scientific article

D ow

nl oa

de d

fr om

h ttp

s: //r

oy al

so ci

et yp

ub lis

hi ng

.o rg

/ o n

29 S

ep te

m be

r 20

21

rspb.royalsocietypublishing.org

Research Cite this article: Walker RH, King AJ, McNutt

JW, Jordan NR. 2017 Sneeze to leave: African

wild dogs (Lycaon pictus) use variable quorum

thresholds facilitated by sneezes in collective

decisions. Proc. R. Soc. B 284: 20170347.

http://dx.doi.org/10.1098/rspb.2017.0347

Received: 18 February 2017

Accepted: 31 July 2017

Subject Category: Behaviour

Subject Areas: behaviour

Keywords: African wild dog (Lycaon pictus), signal,

consensus decision-making,

social communication

Author for correspondence: Reena H. Walker

e-mail: [email protected]

Electronic supplementary material is available

online at https://dx.doi.org/10.6084/m9.

figshare.c.3850912.

& 2017 The Author(s) Published by the Royal Society. All rights reserved.

Sneeze to leave: African wild dogs (Lycaon pictus) use variable quorum thresholds facilitated by sneezes in collective decisions

Reena H. Walker1,2, Andrew J. King3, J. Weldon McNutt1

and Neil R. Jordan1,4,5

1Botswana Predator Conservation Trust, Private Bag 13, Maun, Botswana 2Brown University, Providence, RI 02912, USA 3Department of Biosciences, College of Science, Singleton Park, Swansea University, Swansea SA2 8PP, UK 4Centre for Ecosystem Science, University of New South Wales, Sydney, New South Wales 2052, Australia 5Applied Eco-Logic Group, Taronga Western Plains Zoo, Obley Rd, Dubbo, New South Wales 2830, Australia

RHW, 0000-0003-3584-5399; AJK, 0000-0002-6870-9767; NRJ, 0000-0002-0712-8301

In despotically driven animal societies, one or a few individuals tend to have a

disproportionate influence on group decision-making and actions. However,

global communication allows each group member to assess the relative

strength of preferences for different options among their group-mates. Here,

we investigate collective decisions by free-ranging African wild dog packs in

Botswana. African wild dogs exhibit dominant-directed group living and

take part in stereotyped social rallies: high energy greeting ceremonies

that occur before collective movements. Not all rallies result in collective move-

ments, for reasons that are not well understood. We show that the probability

of rally success (i.e. group departure) is predicted by a minimum number of

audible rapid nasal exhalations (sneezes), within the rally. Moreover, the

number of sneezes needed for the group to depart (i.e. the quorum) was

reduced whenever dominant individuals initiated rallies, suggesting that

dominant participation increases the likelihood of a rally’s success, but is not

a prerequisite. As such, the ‘will of the group’ may override dominant prefer-

ences when the consensus of subordinates is sufficiently great. Our findings

illustrate how specific behavioural mechanisms (here, sneezing) allow for

negotiation (in effect, voting) that shapes decision-making in a wild, socially

complex animal society.

1. Background Group consensus is ubiquitous in social invertebrate and vertebrate animals [1]

and is necessary for individuals to reap the benefits of group living—including

added protection from predators, greater information sharing and better

defence of resources [2]. One of the most obvious instances of group coordi-

nation in social animals is the decision to move off from a resting spot [3].

Signals used by individuals in the pre-departure and foraging stage of group

movement have been described across taxa [4] and often operate in a type of

quorum, where a specific signal has to reach a certain threshold before the

group changes activity [4,5]. This ensures that a minimum number of individ-

uals (the actual quorum number) are ready to move off [4]. Past research in

meerkats, Suricata suricatta, for example, has found that a quorum of at least

two and usually three meerkats emitting ‘moving calls’ are necessary for the

whole group to move to a new foraging patch, and ‘piping signals’ in honey-

bees, Apis mellifera [6], and ‘trills’ in white faced capuchin monkeys, Cebus capucinus, [7] are required for collective departures to occur.

Certain individuals can also have a disproportionate influence on

collective behaviour decisions within social systems that exhibit variation in

15

10

5

0 0.15 0.30 0.45 time (s)

fr eq

ue nc

y (k

H z)

Figure 1. Spectrogram of dominant male African wild dog ‘sneeze’ recorded prior to a group departure event. This example spectrogram was prepared in CoolEdit Pro 2002 (v. 2.0, Syntrillium Software Corporation, Phoenix, AZ), with 44 100 sampling rate visualized in Hamming window, resolution 1024 bands, and linear energy plot at 20% scaling. The spectrogram shows linear bars (likely an intake of breath), followed by atonal high- frequency bandwidth rapid exhalation, or ‘sneeze’. Energy is shown from light (low) to dark (high).

rspb.royalsocietypublishing.org Proc.R.Soc.B

284:20170347

2

D ow

nl oa

de d

fr om

h ttp

s: //r

oy al

so ci

et yp

ub lis

hi ng

.o rg

/ o n

29 S

ep te

m be

r 20

21

inter-individual relationships (e.g. kinship and dominance

structures, see [8,9]). For example, dominance rank and/or

an individual’s social role (measured as social affiliation

strength to others) are often found to correlate with leader-

ship roles, a phenomenon observed pervasively in primates

[10]. In social canids, research on group living has focused

largely on the role of dominants in directing and repressing

subdominant behaviour in group decision-making [11].

Here, we investigate the collective decisions of African wild

dog (Lycaon pictus) packs in Botswana during the transition

from a sedentary resting state to an active moving state. African

wild dogs are the ‘most social canid’ and exhibit uniquely

non-aggressive, dominant directed group living, exemplified

by stereotyped social rallies [12–14]: high energy, socially intri-

cate pre-departure greeting ceremonies that are ‘conspicuous’,

‘highly ritualized’, and are ‘of high adaptive value. . .and serve

to hold the pack together’ [13]. Dominant breeding pairs in an

African wild dog pack affect the behaviour of the pack as a

whole; the dominant-directed social system facilitates feeding

by pups at kill sites [15], suppresses sub-dominant pregnancies

[16], and ensures collective care for a denning female and

pups [17]. However, little is known about the extent to which

dominants, or single individuals, drive behaviour outside the

reproductive realm. Sueur & Petit [3], assert that African wild

dogs (Lycaon pictus) likely use ‘shared consensus’, in which all

group members participate in the decision-making process,

because their ‘open social system’ is defined by pervasive

cooperation. However, no study has systematically investigated

how these social carnivores make collective decisions.

Given that African wild dog packs are characterized by per-

vasive cooperation [12,14] and show intricate pre-departure

greeting ceremonies [12–14], we expected a majority or all

group members to participate in group consensus about depar-

tures. However, because dominant individuals are known to

steer many types of group activities [16–18], we expected

dominants to have a disproportionate influence in this process.

We therefore tested the overall hypothesis that African wild

dogs exhibit ‘partially shared consensus’ decisions [3,18].

It is known in several other animal species that the number

and identity of individuals participating in the decision process

can influence the outcome of collective decisions, and that valu-

able experience may be correlated with age or dominance

[9,19,20]. Moreover, specific recruitment cues or signals may

help guide conspecifics [9] or even be used as a type of voting

mechanism [4]. Therefore, to understand the mechanisms by

which packs reach a consensus [15] we gathered data relating

to the proportion of the pack engaged in social behaviour, indi-

vidual participation, and the role of potential communication

mechanisms to negotiate timing of departure. Since African

wild dogs display dominant-directed group living [20,21] we

examined to what extent individual participation in rallies,

and specifically the dominants’ participation, affected the

likelihood of a successful group movement. Preliminary obser-

vations during rallies indicated that audible, abrupt exhalations

of air through the nose, ‘sneezes’ (figure 1; see electronic sup-

plementary material video), appeared to be frequent during

rallies and may serve as a pre-departure cue or signal [15].

Therefore we investigated the potential for the occurrence of

sneezes to serve as a voting mechanism that determines

whether the pack should depart [4,22] while also considering

the relative importance of other factors: the dominance status

of the initiator [9], the level of social participation [10], and

the number of other departure events that day [11].

2. Methods Data were collected from five packs (�X + s.d. adult group size ¼

8.80+3.63) of African wild dogs in and around the Moremi

Game Reserve in the Okavango Delta from June 2014 to May

2015. At least one individual in each pack was fitted with a VHF

radio collar (ca. 180 g; Sirtrack, Havelock West, New Zealand)

using darting and immobilization procedures described previous-

ly [23]. Collars allowed packs to be located and were replaced

when they failed. Some individuals remained collared following

the completion of this study as they formed part of a long-term

study conducted by the Botswana Predator Conservation Trust

(BPCT) spanning the past 25 years [20]. All individuals (N ¼ 49)

were identified by their unique pelage patterns, and ages and

life histories were known for all individuals except some immi-

grants (N ¼ 10). We estimated the age classes (adult, yearling, or

pup) of these 10 individuals using a combination of body size,

pelage development, testicular development, and tooth and

ear wear. All work was conducted in accordance with the guide-

lines for the treatment of animals in behavioural research and

teaching [24].

To explore the dynamics of collective movement decisions,

packs were observed from a vehicle (N ¼ 52 days; �X + s.e. days/

pack/month¼ 2.03+0.50), and their behaviours were recorded

during rally periods via direct observation (scan and continuous

sampling) and video recordings (Nikon, COOLPIX S7000). Rallies

were initiated when an individual rose from rest in the distinctive

initiation posture: head lowered, mouth open, and ears folded

back [13]. These initiators were identified. Not all rallies resulted

in collective movements, and rallies were considered to have

ended when all individuals either returned to rest or departed

the resting site. We observed 1.92+0.54 (�X + s.e.) rallies per

observation session (N¼ 68 rallies; �X + s.e. per pack ¼ 14.2+6.75).

From video data, we performed behavioural scans every 5 s

from initiation until the end of the rally. We used critical incident

sampling to record the number of audible, abrupt exhalation of

air through the nose, or ‘sneezes’, during rally attempts and cal-

culated the aggregated frequency of sneeze events per minute

before and after the end of rallies. ‘Sneezes’ are atonal high-

frequency bandwidth rapid exhalations that are stereotyped

rspb.royalsocietypublishing.org Proc.R.Soc.B

284:20170347

3

D ow

nl oa

de d

fr om

h ttp

s: //r

oy al

so ci

et yp

ub lis

hi ng

.o rg

/ o n

29 S

ep te

m be

r 20

21

and obvious in rallies (figure 1). We observed sneezes while indi-

viduals were walking with their heads hanging or standing with

their ears alert and tail relaxed (electronic supplementary

material video). Other dogs did not startle in response to these

vocalizations, or look toward the sneezer, as might be expected

if the sounds were associated with a threat display or a sign of

alarm. While it was clear from video data how many sneezes

occurred during a rally, the thick habitat prevented us from

being able to routinely identify which individuals sneezed, and

so we only measure frequency, and not identity, of sneezers.

For each behavioural scan, we recorded which individuals

participated in one or more of three stereotypical social inter-

actions: ‘Greet’, when individuals touched heads or approached

within 1 m of one another; ‘Parallel Run’, when individuals ran

flank to flank; and ‘Mob’, when three or more individuals gathered

within 1 m of one another [12,13]. The proportion of adults parti-

cipating in these interactions ranged from 0 (rallies in which

there was no social behaviour or only yearlings and pups inter-

acted) to 1 (rallies in which all adults were actively engaged at

one point, though not necessarily simultaneously).

In a variety of animal systems, the identity, social status, or

age–sex class of the individual initiating a collective movement

(i.e. moving away from the resting group) can be critical to the

likelihood of a collective departure [9,18,19]. Because relative

rank beneath the dominant pair is not readily decipherable

within African wild dog packs, we used priority of access to

carcasses (POA) as a proxy for dominance: the dominant pair

and their pups (less than 1 year) have first access to kills

(POA1), followed by yearlings, (POA2), and subdominant

adults (more than 2 years) (POA3) [20,21].

We used simple bivariate tests, such as chi-square and the

binomial test for equality of proportions conducted in the pack-

age R with significance level 0.05, to initially explore

relationships between rally success (departure/no departure)

and recorded observations of order of rally attempt, proportion

of adults participating in social behaviour, dominants’ partici-

pation in rallies, number of sneezes, and initiator demographic

[25]. To further investigate the factors affecting whether a social

rally resulted in the pack departing (1) or not (0) from their cur-

rent rest site, we ran a series of binomial generalized linear mixed

effects models (GLMMs) in the package ‘lme4’ [26] in R [25].

Eleven out of 68 rallies were excluded from these specific ana-

lyses as their ultimate success or failure and/or the identity of

the initiator was not determined. Terms included in the model

set were: total number of sneezes in a rally, the initiator’s priority

of access to kills (1, 2, 3), consecutive attempt number per obser-

vation session, and the proportion of adults participating in

social behaviours. Pack identity was included as a random

term in the models to control for repeated measures. We used

Akaike’s information criterion to select the most plausible

model from a set of credible options. All terms and their two-

way interactions were sequentially added to the basic model,

with each retained only if it reduced the AIC by two or more

as lower AIC values correspond to better relative support for

each model [27]. To validate that there was no improvement to

the minimal model, each term was then removed sequentially

from the minimal model. Terms were retained only if their

removal inflated AIC by more than two [28] As the Akaike

weight of the best model was less than 0.9 and several models

had deviance in the AIC lower than seven units [29,30], we con-

ducted model averaging using the MuMIn package [31]. We

selected the top models whose cumulative AIC weights were

more than 0.95 to construct model-averaged estimates of the par-

ameters [28] Model diagnostics were performed by inspection

using the DHARMa package, which uses a simulation-based

approach to create readily interpretable scaled residuals from

fitted GLMMs [32] Data from all top models included in model

averaging met model assumptions.

3. Results and discussion We first explored whether the likelihood of a rally resulting in

the group’s departure increased with every failed rally. We

found that first rallies rarely (26%, 9/34) ended in movement

away from resting spots, but the likelihood of a successful

collective movement increased over successive rallies

(figure 2a); 64% (5/8) of third rallies were successful (table 1).

The �X + s.e. proportion of adult social participation in all

recorded rallies (N ¼ 68) was 0.58+0.36, but variation in pro-

portion of adult participation was not strongly related to rally

success (table 1; supplemental social participation results).

We positively identified the initiator in 84% (57/68) of

total observed rallies, 44% (25/57) of which were successful.

We found that rally success was influenced by initiator demo-

graphics (table 1; figure 2b); rally attempts initiated by POA1

individuals (76.5% successful, N ¼ 17) were significantly

more likely to succeed than rallies initiated by POA2 individ-

uals (27.3% successful, N ¼ 22) and POA3 individuals (33.3%

successful, N ¼ 18) individuals combined (30% successful,

N ¼ 40; binomial test for equality of proportions without

continuity correction:x2 1 ¼ 10:46, p , 0.001). Once packs

were on the move, dominants lead most pack movements

(65%, N ¼ 15/23). Accounting for the relative demographic

proportions of the study population, these POA1 individuals

were significantly more likely to lead movements (n ¼ 15) than

POA2 (n ¼ 1) and POA3 (n ¼ 7) individuals (chi-square test,

X22¼ 31.348, p , 0.001).

We found a statistically significant difference in the total

number of sneezes in successful and unsuccessful rallies

(successful: �X + s.e., sneeze/rally¼ 7.48+1.49; unsuccessful: �X + s.e., sneeze/rally¼ 1.20+0.663; unpaired t-test: t ¼ 5.329, d.f. ¼ 66, p , 0.0001; figure 2c) and sneezes were the

most important factor predicting departures in our GLMM

model sets (table 1). A sneeze has never before been documented

as a major communicative function of African wild dogs (see,

[13]). However, it is not unique in the repertoire of important

signals in canids: Cohen [33] found ubiquitous ‘mechanical’ or

‘unvoiced’ sounds—like ‘panting’ as a play solicitation in dogs

and foxes—used by all canids for short-range communica-

tion [33]. Lehner [34] described a ‘huff’ in coyotes, as a ‘short

range, low intensity threat that is produced by a rapid expulsion

of air through the mouth primarily but also the nose’ [34]. He

parallels the behaviour to the ‘sneeze’ documented in golden

jackals [35]. Deaux & Clark [36] describe the ‘snuff’ of dingoes

as a ‘broadband nasal sound produced by the expulsion of

air through nasal passages’ that is produced in response to

environmental disturbance or in the context of frustration or

anxiousness [36]. While we find broad similarities in the descrip-

tion of the physical aspects of sneezes in African wild dogs

and the ‘huff’, ‘sneeze’, and ‘snuff’ of other canids, the com-

municative function described here seems to be previously

undocumented in the taxon.

In general, noisy, abrupt vocalizations are associated with

aggression or frustration in the behavioural context of signal

production [37]. Robbins, however, documented several

anomalies to general motivational-structure rules of animal

vocalizations in African wild dogs, which he attributed to the

‘muted nature of aggression characteristic of African wild

dog social organization’ [13]. Sneezes, which seem to be phys-

ically similar to signals used by other canid species as threats or

alarms [33–36], are produced in African wild dogs in contexts

absent of anxious postures, such as pacing, or threatening

1.0

0.8

0.6

0.4

0.2

0

1.0

0.8

0.6

0.4

0.2

0

2.5 2.0 1.5 1.0 0.5

0

1 2 3 or more order of rally within observation

–4 –3 –2 –1 depart +1 +2 +3 +4

time (mins) before and after departure

0 5 10 15 20 25 frequency of sneezes during rally

POA2POA1 POA3 initiator dominance category

pr op

or tio

n of

r al

lie s

su cc

es sf

ul

no . r

al lie

s

pr op

or tio

n of

r al

lie s

su cc

es sf

ul m

ea n

± s.

e. sn

ee ze

s/ m

in unsuccessful successful

estimate observed POA1 POA2 POA3

(a) (b)

(c)

(d) 20

15

10

5

0

Figure 2. Effects of attempt number, initiator demographic, and sneeze frequency on rally success. Panel (a) shows that the proportion of successful rallies (those that resulted in collective movements) increased with every failed rally in an observation period. Bars indicate the standard error of the mean. Panel (b) shows the frequency of successful (n ¼ 28) and unsuccessful (n ¼ 40) rallies initiated by individuals from each demographic category of priority of access to carcasses (POA): POA1, dominants and pups (less than 1 year old); POA2 yearlings (1 – 2 years old); and POA3, subdominant adults (more than 2 years old) with bars indicating standard error. Panel (c) shows the mean frequency of sneezes per minute before and after the end of successful and unsuccessful rallies. Hatched lines represent standard error. Data were gathered by critical incident sampling of sneezes from all packs’ (N ¼ 5) successful (n ¼ 28) and unsuccessful (n ¼ 40) rallies. The presented data are limited to the sneezes 4 min before the end of the rally and 4 min after the first individual departed or individuals returned to rest, which includes 80% (209/260) of total observed sneezes in rallies. Panel (d ) contains the plot of the interaction total number of sneezes in a rally and initiator POA effect on rally success. Solid lines show the estimated effects from the GLMM model (table 1) across a range of sneeze counts (increasing along the x-axis and scaled to the range of the actual data 0 – 25). These lines are colour-coded to initiator POA. Raw data are represented by the dashed lines, the lengths of which indicate the four bins into which raw data were directed on the basis of frequency of sneezes observed in the rally (0, 1 – 2, 2 – 9, more than 10).

Table 1. Model averaged generalized linear mixed effects model (GLMM) outputs from all models whose cumulative AIC weights were more than 0.95 showing: (a) effect sizes, relative importance of terms and confidence intervals and (b) AICc model weights for all models in the model set. Terms included in the model set were: total number of sneezes in a rally (‘TotSneeze’), the initiator priority of access to kills (‘InitPOA’), the interaction between sneeze frequency and initiator POA (‘InitPOA:TotSneeze’), consecutive attempt number per observation session (‘Attempt’), and the proportion of adults participating in social behaviours (‘AdSoc’).

(a)

term name term code importance

N containing models estimate s.e. z

CI (2.5 – 97.5%)

Intercept 4 21.69 2.1 0.79 26.04 to 2.64

TotSneeze 1 1 4 2.03 1.66 1.2 21.18 to 5.49

InitPOA 2 0.93 3

InitPOA1 0 0 0 0

InitPOA2 21.29 2.04 0.62 25.59 to 2.8

InitPOA3 22.62 2.17 1.18 27.08 to 1.43

InitPOA:TotSneeze 3 0.93 3

InitPOA1:TotSneeze 0 0 0 0

InitPOA2:TotSneeze 21.66 1.65 0.98 25.1 to 1.52

InitPOA3:TotSneeze 21.18 1.63 0.7 24.61 to 2.07

Attempt 4 0.58 2 0.75 0.9 0.83 20.33 to 2.95

AdSoc 5 0.12 1 20.08 0.65 0.12 24.23 to 2.92

(b)

model (incl. term codes) d.f. logLik AICc delta weight

1234 8 217.32 53.63 0 0.46

123 7 218.94 54.16 0.53 0.35

12345 9 217.25 56.32 2.69 0.12

1 3 225.43 57.32 3.69 0.07

rspb.royalsocietypublishing.org Proc.R.Soc.B

284:20170347

4

D ow

nl oa

de d

fr om

h ttp

s: //r

oy al

so ci

et yp

ub lis

hi ng

.o rg

/ o n

29 S

ep te

m be

r 20

21

rspb.royalsocietypublishing.org Proc.R.Soc.B

284:20170347

5

D ow

nl oa

de d

fr om

h ttp

s: //r

oy al

so ci

et yp

ub lis

hi ng

.o rg

/ o n

29 S

ep te

m be

r 20

21

expressions, such as flattened ears or bared teeth. Further

investigation of the occurrence of sneezes suggested this

signal was explicitly linked to the decision-making process, a

correlation that future research could explore in other species

exhibiting similar mechanical noises.

We also find an interaction between total sneezes and

initiator POA in rallies (table 1) indicating that the number

of sneezes required to initiate a collective movement differed

according to the dominance of individuals involved in the

rally. Specifically, we found that the likelihood of rally

success increases with the dominance of the initiator (i.e. for

lower POA categories) with lower-ranking initiators requiring

more sneezes in the rally for it to be successful (figure 2d ). In

fact, our raw data and the resultant model showed that rallies

never failed when a dominant (POA1) individual initiated and

there were at least three sneezes, whereas rallies initiated by

lower ranking individuals required a minimum of 10 sneezes

to achieve the same level of success. Together these data

suggest that wild dogs use a specific vocalization (the sneeze)

along with a variable quorum response mechanism in the

decision-making process [19].

This quorum response mechanism observed is similar to the

type of vocal coordination observed in foraging meerkats [4],

and appears to be similar to the increased rate of ‘grunts’ that

occur before Mountain gorilla (G. g. beringei) groups moved

off from their resting sites [38]. In African wild dogs, individ-

uals occasionally sneezed while lying down during resting

periods (NR Jordan, RH Walker, JW McNutt 2014, personal

observation), the significant increase in frequency during ral-

lies, especially in successful rallies, suggests this sound carries

contextual meaning (figure 2d). Both sneezes in African wild

dogs and grunts in gorillas occur in multiple behavioural con-

texts, but their rates are significantly positively correlated

with departure events [38]. Note, however, that our analyses

fail to differentiate between whether sneezes act as a true

voting mechanism or reflect a purely physiological response

to a consensus already achieved through other signals that we

did not observe [39,40]. Physiologically, the rapid exhalation

may function to prepare the pack to hunt by clearing the

nasal cavity to make scenting and running easier, but this

does not preclude that sneezing may also be a true voting

mechanism. Further research is required to confirm causality.

Perhaps most interestingly, our data further suggest that

the quorum number (indicated by the number of sneezes) is

variable and not wholly dependent on the involvement of

dominant individuals during rallies. These data are also con-

sistent with evidence of quorum thresholds shifting with

context, described in ants (Temnothorax albipennis) [41]. In

dominant-initiated rallies, a threshold of only three sneezes

was required, a result consistent with a commonly observed

quorum threshold of two to three individuals ‘acting as signal-

lers’ in groups ‘ranging from six to 22 individuals’ found across

taxa [4]. Because the number of adults involved in greeting,

parallel running, or mobbing did not affect rally success,

mutual appeasement seems not to be a crucial motivating

factor for departure even though those behaviours are typical

of rally displays [12,14,15]. We observe that appeasement sig-

nals, generally considered to enforce group cohesion, and

signals intended to motivate the group to move, are separate

features of African wild dog rally behaviour.

4. Conclusion Our study is the first to quantitatively assess behaviour and

decision-making processes in African wild dog pre-departure

rallies. We found that sneezes, a previously undocumented

unvoiced sound in the species, are positively correlated with

the likelihood of rally success preceding group movements

and may function as a voting mechanism to establish group

consensus in an otherwise despotically driven social system.

While our research focused on the decision-making aspect

and movement consequences of social rallies, further work

might include a quantitative study of mutual appeasements

and social bonding characteristics of rallies. Our results

contribute to a growing trend in the literature that finds

voting mechanisms and quorum thresholds used in decision-

making processes across taxa. Further research identifying

specific signals used to establish group consensus will help

us to better understand the evolution of social behaviour in

carnivores and other social mammals.

Ethics. The authors confirm that we have read and abided by the Royal Society publishing ethics policy and that all work was conducted in accordance with the guidelines for the treatment of animals in behavioural research and teaching [24].

Data accessibility. Our data are made publically available via Dryad at: http://dx.doi.org/10.5061/dryad.d6q6b [42].

Authors’ contributions. R.H.W. designed the methods, conducted the fieldwork and data collection, and wrote the paper with input from all authors. N.R.J. conceived the study, helped design the methods, hypotheses, and conducted the GLMM analysis. A.J.K. contributed to methodology, analyses and interpretation of the data. J.W.M. established and directs the research site and funded the study.

Competing interests. We have no competing interests to report that may influence the objective presentation of our manuscript.

Funding. We acknowledge and thank our funding agencies. Funding for this study was generously provided to BPCT by The Paul G Allen Family Foundation, the Myhrvold Family Charitable Fund, the Sunset Fund, Brookfield Zoo, Dallas Zoo, Woodland Park Zoo, and numerous private donors through Tusk Trust and Wild Entrust International.

Acknowledgements. We would like to thank the government of Botswana, particularly the Ministry of the Environment, Wildlife and Tourism, and staff and researchers of Botswana Predator Conservation Trust for the opportunity to conduct this study. In particular R.H.W. acknowl- edges the full and varied education received as a research assistant for the Trust. Krystyna Golabek kindly produced the spectrogram. Thank you to Katherine Smith for her support and input into this project and to Hugh Webster for his valuable comments and feedback.

References

1. Petit O, Bon R. 2010 Decision-making processes: the case of collective movements. Behav. Processes 84, 635 – 647. (doi:10.1016/j.beproc.2010.04.009)

2. Alexander R. 1974 The evolution of social behavior. Annu. Rev. Ecol. Syst. 5, 325 – 383. (doi:10.1146/ annurev.es.05.110174.001545)

3. Sueur C, Petit O. 2008 Shared or unshared consensus decision in macaques? Behav. Processes 78, 84 – 92. (doi:10.1016/j.beproc. 2008.01.004)

4. Bousquet CAH, Sumpter DJT, Manser MB. 2011 Moving calls: a vocal mechanism

underlying quorum decisions in cohesive groups. Proc. R. Soc. B 248, 1482 – 1488. (doi:10.1098/rspb. 2010.1739)

5. Pyritz LW, King AJ, Sueur C, Fichtel C. 2011 Reaching a consensus: terminology and concepts used in coordination and decision-making research.

rspb.royalsocietypublishing.org Proc.R.Soc.B

284:20170347

6

D ow

nl oa

de d

fr om

h ttp

s: //r

oy al

so ci

et yp

ub lis

hi ng

.o rg

/ o n

29 S

ep te

m be

r 20

21

Int. J. Primatol. 32, 1268 – 1278. (doi:10.1007/ s10764-011-9524-9)

6. Makison JC, Beekman M. 2014 Moving without a purpose: an experimental study of swarm guidance in the western honey bee, Apis mellifera. J. Exp. Biol. 217, 2020 – 2027. (doi:10.1242/jeb.103283)

7. Boinski S. 1993 Vocal coordination of troop movement among white-faced capuchin monkeys, Cebus capucinus. Am. J. Primatol. 30, 85 – 100. (doi:10.1002/ajp.1350300202)

8. West-Eberhard MJ. 1979 Sexual selection, social competition, and evolution. Proc. Am. Philos. Soc. 123, 222– 234.

9. King AJ, Cowlishaw G. 2009 Leaders, followers, and group decision-making. Commun. Integr. Biol. 2, 147 – 150. (doi:10.4161/cib.7562)

10. King AJ, Sueur C, Huchard E, Cowlishaw G. 2011 A rule-of-thumb based on social affiliation explains collective movements in desert baboons. Anim. Behav. 82, 1337 – 1345. (doi:10.1016/j.anbehav. 2011.09.017)

11. Holekamp KE, Boydston E, Smale L. 2000 Group travel in social carnivores. In On the move (eds S Boinski, PA Garber), pp. 587 – 627. Chicago, IL: University of Chicago Press.

12. van Lawick-Goodall H, van Lawick-Goodall J. 1970 African wild dog. In Innocent killers, pp. 49 – 104. Ballantine Books.

13. Robbins RL. 2000 Vocal communication in free- ranging African wild dogs (Lycaon pictus). Behaviour 137, 1271 – 1298. (doi:10.1163/156853900501926)

14. Rutten S, Fleissner G. 2004 On the function of the greeting ceremony in social canids—exemplified by African wild dogs Lycaon pictus. Canid News, 7.3. Online.

15. Estes R, Goddard J. 1967 Prey selection and hunting behavior of the African wild dog. J. Wildl. Manage. 31, 52 – 70. (doi:10.2307/3798360)

16. Creel S. 1997 Cooperative hunting and group size: assumptions and currencies. Anim. Behav. 54, 1319 – 1324. (doi:10.1006/anbe.1997.0481)

17. Malcolm JR. 1979 Social organization and the communal rearing of pups in African wild dogs (Lycaon pictus). PhD thesis, Harvard University, Cambridge, MA.

18. Conradt L, Roper TJ. 2005 Consensus decision making in animals. Trends Ecol. Evol. 20, 449 – 456. (doi:10.1016/j.tree.2005.05.008)

19. Camazine S, Deneubourg JL, Franks NR, Sneyd J, Theraulaz G, Bonabeau E. 2001 Self-organization in biological systems. Princeton, NJ: Princeton University Press.

20. McNutt JW. 1996 Sex based dispersal in African wild dogs (Lycaon pictus). Anim. Behav. 52, 1067 – 1077. (doi:10.1006/anbe/1996.0254)

21. Malcom JR, Marten K. 1982 Natural selection and the communal rearing of pups in African wild dogs (Lycaon pictus). Behav. Ecol. Sociobiol. 10, 1 – 13. (doi:10.1007/BF00296390)

22. Condadt L, List C. 2009 Group decisions in humans and animals: a survey. Phil. Trans. R. Soc. B 364, 719 – 742. (doi:10.1098/rstb.2008.0276)

23. Osofsky SA, McNutt JW, Hirsch KJ. 1996 Immobilization of free-ranging African wild dogs (Lycaon pictus) using a ketamine/xylazine/atropine combination. J. Zoo Wildl. Med. 27, 528 – 532.

24. 2012 Guidelines for the treatment of animals in behavioral research and teaching. Anim. Behav. 83, 301 – 309. (doi:10.1016/j.anbeh.2011.10.031)

25. R Core Team. 2016 R: a language and environment for statistical computing. Vienna, Austria: R Foundation For Statistical Computing. (http://www. R-project.org/)

26. Bates D, Maechler M, Bolker B, Walker S. 2015 Fitting linear mixed-effects models using lme4. J. Stat. Softw. 67, 1 – 48. (doi:10.18637/jss.v067.i01)

27. Akaike, H. 1974 New look at statistical-model identification. IEEE Trans. Automat. Contr. 19, 716 – 723. (doi:10.1109/TAC.1974.1100705)

28. Burnham KP, Anderson DR. 2004 Multimodel inference: understanding AIC and BIC in model selection. Sociol. Methods Res. 33, 261 – 304. (doi:10.1177/0049124104268644)

29. Grueber CE, Nakagawa S, Laws RJ, Jamieson IG. 2011 Multimodel inferences in ecology and evolution: challenges and solutions. J. Evol. Biol. 24, 699 – 711. (doi:10.111/j.1420-9101.2010.02210.x)

30. Burnham KP, Anderson DR, Huyvaert KP. 2011 AIC model selection and multimodel inference in behavioral ecology: some background, observations, and comparisons. Behav. Ecol. Sociobiol. 65, 23 – 35. (doi:10.1007/s00265-010-1029-6)

31. Barton K. 2012 MuMIn: multi-model inference. (R package version 1.7.7) 2012. See http://CRAN.R- project.org/package=MuMIn.

32. Hartig F. 2017 DHARMa: residual diagnostics for hierarchical (multi-level/mixed) regression models. R package version 0.1.5. See https://CRAN.R-project. org/package=DHARMa.

33. Cohen JA, Fox MW. 1976 Vocalizations in wild canids and possible effects of domestication. Behav. Processes 1, 77 – 92. (doi:10.1016/0376- 6357(76)90008-5)

34. Lehner PN. 1978 Coyote vocalizations: a lexicon and comparisons with other canids. Anim. Behav. 26, 712 – 722. (doi:10.1016/0003-3472(78)90138-0)

35. Seitz A. 1959 Beobachtugen an handaufgezogenen Goldschaklen (Canis aureus aligirensis). Z. Tierpsychol. 16, 747 – 771. (doi:10.111/j.1439. 1959.tb02190.x)

36. Deaux EC, Clarke JA. 2013 Dingo (Canis lupus dingo) acoustic repertoire: form and contexts. Behaviour 150, 75 – 101. (doi:10.2307/23434982)

37. Morton ES. 1977 On the occurrence and significance of motivation-structural rules in some bird and mammal sounds. Am. Nat. 111, 855 – 869. (doi:10. 1086/283219)

38. Stewart KJ, Harcourt AH. 1994 Gorillas’ vocalizations during rest periods: signals of impending departure? Behaviour 130, 29 – 40. (doi:10.1163/ 156853994X00127)

39. Schassburger RM. 1993 Vocal communication in the timber wolf, Canis lupus, Linnaeus: structure, motivation, and ontology. In Advances in ethology no. 30. Berlin, Germany: Paul Parey Scientific.

40. Sueur C, Jacobs A, Amblard F, Petit O, King AJ. 2011 How can social network analysis improve the study of primate behavior? Am. J. Primatol. 73, 703 – 709. (doi:10.1002/ajp.20915)

41. Franks NR, Deschaume-Moncharmont F, Hanmore E, Reynolds JK. 2009 Speed versus accuracy in decision-making ants: expediting politics and policy implementation. Phil. Trans. R. Soc. B 364, 845 – 852. (doi:10.1098/ rstb.2008.0224)

42. Walker RH, King AJ, McNutt JW, Jordan NR. 2017 Data from: Sneeze to leave: African wild dogs (Lycaon pictus) use variable quorum thresholds facilitated by sneezes in collective decisions. Dryad Digital Repository. (http://dx.doi.org/10.5061/ dryad.d6q6b)