10.1111/j.1365-294X.2007.
48. Yoshida H, Yoshioka M, Shirakihara M, Chow S.
2001 Population structure of finless porpoises
(Neophocaena phocaenoides) in coastal waters
of Japan based on mitochondrial DNA
sequences. J. Mammal. 82, 123–130. (doi:10.
1644/1545-1542(2001)082,0123:PSOFPN .2.
49. Alexander A, Steel D, Slikas B, Hoekzema K,
Carraher C, Parks M, Cronn R, Baker CS. 2013 Low
diversity in the mitogenome of sperm whales
revealed by next-generationsequencing. Genome
Biol. Evol. 5, 113– 129. (doi:10.1093/gbe/evs126)
50. Woods JG, Paetkau D, Lewis D, Mclellan BN,
Proctor M, Strobeck C. 1999 Genetic tagging of
free-ranging black and brown bears. Wildl. Soc.
Bull. 27, 616–627.
51. Evett IW, Weir BS. 1998 Interpreting DNA
evidence: statistical genetics for forensic
scientists. Sutherland, MA: Sinauer Associates.
52. Gero S, Whitehead H. 2007 Suckling behavior in
sperm whale calves: observations and
hypotheses. Mar. Mammal Sci. 23, 398 413.
53. Best PB, Canham PAS, Macleod N. 1984 Patterns
of reproduction in sperm whales, Physeter
macrocephalus. Rep. Int. Whal. Commn Spec.
Issue 6, 51–79.
54. Jones OR, Wang J. 2010 Colony: a program for
parentage and sibship inference from multilocus
genotype data. Mol. Ecol. Resour. 10, 551– 555.
55. Kalinowski ST, Wagner AP, Taper ML. 2006 ML-
RELATE: a computer program for maximum
likelihood estimation of relatedness and
relationship. Mol. Ecol. Notes 6, 576– 579.
56. Pew J, Muir PH, Wang J, Frasier TR. 2015
related: an R package for analysing pairwise
relatedness from codominant molecular
markers. Mol. Ecol. Resour. 15, 557– 561.
57. Van De Casteele T, Galbusera P, Matthysen E.
2001 A comparison of microsatellite-based
pairwise relatedness estimators. Mol. Ecol. 10,
1539–1549. (doi:10.1046/j.1365-294X.2001.
58. Csille
´ry K, Johnson T, Beraldi D, Clutton-Brock T,
Coltman D, Hansson B, Spong G, Pemberton JM.
2006 Performance of marker-based relatedness
estimators in natural populations of outbred
vertebrates. Genetics 173, 2091– 2101. (doi:10.
59. Li CC, Weeks DE, Chakravarti A. 1993 Similarity
of DNA fingerprints due to chance and
relatedness. Hum. Hered. 43, 45– 52. (doi:10.
60. Lynch M, Ritland K. 1999 Estimation of pairwise
relatedness with molecular markers. Genetics
152, 1753–1766.
61. Queller D, Goodnight K. 1989 Estimating
relatedness using genetic markers. Evolution 43,
258– 275. (doi:10.1111/j.1558-5646.1989.
62. Wang J. 2002 An estimator for pairwise
relatedness using molecular markers. Genetics
160, 1203–1215.
63. Mantel N. 1967 The detection of disease
clustering and a generalized regression
approach. Cancer Res. 27, 209– 220.
64. Whitehead H. 2009 SOCPROG programs:
analysing animal social structures. Behav. Ecol.
Sociobiol. 63, 765– 778. (doi:10.1007/s00265-
65. Newman MEJ. 2006 Modularity and community
struct ure in netwo rks. Proc.Natl Acad. Sci. USA 103,
8577– 8582. (doi:10.1073/pnas.0601602103)
66. Alexander A, Steel D, Hoekzema K, Mesnick S,
Engelhaupt D, Kerr I, Payne R, Baker CS. 2016
What influences the worldwide genetic structure
of sperm whales (Physeter macrocephalus)? Mol.
Ecol. 25, 2754–2772. (doi:10.1111/mec.13638)
67. Dabrowski MJ, Pilot M, Kruczyk M, Zmihorski M,
Umer HM, Gliwicz J. 2014 Reliability assessment
of null allele detection: inconsistencies between
and within different methods. Mol. Ecol. Resour.
14, 361– 373. (doi:10.1111/1755-0998.12177)
68. Newman MEJ. 2004 Analysis of weighted
networks. Phys. Rev. E 70, 56131. (doi:10.1103/
69. Gero S, Whitehead H. 2016 Critical decline of
the Eastern Caribbean sperm whale population.
PLoS ONE 11, e0162019. (doi:10.1371/journal.
70. Frantzis A,AlexiadouP, GkikopoulouKC. 2014 Sperm
whale occurrence, site fidelity and population
structure along the Hellenic Trench (Greece,
Mediterranean Sea). Aquat. Conserv. Mar. Freshw.
Ecosyst. 24, 83– 102. (doi:10.1002/aqc.2435)
71. Pace DS, Miragliuolo A, Mariani M, Vivaldi C,
Mussi B. 2014 Sociality of sperm whale off
Ischia Island (Tyrrhenian Sea, Italy). Aquat.
Conserv. Mar. Freshw. Ecosyst. 24, 71– 82.
72. Bigg MA, Olesiuk PF, Ellis GM, Ford JKB,
Balcomb KC. 1990 Social organization and
genealogy of resident killer whales (Orcinus
orca) in the coastal waters of British Columbia
and Washington State. Rep. Int. Whal. Commn
Spec. Issue 12, 383–405. (doi:10.1098/rsbl.
73. Weilgart L, Whitehead H, Payne K. 1996 A
colossal convergence. Am. Sci. 84, 278– 287.
74. Archie EA, Moss CJ, Alberts SC. 2006 The ties
that bind: genetic relatedness predicts the
fission and fusion of social groups in wild
African elephants. Proc. R. Soc. B 273,
513– 522. (doi:10.1098/rspb.2005.3361)
75. Wittemyer G, Okello JBA, Rasmussen HB,
Arctander P, Nyakaana S, Douglas-Hamilton I,
Siegismund HR. 2009 Where sociality and
relatedness diverge: the genetic basis for
hierarchical social organization in African
elephants. Proc. R. Soc. B 276, 3513 –3521.
76. Baird RW, Dill LM. 1996 Ecological and social
determinants of group size in transient killer
whales. Behav. Ecol. 7, 408– 416. (doi:10.1093/
77. Silk JB. 2003 Social bonds of female baboons
enhance infant survival. Science 302,
1231–1234. (doi:10.1126/science.1088580)
78. Kurvers RHJM, Krause J, Croft DP, Wilson ADM,
Wolf M. 2014 The evolutionary and ecological
consequences of animal social networks:
emerging issues. Trends Ecol. Evol. 29,
326– 335. (doi:10.1016/j.tree.2014.04.002)
79. Konrad CM, Gero S, Frasier T, Whitehead H.
2018 Data from: Kinship influences sperm whale
social organization within, but generally not
among, social units. Dryad Digital Repository.
( R. Soc. open sci. 5: 180914
... Whales from different vocal clans have distinct coda type and usage repertoires and generally do not associate with each other, even when they occur in sympatry. Variation in these dialects is not consistent with genetic variation [13,32], indicating that coda repertoires are socially learned. Vocal clans are therefore suggested to be a culturally mediated form of population structure [32,33]. ...
... Female sperm whales generally inhabit tropical pelagic waters while males disperse to high latitudes at the onset of maturity [11]. Females have a multi-level social structure centred around social units of one or two matrilines [13,14]. These social units usually contain 6-12 individuals and are stable over time miles from coast), leeward offshore (15 nautical miles from coast) and windward inshore (5-7 nautical miles from coast) (electronic supplementary material, figure S1). ...
... However, the observations that EC vocal clans were substantially restricted, during our two-year survey period, to specific islands (or pairs of neighbouring islands) in the Lesser Antilles, and that these preferences are maintained in sympatry without evidence of nuclear genetic differentiation [13], suggest that differences in habitat use between EC1 and EC2 are mostly culturally driven. The exact mechanism responsible for such a divide remains unknown and is an important focus for future research, but we consider several potential explanations for this spatio-temporal pattern of behavioural variation below. ...
Full-text available
The sperm whale ( Physeter macrocephalus ) is a deep-diving cetacean with a global distribution and a multi-leveled, culturally segregated, social structure. While sperm whales have previously been described as ‘ocean nomads’, this might not be universal. We conducted surveys of sperm whales along the Lesser Antilles to document the acoustic repertoires, movements and distributions of Eastern Caribbean (EC) sperm whale cultural groups (called vocal clans). In addition to documenting a potential third vocal clan in the EC, we found strong evidence of fine-scale habitat partitioning between vocal clans with scales of horizontal movements an order of magnitude smaller than from comparable studies on Eastern Tropical Pacific sperm whales. These results suggest that sperm whales can display cultural ecological specialization and habitat partitioning on flexible spatial scales according to local conditions and broadens our perception of the ecological flexibility of the species. This study highlights the importance of incorporating multiple temporal and spatial scales to understand the impact of culture on ecological adaptability, as well as the dangers of extrapolating results across geographical areas and cultural groups.
... Mitochondrial and nuclear loci were amplified from all DNA extracts, allowing for an analysis of the polymorphisms at 638 bp of the MCR as well as at 16 microsatellite loci (see the electronic supplementary material, table S3 for the number of alleles per locus). No major variations in the quality of the DNA extracted from the skin samples were noticed, contrasting with previous results [66,75]. This is most likely owing to the fact that the skin samples were taken immediately after their release from the whales' body. ...
... The matrilineality of Irène's group is nevertheless strongly confirmed here by our results. This agrees with the findings of Konrad et al. [75] for the Atlantic Ocean, though they identified only two mtDNA haplotypes in 12 different social groups, which was low to conclude about matrilineality. In Irène's group, only one adult female, Claire, possesses a different haplotype, SW_MCK1, which differs by two mutations from SW_M1. ...
... Two other females, Vanessa and Yukimi, despite sharing the same haplotype as the rest of the group, have no or few second-degree relationships. Konrad et al. [75] showed that some individuals may present no clear genetic relationships with other members of their social groups. Transfers of sperm whales R. Soc. ...
Full-text available
Understanding the organization and dynamics of social groups of marine mammals through the study of kin relationships is particularly challenging. Here, we studied a stable social group of sperm whales off Mauritius, using underwater observations, individual-specific identification, non-invasive sampling and genetic analyses based on mitochondrial sequencing and microsatellite profiling. Twenty-four sperm whales were sampled between 2017 and 2019. All individuals except one adult female shared the same mitochondrial DNA (mtDNA) haplotype—one that is rare in the western Indian Ocean—thus confirming with near certainty the matrilineality of the group. All probable first- and second-degree kin relationships were depicted in the sperm whale social group: 13 first-degree and 27 second-degree relationships were identified. Notably, we highlight the likely case of an unrelated female having been integrated into a social unit, in that she presented a distinct mtDNA haplotype and no close relationships with any members of the group. Investigating the possible matrilineality of sperm whale cultural units (i.e. vocal clans) is the next step in our research programme to elucidate and better apprehend the complex organization of sperm whale social groups.
... One major area of acoustic research focuses within species, often with the goal of identifying the core vocal repertoire of a species or population, in terms of both acoustic characteristics and functions of vocalizations [20][21][22][23]. Vocalizations can vary with demographic features such as sex, age, and breeding status [24,25], behavioral states [26][27][28][29], social structures [30,31], and even species morphology [32][33][34][35]. More broadly, acoustic methods can be used to investigate topics related to population structure such as Figure 1. ...
... One major area of acoustic research focuses within species, often with the goal of identifying the core vocal repertoire of a species or population, in terms of both acoustic characteristics and functions of vocalizations [20][21][22][23]. Vocalizations can vary with demographic features such as sex, age, and breeding status [24,25], behavioral states [26][27][28][29], social structures [30,31], and even species morphology [32][33][34][35]. More broadly, acoustic methods can be used to investigate topics related to population structure such as distribution and speciation across habitats or regions [12,[36][37][38][39][40], and population densities [41][42][43][44]. ...
Full-text available
The field of bioacoustics is rapidly developing and characterized by diverse methodologies, approaches and aims. For instance, bioacoustics encompasses studies on the perception of pure tones in meticulously controlled laboratory settings, documentation of species’ presence and activities using recordings from the field, and analyses of circadian calling patterns in animal choruses. Newcomers to the field are confronted with a vast and fragmented literature, and a lack of accessible reference papers or textbooks. In this paper we contribute towards filling this gap. Instead of a classical list of “dos” and “don’ts”, we review some key papers which, we believe, embody best practices in several bioacoustic subfields. In the first three case studies, we discuss how bioacoustics can help identify the ‘who’, ‘where’ and ‘how many’ of animals within a given ecosystem. Specifically, we review cases in which bioacoustic methods have been applied with success to draw inferences regarding species identification, population structure, and biodiversity. In fourth and fifth case studies, we highlight how structural properties in signal evolution can emerge via ecological constraints or cultural transmission. Finally, in a sixth example, we discuss acoustic methods that have been used to infer predator–prey dynamics in cases where direct observation was not feasible. Across all these examples, we emphasize the importance of appropriate recording parameters and experimental design. We conclude by highlighting common best practices across studies as well as caveats about our own overview. We hope our efforts spur a more general effort in standardizing best practices across the subareas we’ve highlighted in order to increase compatibility among bioacoustic studies and inspire cross-pollination across the discipline.
... MLSs have been reported in a range of marine and terrestrial mammals (e.g. bottlenose dolphins (Tursiops spp.), killer whales, Orcinus orca, sperm whales, Physeter macrocephalus, plains zebras, Equus quagga, feral horses, Equus caballus, and African elephants, Loxodonta africana) and are best known in primates (Archie & Alberts, 2006;Connor et al., 1992;de Stephanis et al., 2008;Feh, 2005;Grueter et al., 2020;Konrad et al., 2018;Rubenstein & Hack, 2004). In various marine species and in African elephants, the core units are composed of closely associated breeding females and their offspring, with occasional male visitors . ...
... This pattern has been reported in other MLS-based species. In sperm whales, kinship does not explain the association preferences between most social units, although matrilineality exists (Konrad et al., 2018). In African elephants, kinship does not predict higher levels of association and associations between units are not strictly kin based (Archie & Alberts, 2006). ...
Full-text available
Various primate species, including golden snub-nosed monkeys, Rhinopithecus roxellana, form highly complex multilevel social systems. Breeding bands of snub-nosed monkeys maintain stability even though multiple one-male units (OMUs) within a band do not appear to engage in affiliative behaviours among units. As such, female dispersal within breeding bands may be a key factor related to social stability in these complex social systems. However, how individual and group behaviours influence both the formation of this complex social system and the maintenance of social stability within it remain to be fully elucidated. In the current study, based on 16 years of accumulated data, we investigated female dispersal in a wild population of golden snub-nosed monkeys in the Qinling Mountains of China. Using social network analysis and dynamic social modelling, we found that male take-over had no influence on the dominance rank of an OMU, whereas the number of breeding females within an OMU had a significant positive effect on rank. Both the number of breeding females and the number of females entering an OMU had significant positive effects on eigenvector centrality of an OMU. Female dispersal between OMUs decreased the clustering coefficient and reduced the risk of subgroup formation. In addition, female kinship matrices between units were not significantly correlated with association behaviour patterns. Thus, this study sheds light on the internal mechanisms that drive social stability in a complex primate social system.
... Male and female sperm whales live in societies that are strongly geographically segregated post-maturity (e.g., Christal, 1998;Gordon et al., 1998;Christal and Whitehead, 1999;Lyrholm et al., 1999;Whitehead et al., 2008;Labadie et al., 2018). Adult females form social units with immatures, stable over time and found all year round in warm waters at low latitudes (Whitehead and Kahn, 1992;Konrad et al., 2018;Sarano et al., 2021). In contrast, males disperse from their natal group after 6-8 years, before their sexual maturity, and move poleward to areas abundant in food (Rice, 1989). ...
... Here, we confirm and extend these observations in the Indian Ocean. The level of this male social fidelity (e.g., for social units, for vocal clans, defined in Konrad et al., 2018) is still to be evaluated. ...
Full-text available
Adult male sperm whales (Physeter macrocephalus) are long distance runners of the marine realm, feeding in high latitudes and mating in tropical and subtropical waters where stable social groups of females and immatures live. Several areas of uncertainty still limit our understanding of their social and breeding behavior, in particular concerning the potential existence of geographical and/or social fidelities. In this study, using underwater observation and sloughed-skin sampling, we looked for male social fidelity to a specific matrilineal sperm whale group near Mauritius. In addition, we captured a wider picture of kin relationships and genetic diversity of male sperm whales in the Indian Ocean thanks to biopsies of eight individuals taken in a feeding ground near the Kerguelen and Crozet Archipelagos (Southern Indian Ocean). Twenty-six adult male sperm whales were identified when socializing with adult females and immatures off Mauritius. Sloughed-skin samples were taken from thirteen of them for genetic analysis. Long-term underwater observation recorded several noteworthy social interactions between adult males and adult females and/or immatures. We identified seven possible male recaptures over different years (three by direct observation, and four at the gametic level), which supports a certain level of male social fidelity. Two probable first- and thirty second-degree kin relationships were highlighted between members of the social unit and adult males, confirming that some of the adult males observed in Mauritian waters are reproductive. Male social philopatry to their natal group can be excluded, as none of the males sampled shared the haplotype characteristic of the matrilineal social group. Mitochondrial DNA control region haplotype and nucleotide diversities calculated over the 21 total male sperm whales sampled were similar to values found by others in the Indian Ocean. Our study strongly supports the existence of some levels of male sperm whale social fidelity, not directed to their social group of birth, in the Indian Ocean. Males sampled in breeding and feeding grounds are linked by kin relationships. Our results support a model of male mediated gene flow occurring at the level of the whole Indian Ocean, likely interconnected with large-scale geographical fidelity to ocean basin, and a small-scale social fidelity to matrilineal social groups.
... Measuring the correlation between social relationships and kinship requires either pedigrees derived from observed maternities, which take decades to estimate with confidence in long-lived mammals and only provides information about maternal relatedness, or genetic data which are often not available in cetacean populations. Studies of kin structuring in our review were limited to pilot whales (Alves et al., 2013;Van Cise et al., 2017), killer whales (e.g., Esteban et al., 2016a;Reisinger et al., 2017), sperm whales (e.g., Gero et al., 2008;Konrad et al., 2018a), and bottlenose dolphins (e.g., . The lack of studies on smaller dolphins and beaked whales means our picture here is incomplete, and our knowledge is clearly taxonomically biased toward species with stable social units. ...
... Unlike these species, however, sperm whale males disperse at maturity (Whitehead, 2003), and social units may contain multiple matrilines (Richard et al., 1996). Variations in kinship drive social association rates within units , however, kinship between units does not appear to predict cross-unit affiliation patterns (Konrad et al., 2018a). ...
Full-text available
Toothed whales (suborder Odontoceti) are highly social, large brained mammals with diverse social systems. In recent decades, a large body of work has begun investigating these dynamic, complex societies using a common set of analytical tools: social network analysis. The application of social network theory to toothed whales enables insight into the factors that underlie variation in social structure in this taxon, and the consequences of these structures for survival, reproduction, disease transmission, and culture. Here, we perform a systematic review of the literature regarding toothed whale social networks to identify broad patterns of social network structure across species, common drivers of individual social position, and the consequences of network structure for individuals and populations. We also identify key knowledge gaps and areas ripe for future research. We recommend that future studies attempt to expand the taxonomic breadth and focus on standardizing methods and reporting as much as possible to allow for comparative analyses to test evolutionary hypotheses. Furthermore, social networks analysis may provide key insights into population dynamics as indicators of population health, predictors of disease risk, and as direct drivers of survival and reproduction.
... Lowland gorillas, for instance, form harems with one silver-back male and several females (Hagemann et al., 2019). Another highly social vertebrate is the sperm whale, a mammal that can form multilevel social structures based on smaller long-term groups called social units (Konrad et al., 2018). Social units are comprised of either a female and younger whales (typically offspring), or a group of mature males (Konrad et al., 2018). ...
... Another highly social vertebrate is the sperm whale, a mammal that can form multilevel social structures based on smaller long-term groups called social units (Konrad et al., 2018). Social units are comprised of either a female and younger whales (typically offspring), or a group of mature males (Konrad et al., 2018). As a final example, Dungan et al. (2016) showed that the social alignment of Indo-Pacific humpback dolphins, a small and isolated population, is centralized around mother-calf rearing groups and that they form both long-term (years) and short-term (hours-days) social associations. ...
Full-text available
• The Cormack–Jolly–Seber (CJS) model and its extensions have been widely applied to the study of animal survival rates in open populations. The model assumes that individuals within the population of interest have independent fates. It is, however, highly unlikely that a pair of animals which have formed a long-term pairing have dissociated fates. • We examine a model extension which allows animals who have formed a pair-bond to have correlated survival and recapture fates. Using the proposed extension to generate data, we conduct a simulation study exploring the impact that correlated fate data has on inference from the CJS model. We compute Monte Carlo estimates for the bias, range, and standard errors of the parameters of the CJS model for data with varying degrees of survival correlation between mates. Furthermore, we study the likelihood ratio test of sex effects within the CJS model by simulating densities of the deviance. Finally, we estimate the variance inflation factor for CJS models that incorporate sex-specific heterogeneity. • Our study shows that correlated fates between mated animals may result in underestimated standard errors for parsimonious models, significantly deflated likelihood ratio test statistics, and underestimated values of for models taking sex-specific effects into account. • Underestimated standard errors can result in lowered coverage of confidence intervals. Moreover, deflated test statistics will provide overly conservative test results. Finally, underestimated variance inflation factors can lead researchers to make incorrect conclusions about the level of extra-binomial variation present in their data.
... Killer whales (Orcinus orca) are one of the best examples of matrilineal societies, where pods are formed of matriarchal hierarchical social structures, stable for decades (Bigg et al., 1990;Baird and Whitehead, 2000;Parsons et al., 2009). Similarly, sperm whales (Physeter macrocephalus) maintain strong bonds between units of females and calves, even when associated with other units Konrad et al., 2018). On the other hand, fission-fusion socialities are dynamic associations of individuals, which vary in group size, composition, and cohesion over time (Aureli et al., 2008). ...
Full-text available
Beluga whales (Delphinapterus leucas) are considered social whales, but like any other cetaceans, the study of social behaviour is challenging to conduct. Due to the wide distribution of the Eastern Beaufort Sea beluga whale population across its summering grounds, little is known about the large-scale grouping behaviour and spatial distribution of groups. The aim of this research is to explore the grouping characteristics and organization of beluga groups, as well as the habitat preference of different social groups in summer. First, we used aerial photographs captured in July 2019 to describe group size, age composition, inter-individual distance, and swimming direction of beluga groups. We compared characteristics between two key summer habitats: the extended offshore of the Beaufort Sea shelf and the inshore of the Mackenzie Estuary. Results showed that group size and inter-individual distance were similar in both habitats. The average distance in a group varied with age composition and the swimming direction varied between the offshore and inshore. Second, we used GPS locations of beluga sightings recorded by visual observers during aerial surveys conducted in July and August 2019. We investigated the distribution of three beluga social group types (individual belugas, groups of adults, and groups with calf) using hierarchical generalized additive models. The sea surface temperature, bathymetry, and slope described best the summer distribution. Areas of high preference were often associated with prey distribution, suggesting foraging as the main driver of habitat preference. We also hypothesized that body size energy requirements contributed to the variation between the group types. This study revealed for the first-time observations of grouping behaviour in the summer habitat of the Eastern Beaufort Sea beluga whales. Although the results do not reflect the extent and complexity of beluga social behaviour, this study now provides an information baseline for this beluga population. We also encourage multidisciplinary research as an opportunity to further collect data and explore other elements of beluga whale sociality. iii
Full-text available
Adult male sperm whales (Physeter macrocephalus) are long distance runners of the marine realm, feeding in high latitudes and mating in tropical and subtropical waters where stable social groups of females and immatures live. Several areas of uncertainty still limit our understanding of their social and breeding behaviour, in particular concerning the potential existence of geographical and/or social fidelities. In this study, using underwater observation and sloughed-skin sampling, we looked for male social fidelity to a specific matrilineal sperm whale group near Mauritius. In addition, we captured a wider picture of kin relationships and genetic diversity of male sperm whales in the Indian Ocean thanks to biopsies of eight unique individuals taken in a feeding ground near the Kerguelen and Crozet Archipelagos (Southern Indian Ocean). Twenty-six adult male sperm whales, of which 13 were sampled, were identified when socializing with adult females and immatures off Mauritius. Long-term underwater observation recorded several noteworthy social interactions between adult males and adult females and/or immatures. We identified seven possible male recaptures over different years (three by direct observation, and four at the gametic level), which supports a certain level of male social fidelity. Several first- and second-degree kin relationships were highlighted between members of the social unit and adult males, confirming that some of the adult males observed in Mauritian waters are reproductive. Male social philopatry to their natal group can be excluded, as none of the males sampled shared the haplotype characteristic of the matrilineal social group. Mitochondrial DNA control region haplotype and nucleotide diversities calculated over the 21 total male sperm whales sampled were similar to values found by others in the Indian Ocean. Our study strongly supports the existence of some levels of male sperm whale social fidelity, not directed to their social group of birth, in the Indian Ocean. Males sampled in breeding and feeding grounds are linked by kin relationships. Our results support a model of male mediated gene flow occurring at the level of the whole Indian Ocean, likely interconnected with large-scale geographical fidelity to ocean basin, and a small-scale social fidelity to matrilineal social groups.
The fully aquatic lifestyle of dugongs means that direct observation of social tusk use is not usually possible. This study used body scarring as an indicator of tusk function by males. Tusk rake scars on 298 live wild dugongs, of both sexes and all sizes, were categorized and counted in over 1,000 photographs, and examined in relation to maturity and reproductive activity over seasons. All dugongs had tusk scars, but adults were the main recipients. Sexually active adults acquired the greatest number of fresh tusk wounds during the mating season. Subadults received fresh rakes at similar numbers year‐round. Adult males had more scars on the mid and posterior dorsum, indicating that males direct combative force to these regions of the male body when competing for females. Adult females had heaviest scarring and more tusk puncture wounds on the anterior‐mid dorsum and head, suggesting that male dugongs use tusks in sexual coercion. Heavy scarring sustained by solitary calves compared to dependent ones, suggests that mothers afford some protection. Body scarring caused by tusks may serve as an indicator of reproductive contribution of the recipients, providing that successful males are involved in more reproductive competitions, and successful females in more mating events.
Full-text available
Sperm whale (Physeter macrocephalus) populations were expected to rebuild following the end of commercial whaling. We document the decline of the population in the eastern Caribbean by tracing demographic changes of well-studied social units. We address hypotheses that, over a ten-year period of dedicated effort (2005–2015), unit size, numbers of calves and/or calving rates have each declined. Across 16 units, the number of adults decreased in 12 units, increased in two, and showed no change in two. The number of adults per unit decreased at -0.195 individuals/yr (95% CI: -0.080 to -0.310; P = 0.001). The number of calves also declined, but the decline was not significant. This negative trend of -4.5% per year in unit size started in about 2010, with numbers being fairly stable until then. There are several natural and anthropogenic threats, but no well-substantiated cause for the decline.
Full-text available
The 'social complexity hypothesis' suggests that complex social structure is a driver of diversity in animal communication systems. Sperm whales have a hierarchically structured society in which the largest affiliative structures, the vocal clans, are marked on ocean-basin scales by culturally transmitted dialects of acoustic signals known as 'codas'. We examined variation in coda repertoires among both individual whales and social units-the basic element of sperm whale society-using data from nine Caribbean social units across six years. Codas were assigned to individuals using photo-identification and acoustic size measurement, and we calculated similarity between repertoires using both continuous and categorical methods. We identified 21 coda types. Two of those ('1+1+3' and '5R1') made up 65% of the codas recorded, were shared across all units and have dominated repertoires in this population for at least 30 years. Individuals appear to differ in the way they produce '5R1' but not '1+1+3' coda. Units use distinct 4-click coda types which contribute to making unit repertoires distinctive. Our results support the social complexity hypothesis in a marine species as different patterns of variation between coda types suggest divergent functions, perhaps representing selection for identity signals at several levels of social structure.
A new method is described for estimating genetic relatedness from genetic markers such as protein polymorphisms. It is based on Grafen's (1985) relatedness coefficient and is most easily interpreted in terms of identity by descent rather than as a genetic regression. It has several advantages over methods currently in use: it eliminates a downward bias for small sample sizes; it improves estimation of relatedness for subsets of population samples; and it allows estimation of relatedness for a single group or for a single pair of individuals. Individual estimates of relatedness tend to be highly variable but, in aggregate, can still be very useful as data for nonparametric tests. Such tests allow testing for differences in relatedness between two samples or for correlating individual relatedness values with another variable.
The interplay of natural selection and genetic drift, influenced by geographic isolation, mating systems, and population size, determines patterns of genetic diversity within species. The sperm whale provides an interesting example of a long-lived species with few geographic barriers to dispersal. Worldwide mtDNA diversity is relatively low, but highly structured among geographic regions and social groups, attributed to female philopatry. However, it is unclear if this female philopatry is due to geographic regions or social groups, or how this might vary on a worldwide scale. To answer these questions, we combined mtDNA information for 1,091 previously published samples with 542 newly obtained DNA profiles (394 bp mtDNA, sex, 13 microsatellites) including the previously un-sampled Indian Ocean, and social group information for 541 individuals. We found low mtDNA diversity (π=0.430%) reflecting an expansion event <80,000 years bp, but strong differentiation by ocean, among regions within some oceans, and among social groups. In comparison, microsatellite differentiation was low at all levels, presumably due to male-mediated gene flow. A hierarchical AMOVA showed that regions were important for explaining mtDNA variance in the Indian Ocean, but not Pacific, with social group sampling in the Atlantic too limited to include in analyses. Social groups were important in partitioning mtDNA and microsatellite variance within both oceans. Therefore, both geographic and social philopatry influence genetic structure in the sperm whale, but their relative importance differs by sex and ocean, reflecting breeding behavior, geographic features, and perhaps a more recent origin of sperm whales in the Pacific. By investigating the interplay of evolutionary forces operating at different temporal and geographic scales, we show that sperm whales are perhaps a unique example of a worldwide population expansion followed by rapid assortment due to female social organization.
To clarity population structure of finless porpoises (Neophocaena phocaenoides) in Japan, a sequence of studies have been conducted: a questionnaire survey and aerial sighting surveys to obtain information on porpoise distribution, and analyses of skull morphology and mitochondrial DNA control region sequences to examine their geographic variation. The first two surveys indicated that finless porpoises are mainly distributed in five coastal regions in Japan (Sendai Bay-Tokyo Bay, Ise-Mikawa Bays, Inland Sea-Hibiki Nada, Omura Bay, and Ariake Sound-Tachibana Bay) and that occurrence of animals is rare in the other areas. The latter two analyses revealed geographic differences in the morphology and sequences among the five areas. Finless porpoises in each of the five regions are considered to belong to distinct populations.