Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Sexual dimorphism in African elephant social rumbles


This study used the source and filter theory approach to analyse sex differences in the acoustic features of African elephant (Loxodonta africana) low-frequency rumbles produced in social contexts (‘social rumbles’). Permuted discriminant function analysis revealed that rumbles contain sufficient acoustic information to predict the sex of a vocalizing individual. Features primarily related to the vocalizer’s size, i.e. fundamental frequency variables and vocal tract resonant frequencies, differed significantly between the sexes. Yet, controlling for age and size effects, our results indicate that the pronounced sexual size dimorphism in African elephants is partly, but not exclusively, responsible for sexual differences in social rumbles. This provides a scientific foundation for future work investigating the perceptual and functional relevance of specific acoustic characteristics in African elephant vocal sexual communication.


Sexual dimorphism, describing morphological, physiological and behavioural differences of sexually mature females and males, is proposed to have evolved either through sexual selection or adaptations for sex-specific niche divergence [1]. In mammals with polygynous mating systems, selective processes producing sexual dimorphism result in different overall body size, with males being generally larger than females (male-biased size dimorphism) [2, 3]. Body size in animals is a crucial component in many aspects of ecology and social behaviour, and plays an important role in determining an individual’s fighting ability and reproductive success [2].

Studies on sound production and usage in various mammals showed that vocalizations encode honest information about a caller’s phenotype, including body size, age and sex (see [4] for review). Moreover, vocalizations play a key role for kin and individual recognition, e.g. in rhesus monkeys [5], wolves [6] and meerkats [7].

According to the source-filter theory, most mammal vocalizations are generated by the larynx and are subsequently filtered by the supra-laryngeal vocal tract (VT) [8]. The fundamental frequency (F0) of a vocalization is affected by the dimensions of the vocal folds (the source) [8, 9]. In nonhuman mammals, the exact underlying mechanisms of hormones in influencing vocal fold dynamics and acoustic structure have not been identified yet, but it has been experimentally shown that androgen stimulation has a direct effect upon the larynx [10]. Further studies demonstrated a correlation between androgen levels and the acoustic structure of vocalizations. For instance, higher androgen levels are correlated with higher-pitched solo songs in male gibbons [11] and with higher F0 modulation rates in male giant panda bleats [12]. Since testosterone is an important factor mediating competitive behaviours among males [13, 14], androgen-induced vocal cues may be generally important for intra-sexual assessment and mate preference.

As the source signal travels through the successive VT (the filter), its length and shape amplifies certain resonant frequencies (formants) selectively [8]: longer VTs produce lower, more closely spaced formants, reflecting larger body size [4]. This, together with the demonstration of formant perception by nonhuman mammals [15, 16], suggests that formants function as a universal cue to body size in terrestrial mammals [17]. In addition, intriguing sexually dimorphic adaptations–such as laryngeal air sacs in reindeer [18] a descended larynx in humans and red deer [19], an extremely enlarged larynx in hammerhead bats [20], or an additional set of vocal folds outside the larynx in koalas [21]–are important for mate selection and form the basis for most sex differences in animal vocalizations [4].

African elephants (Loxodonta africana) live in matrilineal societies in which most female vocalizations are used for family group and inter-individual cohesion [22]. Males leave their natal family at an average age of 14 years, but all-male coalitions and companionship have been reported [23]. Their vocalizations seem to be primarily related to contexts of intra-sexual dominance and reproduction [22]. Compared to females, however, little work has been done on male vocal behaviour in general, and studies addressing gender-dependent differences in vocalizations are mainly restricted to differences in vocal activity and vocalization types [22]. African elephants make use of 8–10 distinct call types [2426], which grade with intermediate cross-forms between and within different call categories; this affects acoustic parameters including formants [2731]. The most common call type is the “rumble”, a low-frequency and harmonically rich vocalization with frequencies in the infrasonic range, but with harmonics that extend well into the audible range [32]. Rumbles were suggested to have multiple social functions, from close-distance social interactions to reinforce social bonds [33] to the coordination of long-distance movements [34] and the maintenance of contact between spatially separated groups [35, 36].

Little is known about how elephants generate their distinct vocalizations, but recent experiments on low-frequency rumbles indicate flow-induced vocal fold vibration as the source for this particular call type [37].

Although elephants apparently follow the basic way of mammalian sound production, they possess an exceptionally elongated nasal VT (proboscis) compared to their oral VT [32]. Sound visualization experiments revealed that elephants control the vocal path from oral to nasal rumble production depending on context and thereby significantly vary the formant structure of their vocalizations [38].

Despite the extreme sexually dimorphic body size in African elephants [22], both sexes produce distinctive ‘sex-specific’ (i.e. produced by only one sex; in the manner of [39]) rumble types in reproductive contexts [34]. When mature males are in ‘musth’, a periodically occurring reproductive state characterized by aggressive behavior and highly elevated androgen levels [40], a distinctive pulsated ‘musth-rumble’ is emitted. Females in oestrus produce an ‘estrous-rumble’ [34]. It is, however, not acoustically distinctive and unique to the ovulatory follicular phase [25], which represents the second wave of follicular development that ultimately leads to ovulation [41]. Both rumble types are probably used to assess or advertise the behavioural and physiological state of/for a sexual partner and a sexual rival [32]. Whether sex differences in the acoustic structure are pronounced in functionally identical ‘social rumbles’, shared by both sexes, is currently unknown and requires clarification.

Cues about sex in social long-distance vocalizations might be particularly important in spatially and socially flexible species for mediating social dynamics. Elephant ranging patterns can cover huge areas, where hundreds of individuals of different age and sex could be encountered opportunistically [42]. Behavioural reactions depend upon the identity and maturity of encountered individuals [43, 44], but the influence of sex has not been explored. Here, we investigate whether source- and filter-related acoustic features encode sex-specific information in social rumbles and discuss their potential functional relevance for the African elephant’s vocal communication system.

Material and methods

Study subjects and housing

Our study subjects were 9 adult female and 10 adult male elephants from two European zoos and four South African privately owned elephant keeping reserves (Table 1). All elephants have social contact throughout the day and spend the night in separated stables (but do have tactile, visual or acoustic contact). Elephants housed in South Africa were allowed to free-roam in areas of 3–45 km2 for several hours per day.

Table 1. Study sites, sex, age and number of calls for each study subject.

Data collection

Members of elephant herds mostly move together, accompanied by brief periods of spatial and temporal separation of individuals for foraging [36]. Rumbles mediate spatial relationships within a herd and thus promote group cohesion [45], even when out of visual range [35]. To allocate social rumbles to the vocalizing elephant, we focused on particular individuals during the following recording situations: (1) feeding: where the focus individual performs exclusively browsing and feeding activities (S1 Video), either within visual range or when segregated from its group members; (2) active approach: rumbling when approaching another group member; (3) passive approach: rumbling when being approached by another group member (S2 Video); (4) in reaction to physical contact: for example, in reaction to a trunk-touch (no agonistic interactions such as pushing or tusking); (5) general locomotion of the group: physical movement of all group members in the same direction, as for example when leaving enclosures together; and (6) spatial separation: vocalizing elephant is spatially separated from the rest of the group due to the daily routine of the respective institution (S3 Video, S1 Table).

We recorded throughout the day between 7 a.m. and 5 p.m. Recordings at both European zoos were made from protected keeper areas 5–20 m from the elephants. In South Africa we followed the animals on foot accompanied by elephant handlers for security reasons, did not interact with the animals, and passively recorded at distances ranging from 5–50 m.

The elephants’ individual identification was based on their physical characteristics, such as body size, genitals, shape and patterns of notches and holes in the ears, length and shape of tusks and tail, and the presence of specific marks on the body (e.g. warts). Moreover, since African elephants often raise and flap their ears while rumbling and, as if listening, sometimes hold very still with their head and ears lifted before or after vocalizing [34] (see S1, S2 and S3 Videos for demonstration), we additionally used behavioural categories such as ear and head movement to identify the calling individual. The identity of the focus elephant was, however, noted only when both authors agreed upon the vocalizing individual. No behavioural or visual signals of oestrus and musth occurred [46].

Shoulder height of the study subjects was measured as the vertical distance from the ground up to the top of the shoulder. Individuals from South Africa were measured using a Telefix 4 telescopic meter. Measurements at Vienna Zoo and Tierpark Berlin were taken by the keepers using a measuring scale attached along the steel-bar of the enclosures. The respective institution provided information about age for each individual.

Acoustic data were obtained using an omni-directional Neumann KM183 condenser microphone, modified for recording frequencies below 20 Hz (flat-recording down to 5 Hz), connected to a Sound Devices 722 (frequency response: 10 Hz–40 kHz) at 48 kHz sampling rate and 16-bit.

Acoustic analysis

Data were annotated in S_Tools-STx 4.2.8 [47] by defining the on- and off-set of each elephant rumble. Prior to in-depth acoustic analysis, we pre-selected calls with specific formant locations, namely formant 1 (F1) and 2 (F2), potentially indicating nasally emitted rumbles. Using an estimated VT length of 0.75 m for oral and 2.5 m for nasal rumbles, the predicted formant locations for oral rumbles are F1 = 116.7 Hz, F2 = 350.0 Hz, and for nasal rumbles F1 = 35.0 Hz, F2 = 105.0 Hz, respectively [32]. Therefore, we classified rumbles as nasally produced if F1 was below 50 Hz and F2 below 150 Hz. We then used a customized semi-automatic Matlab sound analysis tool [48] to extract source-related acoustic features by tracing the F0-contour in Fast Fourier spectrograms (frame size: 300 ms, step size: 40 ms). To analyse filter-related formants, we used linear predictive coding (LPC) in S_Tools-STx to extract frequency values of the first two formants. Definitions of the source- and filter-related acoustic parameters are provided in S2 Table.

Statistical analysis

Our selections yielded 140 rumbles from 10 males and 123 rumbles from 9 female elephants (S1 Table) recorded in different sessions over an overall data collection time of 50 days (mean 7.2 ± 2.9 days per elephant, range 3–17 days).

An initial multivariate analysis of variance (MANOVA) was carried out to identify significantly different acoustic parameters between sexes. To correct for the effects of multiple testing and the occurrence of false positives, all P-values were adjusted using the Benjamini-Hochberg False Discovery Rate (BH FDR) procedure [49]. Acoustic features above the BH FDR corrected significance level were discarded as insignificant and not used for further analysis. To evaluate whether the data were suitable for conducting a principal component analysis (PCA), we used the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and performed the Bartlett’s test of Sphericity. The verified acoustic parameters were then entered into a varimax rotated PCA with Kaiser-normalization for data reduction. Resulting factor scores with eigenvalues greater than one were retained, saved using the regression method and utilized as input for classification analysis. To test the classification of individual elephant rumbles to sex, we performed a permuted discriminant function analysis (pDFA), a randomization procedure used for non-independent two-factorial data sets when one factor is nested in another [50]. The pDFA was conducted using a function (provided by R. Mundry) based on Ida of the R package MASS [51]. We defined sex as test and individuality as control factor. Although age and shoulder height are tightly correlated in both sexes of African elephants [52], we controlled for the effects of both age and size in males (Pearson Correlation, r = 0.944, p < 0.001) and females (Pearson Correlation, r = 0.747, p < 0.001) by conducting 2 separate pDFAs. Five age groups were established (S3 Table) and entered as restriction factor to control for age effects. The second pDFA used 5 shoulder height categories as restriction factor to control for the effect of size (S4 Table). Further, to evaluate the influence of source-related parameters only, we repeated all statistical procedures by excluding filter-related variables. Results are expressed as percentage of correct classification (cross-validated). All statistical tests were performed in SPSS v.22 and RStudio v.1.0.136.


Using the BH FDR corrected significance level (q* = 0.0296), the initial MANOVA verified significant differences in 16 variables: in all filter-related parameters, all absolute frequency parameters (except ‘Min F0’ and ‘Max F0/Mean F0’), and 2 shape- and contour-related parameters (‘coefficient of frequency modulation’ (COFM) and ‘jitter factor’). None of the temporal-related parameters were significantly discriminant between sexes (Table 2). The subsequent PCA was justified as shown by the KMO criterion (0.802) and the Bartlett’s test of Sphericity (X2 = 16,876.96, df = 120, p < 0.001).

Table 2. MANOVA comparing source- and filter-related acoustic features between male and female African elephant social rumbles.

The PCA reduced 16 significant source- and filter-related parameters to 3 principal components (PC) explaining 82.5% of the total variation: most source-related parameters correlated strongly with PC 1 explaining 45.6% of the variance. Shape- and contour-related variables were assigned to PC 2 explaining 21.6% of the variance, whereas filter-related parameters strongly correlated to PC 3 explaining 15.3% of the variance. Loading values for each parameter that loaded strongly to one of the three PC are shown in S5 Table.

The pDFA, controlling for an effect of age, resulted in 85.6% correct cross-validated classification after 100 random selections and 10,000 permutations. This shows that female and male elephants are discriminable based on 16 acoustic parameters of social rumble vocalizations (p = 0.0207). The second pDFA, comparing both sexes using ‘shoulder height’ as restriction factor, resulted in 85.6% correct cross-validated classification (p = 0.0187).

After excluding the filter-related variables, a second MANOVA identified the same 13 source-related parameters (F, df and p values are identical as in Table 2) as being significantly different (q* = 0.0271) between the sexes. The subsequent second PCA was justified by the KMO measure of sampling adequacy (0.798) and the Bartlett’s test of Sphericity (X2 = 7,284.059, df = 78, p < 0.001). The 13 variables were then reduced to 2 PCs explaining 80.9% of the total variation (loading values for each parameter are shown in S6 Table). Controlling for the effect of age resulted in 73.2% (p = 0.0367) correct cross-validated classification. When controlling for size effects, a correct cross-validated classification of 72.9% (p = 0.1297) was achieved.

Fig 1 illustrates comparative spectrograms of a female and male elephant social rumble. The corresponding acoustic recording is provided in S1 Sound, with the female rumbling first.

Fig 1. Comparison of a female (left, Chikwenya) and a male (right, Mike) African elephant.

The spectrogram and power spectra below the photograph provide an example of a social rumble of each sex, indicating formant positions (rumbles uttered first by Chikwenya, followed by Mike).


This study presents the first comparative acoustic analysis of adult female and male African elephant vocalizations. It considers source- and filter-related acoustic features of low-frequency rumbles emitted in non-reproductive social contexts. Our results demonstrate that female and male social rumbles encode reliable information about the physical attributes and sex of the caller.

Elephant vocal folds are voluminous and long, reaching about 10 cm in adult females [37]. The dimensions of the vocal folds determine the vibrational behaviour during phonation and hence determine the F0 [9, 17]. When excluding formants (which are known reliable indicators for body size), source-related features, i.e. absolute frequency, shape and contour parameters, showed a considerable and significant classification success between the sexes when controlling for age, but a non-significant result controlling for size. Due to the size dimorphism, in our data set the same individuals assigned to their respective age category are not necessarily found in the same shoulder height category. This non-significant p-value could reflect the small sample size in some size categories. Notwithstanding, this might mean that most of our analysed source-related parameters are influenced by size effects. Our findings go along with previous studies showing that source-related vocal features reflect size-related differences between age categories in many species (e.g., baboons [55] and goats [56]), between sexes (baboons [57]) and within species exhibiting a large morphological range (domestic dogs [58]). Beyond size, increased male androgen levels might affect fundamental frequency. Vocal folds are highly sensitive to testosterone [10], leading to lower F0 in human males (by potentially changing vocal fold bulk, length or tension of the vocal folds) [59]. Accordingly, increased levels of circulating testosterone affecting vocal fold dynamics might further cause a sexual difference of the F0 in elephant social rumbles. Also, COFM and jitter factor–acoustic features representing the shape and contour features of the F0 –differed significantly between the sexes. Jitter, typically an index for F0-variability, was higher in male elephants than in females. Females, however, had higher COFM values, indicating high frequency modulation [60], and thus more modulated rumbles than those of males. A similar result was observed in giant pandas, where circulating oestrogens and testosterone were suggested to affect vocal fold oscillations in each sex [61]. Elephant larynx anatomy differs proportionally in size and structure compared to humans, and a highly complex vocal fold vibratory pattern, previously not documented in other species, has been reported [37]. This complicates data interpretation of F0 shape and contour features. Whether these parameters are related to testosterone levels [61] requires anatomical and physiological investigations.

Elephants are among the most size dimorphic of all mammals: males can be twice the mass of a female. Also, secondary sexual characteristics between both sexes such as dimorphism of tusk and skull size exist [22]. Particularly, skull and mandible size are larger and more pronounced in male elephants, including more robust muscle markings and attachments [62]. The most rostral part of the skull is formed by the paired incisive bones, Os incisivum (located at the tip of the maxilla), which are more developed and rostrally expanded in bulls [63]. Nonetheless, it remains to be determined whether a morphological dimorphism in the vocal apparatus (e.g. in the howler monkey [64] and the Mongolian gazelle [65]) occurs other than size-related differences. In elephants it is argued that pharyngeal pouches present in both sexes might have an effect on the vocal output [66]. The acoustic role and method of sound production for these anatomical structures remain unexplored in both sexes.

Note also that social rumbles were recorded from individuals that were either born in captivity, wild-born orphans but raised in captivity from a young age on, or wild-caught as adults. Ultimately, (apart from captive-borns) all originate from different regions in Africa, where elephants occur in varying sizes across the continent [52, 67]. This inter-population variability in body sizes or stature has been suggested to be influenced by resource availability in the elephants’ environment and as a result of genetic variation because some populations differ considerably in certain haplotypes [52]. Our elephants had access exclusively to a highly nutritious diet, which in combination with little physical activity (some of our zoo elephants) leads to faster growth rates and higher weight compared to their wild counterparts [68]. We therefore emphasize that different life histories and life styles of study animals, and inter-population variances in certain demographic and morphometric parameters, should not be disregarded because they may play a relevant role for data interpretation.

Apart from comparing the acoustic structure between different cycle stages in females, no correlational studies have been conducted between hormone levels and acoustic parameters of these two rumble types. This calls for future research correlating vocalizations with sex steroid levels in both sexes to better understand the effect of testosterone. Most interestingly, this includes the transition to and/or the onset of the distinctive ‘musth-rumble’ during periods of extremely elevated androgen levels in males [69].

In conclusion, our results provide a clear distinction between male and female African elephant social rumbles. This is evidence that certain acoustic features facilitate the recognition of sex and are likely to be important for elephant social dynamics and inter-sexual vocal communication, which is, so far, poorly understood. In addition, interest in automatic acoustic monitoring of elephant populations is increasing [7072], and early warning systems for people living in human-elephant conflict areas have become an issue [48, 73]. Vocal cues can be used as indicators of group size [70] and to categorize age groups based on vocalizations. This has considerable potential to help assess the demography of a monitored population [74]. Information on acoustic cues that reliably categorize sex could optimise algorithm development and make acoustic monitoring and detection systems more sophisticated. Accordingly, deterrent methods could be adjusted depending on the sex and the age of approaching elephants (e.g. a female group with infants versus males).

Finally, this study provides the framework for future research to further assess the role of particular acoustic features for inter- as well as for intra-sexual elephant communication by using, for example, re-synthesis techniques in acoustic playback experiments.

Ethics statement

This non-experimental research meets all applicable European Union and South African laws and was conducted din accordance with the Guidelines for the Treatment of Animals in Behavioural Research and Teaching [75]. All participating institutions mentioned in this manuscript approved data collection for the study. The nature of the study was purely observational: No invasive methodologies were applied at any point of the study. The research did not affect the housing, daily routine, behaviour, diet or management of the animals. Therefore, no ethics committee approval was required.

Supporting information

S1 Video. Male African elephant rumbling during browsing and feeding activity.


S2 Video. Male African elephant rumbling while being approached.


S3 Video. Male African elephant rumbling in a spatial separation situation.


S2 Table. Description of the source- and filter-related parameters measured.


S5 Table. Extracted factors of the principal component analysis.


S6 Table. Extracted factors of the principal component analysis.


S1 Sound. Sound sample of a low-arousal social rumble of each sex, with the female rumbling first.

Maximum frequency of the sound file was set to 200 Hz and the absolute peak of the amplitude was scaled to 0.99.



We thank all participating institutions. We thank Sean Hensman and the Rory Hensman Elephant Research Unit (Adventures with Elephants), Brett Mitchel (Pilanesberg Back Safaris), Andre Kotze (Elephant Whisperers), Dr. Debbie Young (African Elephant Research Unit, Knysna Elephant Park and Addo Elephant Back Safaris), Dr. Harald Schwammer (Vienna Zoo) and Dr. Florian Sicks (Tierpark Berlin) for enabling and supporting our research at their institutions. We further thank our long-term collaborator Matthias Zeppelzauer for developing the annotation and contour analyses tool, and Dr. Michael Stachowitsch for editing the text. We are grateful to Prof. Tecumseh Fitch and Prof. Thomas Bugnyar for strongly supporting our research at the Department of Cognitive Biology. This project was supported by the FWF Austrian Science Fund (P26448).

Author Contributions

  1. Conceptualization: AB.
  2. Data curation: AB ASS.
  3. Formal analysis: AB ASS.
  4. Funding acquisition: ASS.
  5. Investigation: AB ASS.
  6. Methodology: AB.
  7. Project administration: AB ASS.
  8. Resources: AB ASS.
  9. Software: AB ASS.
  10. Supervision: ASS.
  11. Validation: AB ASS.
  12. Visualization: AB.
  13. Writing – original draft: AB.
  14. Writing – review & editing: ASS.


  1. 1. Darwin C. The descent of man, and selection in relation to sex. By Darwin Charles. New York: D. Appleton and Company; 1871.
  2. 2. Fairbairn DJ. Allometry for Sexual Size Dimorphism: Pattern and Process in the Coevolution of Body Size in Males and Females. Annual Review of Ecology and Systematics. 1997;28(1):659–87.
  3. 3. McPherson FJ, Chenoweth PJ. Mammalian sexual dimorphism. Animal Reproduction Science. 2012;131(3):109–22.
  4. 4. Taylor AM, Reby D. The contribution of source–filter theory to mammal vocal communication research. Journal of Zoology. 2010;280(3):221–36.
  5. 5. Rendall D, Rodman PS, Emond RE. Vocal recognition of individuals and kin in free-ranging rhesus monkeys. Animal Behaviour. 1996;51(5):1007–15.
  6. 6. Palacios V, Font E, Márquez R, Carazo P. Recognition of familiarity on the basis of howls: a playback experiment in a captive group of wolves. Behaviour. 2015;152(5):593–614.
  7. 7. Townsend SW, Allen C, Manser MB. A simple test of vocal individual recognition in wild meerkats. Biology Letters. 2012;8(2):179–82. pmid:21992821
  8. 8. Titze IR. Principles of Voice Production. Englewood Cliffs, New Jersey: Prentice Hall; 1994.
  9. 9. Titze IR. Vocal Fold Mass Is Not A Useful Quantity for Describing F0 in Vocalization. Journal of Speech, Language, and Hearing Research. 2011;54(2):520–2.
  10. 10. Beckford NS, Rood SR, Schaid D, Schanbacher B. Androgen stimulation and laryngeal development. The Annals of Otology, Rhinology and Laryngology. 1985;94(6):634–40.
  11. 11. Barelli C, Mundry R, Heistermann M, Hammerschmidt K. Cues to Androgens and Quality in Male Gibbon Songs. PLoS ONE. 2013;8(12):e82748. pmid:24367551
  12. 12. Charlton BD, Keating JL, Kersey D, Rengui L, Huang Y, Swaisgood RR. Vocal cues to male androgen levels in giant pandas. Biology Letters. 2011;7(1):71–4. pmid:20810426
  13. 13. Kelley DB, Brenowitz E. Hormonal influences on courtship behaviours. In: Becker JB, Breedlove SM, Crews D, McCarthy MM, editors. Behavioral endocrinology. 2nd ed. Massachusetts: MIT Press; 2002. p. 289–325.
  14. 14. Zielinski WJ, Vandenbergh JG. Testosterone and competitive ability in male house mice, Mus musculus: laboratory and field studies. Animal Behaviour. 1993;45(5):873–91.
  15. 15. Reby D, McComb K, Cargnelutti B, Darwin C, Fitch WT, Clutton-Brock T. Red deer stags use formants as assessment cues during intrasexual agonistic interactions. Proceedings of the Royal Society B: Biological Sciences. 2005;272(1566):941–7. pmid:16024350
  16. 16. Charlton B, Ellis WH, Larkin R, Tecumseh Fitch W. Perception of size-related formant information in male koalas (Phascolarctos cinereus). Animal Cognition. 2012;15(5):999–1006. pmid:22740017
  17. 17. Charlton BD, Reby D. The evolution of acoustic size exaggeration in terrestrial mammals. Nature Communications. 2016;7:12739.—supplementary-information. pmid:27598835
  18. 18. Frey R, Gebler A, Fritsch G, Nygrén K, Weissengruber GE. Nordic rattle: the hoarse vocalization and the inflatable laryngeal air sac of reindeer (Rangifer tarandus). Journal of Anatomy. 2007;210(2):131–59. pmid:17310544
  19. 19. Fitch WT, Reby D. The descended larynx is not uniquely human. Proceedings of the Royal Society of London Series B: Biological Sciences. 2001;268(1477):1669–75. pmid:11506679
  20. 20. Schneider R, Kuhn H-J, Kelemen G. Der Larynx des männlichen Hypsignathus monstrosus Allen, 1861 (Pteropodidae, Megachiroptera, Mammalia). Zeitschrift für Wissenschaftliche Zoologie. 1967;175:1–53.
  21. 21. Charlton BD, Frey R, McKinnon AJ, Fritsch G, Fitch WT, Reby D. Koalas use a novel vocal organ to produce unusually low-pitched mating calls. Current Biology. 2013;23(23):R1035–R6. pmid:24309276
  22. 22. Poole JH. Sex differences in the behviour of African elephants. In: Short RV, Balaban E, editors. The differences between the sexes. Cambridge, U.K.: Cambridge University Press; 1994. p. 331–246.
  23. 23. O'Connell-Rodwell CE, Wood JD, Kinzley C, Rodwell TC, Alarcon C, Wasser SK, et al. Male African elephants (Loxodonta africana) queue when the stakes are high. Ethology Ecology & Evolution. 2011;23(4):388–97.
  24. 24. Stoeger-Horwath AS, Stoeger S, Schwammer HM, Kratochvil H. Call repertoire of infant African elephants: First insights into the early vocal ontogeny. The Journal of the Acoustical Society of America. 2007;121(6):3922–31. pmid:17552738
  25. 25. Leong KM, Ortolani A, Graham LH, Savage A. The use of low-frequency vocalizations in African elephant (Loxodonta africana) reproductive strategies. Hormones and Behavior. 2003;43(4):433–43. pmid:12788289
  26. 26. Poole JH. Behavioral contexts of elephant acoustic communication. In: Moss CJ, Croze H, Lee PC, editors. The Amboseli elephants: a long-term perspective on a long-lived mammal. Chicago, USA: The University of Chicago; 2011. p. 125–61.
  27. 27. Soltis J. Emotional communication in African elephants. In: Altenmüller E, Schmidt S, Zimmermann E, editors. The Evolution of Emotional Communication: From Sounds in Nonhuman Mammals to Speech and Music in Man. Oxford, UK: Oxford University Press; 2013. p. 105–15.
  28. 28. Soltis J, Blowers , Tracy E., Savage A. Measuring positive and negative affect in the voiced sounds of African elephants (Loxodonta africana). The Journal of the Acoustical Society of America. 2011;129(2):1059–66. pmid:21361461
  29. 29. Soltis J, King LE, Douglas-Hamilton I, Vollrath F, Savage A. African Elephant Alarm Calls Distinguish between Threats from Humans and Bees. PLoS ONE. 2014;9(2):e89403. pmid:24586753
  30. 30. Stoeger AS, Charlton BD, Kratochvil H, Fitch WT. Vocal cues indicate level of arousal in infant African elephant roars. The Journal of the Acoustical Society of America. 2011;130(3):1700–10. pmid:21895107
  31. 31. King LE, Soltis J, Douglas-Hamilton I, Savage A, Vollrath F. Bee Threat Elicits Alarm Call in African Elephants. PLoS ONE. 2010;5(4):e10346. pmid:20436682
  32. 32. Soltis J. Vocal communication in African Elephants (Loxodonta africana). Zoo Biology. 2010;29(2):192–209. pmid:19434672
  33. 33. Soltis J, Leong K, Savage A. African elephant vocal communication I: antiphonal calling behaviour among affiliated females. Animal Behaviour. 2005;70(3):579–87.
  34. 34. Poole JH, Payne K, Langbauer WR, Moss CJ. The social contexts of some very low frequency calls of African elephants. Behavioral Ecology and Sociobiology. 1988;22(6):385–92.
  35. 35. McComb K, Reby D, Baker L, Moss C, Sayialel S. Long-distance communication of acoustic cues to social identity in African elephants. Animal Behaviour. 2003;65:317–29.
  36. 36. Charif RA, Ramey RR, Langbauer WR, Payne KB, Martin RB, Brown LM. Spatial relationships and matrilineal kinship in African savanna elephant (Loxodonta africana) clans. Behavioral Ecology and Sociobiology. 2005;57(4):327–38.
  37. 37. Herbst CT, Švec JG, Lohscheller J, Frey R, Gumpenberger M, Stoeger AS, et al. Complex vibratory patterns in an elephant larynx. Journal of Experimental Biology. 2013;216(21):4054–64.
  38. 38. Stoeger AS, Heilmann G, Zeppelzauer M, Ganswindt A, Hensman S, Charlton BD. Visualizing Sound Emission of Elephant Vocalizations: Evidence for Two Rumble Production Types. PLoS ONE. 2012;7(11):e48907. pmid:23155427
  39. 39. Gahr M. Sexual Differentiation of the Vocal Control System of Birds. Advances in Genetics. Volume 59: Academic Press; 2007. p. 67–105. pmid:17888795
  40. 40. Ganswindt A, Rasmussen HB, Heistermann M, Hodges JK. The sexually active states of free-ranging male African elephants (Loxodonta africana): defining musth and non-musth using endocrinology, physical signals, and behavior. Hormones and Behavior. 2005;47(1):83–91. pmid:15579269
  41. 41. Lueders I, Hildebrandt TB. Chapter 66—Female Elephant Reproduction A2—Miller, R. Eric. In: Fowler M, editor. Fowler's Zoo and Wild Animal Medicine. Saint Louis: W.B. Saunders; 2012. p. 503–14.
  42. 42. Lee PC, Poole JH, Njiraini N, Moss CJ. Male social dynamics: independence and beyond. In: Moss CJ, Croze HJ, Lee PC, editors. The Amboseli Elephants: A longterm perspective on a long-lived mammal. Chicago: University of Chicago Press; 2011. p. 260–71.
  43. 43. McComb K, Moss C, Durant SM, Baker L, Sayialel S. Matriarchs As Repositories of Social Knowledge in African Elephants. Science. 2001;292(5516):491–4. pmid:11313492
  44. 44. Shannon G, Slotow R, Durant SM, Sayialel KN, Poole J, Moss C, et al. Effects of social disruption in elephants persist decades after culling. Frontiers in Zoology. 2013;10(1):62. pmid:24152378
  45. 45. Leighty KA, Soltis J, Wesolek CM, Savage A. Rumble vocalizations mediate interpartner distance in African elephants, Loxodonta africana. Animal Behaviour. 2008;76(5):1601–8.
  46. 46. Ortolani A, Leong K, Graham L, Savage A. Behavioral indices of estrus in a group of captive African elephants (Loxodonta africana). Zoo Biology. 2005;24(4):311–29.
  47. 47. Noll A, White J. S_Tools-STx Online Manual 2017. Available from:
  48. 48. Zeppelzauer M, Stoeger AS. Establishing the fundamentals for an elephant early warning and monitoring system. BMC Research Notes. 2015;8(1):1–15.
  49. 49. Benjamini Y, Hochberg Y. Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society, Series B (Methodological). 1995;57(1):289–300.
  50. 50. Mundry R, Sommer C. Discriminant function analysis with nonindependent data: consequences and an alternative. Animal Behaviour. 2007;74(4):965–76.
  51. 51. Venables WN, Ripley BD. Modern Applied Statistics with S. 4th ed. New York: Springer; 2002.
  52. 52. Shrader AM, Ferreira SM, McElveen ME, Lee PC, Moss CJ, Van Aarde RJ. Growth and age determination of African savanna elephants. Journal of Zoology. 2006;270(1):40–8.
  53. 53. McCowan B, Reiss D. Whistle contour development in captive-born infant bottlenose dolphins (Tursiops truncatus): Role of learning. Journal of Comparative Psychology. 1995;109(3):242–60.
  54. 54. Mitani JC, Brandt KL. Social Factors Influence the Acoustic Variability in the Long-distance Calls of Male Chimpanzees. Ethology. 1994;96(3):233–52.
  55. 55. Pfefferle D, Fisher J. Sounds and size: identification of acoustic variables that reflect body size in hamadryas baboons, Papio hamadryas. Animal Behaviour. 2006;72:43–51.
  56. 56. Briefer E, McElligott AG. Indicators of age, body size and sex in goat kid calls revealed using the source–filter theory. Applied Animal Behaviour Science. 2011;133(3):175–85.
  57. 57. Rendall D, Kollias S, Ney C, Lloyd P. Pitch (F0) and formant profiles of human vowels and vowel-like baboon grunts: The role of vocalizer body size and voice-acoustic allometry. The Journal of the Acoustical Society of America. 2005;117(2):944–55. pmid:15759713
  58. 58. Taylor AM, Reby D, McComb K. Human listeners attend to size information in domestic dog growls. The Journal of the Acoustical Society of America. 2008;123(5):2903–9. pmid:18529206
  59. 59. Dabbs JM Jr, Mallinger A. High testosterone levels predict low voice pitch among men. Personality and Individual Differences. 1999;27(4):801–4.
  60. 60. Morisaka T, Shinohara M, Nakahara F, Akamatsu T. Effects of Ambient Noise on the Whistles of Indo-Pacific Bottlenose Dolphin Populations. Journal of Mammalogy. 2005;86(3):541–6.
  61. 61. Charlton B, Zhihe Z, Snyder R. The information content of giant panda, Ailuropoda melanoleuca, bleats: acoustic cues to sex, age and size. Animal Behaviour. 2009;78(4):893–8.
  62. 62. Todd NE. Qualitative Comparison of the Cranio-Dental Osteology of the Extant Elephants, Elephas Maximus (Asian Elephant) and Loxodonta africana (African Elephant). The Anatomical Record: Advances in Integrative Anatomy and Evolutionary Biology. 2010;293(1):62–73.
  63. 63. Van der Merwe NJ, Bezuidenhout AJ, Seegers CD. The skull and mandible of the African elephant (Loxodonta africana). Onderstepoort Journal of Veterinary Research. 1995;62:245–60. pmid:8668323
  64. 64. Youlatos D, Couette S, Halenar LB. Morphology of Howler Monkeys: A Review and Quantitative Analyses. In: Kowalewski MM, Garber PA, Cortés-Ortiz L, Urbani B, Youlatos D, editors. Howler Monkeys: Adaptive Radiation, Systematics, and Morphology. New York, NY: Springer New York; 2015. p. 133–76.
  65. 65. Frey R, Riede T. Sexual Dimorphism of the Larynx of the Mongolian Gazelle (Procapra gutturosa Pallas, 1777) (Mammalia, Artiodactyla, Bovidae). Zoologischer Anzeiger—A Journal of Comparative Zoology. 2003;242(1):33–62.
  66. 66. Shoshani J. Understanding proboscidean evolution: a formidable task. Trends in Ecology & Evolution. 13(12):480–7.
  67. 67. Lee PC, Moss CJ. Statural growth in known-age African elephants (Loxodonta africana). Journal of Zoology. 1995;236(1):29–41.
  68. 68. Sukumar R. Appendix 2: Statural Growth in Elephants. The Living elephants: Evolutionary Ecology, Behavior and Conservation. Oxford, New York: Oxford University Press; 2003. p. 409–14.
  69. 69. Poole JH. Rutting Behavior in African Elephants: The Phenomenon of Musth. Behaviour. 1987;102(3/4):283–316.
  70. 70. Payne KB, Thompson M, Kramer L. Elephant calling patterns as indicators of group size and composition: the basis for an acoustic monitoring system. African Journal of Ecology. 2003;41(1):99–107.
  71. 71. Thompson ME, Schwager SJ, Payne KB. Heard but not seen: an acoustic survey of the African forest elephant population at Kakum Conservation Area, Ghana. African Journal of Ecology. 2010;48(1):224–31.
  72. 72. Thompson ME, Schwager SJ, Payne KB, Turkalo AK. Acoustic estimation of wildlife abundance: methodology for vocal mammals in forested habitats. African Journal of Ecology. 2010;48(3):654–61.
  73. 73. Zeppelzauer M, Hensman S, Stoeger AS. Towards an automated acoustic detection system for free-ranging elephants. Bioacoustics. 2014;24(1):13–29.
  74. 74. Stoeger AS, Zeppelzauer M, Baotic A. Age-group estimation in free-ranging African elephants based on acoustic cues of low-frequency rumbles. Bioacoustics. 2014:1–16.
  75. 75. Guidelines for the treatment of animals in behavioural research and teaching. Animal Behaviour. 1997;53(1):229–34.