universidade federal do rio grande do norte … · domo arigato gozaimashita ao meu sensei james!...
TRANSCRIPT
*London J. 1903. The call of the wild. New York: Macmillian Publishers.
UNIVERSIDADE FEDERAL DO RIO GRANDE DO NORTE
PROGRAMA DE PÓS-GRADUAÇÃO EM PSICOBIOLOGIA
The call of the (Neotropical) wild*: maned wolf long-range acoustic
ecology
LUANE MARIA STAMATTO FERREIRA
Orientadora: Profª. Drª. Renata Santoro de Sousa Lima
Co-orientador: Prof. Dr. Flávio Henrique Guimarães Rodrigues (UFMG)
NATAL – RN
2019
LUANE MARIA STAMATTO FERREIRA
The call of the (Neotropical) wild: maned wolf long-range acoustic
ecology
Tese de doutorado apresentada ao Programa
de Pós-graduação em Psicobiologia do
Centro de Biociências da Universidade
Federal do Rio Grande do Norte, como parte
dos requisitos para a obtenção do título de
Doutor.
Orientadora: Profª. Drª. Renata Santoro de Sousa Lima
Co-orientador: Prof. Dr. Flávio Henrique Guimarães Rodrigues (UFMG)
NATAL – RN
2019
UNIVERSIDADE FEDERAL DO RIO GRANDE DO NORTE
Título da tese: The call of the (Neotropical) wild: maned wolf long-range acoustic ecology
Autora: Luane Maria Stamatto Ferreira
Orientadora: Profa. Renata Sousa-Lima
Coorientador: Prof. Flávio Rodrigues
Data da defesa: 28 de fevereiro de 2019
Parecer:
Banca examinadora:
Profa. Dr. Renata Sousa-Lima (presidente)
Universidade Federal do Rio Grande do Norte (UFRN)
Prof. Dr. Jeff Podos
University of Massachusetts Amherst
Prof. Dr. Holger Klinck
Cornell University
Profa. Dra. Julie Patris
Aix-Marseille Université
Profa. Dra. Susan Parks
Syracuse University
Agradecimentos
A aparente solitária e monumental tarefa de escrever uma tese é, na verdade, o
trabalho de um exército e eu agradeço a todos que apoio por apoio, ideia por ideia,
construíram esse projeto. Assim como o lobo-guará parece solitário, mas está inserido
em uma rede social de comunicações, eu agradeço a vocês que foram, mesmo que
alguns à longas distâncias, minha alcateia.
Agradeço primeiramente aos meus pais, Inês e Adir, pelo excepcional cuidado
parental. Agradeço ao meu parceiro Carlos Carvalheira, que caminhou sempre ao meu
lado. Agradeço a meu irmão Alessandro e sua parceira Sarah, que já dispersaram para
Curitiba, mas sempre serão parte da alcateia. E também aos meus canídeos não
selvagens, Neblina e Luke Luke.
Agradeço à minha família estendida, o Laboratório de Bioacústica da UFRN.
Em especial aos lobos Victor Sábato (o primeiro lobo), Luciana Rocha (a primeira loba
selvagem), Danielly Duarte, Edvaldo Neto, Thiago Pinheiro, Rafael Frigo, Flávio
Rodrigues (o lobo-mor) e claro, à líder do grupo de caça, Renata Sousa Lima. Sem sua
visão e orientação esse projeto não seria possível e nenhum lobo teria chegado tão
longe.
Meus agradecimentos também vão a todos outros integrantes do LaB, mesmo os
que ficaram por tempo breve. Vocês foram mais que colegas, foram amigos e família.
Daqueles de mar e de terra, de rio e de ar, desde as intérpretes de baleiês, aos “ouvidores
de tudo” (paisagem acústica), as lontras (e ariranhas), tartarugas, golfinhos, focas,
saguis, passarinhos e gaviões. Meus agradecimentos especiais aos companheiros de
aventura: Eliziane, Divna, Lara, Marcos e Letícia. E a todos colaboradores, Milagros
Villavicencio, Júlio Baumgarten, Eduardo Venticinque e Mauro Pichorim.
Agradeço as pessoas que se disponibilizaram a ler e também contribuir com suas
vozes no trabalho: Jeff Podos, Holger Klink, Julie Patris, Paulo Cordeiro, Susan Parks e
Gustavo Zampier.
Agradeço as pessoas cujo apoio logístico e de know-how foi essencial ao projeto,
em particular Jean Piere (que é afiliado do Dietz!), Ricardo, Marcello, Júlia Simões
(time 3m), Rogério Cunha de Paula e Flávio Rodrigues. Agradeço as pessoas fantásticas
que conheci em São Roque de Minas, incluindo Adriano Gambarini e Flávia Ribeiro.
Também aqueles que me ofereceram estadia nos momentos mais críticos: Renilda e
funcionários da pousada Chapadão da Canastra, Pavel (melhor flash-touristic-guide de
Ilhéus), Maria Inês Santoro (a mãe da Renata, que tem o mesmo nome da minha!), e
Rachel e Gustavo, amigos de longa data (saudades). Me senti acolhida em todos
sentidos.
Agradeço a CAPES pela bolsa, ao Programa de Pós-Graduação em
Psicobiologia da UFRN pelo auxílio financeiro, ao ICMBio pela licença concedida e
aos funcionários do centro administrativo do PNSC, em São Roque de Minas, por terem
sido tão compreensivos e prestativos.
Agradeço de coração aos amigos que me mantiveram sã no processo, os grupos
do RPG, boardgames, aikidô, séries (Lúcifer!), biologia (As Cobaias) e tantos outros.
Um obrigada muito especial a Rodrigo, Thieza, Amanda, Aja, Fred Dimitrius, Geórgia,
Girão, Ramon, Moal, Nicolau, Walles, Dinara, Suelen, Naíra, Carol e Nelson. E um
domo arigato gozaimashita ao meu sensei James!
Por fim, agradeço aos lobos-guará do Parque Nacional da Serra da Canastra que,
mesmo sem terem assinado termo de consentimento livre esclarecido, tiveram suas
conversas pessoais gravadas, dando, literalmente, voz a esse projeto. (Também espero
que um dia eles tragam meu celular de volta).
Enfim, obrigada a todos.
Passamos por fogo e foi preciso ser ninja,
mas esta etapa está concluída!
Luane Maria Stamatto Ferreira
2019
For the strength of the Pack is the Wolf, and the strength of the Wolf is the Pack.
– Rudyard Kipling
Abstract
Maned wolves are difficult to observe in the wild because of their low densities
and their cryptic and crepuscular-nocturnal habits. Exploring their long-range call – the
roar-bark – is an efficient alternative for studying the species. We used a combination of
methodologies: we played back roar-barks in the wolves’ natural habitat to test how
free-ranging animals would respond and to understand the propagation properties of this
vocalization in the wild; we recorded spontaneous roar-bark sequences of wild maned
wolves using a grid of autonomous recorders for eight months to reveal long term
temporal patterns; and we used captive records to access sex and individuality in the
roar-bark and to test its application to natural recordings. We found that maned wolves
vocalize more during the beginning of the night, and this was the only period we
obtained responses during the playback experiment, despite both twilights having
efficient propagation of roar-barks. Social factors may be influencing the timing of the
wolves’ long-range vocal activity. We suggest that roar-barks may be an honest
advertisement of quality for territorial defense. Maned wolves vocalize more on better
moonlit nights, especially when the first half of the night is illuminated, likely as a
consequence of reduced foraging time and therefore having more time to invest in
acoustic communication. It was possible to identify the mating and circa-parturition
period in our natural recordings by an increase in solo and group vocal activity, which
suggests a role of roar-barks in partner attraction/guarding and intra-familiar-group
communication. In captivity, male roar-barks were distinguishable by their longer
duration, also indicating a sexual function and suggesting a higher energy investment to
advertise motivation. Roar-barks were also individually distinct. However, site
characteristics, such as presence of vegetation, drastically affected both the propagation
of broadcasted roar-barks and most identity and sexual parameters’ transmission in the
wild. Elevating the speaker 45° upward to simulate the head/muzzle position during
vocalization lead to lower recorded sound intensities, but partially counteracted the
negative effects of vegetation on signal transmission. The few stable parameters were
able to discriminate individuals, although with lower success rate. In wild recordings
the variation of parameters due to propagation was larger than the variation due to
individual differences, therefore limiting passive acoustic monitoring as a means of
counting individuals in their natural habitats. Despite the present limitation of vocal
identification in the wild, bioacoustic tools proved efficient in revealing the secretive
behavior ecology of maned wolves.
Key-words: Chrysocyon brachyurus, canid, vocalization, sound propagation, passive
acoustic monitoring, temporal patterns, playback.
Resumo
Os lobos-guará são difíceis de serem observados na natureza devido as suas
baixas densidades e hábitos crípticos e noturno-crepusculares. Explorar seu chamado de
longa distância – o aulido – pode ser uma alternativa eficiente para estudar a espécie.
Usando uma combinação de metodologias: reproduzindo aulidos no ambiente natural da
espécie para testar como animais de vida livre responderiam e para entender as
propriedades de propagação dessa vocalização; gravando sequências de aulidos
espontâneas de lobos-guará selvagens através de uma rede de gravadores autônomos por
oito meses para revelar padrões temporais de longo prazo; e registramos os sons
produzidos em cativeiro para conferir a discriminação de gênero e individualidade no
aulido e testar sua aplicação em gravações de ambiente natural. Nós descobrimos que os
lobos-guará vocalizam mais no início da noite, e esse foi o único período em que
obtivemos respostas durante o experimento de playback, apesar de ambos crepúsculos
apresentarem uma propagação eficiente deste tipo de som. Fatores sociais podem estar
influenciando esse padrão temporal, como o anúncio honesto de qualidade para defesa
territorial. Lobos-guará vocalizam mais em noites de maior iluminação lunar,
especialmente quando a primeira metade da noite está iluminada, provavelmente como
consequência de uma redução no tempo de forrageio e, portanto, mais tempo para
investir na comunicação acústica. Foi possível identificar o período de acasalamento e
aquele em torno do parto nas nossas gravações de ambiente natural através do aumento
na atividade vocal solo e de grupo, o que indica um papel dos aulidos na atração/guarda
de parceiros e na comunicação intra grupo familiar. Em cativeiro, os aulidos dos
machos foram distinguíveis principalmente por sua duração mais longa, também
indicando uma função sexual e sugerindo um investimento energético mais alto para
anunciar motivação. Aulidos também foram distintos individualmente. Porém,
características locais afetaram dramaticamente tanto a propagação dos aulidos
reproduzidos quanto quase todos parâmetros que conferem identidade e gênero aos sons
emitidos. Elevar a caixa de som 45° para cima para simular a posição da cabeça/focinho
durante a vocalização resultou em intensidades sonoras mais baixas nas gravações, mas
compensou parcialmente os efeitos negativos da vegetação na transmissão do sinal
acústico. Os poucos parâmetros estáveis durante a propagação em ambiente natural
foram capazes de discriminar indivíduos, embora com menor taxa de sucesso.
Infelizmente, nas gravações obtidas na natureza a variação dos parâmetros devido à
propagação foi maior que as diferenças individuais observadas. Apesar da presente
inaplicabilidade da identificação vocal em gravações de aulidos na natureza, as
ferramentas bioacústicas se provaram eficientes em revelar a elusiva ecologia
comportamental dos lobos-guará.
Palavras-chave: Chrysocyon brachyurus, canídeo, vocalização, propagação sonora,
monitoramento acústico passivo, padrões temporais, playback.
Figure list
Figure A. Maned wolf. From: Paula & Gambarini 2013 (book) ………………………….……………. 20
Figure B I. Maned wolf roar-barking. From: Paula & Gambarini 2013 (book) ……….……………..… 23
Figure B II. Maned wolf (male “Nopal”) roar-barking at the Endangered Wolf Center in St. Louis,
MO/U.S.A. Photo: Michelle Steinmeyer, 2015. ……………………………………….………….…….. 24
Figure B III. Radio collared maned wolf roar-barking. Photo: Flávio H. G. Rodrigues, 2011. .............. 25
Figure C. Broadcasting (Pioneer S-DJ50X speaker) maned wolf roar-barks and re-recording them
(SongMeter SM2+) at different distances (top: 03/07/2017) and deploying an autonomous recorder (from
13) to passively register spontaneous roar-barks sequences (bottom: 03/09/2016). Serra da Canastra
National Park, Minas Gerais, Brazil. …………………………………………………………….……… 27
Chapter 1
Figure 1. Location of passive autonomous recorders and playback sites to study maned wolves at Serra
da Canastra National Park, Minas Gerais, Brazil. Imagery ©2018 CNES / Airbus, Map data ©2018
Google…………...………………………………………………………………………………….……. 34
Figure 2. Edited maned wolf roar-bark sequences used as stimuli for playback studies of maned wolves
in the wild (Serra da Canastra National Park, Brazil). GA and SH are males, SA and JU are females. Top
spectrograms are the original files (96 kHz sample rate, 32 bit wav, 4000 windows size, 56% brightness
and 50% contrast) and the bottom a recording extracted from one autonomous recorder (Song Meter
SM2+; Wildlife Acoustics) 80 meters from the playback speaker (8 kHz sample rate, 16 bit wav, 512
windows size, 50% brightness and contrast). Spectrogram made on Raven Pro 1.5…………….…..….. 37
Figure 3. Distribution of wild maned wolf roar-bark sequences registered between March 04 and 11 2017
during a playback experiment at Serra da Canastra National Park, MG/Brazil. Each sequence is named by
its start time and the size of the bar shows the time elapsed from the last broadcasted playback sequence.
This time is also discriminated on tags above the sequences considered responses to the playback, i.e.
those within 10 minutes after the end of any broadcasted sequence. ……………………….…..…......... 42
Figure 4. Temporal distribution of maned wolf roar-bark sequences recorded at Serra da Canastra
National Park, MG/Brazil, with autonomous recorders (Song Meter SM2+; Wildlife Acoustic). March
2017 (black line, left axis): percentage relative to the total (30 sequences) of vocal activity registered on
continuous recordings of the 6 days in which the roar-bark playback experiment was conducted. March
2016 (dark gray bars, right axis): absolute number of sequences (total 224), 13 recorders, 20 nights, from
5 PM to 5 AM. April 2014 (light gray bars, right axis): absolute number of sequences (total 192), 12
recorders, 25 nights, from 6 PM to 6 AM (Rocha et al. 2016 dataset, used with permission). …...….…. 47
Supplementary Data SD1.Wild maned wolf roar-bark sequences recorded at Serra da Canastra National
Park, MG/Brazil. a and b: sequence in response to a playback stimulus registered on 18:55:37 March 09
2017. c and d: two individuals (note the change in spectral characteristics after 33 s) recorded passively
on 19:15:00 March 17 2016. Recordings made with autonomous recorders (Song Meter SM2+; Wildlife
Acoustic), at 8 kHz sample rate and 16-bit wav file format. Spectrogram made on Raven pro 1.5 (Cornell
Bioacoustics Lab, Ithaca, NY, USA), Hann window, 512 window size, 50% brightness and contrast, 50%
overlap, smoothing on. .…………………………………….. ……………………………………...…… 59
Chapter 2
Figure 1. Study site at Serra da Canastra National Park, MG, Brazil. a - site Flat; b - site Low to high; c -
site Vegetation; d - site High to low. Horizontal distance to the speaker is discriminated on the left side of
the / and altitude on the right side. Maps constructed with QGIS 3.4.0-Madeira (QGIS Development
Team, 2018. QGIS Geographic Information System. Open Source Geospatial Foundation
Project. http://qgis.osgeo.org) and Google Satellite images (Map data ©2018 Google, Imagery ©2018
TerraMetrics). ……………………………………………………………………………..…………….. 70
Figure 2. Captive maned wolves roar-barks sequences broadcasted at Serra da Canastra National Park,
MG/Brazil. GA and SH are males, SA and JU females. Red selection boxes on the first roar-bark of each
animal exemplifies the ones used to measure roar-bark intensity (peak power, dB). Selections near the
second roar-bark of each animal exemplifies the ones used to measure noise intensity (average power,
dB). Spectrograms and measures were made on Raven Pro 1.5 (Cornell Bioacoustics Lab, Ithaca, NY,
USA), Hann window, 512 window size, 50% brightness and contrast, 50% overlap, smoothing “on”... 76
Figure 3. Propagation of broadcasted roar-barks from captive maned wolves at Serra da Canastra
National Park, MG/Brazil. Re-recordings made with autonomous recorders (Song Meter SM2+; Wildlife
Acoustics, Inc., Concord, Massachusetts). The intensity loss is relative to the re-recording at 1.25m …. 80
Figure 4. Propagation of broadcasted captive records of maned wolves roar-barks at 4 sites at Serra da
Canastra National Park, MG/Brazil. Re-recordings made with autonomous recorders (Song Meter SM2+;
Wildlife Acoustics, Inc., Concord, Massachusetts). The intensity loss is relative to the re-recording at
1.25m. ……………………………………………...…………………………...………...…………..…. 82
Figure 5. Propagation of broadcasted captive records of maned wolves roar-barks at 4 sites at Serra da
Canastra National Park, MG/Brazil. We conducted broadcasts with the speaker box positioned straight
forward (Straight) and with the speaker box inclined 45o upward (Inclined) to simulate the inclination of
the head/muzzle seen when animals roar-bark. Re-recordings made with autonomous recorders (Song
Meter SM2+; Wildlife Acoustics, Inc., Concord, Massachusetts). The intensity loss is relative to the
straight re-recording at 1.25m. …………………………………………………………………..….…… 84
Figure 6. Propagation of broadcasted captive records of maned wolves roar-barks at 6 time intervals at
Serra da Canastra National Park, MG/Brazil. The time shown is the beginning of a 1 hour interval in
which broadcasts were made. Re-recordings made with autonomous recorders (Song Meter SM2+;
Wildlife Acoustics, Inc., Concord, Massachusetts). The intensity loss is relative to the re-recording at
1.25m……………………………………………….…………………………...……………………….. 85
1 – Preliminaty data exploration (supplementary material) ……………………..…………………….… 94
2 – Normality and homogeneity of residuals (supplementary material) ……………………...…………. 95
3 – Predicted x observed values of the model (supplementary material) …………………...……..……. 96
Chapter 3
Figure 1. Study region at the Serra da Canastra National Park, MG/Brazil. Yellow squares indicate
autonomous recorder (SongMeter SM2+) sites used only in 2014, pink triangles sites used only in 2016,
and white circles sites used in both years. …………...……………………………………….….......… 107
Figure 2. Wild maned Wolf roar-bark sequences recorded passively with a grid of 12/13 autonomous
recorders (Song Meter SM2+) at Serra da Canastra National Park, MG/Brazil. a) solo roar-bark sequence
with detection boxes from XBAT (Figueroa 2007) extension for Matlab (MathWorks, Inc.). Spectrogram
parameters: 512 window size, Hann window, 100% brightness, 43% contrast, 50% overlap on an 8 kHz
sampling rate wav file. b) group vocalization consisting of two animals alternating roar-barks.
Spectrogram made on Raven pro 1.5 (Cornell Bioacoustics Lab, Ithaca, NY, USA), Hann window, 512
window size, 50% brightness and contrast, 50% overlap, smoothing on, on an 8 kHz sampling rate wav
file. ...…………………………………………………………………................................................… 111
Figure 3. Histogram of the number of roar-barks on each sequence of maned wolves’ vocalizations
recorded passively with a grid of 12/13 autonomous recorders (Song Meter SM2+) at Serra da Canastra
National Park, MG/Brazil. One sequence was defined by one or a bout of roar-barks not separated by
more than 10 seconds. ………………………………………………………………………….....……. 115
Figure 4. Seasonal variation in the maned wolf vocal activity recorded passively with a grid of 12/13
autonomous recorders (Song Meter SM2+) at Serra da Canastra National Park, MG/Brazil. Each point is
a sum of 5 nights. Photos: Endangered Wolf Center, St.Louis, and Adriano Gambarini ……………… 116
Figure 5. Maned wolf roar-bark sequences distribution over the lunar phases (gray = total). Records were
made from April to July on 2014 (blue) and from March to June on 2016 (red) with a grid of 12/13
autonomous recorders (Song Meter SM2+) at Serra da Canastra National Park, MG/Brazil. The radial line
represents the mean angle and the concentric bar at the end of the line the 95% confidence interval. …119
Figure 6. Maned Wolf roar-barks registered between 17-19h on passive audio recordings made with a
grid of 12/13 autonomous recorders (Song Meter SM2+) at Serra da Canastra National Park, MG/Brazil.
In 2014 there was no recordings in March, and in 2016 no recordings in July. ………….……….…… 121
Figure 7. Maned wolf nightly vocal activity relative to sunset. Recordings were made with a grid of
12/13 autonomous recorders (Song Meter SM2+) at Serra da Canastra National Park, MG/Brazil…… 122
Figure 8. Maned wolf roar-bark sequences recorded passively with 12 autonomous recorders (Song
Meter SM2+) at Serra da Canastra National Park, MG/Brazil (gray contour). Circles have approximately
0.5 km radius with the center point being the recorder site. Heat colors represent the intensity of vocal
activity (number of sequences). Letters indicates the site name and numbers following them on the circles
indicates the amount of group vocalizations. * Indicates at least one sequence involved 3 animals
(otherwise group vocalizations involve 2 animals).…………………………………………...…...…... 124
Figure 9. Maned wolf roar-bark sequences recorded passively with 13 autonomous recorders (Song
Meter SM2+) at Serra da Canastra National Park, MG/Brazil (gray contour). Circles have approximately
0.5 km radius with the center point being the recorder site. Heat colors represent the intensity of vocal
activity (number of sequences). Letters indicates the site name and numbers following them indicates the
amount of group vocalizations (sequences involving 2 animals). ….…………...…………………..…. 125
Figure S1. Nightly maned wolf roar-barks recorded passively with 12 autonomous recorders (Song Meter
SM2+) at Serra da Canastra National Park, MG/Brazil (gray contour). From top-left to bottom-right:
April 05/06/07/08 2014, April 15/16/17/18 2014, March 19/20/21/22/23 2016, and May 01/02/03/04
2016. **indicates at least one sequence of roar-barks involved two animals. *** indicates at least one
sequence of roar-barks involved three animals. ……………………………………………...………… 140
Chapter 4
Figure 1. Maned wolves roar-barks recorded in Minas Gerais, Brazil. a. One example of roar-bark from
each of 10 individuals (letters) recorded with unidirectional microphone and a hand recorder in two
captivity facilities. b. Roar-barks of GA and JU broadcasted and re-recorded with autonomous recorders
at 7 different distances from the speaker on site “Flat” at the Serra da Canastra National Park. c. Free-
ranging animals spontaneous roar-bark sequences recorded with autonomous recorders at the same park:
top spectrogram shows some roar-barks (numbers) from a sequence involving two animals (letters);
bottom spectrogram shows the same sequence recorded by another autonomous recorder 2.41 km away,
roar-barks from “b” reach the recorders on different times because animals are at different positions.
Spectrogram parameters: a. 96 kHz sample rate, 3080 window size, Hann window, 55% brightness, 60%
contrast, 24-bit wav; b. and c. 8 kHz sample rate, 512 window size, Hann window, 50% brightness, 60%
contrast, 16-bit wav. ……………………………………………………………….............................… 152
Figure 2. First 3 linear discriminant functions for identity discrimination of 10 captive maned wolves
(colors) roar-barks recorded from two facilities at Minas Gerais, Brazil. …………………....…..……. 159
Figure 3. Differences in the roar-barks parameters between females and males maned wolves recorded in
two facilities in Minas Gerais, Brazil. ……………………………………………………..………..…. 161
Figure 4. Variation of 8 selected parameters of broadcasted maned wolves roar-barks re-recorded at 7
distances (1.25-640m) at the Serra da Canastra National Park, MG/Brazil. ……………….….………. 163
Figure 5. LDA percentage of correctly identity classification of broadcasted roar-barks of 4 maned
wolves (bottom) re-recorded at 7 distances (1.25-640m) in 4 sites (top) at the Serra da Canastra National
Park, MG/Brazil. ……………………………………………………………………..…...............……. 164
Figure 6. Figure 6. Signal-to-noise ratio of broadcasted roar-barks of maned wolves re-recorded at 7
distances (1.25-640m) in 4 sites at the Serra da Canastra National Park, MG/Brazil. The signal-to-noise
ratio was calculated subtracting from the in-band power of each roar-bark (150-2000 Hz) and the same
measurement taken from an equal sized spectrogram portion immediately before the vocalization
(measure of the background noise level). ………………………………………………...………….… 165
Figure 7. Two different roar-bark sequences involving the same two maned wolves each (top), and their
roar-bark parameters (bottom). Recordings made passively by two different autonomous recorders
(SongMeter SM2+) at the Serra da Canastra National Park, MG/Brazil …………..…………….…….. 168
Appendix I
Maned wolves do not emit more roar-barks than expected by chance during the illuminated versus the
non-illuminated portion of the night, except from new to waxing crescent phase. In this phase only the
first part of the night is illuminated, and thus the difference may be caused by the species preference to
vocalize on this time. *t=2.906, df=33, p=0.006. Graph extracted and translated from: ÁRAUJO, D.D.,
FERREIRA, L.S., ROCHA, L.H.S., & SOUSA-LIMA, R.S. 2016. Influência do ciclo lunar nas
vocalizações de lobo guará. Abstract and poster presentation at the III Conferência e VIII Simpósio de
Psicobiologia, UFRN, Natal, Rio Grande do Norte, Brazil. ................................................... ................. 191
Appendix II
Individual variation of the time interval between the start of one maned wolf roar-bark to the next one in
the sequence. Potential for Identity Coding (PIC; as in Robisson et al. 1993) for this parameter is 1.04.
Data from 10 captive maned wolves, 24-124 roar-barks by individual, 897 roar-barks in total, recorded in
2010 by V.S. Rocha, at 2 facilities in Minas Gerais, Brazil …………………………………………… 192
Annex I
A sedated lactating maned wolf being examined by the Lobos da Canastra team. This female (known as
“Rose”) was captured on the night between July 12-13 2016 at the Serra da Canastra National Park,
around site F (Figure 1, Chapter 3). At this year she was already without a VHS collar, but her data shows
she lived on the present study area since at least 2014, when she was first captured, also lactating. Photo:
Rogério Cunha de Paula. Used with permission. ………………………………………………...……. 193
Table list
Chapter 1
Table 1. Wild maned wolves’ vocal activity recorded by autonomous recorders (Song Meter SM2+;
Wildlife Acoustics) at Serra da Canastra National Park, MG/Brazil, during the playback experiment days
on 2017. Each row is a different roar-bark sequence. Those considered playback responses are
underlined. Each playback session consisted of 8 broadcasted roar-bark sequences. The column “Heard?”
indicates if the researchers, when present in loco, heard maned wolves’ calls …………………………. 43
Chapter 2
Table 1. ANOVA test for the fixed factors of the main model for the intensity (dB) loss of maned wolves
roar-barks broadcasted on their natural environment. ...…………...……………………………...…….. 78
Table 2. Approximate 95% confidence intervals for the estimate factor effects of the main model for the
intensity (dB) loss of maned wolves roar-barks broadcasted on their natural environment. Base categories
are specified under parenthesis. Factors/levels with positive and negative estimates, which indicates they
are not influential on the model or not significantly different from the base category, are
underlined………………...…………………………………………………………………..…...…...… 79
Table 3. Simultaneous tests for general linear hypotheses using Tukey contrasts for multiple comparisons
of means. The reported p values are adjusted by single-step method. Significance codes: 0 '***', 0.001
'**', 0.01 '*', 0.05 '.', 0.1 ' '. Only comparisons of consecutive distances are shown. ………...………… 81
Table S2. ANOVA test for the fixed factors of the secondary model for the intensity propagation of
maned wolves roar-barks broadcasted on their natural environment. …………………………….…….. 97
Table S3. Secondary model: simultaneous tests for general linear hypotheses using Tukey contrasts for
multiple comparisons of means. The reported p values are adjusted by single-step method. Significance
codes: 0 '***', 0.001 '**', 0.01 '*', 0.05 '.', 0.1 ' '. Only com comparisons of consecutive distances are
shown. ………………………………………………………………………...……………….……..….. 97
Chapter 3
Table 1. Summary of maned wolf’s vocal activity recorded passively with a grid of 12/13 autonomous
recorders (Song Meter SM2+) at Serra da Canastra National Park, MG/Brazil. ………………...…..… 114
Table 2. Maned wolf vocal activity recorded passively with a grid of 12/13 autonomous recorders (Song
Meter SM2+) at Serra da Canastra National Park, MG/Brazil. Values reported are mean by night ± SD.
…………………………………………………………………………………………………………... 117
Table 3. Concentration of maned wolf vocal activity on each of eight moon cycles recorded passively
with a grid of 12/13 autonomous recorders (Song Meter SM2+) at Serra da Canastra National Park,
MG/Brazil. When the concentration is significant, the mean moon phase, mean angle ± circular standard
deviation, and Rayleigh test statistics are reported. ……………………………………….....……....… 120
Chapter 4
Table 1. Maned wolves recorded on 2010 at two facilities in Minas Gerais, Brazil. *estimated age. mx are
non-participant males. m3 is GA/GI half-brother. ………………………………………………..……. 148
Table 2. Selected parameters on maned wolves roar-barks (LSF analyst). “Full” measures refer to the
roar-bark from 150 Hz to 2000 Hz, while “First” and “Second” bands refer to portions from lower to
higher frequencies. Means±SD are for all 10 individuals, 20 roar-barks each. PIC = Potential for Identity
Coding (Robisson et al. 1993). (A) = unitless: proportion relative to entire duration. Parameters detailing
can be found on Raven’s manual (Charif et al. 2010). ……………………………………………….... 158
Table 3. Confusion matrix for the cross validation classification of 10 captive maned wolves by their
roar-barks. Average from the classification results of two analysts (LF, VS), each constructed by means
of the results from 1000 randomizations of 100 roar-barks (10 for each individual) from a total of 200.
Extracted by the pDFA R function written by Roger Mundry (2015 version). ……………..…………. 160
Table 4. ANOVA effect sizes (F) for the 8 selected parameters of 20 roar-barks from 4 maned wolves
broadcasted and re-recorded at 5 distances (1.25-160m) in 4 sites. (A) = unitless: proportion relative to
entire duration. α=0.0012. ……………………………………………………………………..…….…. 162
Table 5. Effect size (t) for the absolute difference between parameters of roar-barks from two different
maned wolves vocalizing together (2 wolves) and roar-barks simultaneously recorded by two different
autonomous recorders (2 recorders; SongMeter SM2+). 2 wolves: one sample t-test, df=91. 2 recorders:
paired t-test, df=82. α=0.0012. Cor. SNR: Pearson’s correlation coefficient with the signal-to-noise ratio
difference. (A) = unitless: proportion relative to entire duration. The vocalizations were recorded
passively with a grid of 12/13 autonomous recorders at Serra da Canastra National Park, MG/Brazil... 167
Table 6. Mean±SD difference of the absolute difference between parameters of roar-barks from two
different maned wolves vocalizing together (2 wolves) and roar-barks simultaneously recorded by two
different autonomous recorders (2 recorders; SongMeter SM2+). Those differences were compared with a
Welch two sample t-test. α=0.0012. (A) = unitless: proportion relative to entire duration. The
vocalizations were recorded passively with a grid of 12/13 autonomous recorders at Serra da Canastra
National Park, MG/Brazil. ………………………………………...…………...…………………....…. 167
Table S1. All measured parameters on maned wolves roar-barks (LSF analyst). “Full” measures refer to
the roar-bark from 150 Hz to 2000 Hz, while “First” and “Second” bands refer to portions from lower to
higher frequencies. Means±SD are for all 10 individuals, 20 roar-barks each. PIC = Potential for Identity
Coding (Robisson et al. 1993). # selected for the identity classification. (A) = unitless: proportion relative
to entire duration. *the selection box of the 2nd band is limited below by the selection box of the 1st band,
therefore we manually measured the 2nd band true lower frequency. Parameters detailing can be found on
Raven’s manual (Charif et al. 2010). …………………………………..………..……………..….…… 183
Table S2. Explained variance and coefficients of each linear discriminant function for identity
discrimination of 10 captive maned wolves roar-barks. Non-normal parameters were transformed (Yeo
Johnson). All parameters were centralized and scaled. ………………………....………….………….. 184
Table S3. ANOVA effect sizes (F) for all measured parameters on maned wolves roar-barks broadcasted
and re-recorded at 5 distances (1.25-160m). “Full” measures refer to the entire roar-bark (from 150 Hz to
2000 Hz), while “First” and “Second” bands refer to portions from lower to higher frequencies. (A) =
unitless: proportion relative to entire duration. α=0.0012. ……………………………………...……... 185
Summary
General introduction ....................................................................................................... 20
Chapter 1......................................................................................................................... 28
Using playbacks to monitor and investigate the behavior of wild maned wolves ......... 28
Abstract ....................................................................................................................... 29
Resumo ....................................................................................................................... 30
Introduction ................................................................................................................. 31
Materials and methods ................................................................................................ 34
Results ......................................................................................................................... 42
Discussion ................................................................................................................... 49
References ................................................................................................................... 55
Supplementary data ..................................................................................................... 59
Chapter 2......................................................................................................................... 60
Maned wolf long range call propagation and its implication for the species’
communication ............................................................................................................... 60
Abstract ....................................................................................................................... 61
Resumo ....................................................................................................................... 62
Introduction ................................................................................................................. 64
Materials and methods ................................................................................................ 69
Results ......................................................................................................................... 78
Discussion ................................................................................................................... 86
References ................................................................................................................... 90
Supplementary material .............................................................................................. 94
Chapter 3......................................................................................................................... 99
Temporal and spatial patterns of the long-range calls of maned wolves ........................ 99
Abstract ..................................................................................................................... 100
Resumo ..................................................................................................................... 101
Introduction ............................................................................................................... 103
Material and Methods ............................................................................................... 107
Results ....................................................................................................................... 114
Discussion ................................................................................................................. 127
References ................................................................................................................. 136
Supplementary material ............................................................................................ 140
Chapter 4....................................................................................................................... 141
Identity and sex discrimination of roar-barks for captive and free-ranging maned wolves
...................................................................................................................................... 141
Abstract ..................................................................................................................... 142
Resumo ..................................................................................................................... 143
Introduction ............................................................................................................... 144
Material and Methods ............................................................................................... 148
Results ....................................................................................................................... 158
Discussion ................................................................................................................. 169
References ................................................................................................................. 177
Supplementary material ............................................................................................ 183
Final remarks ................................................................................................................ 186
Extra-textual references ................................................................................................ 188
Appendix I .................................................................................................................... 191
Appendix II ................................................................................................................... 192
Annex I ......................................................................................................................... 193
Annex II ........................................................................................................................ 194
20
General introduction
The maned wolf (Chrysocyon brachyurus, Illiger, 1815; Figure A) is the only
large canid of South America (Queirolo et al. 2007). The species is listed as Near
Threatened by IUCN (IUCN 2015), Vulnerable in the Brazilian list of endangered
species (Paula et al. 2013), and models predict a 30% reduction in populations in only 3
generations (21 years; Paula et al. 2008). The main threat to the species is the
destruction of its habitat, the Cerrado neotropical savanna, which is one of the world’s
most important biodiversity hotspots (Silva & Bates 2002).
Figure A. Maned wolf. From: Paula & Gambarini 2013 (book).
The maned wolfs’ iconic characteristics make it appealing to the public and a
potential flagship species for the conservation of this biome (Myers et al. 2000). Its
slender constitution and long legs are adapted to walk over the bushes and tall grass of
its environment, while its large ears are used to detect hidden rodents and birds (Rodden
21
et al. 2004). This last characteristic also suggests that acoustic communication among
individuals of this species over long distances might play a significant role.
Maned wolves present many typical canid characteristics, including
nocturnal/crepuscular habits, some degree of territoriality, monogamous breeding with
the formation of stable pairs, and biparental care of young (Rodden et al. 2004).
However, “canid rules” state that larger canids (>13 kg) tend to be: more social,
forming larger groups composed of the breeding pair and grown offspring; to have
young cared by parents and helpers; to have large litter sizes; and to hunt cooperatively
fot larger preys (Kleiman & Eisemberg 1973; Moehlamn 1987, 1989). Despite their size
(20-30 kg, 70-90 cm; Silveira 1999), maned wolves tend to behave opposite than
expected (Rodden et al. 2004): they forage alone for small vertebrates and fruits, the
pair is rarely seen together, the presence of helpers have never been confirmed, and the
litter size is small (captivity mean is 3; Maia & Gouveia 2002). They are normally
described as less social than many small canids for which at least the breeding pair
associates extensively, as with artic foxes (Frommolt et al. 2003), and crab eating foxes
(Courtenay & Maffei 2004).
In comparison, the maned wolf closest living relative, the bush dog (Speothos
venaticus; Slater et al. 2009), is highly social and also an exception to canid rules, in the
other direction (as it is small: 4-7 kg; Beisiegel & Ades 2002). They are hypercarnivores
and hunt cooperatively for large prey (Kleiman 1972; Beisiegel & Ades 2002), which
highlights the relationship of feeding habits and canid social systems and indicates
canid rules may not be widespread. The small items’ diet of the maned wolf makes
sharing unprofitable and the presence of conspecifics may interfere in foraging (McNab
1986). Accordingly, Melo et al. (2007) reports that a breeding pair and a juvenile slept
relatively close during the day but stayed far apart during the active hours (i.e., night).
22
This last study also suggests a higher intra-group (pair and occasional offspring)
association than previously though. Despite telemetry studies usually not showing so
(Jácomo et al. 2009), there are several instances of group bonding/socialization: reports
of a breeding pair sleeping together (Melo et al. 2009; Emmons 2012); cooperative
hunting (Jácomo et al. 2009); males accompanied by young (Rodrigues 2002); males
providing or regurgitating to the female and pups (Dietz 1984; Jácomo et al. 2009),
including regurgitation to 5-9 month young in captivity (Rasmussen & Tilson 1984);
and a group composed of the breeding pair and 3 juveniles that interacted often
(Emmons 2012). On Emmons’ (2012) work, the long-range acoustic communication
was very important for these interactions.
Maned wolves use multiple communication modalities (Rodden et al. 2004), as
visual (e.g. piloerection, gape, ear and tail positioning) and chemical (e.g. urine, feces,
scent marks). The acoustic communication channel is also expected to be exploited by
the species, especially considering the limitations of visual signaling in a
crepuscular/nocturnal and solitary animal (Fox 1975). The species has a complex
acoustic repertoire, including at least 10 types of vocalizations and combinations of
vocalizations (Sábato 2011). Maned wolves’ vocal repertoire comprises almost all canid
broad categories of vocalizations (Tembrock 1976), which points to complex social
interactions, as social and vocal complexity are generally linked (Freeberg 2006). The
most frequent type of vocalization recorded in captivity is the long-range roar-bark
(Sábato 2011; Figure B I-III), a call heard throughout the year in the wolves’ natural
habitat (Rodden et al. 2004). Thus, maned wolves may be solitary, but seem to maintain
social acoustic contact over distance.
Roar-barks are emitted in sequences of 5-15 units, spaced by 3-5 seconds, and
are proposed to have multiple functions. One of the most cited is territorial
23
announcement (Kleiman 1972; Rocha 2011), especially for same-sex spacing (Brady
1981). Maned wolves from adjacent ranges have been heard exchanging roar-barks and
emitting them when facing threats (conspecifics and humans; Dietz 1984), and in
captivity, same sex individuals often exchange roar-barks (Brady 1981; Sábato 2011).
Figure B I. Maned wolf roar-barking. From: Paula & Gambarini 2013 (book).
24
Figure B II. Maned wolf (male “Nopal”) roar-barking at the Endangered Wolf Center in St. Louis,
MO/U.S.A. Photo: Michelle Steinmeyer, 2015.
25
Figure B III. Radio collared maned wolf roar-barking. Photo: Flávio H. G. Rodrigues, 2011.
The other function often cited for the roar-bark is in intra-pair communication
(Rocha et al. 2016; Balieiro & Monticelli 2019), being more important during the
breeding season (Dietz 1984), even in captivity (Sábato 2011). Researchers report that
between roar-barks maned wolves aurally attend to answers (Emmons 2012) and search
visually for the partner (Bestelmeyer 2000), that during estrous they emit roar-barks
whenever the partner is outside of visual range (Rodden et al. 2004), that often the
partner appears after the vocalization or move towards it (Bestelmeyer 2000; Emmons
2012), and pairs have been heard exchanging roar-barks many times (Dietz 1984;
Sábato 2011; Emmons 2012; Balieiro & Monticelli 2019). Some authors thus propose
the intra-group (pair and occasional offspring) communication would be the main roar-
bark function (Emmons 2012).
Roar-barks are considered far ranging, with Brady (1981) stating a human could
discriminate individuals over 1 km. Indeed, it was the only maned wolf vocalization we
could detect with passive autonomous recorders in the wild (Rocha et al. 2015, 2016).
26
Maned wolves are more easily heard than visualized (Emmons 2012; personal
observation), and usually very hard to follow due to their shyness, habitat composition,
and crepuscular/nocturnal habits (Rodden et al. 2004). Thus, exploring their long-range
acoustic communication might be an efficient alternative to monitor populations and
elucidate the species behavior ecology. The potential for vocal individualization makes
the bioacoustics approach even more interesting.
In this work I explored the maned wolf long range acoustic communication to
better understand the species’ behavioral ecology. I did it through three complimentary
methodologies (Figure C). First, I conducted a roar-bark playback experiment in the
wolves’ natural habitat. The experiment had a dual objective: to test if, and how, wild
maned wolves would respond to roar-bark playbacks, assessing both evidence of its
function and monitoring applicability (chapter 1); and to understand how the roar-bark
propagates over distance, also testing if some behaviors, like period of the day and night
and head elevation, were related to acoustic propagation (chapter 2). Second, I passively
recorded spontaneous (non-playback elicited) roar-barks sequences of wild maned
wolves using 12/13 autonomous recorders during eight months over two years (2014
and 2016). Those recordings generated an enormous dataset (over 3.5 TB), that was
processed by automatically detecting roar-barks through a previously established
method (Rocha et al. 2015: 100% of detections in half the time x 92% for visual
inspection of spectrograms). The primary goal of those wild recordings was to
characterize the maned wolf seasonal, lunar and nightly roar-bark emission patterns
(chapter 3). Finally, I used the experimentally broadcasted roar-barks, the naturally
spontaneous vocalization recordings, and roar-barks recorded in captivity by V. Sábato
(2011) to confirm roar-bark individual and sexual discriminability through permuted
27
discriminant analyses and, more important, test the applicability of vocal
individualization in recordings from the wild (chapter 4).
Figure C. Broadcasting (Pioneer S-DJ50X speaker) maned wolf roar-barks and re-recording them
(SongMeter SM2+) at different distances (top: 03/07/2017) and deploying an autonomous recorder (from
13) to passively register spontaneous roar-barks sequences (bottom: 03/09/2016). Serra da Canastra
National Park, Minas Gerais, Brazil.
Examples of maned wolf roar-bark sequences recorded in this work can be heard
in https://soundcloud.com/luane-ferreira-327256713. Visual and acoustic data collection
at this park was authorized by Instituto Chico Mendes de Conservação da
Biodiversidade (ICMBio; SISBIO license number 41329-2, annexed in the end of this
text).
28
Chapter 1
Using playbacks to monitor and investigate the
behavior of wild maned wolves
29
Using playbacks to monitor and investigate the behavior of wild maned
wolves
Luane Stamatto Ferreira, Júlia Simões Damo, Victor Sábato, Júlio Baumgarten, Flávio
H. G. Rodrigues, Renata S. Sousa-Lima
Intended for submission to: Mastozoologia neotropical
Abstract
Maned wolves are difficult to observe in the wild because of their low densities
and their cryptic and crepuscular-nocturnal habits. Exploring their long-range acoustic
communication may offer an efficient alternative to study the species. Here we
evaluated the applicability of playbacks to study maned wolves in the wild and compare
the results with 20 nights of passive recordings on the same area and month during the
previous year. We obtained vocal responses on 3 of 6 nights tested, including responses
involving two animals and an approach after an interactive playback. Although we
conducted 3-6 playback sessions each day at different times, we only obtained vocal
responses during sessions between 17:00 and 19:40. During our passive recordings we
detected on average 0.86 roar-bark sequences per recorder per night, mostly during the
first half of the night. Vocal activity – responses and spontaneous roar-bark sequences –
during playback nights was nearly 4 times greater than during the passive recordings.
We conclude that playbacks stimulate maned wolves to emit roar-barks and that this
method is applicable to test hypotheses about maned wolf behavior and aid in their
monitoring.
30
Resumo
Lobos guará são difíceis de serem observados na natureza devido às suas baixas
densidades e hábitos crípticos e crepusculares-noturnos. Explorar sua comunicação
acústica de longo alcance pode oferecer uma alternativa eficiente para estudar a espécie.
Neste trabalho nós avaliamos a aplicabilidade de usar playbacks para estudar lobos
guará na natureza e comparamos estes resultados com 20 noites de gravações passivas
na mesma área e mês durante o ano anterior. Obtivemos respostas vocais em 3 das 6
noites testadas, incluindo respostas envolvendo dois animais e uma aproximação depois
de um playback interativo. Apesar de termos conduzido 3-6 sessões por dia em
diferentes horários, nós só obtivemos respostas vocais em sessões entre 17:00 e 19:40.
Durante as gravações passivas nós detectamos em média 0.86 sequências de aulidos por
gravador por noite, a maioria na primeira metade da noite. A atividade vocal – respostas
e sequências de aulidos espontâneas – durante as noites de playback foi quase 4 vezes
maior que durante as gravações passivas. Nós concluímos que playbacks estimulam os
lobos guará a emitir aulidos e que o emprego deste método e viável para testar hipóteses
sobre o comportamento do lobo guará e auxiliar seu monitoramento.
Key words: Chrysocyon brachyurus, maned wolf, passive acoustic monitoring,
playback, roar-bark, vocal activity, vocalization time
31
Introduction
The maned wolf (Chrysocyon brachyurus; Illiger, 1815) is South America's
largest canid (80-90 cm shoulder height and 20-30 kg in weight; Rodden et al. 2004;
Jácomo et al. 2009). These animals forage alone for fruits and small vertebrates
(Queirolo and Motta-Junior 2007). Maned wolves form stable breeding pairs that share
an extensive home range (15-115 km2; Rodrigues 2002; Azevedo 2008). Unlike other
large canids, the pair is thought to rarely interact outside the breeding season (Dietz
1984; Rodden et al. 2004; Jácomo et al. 2009). From estrus to the weaning of young, the
pair may sleep together (Melo et al. 2007), encounter frequently, and travel or forage
together for several hours (Rodden et al. 2004; Emmons 2012).
Maned wolves communicate acoustically throughout the year using a long-
distance vocalization called the roar-bark, normally uttered in sequences (bouts) of 5-15
repetitions (see spectrograms in Materials and Methods; Rodden et al. 2004; Rocha et
al. 2016). The proposed functions of roar-barks are intra-pair and parent-offspring
communication, opposite-sex attraction and same-sex repelling, and/or territorial
announcement (Brady 1981; Dietz 1984; Sábato 2011; Emmons 2012; Rocha et al.
2016).
Few studies have been conducted observing maned wolves in the wild because
of their low densities and of their cryptic and crepuscular-nocturnal habits (Jácomo et al.
2004; Melo et al. 2007; Trolle et al. 2007). As the species is difficult to visualize and
follow, exploring their long-range acoustic communication may be an efficient
alternative to fill the many gaps in the knowledge of their behavior, and to monitor their
distribution and population trends (Marion et al. 1981; Rodden et al. 2004; Blumstein et
al. 2011).
32
Grids of autonomous audio recorders mounted for extended periods have been
used to investigate many aquatic species that cannot be easily observed, as most
cetaceans (e.g. humpback whales - Megaptera novaeangliae: Sousa-Lima and Clark
2008; beaked whales - Mesoplodon densirostris: Marques et al. 2009; minke whales -
Balaenoptera acutorostrata: Risch et al. 2013). For terrestrial environments, birds
dwelling in dense forests (Mennill and Vehrencamp 2008), nocturnal remote-nesting
birds (Oppel et al. 2014), and widely dispersed forest mammals (e.g. elephants -
Loxodonta africanacyclotis: Thompson et al. 2010), have also been studied with passive
acoustic monitoring. This methodology has already proven useful in the investigation of
wild maned wolves. The recording of 32 nights (8/month) suggests they vocalize more
often in the mating season, in the beginning of their activity period (first hour of the
night), and less often during new moon nights (Rocha et al. 2016).
Playbacks, different from passive acoustic monitoring, offers a more direct way
of testing hypothesis about the species behavior, and can possibly reduce monitoring
effort by stimulating immediate and higher vocal activity and inducing approaches to
facilitate detection or captures. Playbacks have long been used to study the behavior and
monitor populations of primates (Radick 2005) and birds (e.g. Lanyon 1963; Marion et
al. 1981). Today its use is widespread across taxa including insects, anurans (Greenfield
1994), and carnivores, such as lions and hyenas (Panthera leo, Crocuta crocuta; Cozzi
et al. 2013). This technique has been successfully employed to study other canids in the
wild, including wolves (Canis lupus; Harrington 1987; Brennan et al. 2013), coyotes
(Canis latrans; Petroelje et al. 2013), swift foxes (Vulpes velox; Darden and Dabelsteen
2008), and bush dogs (Speothos venaticus; DeMatteo et al. 2004), but remained untested
for maned wolves.
33
Here we evaluated the applicability of playbacks to study maned wolves in the
wild through the broadcast of roar-bark sequences to elicit responses from free-ranging
animals. We tested different broadcasting hours aiming to guide the planning of future
playback efforts, and to identify the best period of the day for conducting an interactive
playback aiming to escalate the animals’ response and induce approaches to the speaker.
If maned wolves are responsive to playbacks, future investigations could test how the
species use roar-barks, e.g. if sexes respond equally to male and female broadcasted
roar-barks, or survey maned wolves’ presence and distribution in a practical way. If
maned wolves are also shown to be attracted to the playback, this can be used to permit
their detection, even if they do not respond vocally, count individuals, and improve
capture chances. Additionally, we tested the use of autonomous recorders to increase the
probabilities of registering responses and identifying local roar-bark range (and other
propagation effects whose results are not shown here). Determining the roar-bark range
is important to make playback sites independent and to estimate the surveyed area,
while estimating our roar-bark detection capabilities will serve to plan how effectively
we can survey an area.
Playback results were compared with nightly detections of roar-barks from
passive acoustic recordings in the same area on the same month of the previous year.
Although there are no residency studies for the species, it is likely that the same
individuals were residing the area. Long term research suggests that individuals occupy
the same territories from 3 to 5 years (ranging from 1 to over 9 years; Emmons 2012).
The passive acoustic pattern of detection were used to establish a baseline level of roar-
bark spontaneous emissions along the night to test if playbacks would increase vocal
activity or change the temporal emission pattern.
34
Materials and methods
1. Study area
We conducted this study at Serra da Canastra National Park, Minas Gerais state,
Brazil (Figure 1). The park is mainly composed of Cerrado open savannas with a cold,
dry season (April-September) and a hot, rainy season (October-March; Queirolo and
Motta-Junior 2007).
Wild maned wolf acoustic data collection at this park was authorized by Instituto
Chico Mendes de Conservação da Biodiversidade (ICMBio; SISBIO license number
41329-2) and playback experiments were done in accordance with the ASM guidelines
(Sikes et al. 2016).
Figure 1. Location of passive autonomous recorders and playback sites to study maned wolves at Serra
da Canastra National Park, Minas Gerais, Brazil. Imagery ©2018 CNES / Airbus, Map data ©2018
Google.
35
2. Playbacks
Roar-barks used as stimuli were obtained from two facilities in Minas Gerais
state that keeps captive maned wolves: Criadouro Científico de Fauna Silvestre para
Fins de Conservação da Companhia Brasileira de Metalurgia e Mineração, and
Zoológico da Associação Esportiva e Recreativa dos Funcionários das Usinas
Siderúrgicas de Minas Gerais. Recordings were conducted between April and June
2010 and November 2010, respectively. The sounds were recorded 5-8 m from the
animals with a unidirectional microphone Sennheiser K6 coupled to a Sennheiser ME-
66 module and connected with a solid-state recorder Marantz PMD-661, using a 96 kHz
sample rate, and a 24-bit wav coding form.
To set the intensity level we used the playback equipment (described below) to
broadcast the captivity recordings and then re-record the played back sounds with the
same equipment and settings used for the original captivity recordings at the same
distance the focal animals were. We then changed the speaker volume until the roar-
bark intensity measured in the re-recordings matched the intensity measured in the
original captivity recordings (measures made in Raven Pro 1.5 software: Bioacoustics
Research Program, 2014. Raven Pro: Interactive Sound Analysis Software. Ithaca, NY:
The Cornell Lab of Ornithology). This was our solution to achieve a playback intensity
level as similar as maned wolves' roar-bark emission. During the recordings in captivity
we did not have a direct way to measure sound pressure levels, and thus no means of
calculating absolute source intensity levels. Two experienced researchers, including the
author of the original captivity recordings (VS), reported the playback sounded as
strong as heard from the animals in captivity and in the wild (FHR).
Playbacks sessions were conducted in three different sites in the park (Figure 1)
between March 4 and 9, 2017. We used an Acer AspireOne notebook to broadcast the
36
sounds using Raven Pro 1.5 software and a Pioneer S-DJ50X speaker (class A/B Bi-
amp, 80 W output, 50-20000 Hz frequency range) 86 cm above the ground to simulate
the height of a maned wolf.
We used 4 edited roar-bark sequences including sounds from two males and two
females (Figure 2). Recordings of both sexes were used to maximize the chance of
response. Each sequence was composed of 5 roar-barks separated by 2.9-6.2 seconds
(similar to the natural emission for an individual) and intervals between sequences
varied from 10 seconds to 10 minutes depending on weather conditions. In each
playback session, all four individual sequences were played in random order and then
repeated once, resulting in complete sessions being composed of 8 sequences of 5 roar-
barks each and lasting 5-25 minutes in total. If we heard an answer from a wild maned
wolf before finishing the broadcast of all sequences we broadcasted the next sequence
right after the response aiming to create an interactive playback. This was done trying to
stimulate the responsive animal’s approach.
37
Figure 2. Edited maned wolf roar-bark sequences used as stimuli for playback studies of maned wolves
in the wild (Serra da Canastra National Park, Brazil). GA and SH are males, SA and JU are females. Top
spectrograms are the original files (96 kHz sample rate, 32 bit wav, 4000 windows size, 56% brightness
and 50% contrast) and the bottom a recording extracted from one autonomous recorder (Song Meter
SM2+; Wildlife Acoustics) 80 meters from the playback speaker (8 kHz sample rate, 16 bit wav, 512
windows size, 50% brightness and contrast). Spectrogram made on Raven Pro 1.5.
We conducted playbacks on two nights at each site, except on site A where we
could only do it once (March 04 to 05) due to logistical issues. Sessions were conducted
three times each night at the following moments: After Sunset (between 15 and 75
38
minutes after sunset), Midnight (between 23:00 and 00:00), and Before Sunrise
(between 15 and 75 minutes before sunrise). Due to weather conditions, on the first
night (March 07 to 08) at site C we could not do the After Sunset session, and
compensated for it by conducting a session on a third consecutive night (March 09). At
site C we conducted additional diurnal sessions on two days (March 08 and 09) three
times each day to tests the animals’ responsiveness during the light period: After
Sunrise (between 15 and 75 minutes after sunrise), Midday (between 11:00 and 12:00),
and Before Sunset (between 15 and 75 minutes before sunset). Mean local sunset was
18:28 and sunrise 06:05 during the playback days (calculated on
https://www.sunearthtools.com/pt/solar/sunrise-sunset-calendar.php; access Set/16
2018). This experimental design resulted in a total of 21 playback sessions over 6 days.
During the conducted playback sessions wind speed was 0.37 ± 0.61 m/s (mean ± SD,
maximum 2.4 m/s), which based on our previous work should not significantly impact
roar-bark detection and propagation (Rocha et al. 2016).
Roar-bark sequences of free-ranging maned wolves detected up to 10 minutes
after the end of any broadcasted sequence were considered responses to the playback.
Although it is impossible to differentiate a vocal response from a spontaneous
vocalization, we consider 10 minutes a plausible interval to assume the vocalization is
an playback-elicited answer as maned wolves vocalize 0.41 roar-bark sequences per
night per recorder in the wild (Rocha et al. 2016) and 0.68 sequences per individual per
night in captivity (Sábato 2011). The chance of a maned wolf spontaneously emitting a
roar-bark sequence within any given 10 minutes each night is less than 1% (0.57%
based on Rocha et al. 2016; 0.94% based on Sábato 2011).
For each response we recorded the time of emission and, based on the
vocalization source direction determined by ear, if there were more than one animal
39
vocalizing. Additionally, we waited at least 30 minutes after the end of each session
near the playback site to verify possible approaches by responsive animals.
3. Autonomous recordings during the playback experiment
We successively mounted at each playback site a line of 8 autonomous recorders
(Song Meter SM2+; Wildlife Acoustics, Inc., Concord, Massachusetts) with one
omnidirectional weatherproof microphone each (SMX-II; Wildlife Acoustics, Inc.;
sensitivity -36±4dB [0dB=1V/pa@1kHz]; 20Hz-20kHz flat response frequency).
Recorders were set on the road side in a single direction from the playback speaker
positioned at distances of 1.25 m, 20 m, 40 m, 80 m, 160 m, 320 m, 640 m, and 1280 m.
Distances were measured using a tape measure (1.25 to 80 m) and a GPS (Garmin
GPSMAP® 76S; accuracy < 15 m). The autonomous recorders were attached on stakes
of the same height of the speaker (86 cm) with the omnidirectional microphone in a
perpendicular position in relation to the speaker. Recordings were made continuously,
with an 8 kHz sample rate, 16-bit wav files, and partitioned in 30 minutes files.
At site A the autonomous recorders were active from 16:34 March 04 to 06:04
March 05 (total 13 h 30 min), at site B from 16:03 March 05 to 08:33 March 07 (total
40 h 30 min), and at site C from 19:03 March 07 to 06:33 March 11 (total 83 h 30 min).
After the end of the experiment we left the recorders active for an extra day and night
(March 10 to 11), resulting in the larger amount of recording hours at site C. We used
this extra day and night to evaluate if maned wolves would continue to vocalize
spontaneously without the playback stimuli.
We automatically detected roar-barks on the files using the methodology
detailed in Rocha et al. (2015). This methodology uses XBAT_R7 (Extensible
Bioacoustics Tool; Figueroa 2007) extension for MATLAB (R2010a version;
MathWorks, Inc., Natick, MA, USA) to generates spectrograms of the files that are
40
subjected to a cross correlation tool employing 4 roar-bark templates. Matches above a
pre-defined threshold (0.21) are then manually verified for false positives and
undetected roar-barks within 24 seconds of the detected ones.
We detected both free-ranging maned wolves roar-barks and the roar barks we
broadcasted during the playback. For the broadcasted roar-barks we noted all recorders
that registered the calls (e.g. 1.25 m, 20 m, 40 m, etc.). We used this information to
estimate roar-bark range. To do so we used only the playbacks conducted at night, as
maned wolves rarely vocalize during the daylight period (Brady 1981; Emmons 2012).
For free-ranging maned wolves roar-bark sequences we noted the time of
emission, time from the end of the last broadcasted roar-bark sequence (equivalent to
the latency in the cases considered responses), the number of roar-barks in the sequence,
the distance of the autonomous recorder that the roar-bark recording was most intense
(measured by the peak power function of Raven pro 1.5), and if the vocalizations were
heard by us in the cases we were present at the site near the speaker. The maximum
interval between roar-barks that we considered a single sequence was 10 seconds,
longer intervals were considered the beginning of a different sequence (based on Rocha
et al. 2016 dataset and Bender et al. 1996).
4. Comparative passive acoustic monitoring
In 2016 we deployed 13 autonomous recorders (Song Meter SM2+, Wildlife
Acoustics; 8 of which were the same ones used during the 2017 playback experiment) at
Serra da Canastra National Park on the same region that we conducted the playback
experiments (Figure 1). Mean distance between recorders was 3.03 km (minimum 2.16
km, maximum 3.90 km). Recorders were attached onto 1.4 m wooden stakes to
maximize detections of roar-barks and were set with the same recording configurations
as the playback experiment but programed to record from 5 PM to 5 AM each day.
41
Considering our previous study in the area (Rocha et al. 2016) we expected this period
to comprise most of the maned wolf vocal activity. Recordings were made during 20
consecutive nights between March 09 and 28, 2016.
Roar-bark sequences were automatically detected on the audio files the same
way we did with the playback recordings. For each roar-bark sequence found we noted
time of emission, the number of roar-barks in the sequence, recorder location, and if the
sequence was recorded by more than one recorder. That last information was verified
comparing the sequences´ time of emission and the inter roar-barks intervals, which is
unique for each sequence and ensures it is the same sequence recorded in two (or more)
recorders and not two independent sequences of different animals vocalizing at the same
moment. We only counted the most intense record of a sequence that has been detected
in multiple sensors. Finally, we measured the time interval between sequences emitted
in the same recorder during the same night.
In some cases we could identify a second animal started a roar-bark sequence
before the end of another animal sequence, resulting in intercalated roar-barks. We
could recognize the presence of a second animal based on differences in roar-bark
cadence, intensity, spectral shape, and occasional overlap of roar-barks. If roar-bark
sequences were separated (no intercalated roar-barks) and there was no striking spectral
shape difference between them we could not tell if the sequences were emitted by the
same or different animals. In cases of two animals vocalizing at the same moment we
considered the interval between the sequences as 0 seconds.
42
Results
1. Playback experiment and recordings
The results from the playback experiment are summarized in Table 1 and
Figure 3. All roar-bark sequences heard in loco by the researchers were recorded by at
least one autonomous recorder. We were not able to visualize the animals.
We obtained vocal responses in only 4 of 21 playback sessions (see
Supplementary data SD1 and SD 2 for examples of such responses). Responses
occurred in 3 of 6 nights in which there was at least one playback session. On 2 nights
the “After Sunset” playback session was answered and on one night both the “Before
Sunset” and “After Sunset” sessions elicited vocal responses. Responses consisted of 1-
3 roar-bark sequences on average 02:12 ± 01:56 minutes (X ± SD; N = 8) after the end
of the broadcasted sequence.
Figure 3. Distribution of wild maned wolf roar-bark sequences registered between March 04 and 11 2017
during a playback experiment at Serra da Canastra National Park, MG/Brazil. Each sequence is named by
its start time and the size of the bar shows the time elapsed from the last broadcasted playback sequence.
This time is also discriminated on tags above the sequences considered responses to the playback, i.e.
those within 10 minutes after the end of any broadcasted sequence.
43
Ro
ar-B
ark
Seq
uen
ce
Sit
e
Da
te
Sta
rt t
ime
En
d t
ime
Nu
mb
er o
f
roa
r-b
ark
s
Tim
e fr
om
la
st
pla
yb
ack
seq
uen
ce
Pla
yb
ack
ses
sio
n
Pla
yb
ack
seq
. in
the
sess
ion
H
eard
?
Mo
st i
nte
nse
reco
rd
1.
B
mar
/05
20
:37
:54
20
:38
:55
15
00
:56
:10
- -
no
t p
rese
nt
12
80
m
2.
B
mar
/05
20
:39
:05
20
:39
:06
1
00
:57
:21
- -
no
t p
rese
nt
12
80
m
3.
B
mar
/06
18
:45
:56
18
:46
:47
14
00
:00
:26
Aft
er S
unse
t 2
yes
6
40
m
4.
B
mar
/06
18
:53
:48
18
:54
:50
16
00
:02
:29
Aft
er S
unse
t 8
no
1
280
m
5.
B
mar
/07
00
:04
:07
00
:05
:25
21
00
:54
:26
- -
no
t p
rese
nt
12
80
m
6.
B
mar
/07
00
:38
:07
00
:38
:52
11
01
:28
:26
- -
no
t p
rese
nt
64
0 m
7.
B
mar
/07
00
:49
:50
00
:49
:51
1
01
:40
:09
- -
no
t p
rese
nt
64
0 m
8.
B
mar
/07
00
:50
:01
00
:50
:51
12
01
:40
:20
- -
no
t p
rese
nt
64
0 m
9.
C
mar
/08
17
:21
:06
17
:21
:43
11
00
:00
:00
Bef
ore
Su
nse
t 3
no
1
280
m
10
. C
m
ar/0
8
18
:26
:47
18
:27
:50
13
01
:02
:57
- -
no
1
280
m
11
. C
m
ar/0
8
18
:28
:04
18
:28
:05
1
01
:04
:14
- -
no
1
280
m
12
. C
m
ar/0
8
18
:36
:28
18
:37
:30
16
01
:12
:38
- -
no
1
280
m
13
. C
m
ar/0
8
18
:44
:06
18
:45
:00
13
01
:20
:16
- -
no
1
280
m
14
. C
m
ar/0
8
18
:55
:46
18
:56
:57
15
00
:03
:36
Aft
er S
unse
t 8
yes
3
20
m
15
. C
m
ar/0
8
18
:57
:07
18
:57
:40
7
00
:04
:57
Aft
er S
unse
t 8
yes
4
0 m
16
. C
m
ar/0
8
19
:02
:19
19
:03
:12
13
00
:10
:09
- -
yes
4
0 m
17
. C
m
ar/0
9
18
:50
:20
18
:50
:55
10
00
:04
:12
Aft
er S
unse
t 1
yes
4
0 m
18
. C
m
ar/0
9
18
:55
:37
18
:57
:06
23
00
:00
:08
Aft
er S
unse
t 3
yes
6
40
m
19
. C
m
ar/0
9
18
:59
:43
19
:00
:12
8
00
:01
:52
Aft
er S
unse
t 4
yes
4
0 m
Tab
le 1
. W
ild m
aned
wolv
es’
voca
l ac
tivit
y r
ecord
ed b
y a
uto
nom
ous
reco
rder
s (S
ong M
eter
SM
2+
; W
ildli
fe A
coust
ics)
at
Ser
ra d
a C
anas
tra
Nat
ional
Par
k,
MG
/Bra
zil,
duri
ng t
he
pla
ybac
k e
xper
imen
t day
s on 2
017.
Eac
h r
ow
is
a dif
fere
nt
roar
-bar
k s
equen
ce.
Those
consi
der
ed p
laybac
k r
esponse
s ar
e
under
lin
ed.
Eac
h p
laybac
k s
essi
on c
onsi
sted
of
8 b
road
cast
ed r
oar
-bar
k s
equen
ces.
The
colu
mn “
Hea
rd?”
indic
ates
if
the
rese
arch
ers,
when
pre
sent
in
loco
, hea
rd m
aned
wolv
es’
call
s.
44
18
. C
m
ar/0
9
18
:55
:37
18
:57
:06
23
00
:00
:08
Aft
er S
unse
t 3
yes
6
40
m
19
. C
m
ar/0
9
18
:59
:43
19
:00
:12
8
00
:01
:52
Aft
er S
unse
t 4
yes
4
0 m
20
. C
m
ar/0
9
19
:40
:39
19
:41
:23
14
00
:33
:34
- -
yes
4
0 m
21
. C
m
ar/0
9
19
:41
:44
19
:42
:05
5
00
:34
:39
- -
yes
4
0 m
22
. C
m
ar/0
9
19
:45
:48
19
:46
:41
15
00
:38
:43
- -
yes
4
0 m
23
. C
m
ar/0
9
19
:55
:08
19
:56
:21
20
00
:48
:03
- -
yes
6
40
m
24
. C
m
ar/0
9
20
:03
:40
20
:04
:25
12
00
:56
:35
- -
yes
3
20
m
25
. C
m
ar/0
9
20
:04
:36
20
:05
:03
7
00
:57
:31
- -
yes
4
0 m
26
. C
m
ar/0
9
20
:07
:39
20
:09
:12
22
01
:00
:34
- -
yes
4
0 m
27
. C
m
ar/0
9
21
:05
:08
21
:07
:22
25
01
:58
:03
- -
no
4
0 m
28
. C
m
ar/0
9
21
:28
:15
21
:29
:45
21
02
:21
:10
- -
no
t p
rese
nt
40
m
29
. C
m
ar/0
9
21
:29
:59
21
:30
:20
5
02
:22
:54
- -
no
t p
rese
nt
40
m
30
. C
m
ar/0
9
21
:32
:46
21
:33
:21
7
02
:25
:41
- -
no
t p
rese
nt
1.2
5 m
45
In 4 of 6 nights there was additional vocal activity initiating around 30 minutes
to 1 hour after the playback and consisted of 2-7 shortly spaced sequences (Figure 3).
We did not always register a response to the preceding playback session, for instance,
the March 06-07 Midnight session was not answered (to our best detection capabilities)
despite the vocal activity one hour later. We did not register any roar-bark sequence on
the extra day and night recorded at site C following the end of the playback experiment.
Based on differences in the direction of the vocalization source detected aurally
we could identify two occasions in which the vocal response involved two different
wild maned wolves. On March 08, 10 seconds after one animal responded to the “After
Sunset” playback session (sequence no. 14, Table 1), a second animal emitted a
sequence (no. 15, Table 1) and then another sequence 04:39 minutes thereafter (no. 16,
Table 1). Although very close to the pre-established 10 minutes window (10:09 minutes
after the playback), this last roar-bark sequence was not considered a playback response
to our criterion. It was, however, emitted 05:22 minutes after the first animal response
and could have been an answer to that animal.
The second occasion occurred following the interactive playback we were able
to produce during our last playback session (March 09 “After Sunset”). There were
three callback response sequences, after the first, second and fourth sequence we
broadcasted (no. 17-19, Table 1). This playback timing scheme elicited the most intense
behavioral response: 40 minutes later we heard two animals exchanging roar-bark
sequences and moving fast toward the playback site. They reached the location and then
continued moving past it, still exchanging roar-bark sequences (no. 20-26, Table 1).
After a one-hour interval there was another roar-bark sequence nearby (no. 27, Table 1;
not heard, we left the site soon thereafter) and 20 minutes another three (no. 28-30,
Table 1; not present). The last sequence was most intense at the autonomous recorder
46
1.25 m from the position where the speaker was during the playback. The time pattern
sequence of this night can be seen in black on Figure 3.
For comparison purposes, we considered the area covered by the 8 autonomous
recorders (1280 m in line) during the playback experiment as equivalent of the area
covered by 1.5 autonomous recorders during the 2016 passive recordings, as the average
distance between then was 3 km. That results in 3.33 roar-bark sequences per
‘autonomous recorder’ (1.5) per night (6).
All broadcasted roar-barks were detectable in the recordings up to 80 m, 96.2%
were detectable in the 160 m autonomous recorder, 47.7% in the 320 m recorder, 26.7%
in the 640 m recorder, and 0.8% in the 1280 m recorder. From the 21 recorded
sequences emitted by free-ranging animals while we were present at the site, 7 we had
not heard. Six of those sequences had the most intense register in the recorder at 1280
m.
2. Comparison with passive acoustic monitoring detections
We registered 2610 roar-barks distributed in 224 sequences (11.65 ± 7.09 roar-
barks per sequence, X ± SD; minimum 1, maximum 50) during the 20-night recording
period in 2016. Each night had an average of 11.20 ± 8.45 sequences (X ± SD;
minimum 0, maximum 25). The mean number of autonomous recorders that registered
roar-bark sequences each night was 3.65 ± 2.01 (X ± SD; minimum 0, maximum 9),
each of those with an average of 3.07 ± 2.64 sequences (X ± SD; minimum 1, maximum
13). Overall, we obtained an average of 0.86 roar-bark sequences (total 224) per
autonomous recorder (13) per night (20). Supplementary data SD1 and SD3 contains
an example of sequence recorded passively.
We measured a total of 173 intervals between sequences, including 22 times
when there was overlap of sequences, i.e., when the second animal vocalized before the
47
first animal ended its sequence (considered as 0 second interval). Sequences on a single
recorder and night were in general separated by short intervals (6 min 40 s, median; 51 s
– 21 min 47 s, 1st – 3rd quartiles). A recurrent emission pattern consisted of 1-3
sequences separated by short intervals followed by a longer interval (20-80 min) and
another set of 1-3 sequences separated by short intervals.
The diel roar-bark emission temporal pattern found is shown in Figure 4. A
small percentage (8.04%) of the sequences occurred between 17:00 (recordings start)
and sunset (sunset time varied from 18:09 to 18:26), 34.82% in the first 3 hours of the
night, 68.75% on the first half of the night, and 23.21% on the second half of the night.
Except for a peak between 22:00-23:00 hours, the pattern was very similar to the April
2014 temporal pattern described by Rocha et al. (2016; recordings started at 18:00 and
sunset varied from 17:44 to 18:03 in April 2014).
Figure 4. Temporal distribution of maned wolf roar-bark sequences recorded at Serra da Canastra
National Park, MG/Brazil, with autonomous recorders (Song Meter SM2+; Wildlife Acoustic). March
2017 (black line, left axis): percentage relative to the total (30 sequences) of vocal activity registered on
continuous recordings of the 6 days in which the roar-bark playback experiment was conducted. March
2016 (dark gray bars, right axis): absolute number of sequences (total 224), 13 recorders, 20 nights, from
5 PM to 5 AM. April 2014 (light gray bars, right axis): absolute number of sequences (total 192), 12
recorders, 25 nights, from 6 PM to 6 AM (Rocha et al. 2016 dataset, used with permission).
48
Some of the sequences (54 of 224) were detected in more than one autonomous
recorder. All of those cases (except 1) happened on 5 combinations of two autonomous
recorders that were on average 2.40 ± 0.15 km (X ± SD) apart. On average 45.0 ± 18.6
% (X ± SD) of the times one of those recorders registered a sequence it was also
detected on another recorder. The maximum distance between autonomous recorders
that registered a given sequence was 4.85 km (distance from the first and third recorder
that registered that particular sequence). Considering the case of an animal exactly in
the midpoint between those recorders, that means the calls from this sequence
propagated at least 2.4 km before reaching the recorders.
49
Discussion
Here we tested the playback of roar-barks in a protected area where wild maned
wolves inhabit. They responded vocally, proving this technique is applicable to monitor
and study the species. We were able to describe their response and compare it to a 20-
night passive acoustic monitoring effort on the same area and period of the year. We
also identified the best broadcasting time and feasibility of conducting interactive
playbacks. Although we could not define with precision the roar-bark range, we were
able to acquire some information on the response distance and our hearing capabilities.
The vocal response rate obtained (4/21 sessions) may seem low, but maned
wolves are not highly vocally active (Rocha et al. 2016; see also Results), occur in very
low densities (Trolle et al. 2007), and we did not know a priori if there were any
animals on the region at that time. Considering the aforementioned factors, we evaluate
that responses in 3 out of the 6 nights tested was a high response rate.
The average of 0.86 sequences per recorder per night obtained in our
comparative passive recordings of March 2016 was greater than the previous values
obtained in the 2014 passive recording study (April to July 0.41 and only April 0.52;
Rocha et al. 2016). Even so, the average vocal response recorded during our playback
experiment in 2017 (estimated 3.33 sequences per recorder per night, including
responses and non-responses) was near 4 times higher than the spontaneous vocalization
rate one obtained in March 2016.
This higher vocal activity and the complete lack of roar-bark sequences on the
night after the playback suggests most of the vocal activity recorded was at least
indirectly influenced by the playback. That may be true even for sequences emitted
several minutes later (30-60 min). The animals´ approach to the speaker’s site following
the interactive playback vocal responses, although evidently influenced by it, happened
50
40 minutes later with a second set of uttered sequences. A third set of roar-bark
sequences was registered 1 hour after the second set and the last sequence was most
intense on the 1.25 m recorder, suggesting the emitter(s) may still be investigating the
playback area after the stimuli had ceased. Sets of sequences separated by similar
intervals (20-80 min) on the same night were also observed during the passive acoustic
monitoring. Although we do not know what motivated the vocalizations recorded
passively, this suggests that when there is a second set of sequences within such time
intervals, they may have been elicited by the same stimulus (similar motivation) than
the first set, i.e., a redundant message, a reiteration.
We cannot rule out the possibility the roar-bark sequences we are calling
responses were, in fact, spontaneous vocalizations. However, we reinforce that maned
wolves emit on average less than one roar-bark sequence each night (Sábato 2011;
Rocha et al. 2016), making a spontaneous vocalization within 10 minutes of the
playback unlikely. Even considering the 0.86 sequences per recorder per night obtained
here and the increased detection probability between 18:00 and 19:00 (12.05% on the
passive monitoring), the probability of a spontaneous roar-bark sequence being emitted
during our 10-minute response time window would be 1.73%. Additionally, during the
playback experiment, there were instances when more than one sequence within the
same 10 minutes were detected and the maned wolves that emitted those sequences
intercalated them with our playback sequences, which would make the time coincidence
even more unlikely.
We registered two occasions where vocal responses involved two animals, but it
was not possible to confirm their sexes or territory ownership. As maned wolves form
stable breeding pairs that share the same area (Azevedo 2008), and it was the beginning
of the mating season (March to May on this park; Rodden et al. 2004), which mean the
51
pair could be maintaining proximity (Emmons 2012), we speculate it was a mated male
and female. However, it could have been two non-mated individuals from the same or
different territories, or even non-residents, independently responding to the playback.
During a pilot playback experiment at the same park we also obtained delayed
responses involving two animals (conducted in May 2016), indicating this behavior is
not uncommon. Further studies are needed to clarify which animals participate in such
responses.
Although the species increase its physical activity around both twilights (Jácomo
et al. 2004; Melo et al. 2007), responses and vocal activity were concentrated only
closer to dusk (1h before, 3h later). Our passive acoustic monitoring confirms that
animals naturally vocalize more on the first half of the night and thus could be more
prompt to respond also during that period. Despite our playback sessions near dawn
(05:00-06:00 and 06:00-07:00), there was no response or vocal activity on this period.
We cannot know if no animal heard the playback or if they just chose not to respond.
More studies are needed to verify if there is a biological reason for this lack of
responsiveness. Nevertheless, for studies not aiming at this question, it seems
reasonable to concentrate playback efforts on the 17:00 to 22:00 period.
While less than half of the broadcasted roar-barks were detectable at 320 m, a
large portion of naturally emitted roar-bark sequences were detectable in two or more
recorders 2.16 to 4.85 km apart during the passive monitoring (meaning the maned wolf
was at least halfway between those distances). We believe this discrepancy is a product
of the recording height. The 86 cm used during the playback was much lower than the
1.4 m used during the passive monitoring. We choose this lower height to simulate a
maned wolf position, but the range obtained this way does not seem to reflect neither
the maned wolf's nor our own roar-bark hearing capability (see below). Another non-
52
exclusive explanation for this discrepancy in detectability would be that the captive
individuals, whose recordings were used to set the playback intensity, roar-barked at a
lower intensity level than wild maned wolves normally do. This could have happened
because in captivity individuals are held in much closer proximity to conspecifics than
in natural conditions and thus, they would not need to vocalize as loud to reach others,
optimizing their vocal output.
We obtained responses that were most intense in the 1280 m recorder,
suggesting the animal was at least 960 m away (between the 640 m and 1280 m
recorders; sequences no. 4 and 9, Table 1). This was slightly greater than our own
hearing limit as no sequence with the most intense signal in the 1280 m recorder was
heard by us, while all most intense signals on the 640 m were. Some caution with those
distances is needed because terrain composition and wind currents and turbulence can
drastically affect the way the sound reach each recorder, especially in open fields
(Brown and Handford 2000). Besides, the autonomous recorders were not calibrated
together so some intensity discrepancies between the recordings is expected.
Nevertheless, those distances are in agreement with Brady (1981) study that
states a roar-bark could be heard 1 km away, although the author mentioned they could
be even discriminated individually at this distance. Considering the above, future
playbacks should be conducted at sites 2 km apart at minimum (1 km for each side).
Greater distances would be ideal to ensure site independency as we obtained evidence
for the minimum maned wolf roar-bark hearing range, but not the maximum. If it is as
good as a well-positioned autonomous recorder, then minimum site distances should be
4.85 km.
Playbacks could be used to survey maned wolf presence and distribution in a
relatively fast way, considering the increase in vocal activity and smaller data volume to
53
analyze compared to passive acoustic monitoring (Rocha et al. 2015). Playback
methodology has been used, for instance, to monitor the expansion of golden jackals’
range in Europe and count territorial groups (Acosta et al. 2018). Acoustic stimulation
can also be used to attract maned wolves to facilitate their capture, as suggested for
African wild dogs and bush dogs (Robbins and McCreery 2003; DeMatteo et al. 2004),
to detect their presence even without vocal responses (directly or indirectly, e.g. by
footprints: Brennan et al. 2013), or to visually identify individuals for counting, as done
with hyaenas and lions (Cozzi et al. 2013). To estimate absolute population numbers,
we need to conduct future research on the maximum distance, probabilities, and other
factors influencing the emission of maned wolves’ vocal responses and approaches, as
Cozzi et al. (2013) did for their target species. Another possibility is using playback-
elicited vocal responses to create a vocal identity catalog to then count and recognize/
“recapture” individuals on recordings, as for African wild dogs (Hartwing 2005). Roar-
barks have been demonstrated to be individually distinct (Sábato 2011), however, to
date, our efforts have been insufficient to identify individuals on recordings in their
natural habitat (see Chapter 4). Finally, different from the passive acoustic monitoring,
playbacks allow direct test of hypothesis (e.g. Marshall-Ball et al. 2006). This technique
could be used, for instance, to test if maned wolves indeed use roar-barks to defend
territories and/or mates, and what is the male and female role in territorial defense.
On the other hand, passive acoustic monitoring is a non-invasive approach that
result in a less biased index of population size. By counting acoustic cues (e.g. number
of roar-bark sequences) and comparing different years it is possible to monitor relative
population trends (e.g. seabirds: Oppel et al. 2014). Cue counting also depends on
auxiliary information about vocal behavior, sender identity and signal ranging that is not
yet available for maned wolves (Marques et al. 2013). To monitor absolute population
54
size, we would also need more information, such as individual emission rate,
estimatives of caller distance, and/or vocal individual identification. Nonetheless,
passive acoustic monitoring has the advantage of revealing natural multi-individual
activity and interaction patterns (Blumstein et al. 2011). Field time and costs are smaller
than other methodologies (Jorge et al. 2018), allowing for long term studies covering
larger areas.
The total investment cost, however, can be high for both approaches.
Autonomous recorders are expensive (for most underdeveloped countries) and can
represent over 90% of a passive monitoring study cost (Jorge et al. 2018). Maned
wolves occur in low densities on very extensive areas and thus playbacks would need to
be done over many distant sites to have reliable sample size, which can increase field
time and cost manifold. In the end the method of choice will depend on the question to
be answered or problem to be solved. Combining methodologies will be advantageous
as no single approach can reveal every aspect of a given system.
Is conclusion, for future playbacks to maned wolves we recommend: waiting or
recording at least one hour after the playback as vocal activity and approaches may be
delayed; using interactive playbacks for maximum response and chance of approach;
expect multi individual responses; concentrating efforts on the first portion of the night;
choosing distant sites (2-5 km minimum); and consider using autonomous recorders to
increase detection of responses. Our work shows that playbacks, and also passive
acoustic monitoring, have good potential to be used in conjunction during investigations
of behavior ecology even if the target species is not seen in the field and is not highly
vocal, as is the case for many other mammals.
55
References
ACOSTA-PANKOV, I., SPASSOV, N., & BANEA, O.C. 2018. Seasonal differences in
howling response of golden jackals Canis aureus L., 1758 (Mammalia: Canidae) in
eastern Bulgaria. Acta Zoologica Bulgarica, 70(4), 539-545.
AZEVEDO, F.C. 2008. Home range and spatial organization of maned wolf
(Chrysocyon brachyurus) in the Serra da Canastra National Park region, Minas Gerais,
Brazil. M.S. thesis, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil.
BLUMSTEIN, D.T., MENNILL, D.J., CLEMINS, P., GIROD, L., YAO, K.,
PATRICELLI, G., ... & HANSER, S.F. 2011. Acoustic monitoring in terrestrial
environments using microphone arrays: applications, technological considerations and
prospectus. Journal of Applied Ecology, 48(3), 758-767.
BRADY, C.A. 1981. The vocal repertoires of the bush dog (Speothos venaticus), crab-
eating fox (Cerdocyon thous), and maned wolf (Chrysocyon brachyurus). Animal
Behaviour, 29(3), 649-669.
BRENNAN, A., CROSS, P.C., AUSBAND, D.E., BARBKNECHT, A., & CREEL, S.
2013. Testing automated howling devices in a wintertime wolf survey. Wildlife Society
Bulletin, 37(2), 389-393.
BROWN, T.J., & HANDFORD, P. 2000. Sound design for vocalizations: quality in the
woods, consistency in the fields. The Condor, 102(1), 81-92.
COZZI, G., BROEKHUIS, F., MCNUTT, J.W., & SCHMID, B. 2013. Density and
habitat use of lions and spotted hyenas in northern Botswana and the influence of
survey and ecological variables on call-in survey estimation. Biodiversity and
Conservation 22:2937-2956.
DARDEN, S.K. & DABELSTEEN T. 2008. Acoustic territorial signaling in a small,
socially monogamous canid. Animal Behavior 75:905-912.
DeMATTEO, K.E., CARRILLO, O., ZUERCHER, G.L., RAMÍREZ, S., SMITH, K., &
PORTON., I.J. 2004. A technique for attracting bush dogs (Speothos venaticus) in the
wild. Canid News 7.
DIETZ, J.M. 1984. Ecology and social organization of the maned wolf (Chrysocyon
brachyurus). Smithsonian Contribution to Zoology. 1-51.
EMMONS, L.H. 2012. The maned wolves of Noel Kempff Mercado National Park.
Smithsonian Contributions to Zoology 639:1–135.
FIGUEROA, H. 2007. Extensible BioAcoustic Tool. XBAT R7. Cornell Bioacoustics
Research Program, Ithaca, New York.
GIANNATOS, G., MARINOS, Y., MARAGOU, P., & CATSADORAKIS, G. 2005.
The status of the golden jackal (Canis aureus L.) in Greece. Belgian Journal of Zoology
135:145–149.
56
GREENFIELD, M.D. 1994. Synchronous and alternating choruses in insects and
anurans: common mechanisms and diverse functions. American Zoologist 34:605-615.
HARRINGTON, F.H. 1987. Aggressive howling in wolves. Animal Behaviour 35:7-12.
HARTWIG, S. 2005. Individual acoustic identification as a non-invasive conservation
tool: an approach to the conservation of the African wild dog an approach to
the conservation of the African wild dog Lycaon pictus (Temminck, 1820).
Bioacoustics, 15, 35-50.
JÁCOMO, A.T., SILVEIRA, L., & DINIZ-FILHO, J.A.F. 2004. Niche separation
between the maned wolf (Chrysocyon brachyurus), the crab-eating fox (Dusicyon
thous) and the hoary fox (Dusicyon vetulus) in central Brazil. Journal of
Zoology, 262(1), 99-106.
JÁCOMO, A.T., KASHIVAKURA, C.K., FERRO, C., FURTADO, M.M., ASTETE,
S.P., TÔRRES, N.M., ... & SILVEIRA, L. 2009. Home range and spatial organization
of maned wolves in the Brazilian grasslands. Journal of Mammalogy, 90(1), 150-157.
JORGE, F.C., MACHADO, C.G., DA CUNHA NOGUEIRA, S.S., & NOGUEIRA-
FILHO, S.L.G. 2018. The effectiveness of acoustic indices for forest monitoring in
Atlantic rainforest fragments. Ecological Indicators, 91:71-76.
LANYON, W.E. 1963. Experiments on species discrimination in Myiarchus
flycatchers. American Museum Novitates 2126.
MARQUES, T.A., THOMAS, L., MARTIN, S.W., MELLINGER, D.K., WARD, J.A.,
MORETTI, D.J., ... & TYACK, P.L. 2013. Estimating animal population density using
passive acoustics. Biological Reviews, 88(2), 287-309.
MARION, W.R., O’MEARA, T.E., & MAEHR, D.S. 1981. Use of playback recordings
in sampling elusive or secretive birds. Studies in Avian Biology 6:81-85.
MARSHALL-BALL, L., MANN, N., & SLATER, P.J.B. 2006. Multiple functions to
duet singing: hidden conflicts and apparent cooperation. Animal Behavior 71:823–831.
MELO, L.F., LIMA SÁBATO, M.A., VAZ MAGNI, E.M., YOUNG, R.J., &
COELHO, C.M. 2007. Secret lives of maned wolves (Chrysocyon brachyurus Illiger
1815): as revealed by GPS tracking collars. Journal of Zoology, 271(1), 27–36.
MENNILL, D.J., & VEHRENCAMP, S.L. 2008. Context-dependent functions of avian
duets revealed by microphone-array recordings and multispeaker playback. Current
Biology 18:1314-1319.
OPPEL, S., HERVIAS, S., OLIVEIRA, N., PIPA, T., SILVA, C., GERALDES, P., ... &
MCKOWN, M. 2014. Estimating population size of a nocturnal burrow-nesting seabird
using acoustic monitoring and habitat mapping. Nature Conservation, 7, 1.
PETROELJE, T.R., BELANT, J.L., & BEYER JR., D.E. 2013. Factors affecting the
elicitation of vocal responses from coyotes Canis latrans. Wildlife Biology 19:41–47.
57
QUEIROLO, D., AND J. C. MOTTA-JUNIOR. 2007. Prey availability and diet of
maned wolf in Serra da Canastra National Park, southeastern Brazil. Acta Theriologica
52:391–402.
RADICK, G. 2005. Primate language and the playback experiment, in 1890 and
1980. Journal of the History of Biology, 38:461-493.
RISCH, D., CLARK, C.W., DUGAN, P.J., POPESCU, M., SIEBERT, U., & VAN
PARIJS, S.M. 2013. Minke whale acoustic behavior and multi-year seasonal and diel
vocalization patterns in Massachusetts Bay, USA. Marine Ecology Progress
Series 489:279-295.
ROBBINS, R.L., & MCCREERY, E.K. 2003. Acoustic stimulation as a tool in African
wild dog conservation. Biological conservation, 111:263-267.
ROCHA, L.H., FERREIRA, L.S., PAULA, B.C., RODRIGUES, F.H., & SOUSA-
LIMA, R.S. 2015. An evaluation of manual and automated methods for detecting
sounds of maned wolves (Chrysocyon brachyurus Illiger 1815). Bioacoustics, 24(2),
185-198.
ROCHA, L.H., FERREIRA, L.S., VENTICINQUE, E.M., RODRIGUES, F.H., &
SOUSA-LIMA, R.S. 2016. Temporal and environmental influences on long-distance
calling by free-ranging maned wolves (Chrysocyon brachyurus). Journal of
Mammalogy, 98(1), 302-311.
RODDEN, M., RODRIGUES F., & BESTELMEYER, S. 2004. Maned wolf
(Chrysocyon brachyurus). Canids: Foxes, Wolves, Jackals and Dogs, Status Survey and
Conservation Action Plan. 1st ed. IUCN Canid Specialist Group, Cambridge, United
Kingdom 38–43.
RODRIGUES, F.H.G. 2002. Biologia e conservação do lobo-guará na Estação
Ecológica de Águas Emendadas, DF. Ph.D Thesis, Universidade Estadual de
Campinas, Campinas, Brazil.
SÁBATO, V. 2011. Aspectos do comportamento acústico do lobo-guará Chrysocyon
brachyurus (Illiger 1815).
SIKES, R.S., & THE ANIMAL CARE AND USE COMMITTEE OF THE
AMERICAN SOCIETY OF MAMMALOGISTS. 2016. 2016 Guidelines of the
American Society of Mammalogists for the use of wild mammals in research and
education. Journal of Mammalogy 97:663−688.
SOUSA-LIMA, R.S., & CLARK, C.W. 2008. Modeling the effect of boat traffic on the
fluctuation of humpback whale singing activity in the Abrolhos National Marine Park,
Brazil. Canadian Acoustics 36:174-181.
THOMPSON, M.E., SCHWAGER, S.J., & PAYNE, K.B. 2010. Heard but not seen: an
acoustic survey of the African forest elephant population at Kakum Conservation Area,
Ghana. African Journal of Ecology 48:224-231.
58
TROLLE, M., NOSS, A.J., LIMA, E.S., & DALPONTE, J.C. 2007. Camera-trap
studies of maned wolf density in the Cerrado and Pantanal of Brazil. Biodiversity and
Conservation, 16, 1197-1204.
59
Supplementary data
Supplementary Data SD1.Wild maned wolf roar-bark sequences recorded at Serra da Canastra National
Park, MG/Brazil. a and b: sequence in response to a playback stimulus registered on 18:55:37 March 09
2017. c and d: two individuals (note the change in spectral characteristics after 33 s) recorded passively
on 19:15:00 March 17 2016. Recordings made with autonomous recorders (Song Meter SM2+; Wildlife
Acoustic), at 8 kHz sample rate and 16-bit wav file format. Spectrogram made on Raven pro 1.5 (Cornell
Bioacoustics Lab, Ithaca, NY, USA), Hann window, 512 window size, 50% brightness and contrast, 50%
overlap, smoothing on.
Supplementary data SD2.Playback response audio file.
Supplementary data SD3.Passive recorded audio file.
60
Chapter 2
Maned wolf long range call propagation and its
implication for the species’ communication
61
Maned wolf long range call propagation and its implication for the
species’ communication
Luane Stamatto Ferreira, Lourdes M.M. Villavicencio, Júlio Baumgarten, Flávio H.
Rodrigues, Renata S. Sousa-Lima
Intended for submission on: Journal of the Acoustic Society of America
Abstract
There are several major open questions about the behavior of maned wolves due
to the difficulties in finding and following individuals in the wild. The investigation of
the species’ acoustic communication has a great potential to fill those gaps.
Understanding the propagation properties of the species’ long-range roar-bark call is the
first step to accomplish this fascinating challenge of learning about their acoustic
communication. We broadcasted roar-barks and re-recorded them simultaneously at
different distances at the Serra da Canastra National Park, MG/Brazil. Roar-barks
broadcasted and re-recorded from a higher to a lower altitude (down-slope) lost more
intensity than when propagated from a lower to a higher altitude (up-slope). We argue
this could have been wave interaction effects and/or an effect of acoustic shadows. On
the site with more vegetation sound attenuated more than at all other sites, which was in
accordance with our excess attenuation prediction as there are more objects that can
absorb sound energy. We found that inclining the speaker 45o upward to simulate the
head/muzzle position of maned wolves during vocalization had a negative effect on
sound intensity instead of enhancing propagation. However, this angle did counteract
62
partially the negative effects of vegetation, and, coupled with other possible positive
effects of this position, may compensate the observed intensity loss. Finally, we found
that around dusk and before dawn transmission loss was lower. Maned wolves vocalize
more on the first hours of the night, but only rarely near dawn. Thus, the species’ time
choice of vocalization is not only influenced by sound propagation, as our results
suggest factors such as social interactions and weather may explain why they do not
take advantage of both twilights to communicate acoustically.
Resumo
Existem muitas perguntas em aberto em relação ao comportamento do lobo-
guará devido às dificuldades em encontrar e acompanhar os indivíduos na natureza. A
investigação da comunicação acústica da espécie tem grande potencial de preencher
essas lacunas. Entender como o aulido, o chamado de longa distância da espécie, se
propaga em ambientes naturais é o primeiro passo para compreender esse sistema de
comunicação. Nós reproduzimos aulidos e os re-gravamos simultaneamente à diferentes
distâncias no Parque Nacional da Serra da Canastra, MG/Brasil. Diferente do que
previmos, aulidos propagados de uma maior altitude para menor (morro abaixo)
perderam mais intensidade que os propagados de uma menor altitude para uma maior
(morro acima). Nós argumentamos que isso pode ter sido resultado de sombras
acústicas e/ou efeitos da interação entre ondas. No local com mais vegetação o som
atenuou mais do que em todos os outros locais, o que estava de acordo com nossa
predição já que há mais atenuação excessiva devido a presença de mais objetos que
podem absorver energia acústica. Nós descobrimos que inclinar a caixa de som 45o para
cima, para simular a posição do focinho dos lobos-guará durante a vocalização, teve um
63
efeito negativo na intensidade do som, ao invés de melhorar a sua propagação.
Entretanto, esse ângulo neutralizou parcialmente os efeitos negativos da vegetação e,
em conjunto com outros possíveis efeitos positivos dessa posição, pode compensar a
perda de intensidade observada. Por fim, encontramos que em torno do entardecer e
logo antes do amanhecer a atenuação foi menor. Lobos guará vocalizam mais nas
primeiras horas da noite, mas apenas raramente perto do amanhecer. Logo, a escolha de
horário de vocalização da espécie não é influenciada apenas pela propagação do som e
os resultados sugerem que fatores como interações sociais e clima, podem explicar por
que lobos-guará não aproveitam ambos crepúsculos para comunicarem-se
acusticamente.
Key-words: Chrysocyon brachyurus, bioacoustic, acoustic adaptation, sound
propagation
64
Introduction
Maned wolves (Chrysocyon brachyurus; Illiger 1815) are nocturnal/crepuscular
and solitary foragers (Rodden et al. 2004). As other canids, the species is monogamous
and the breeding pair shares the same home range, which is usually very extensive
(mean ± SD: 80 ± 53 km2; Jácomo et al. 2009). However, it is believed that the pair
rarely interacts physically, getting close only during the reproductive season (Dietz
1984). Those habits make visual communication impractical, and a compensation by
other sensory channels is expected, as acoustic communication (Fox 1975). Studies in
wild population indicates that maned wolves indeed interact vocally at long distances in
a regular basis (Rocha et al. 2016). There are many gaps in our knowledge of the
species’ behavior (Rodden et al. 2004), and few studies on natural environments due to
the difficulties in finding and following individuals (Bestelmeyer 2000). Therefore, the
investigation of the species acoustic communication has a great potential to fill those
knowledge gaps.
Understanding the propagation of the acoustic signals in the natural environment
is one way to investigate this communication modality. According to the Acoustic
Adaptation Hypothesis (AAH) long range acoustic signals are shaped by selective
pressures to effectively propagate through the natural habitat of the species (Morton
1975; reviewed in Ey & Fischer 2009). The maned wolf inhabits mainly open
neotropical savannas (Cerrado biome; Coelho et al. 2008). The predicted sound
characteristics for effective propagation in this type of environment are high intensity,
low frequency, and tonal structure (Wiley & Richards 1978, 1982). In agreement, the
long-range vocalization of the maned wolf, the roar-bark (Brady 1981; Sábato 2011),
has high intensity and low frequency. However, the signal energy occupies a broad
frequency band which gives the call a harsh, noisy quality. This characteristic can favor
65
the localization of the emitter (Naguib & Wiley 2001), while redundancy, in the form of
bouts of roar-barks (sequences), can favor signal reception (Morton 1975; Brown &
Handford 2000).
A recurrent point cited in terrestrial sound propagation studies is that the vertical
positioning of the emitter influences signal transmission, with higher positions generally
improving emission and reception (Morton 1975; Brown & Handford 2000; Kime at al.
2000; Schwartz et al. 2015). Singing at positions higher than 1 meter from the ground is
enough to minimize ground effects (Marten & Marler 1977) such as destructive
interference (Wiener & Keast 1959). The maned wolf height ranges from 70 to 90
centimeters at the shoulder (Silveira 1999), with the adult average closer to 84
centimeters (Jácomo et al. 2009). That places the species near the limit to experience
those ground effects (or not). To counter those effects a maned wolf could vocalize on
higher grounds or raise its head the maximum possible while vocalizing. In fact,
captivity studies describe that the species elevate the head (pointing the nose upward)
while emitting roar-barks (Sábato 2011; Balieiro 2015; see Figure B in the General
Introduction).
Additionally, vocalizing with the head/muzzle elevated can broadcast body size
information. There are some cues on the vocalization of animals that are correlated with
the vocal tract length, which is normally correlated to body size (like formants; Fitch
1997). Pointing the nose upward maximize the vocal tract length at the moment of
vocalization, and this can pass a larger body acoustic impression (Charlton & McComb
2007; Pisanski & Rendall 2011).
Another form of understanding a communication system is identifying its
temporal pattern. Classical examples are the dawn and dusk choruses, consisting of a
concentration of vocal activity around twilight (Burt & Vehrencamp 2005; Berg et al.
66
2006). This phenomenon is typical of birds (Sedgwick 1941) but has been described in
several other taxa as primates (Colobus guereza—Schel & Zuberbühler 2012) and
invertebrates (Cato 1978). There are at least 12 non-exclusive hypotheses to explain this
time choice (Staicer et al. 1996), especially a possibly optimized sound propagation
(Brown & Handford 2003), and a period not yet ideal to forage due to illumination but
good to communicate while avoiding predators (Berg et al. 2006).
Temperature gradients are important to define sound propagation properties
during transmission. When the sun is up it heats the surface and the soil heats the above
air layers by irradiation, creating a temperature gradient where the ground and lower air
layers are hotter than the above air layers. As sound travel faster in higher temperatures,
this gradient refracts the sound upward making it harder to be detected by a receptor at
ground level (Wiley & Richards 1978; Embleton 1996). When the sun sets the ground
loses heat irradiating to the air above until eventually the ground becomes cooler than
the air. Thus, at night the sound will bend downward and more of its energy will reach a
ground level receptor (Wiley & Richards 1978; Embleton 1996). Finally, when the
night is ending the ground is at its coldest moment and the rising sun heats the above air
layers, amplifying this gradient and in this case its positive effects on sound
transmission (Wiley & Richards 1978; Embleton 1996).
Many canids vocalize most often during twilight, including grey wolves (Canis
lupus; Harrington & Mech 1979), coyotes (Canis latrans; Walsh & Inglis 1989), and
dingos (Canis dingo; Corbett 2001). Maned wolves are most physically active during
twilight (Melo et al. 2007; Jácomo et al. 2004) but seem to concentrate their vocal
activity during the beginning of the night (Rocha et al. 2016). Nocturnal recordings of
four months indicated that the first hour of the night is when they vocalize most (Rocha
et al. 2016).
67
Here we investigate how the long-range maned wolf roar-bark propagates
through its environment by broadcasting calls and re-recording them at different
distances. We tested the propagation in four different sites to investigate a possible
terrain effect. Our prediction was that on the site with more vegetation the propagation
would be less effective due to the increased number of obstacles (leaves) absorbing and
deflecting sound in multiple directions (Bradbury & Vehrencamp 1998). We also
broadcasted sound from a higher to a lower altitude (down-slope) and vice-versa (up-
slope). In these situations, there are two possible predicted outcomes (Bradbury &
Vehrencamp 1998): 1) that sound will propagate more effectively up-slope, due to
positive interference of the sound wave reflected in the same phase by the smooth
ground surface, contrasting with a sound over-spreading in the open down-slope
situation; or 2) that sound will propagate less effectively up-slope, due to ground effects
as negative interference of the sound wave reflected in the opposite phase and
absorption by the ground and obstacles, contrasted with a sound path free of ground
effects and obstacles in the open down-slope situation.
Although we have no information on the directionality of maned wolves roar-
barks, here we assumed both the natural vocalization and the speaker have a broad
conical-like shape sound volume (similar to Lehner 1982). Therefore, by inclining the
speaker 45o upward we also tested whether the head positioning seen in captivity
enhances propagation, and if the position interacted with site differences such as the
presence of vegetation.
Finally, we tested if the maned wolves’ time choice of vocalization was related
to sound propagation. Our hypothesis was that maned wolves vocalize more in the
beginning of the night (Rocha et al. 2016) because this is the period in which roar-barks
68
propagates more efficiently in their habitat. Therefore, we predict playbacks conducted
at this time would propagate more efficiently than at other times of the day.
69
Materials and methods
We conducted this study at Serra da Canastra National Park, located southwest
of Minas Gerais state, in Brazil (Figure 1). The habitat consists mainly of Cerrado
biome, with morphologies varying from open grasslands, mix compositions of tall grass
and shrubs, and areas with dense shrub and low trees (MMA/IBAMA2005). The park
has many rocky outcrops and water bodies, including souterrain rivers, that are also
typical of Cerrado biomes (MMA/IBAMA 2005). Audio data collection in this park
was authorized by Instituto Chico Mendes de Conservação da Biodiversidade (ICMBio;
SISBIO license number 41329-2).
1. Playback and re-recording procedure
The roar-barks used were recordedby Sábato (2011) at two captive facilities in
Minas Gerais: the Criadouro Científico de Fauna Silvestre para Fins de Conservação
da Companhia Brasileira de Metalurgia e Mineração at Araxá; and the Zoológico da
Associação Esportiva e Recreativa dos Funcionários das Usinas Siderúrgicas de Minas
Gerais S/A at Ipatinga. Recordings were made using a Marantz PMD-661 solid state
recorder, a Sennheiser ME-66 module, and a directional Sennheiser K6 microphone.
70
71
We used sounds recorded from 4 captive individuals, 2 males and 2 females
(Figure 2), which were 6-8 meters away from the microphone. We chose 5 good quality
roar-barks coming from the same sequence (bout) for each of the 4 individuals (total: 20
roar-barks) aiming to include roar-barks that were as spectrally diverse as possible. All
roar-barks were broadcasted on each site and condition (time and speaker position; see
below) with a 2.9-6.2 second interval between roar-barks of the same animal and 10
seconds to 10 minutes between animals. The 5 roar-barks of each animal were always
broadcasted together and in the same order (they were on the same file), but the 4
animals order was randomized for each site.
To set the intensity level we used the playback equipment (described below) to
broadcast the captivity recordings and then re-record the played back sounds with the
same equipment and settings used for the original captivity recordings at the same
distance the focal animals were. We then changed the speaker volume until the roar-
bark intensity measured in the re-recordings matched the intensity measured in the
original captivity recordings (measures made in Raven Pro 1.5 software: Bioacoustics
Research Program, 2014. Raven Pro: Interactive Sound Analysis Software. Ithaca, NY:
The Cornell Lab of Ornithology). This was our solution to achieve a playback intensity
level as similar as maned wolves' roar-bark emission. During the recordings in captivity
we did not have a direct way to measure sound pressure levels, and thus no means of
calculating absolute source intensity levels. Two experienced researchers, including the
Figure 1. Study site at Serra da Canastra National Park, MG, Brazil. a - site Flat; b - site Low to high; c -
site Vegetation; d - site High to low. Horizontal distance to the speaker is discriminated on the left side of
the / and altitude on the right side. Maps constructed with QGIS 3.4.0-Madeira (QGIS Development
Team, 2018. QGIS Geographic Information System. Open Source Geospatial Foundation
Project. http://qgis.osgeo.org) and Google Satellite images (Map data ©2018 Google, Imagery ©2018
TerraMetrics).
72
author of the original captivity recordings (VS), reported the playback sounded as
strong as heard from the animals in captivity and in the wild (FHR).
It should be noted that Raven do not give absolute sound levels, instead
amplitude is displayed in a decibel measurement relative to an arbitrary reference value
of 1 (Charif et al. 2010). However, spectrograms generated by recordings made with the
same equipment, configurations, and broadcasted files, should render comparable
amplitudes (although small variations could happen). That is, the amplitude of the
captivity records and calibration can be compared among them, but they cannot be
compared to the files of the re-recorded broadcasts using autonomous recorders
(detailed below). In the same way, the amplitude of the re-recordings can be compared
among them, but not to the captivity and calibration records.
We used an Acer AspireOne notebook to broadcast the sounds using Raven pro
1.5 software and a Pioneer S-DJ50X speaker (class A/B Bi-amp, 80 W output, 50-
20000 Hz frequency range) 86 centimeters above the ground to simulate the height of a
maned wolf.
To re-record the broadcasts we successively mounted at each site a line of 7
autonomous recorders (Song Meter SM2+; Wildlife Acoustics, Inc., Concord,
Massachusetts) with one omnidirectional weatherproof microphone each (SMX-II;
Wildlife Acoustics, Inc.; sensitivity -36±4dB [0dB=1V/pa@1kHz]; 20Hz-20kHz flat
response frequency). Recorders were set on the road side on the most straight and open
portion possible (except site “Vegetation”, see below) in a single direction from the
speaker position at distances 1.25 m, 20 m, 40 m, 80 m, 160 m, 320 m, and 640 m
(Figure 1a-d). Distances were measured by measuring tape (1.25 to 80 m) and GPS
(Garmin GPSMAP® 76S; accuracy < 15 m). The autonomous recorders were attached
on stakes of the same height of the speaker (86 cm) with the omnidirectional
73
microphone in a perpendicular position in relation to the speaker. Recordings were
made continuously, with an 8 kHz sample rate, 16 bits wav files, 36dB gain, and
partitioned in 30 minutes files (same configuration used in Rocha et al. 2015, 2016).
We had an eighth recorder placed at and 1280 m, but as only once it registered
the roar-barks broadcasted we decided to exclude it from all analysis.
Each recorder unit was placed at the same distance at all sites, e.g. the recorder
placed at 160 m at the first site was always placed at that distance at the following sites.
This was done to render the measures among the same distance more comparable, as the
units were not calibrated together and present some variation between their intensity
reading. A field test indicated our SongMeters units varied 1.91 ± 1.62 dB (mean ± SD)
in their intensity measures.
2. Broadcast sites
The sites were chosen based on the points with the highest spontaneous roar-barks
records of Rocha et al. (2016) study. Those sites also had the different characteristics we
were looking to test the terrain influence.
From March 04 to 05 2017 we conducted the broadcasts on the first site (Figure
1a: “Flat”). From March 05 to 06 2017 we conducted the broadcasts on the second site
(Figure 1b: “Low to High”). After we transferred the autonomous recorders 20 meters
laterally, so they would be outside the road and inside the tall grass, shrubs, and rocky
outcrop area, and conducted broadcasts from March 06 to 07 2017 (Figure 1c:
“Vegetation” site; see top photo in General Introduction Figure C). The vegetation
height was, in general, slightly below the speaker (86 cm). Altitudes differed
approximately 0-2 m from the “Low to High” site. From March 07 to 09 2018 we
conducted the broadcasts on the last site, “High to Low” (Figure 1d).
74
3. Playback times
At each site we conducted playbacks three times a night. We chose the following
moments: between 18:40 and 19:40 (15-75 minutes after sunset), when wild maned
wolves are more vocally active (Rocha et al. 2016); between 23:00 and 00:00, an
intermediary time point; and between 05:00 and 06:00 (15-75 minutes before sunrise),
when wild maned wolves are less vocally active (Rocha et al. 2016).
At site “High to Low” we conducted extra broadcasts to test roar-bark
propagation during the light period: between 06:00 and 07:00 (15-75 minutes after
sunrise); between 11:00 and 12:00; and between 17:00 and 18:00 (15-75 minutes before
sunset). At this site we also conducted one replicate of all broadcasts (day and night) on
the next day.
Mean local sunset was 18:28 and sunrise 06:05 during the experiment days
(calculated on https://www.sunearthtools.com/pt/solar/sunrise-sunset-calendar.php;
access Set/16 2018).
4. Speaker position
To simulate the maned wolf inclination of the head/muzzle seen on captivity we
conducted broadcasts with the speaker box inclined 45o upward. This position was
named “Inclined” while the normal straight forward position was referred as “Straight”.
We conducted a “Straight” and “Inclined” broadcast, in random order, for all sites and
times mentioned above.
75
5. Measurements
All recordings were analyzed on spectrograms generated by Raven Pro 1.5
software with the following characteristics: Hann window type, 512 window size,
grayscale, 50% of brightness and contrast, 50% overlap, and smoothing on.
We created selection boxes including the first two frequency bands
(Figure 2) as this is the only portion normally visible on wild maned wolf recordings
(Rocha et al. 2015). We extracted the roar-bark intensity through the peak power (dB)
measure on Raven Pro 1.5. We then calculated the relative loss in dB compared to the
measure of the same roar-bark re-recorded at 1.25 m. The dB loss was used as the
response variable of the model (see next section). A box of equal dimension was made
immediately before each roar-bark and the average power (dB) measure was taken on
Raven. This was used as a measure of the background noise level to be inserted in the
model as a control.
The temperature (oC), relative air humidity (%RH), and wind speed (m/s) was
measured for each playback session with a digital termo-higro-anemo-luximether
(SKTHAL 01, Skill-tec) at the speaker position and height. We used a single measure
for the “Straight” and “Inclined” broadcasts as they were done in succession (both
broadcasts together lasted between 5 and 25 minutes).
76
Figure 2. Captive maned wolves roar-barks sequences broadcasted at Serra da Canastra National Park,
MG/Brazil. GA and SH are males, SA and JU females. Red selection boxes on the first roar-bark of each
animal exemplifies the ones used to measure roar-bark intensity (peak power, dB). Selections near the
second roar-bark of each animal exemplifies the ones used to measure noise intensity (average power,
dB). Spectrograms and measures were made on Raven Pro 1.5 (Cornell Bioacoustics Lab, Ithaca, NY,
USA), Hann window, 512 window size, 50% brightness and contrast, 50% overlap, smoothing “on”.
6. Statistical analysis
All statistical analyses were conducted with R software (R project version 3.4.4).
The script used and the resulting output is in the supplementary material.
To evaluate the effect of the different sites, times, and speaker position on sound
propagation we constructed a linear mixed model. We opted for a linear model as the
sampled distances increased in doubles resulting in a predicted loss of 6 dB by spherical
spreading only at each distance of re-recording (although some variation in the
measures are expected between different recording units). We used a mixed model to
control for the individual and particular roar-bark differences as they repeated
themselves across distance, sites, times, and speaker position. The method of restricted
maximum likelihood was used in order to produce unbiased variance estimates.
77
The main model was a function of the relative dB loss and we included as fixed
factors distance (7), site (4), speaker position (2), an site*speaker.position interaction,
time (6), wind speed (m/s), and background noise (dB). Neither temperature nor
humidity were included on the main model as they had a conspicuous daily cycle, and
thus would be highly related to the broadcasting time. As our goal was to evaluate
propagation on the different broadcasting times in general, we chose to maintain all
daily cycles variation inside the factor “time” instead of separating the effect of time
and temperature/ humidity. For comparative purposes, we build a secondary model
including temperature (°C) on the fixed factors to control for days with overall different
temperatures, which would bias the factor “site” as each site was tested on a different
day. Only temperature was included on the secondary model as it was significantly
correlated with humidity (-0.738 spearman correlation coefficient). As random factors
for the intercept we included in the models the individual (4) and as sublevels the
different roar-barks (5) of each individual.
To validate the main model, we tested if the residuals had a normal distribution
and if they were symmetrically distributed in relation to the fitted values. We further
evaluated the model by comparing predicted and observed values. We used an ANOVA
to test if the models fixed factors had a significant effect. To test for differences
between levels of the fixed factors we compared multiple means on the models with
Tukey contrasts.
78
Results
An exploratory analysis of the continuous factors (Supplementary material 1)
revealed that temperature and humidity fluctuated according to time: the 11-12h period
had the highest temperature and lower humidity, and 23-00h period the contrary. The
night at the site “Low to High” was on average colder and more humid than the other
nights. Wind speed was very low (0.37 ± 0.61 m/s, mean ± SD), being higher during the
day (maximum 2.4 m/s). The site High-to-low had more intense background noise than
other sites and there was a tendency for recorders positioned farther away to have less
intense background noise.
The main model residuals had a normal distribution and were symmetrically
distributed in relation to the predicted values (Supplementary material 2). The
predicted and the observed values correlated at a coefficient of 0.946 (Supplementary
material 3). An ANOVA revealed the intercept and all fixed factors were significant on
the model, including the interaction (Table 1). Confidence intervals for fixed and
random factors are reported on Table 2.
Table 1. ANOVA test for the fixed factors of the main model for the intensity (dB) loss of maned wolves
roar-barks broadcasted on their natural environment.
numDF denDF F-value p-value
Intercept 1 3736 41584.23 <.0001
Distance 5 3736 14.82 <.0001
Site 3 3736 11114.84 <.0001
Speaker position 1 3736 132.76 <.0001
Site:Speaker.position 3 3736 86.25 0.0008
Time 5 3736 1144.79 <.0001
Wind speed 1 3736 3335.92 0.0001
Background Noise 1 3736 5.57 <.0001
79
Table 2. Approximate 95% confidence intervals for the estimate factor effects of the main model for the
intensity (dB) loss of maned wolves roar-barks broadcasted on their natural environment. Base categories
are specified under parenthesis. Factors/levels with positive and negative estimates, which indicates they
are not influential on the model or not significantly different from the base category, are underlined.
FIXED EFFECTS:
lower est. upper
Intercept -36.709 -34.9073 -33.1052
-36.709 -34.9073 -33.1052
-36.709 -34.9073 -33.1052
Distance (x 20m)
40m -6.955 -6.474 -5.992
80m -12.877 -12.336 -11.794
160m -27.953 -27.375 -26.798
320m -34.880 -34.183 -33.485
640m -39.183 -38.283 -37.384
Site (x Flat)
Low to High -4.691 -3.906 -3.121
Vegetation -10.985 -10.220 -9.454
High to Low -6.229 -5.476 -4.723
Speaker position (x Inclined)
Straight -0.878 -0.176 0.527
Site:Speaker.position (x Inclined:Flat)
Straight:Low to High -1.365 -0.265 0.835
Straight:Vegetation -2.953 -1.891 -0.830
Straight:High to Low -2.061 -1.241 -0.421
Time (x 05-06h)
06-07h -2.138 -1.451 -0.764
11-12h -3.560 -2.667 -1.774
17-18h -0.793 0.335 1.463
18:40-19:40h -0.680 -0.241 0.199
23-00h -2.933 -2.489 -2.045
Wind speed -1.396 -0.860 -0.324
Background Noise 0.532 0.592 0.652
RANDOM EFFECTS:
Between-group: Individual 0.018 0.227 2.884
Between-group: Roar-bark 0.407 0.634 0.988
Within-group standard error 4.695 4.802 4.913
80
When compared to the main model, the secondary model had minimal
differences of estimated values for all factors except Time (Supplementary material
4). This mean days with overall different temperatures did not biased the site differences
result.
As expected, Distance had the larger effect on the model sound intensity (Table
2). Mean estimate difference between consecutive distances was -7.66. The values were
very close to the predicted for spherical spreading loss alone (-6dB), except for the
difference between 80m and 160m (Figure 3 and Table 3). That mean the factor
Distance on the model was able to reflect almost only the spherical loss, leaving the
excess of attenuation (i.e. other negative influences besides the spherical spreading) to
be explained by the remaining factors, for instance Site and Time.
Figure 3. Propagation of broadcasted roar-barks from captive maned wolves at Serra da Canastra
National Park, MG/Brazil. Re-recordings made with autonomous recorders (Song Meter SM2+; Wildlife
Acoustics, Inc., Concord, Massachusetts). The intensity loss is relative to the re-recording at 1.25m.
81
Table 3. Simultaneous tests for general linear hypotheses using Tukey contrasts for multiple comparisons
of means. The reported p values are adjusted by single-step method. Significance codes: 0 '***', 0.001
'**', 0.01 '*', 0.05 '.', 0.1 ' '. Only comparisons of consecutive distances are shown.
Hypothesis Estimate Std. Error z value Pr(>|z|)
D2(40m) - D1(20m) = 0 -6.474 0.246 -26.350 <2e-16 ***
D3(80m) - D2(40m) = 0 -5.862 0.253 -23.202 <2e-16 ***
D4(160m) - D3(80m) = 0 -15.040 0.240 -62.748 <2e-16 ***
D5(320m) - D4(160m) = 0 -6.808 0.318 -21.394 <2e-16 ***
D6(640m) - D5(320m) = 0 -4.100 0.471 -8.707 <2e-16 ***
Low to High - Flat = 0 -3.906 0.400 -9.760 <0.001 ***
Vegetation - Flat = 0 -10.220 0.390 -26.182 <0.001 ***
High to Low - Flat = 0 -5.476 0.384 -14.258 <0.001 ***
Vegetation - Low to High = 0 -6.314 0.419 -15.064 <0.001 ***
High to Low - Low to High = 0 -1.570 0.411 -3.824 <0.001 ***
High to Low - Vegetation = 0 4.743 0.414 11.450 <0.001 ***
Straight - Inclined = 0 -0.176 0.359 -0.490 0.624
06h - 05h = 0 -1.451 0.351 -4.139 <0.001 ***
11h - 05h = 0 -2.667 0.456 -5.854 <0.001 ***
17h - 05h = 0 0.335 0.576 0.582 0.991
18h - 05h = 0 -0.241 0.224 -1.074 0.878
23h - 05h = 0 -2.489 0.227 -10.991 <0.001 ***
11h - 06h = 0 -1.216 0.418 -2.910 0.036 *
17h - 06h = 0 1.786 0.495 3.612 0.004 **
18h - 06h = 0 1.211 0.351 3.451 0.006 **
23h - 06h = 0 -1.038 0.348 -2.986 0.029 *
17h - 11h = 0 3.002 0.430 6.986 <0.001 ***
18h - 11h = 0 2.426 0.444 5.459 <0.001 ***
23h - 11h = 0 0.177 0.446 0.398 0.998
18h - 17h = 0 -0.576 0.563 -1.022 0.898
23h - 17h = 0 -2.824 0.562 -5.029 <0.001 ***
23h - 18h = 0 -2.249 0.231 -9.727 <0.001 ***
82
Regarding sites (Table 3 and Figure 4), Tukey contrasts revealed that sound
propagated more efficiently, i.e. lost less intensity, at the Flat site than all other sites.
Sound propagated more efficiently at the Low-to-high site than at the High-to-low site,
in accordance with our first predicted outcome that positive interference would be more
influential than negative ground effects and obstacles. In accordance with our
prediction, sound propagated less efficiently at the Vegetation site than all other sites.
The estimate difference when compared with the same place at the road (site Low to
High) was -6.31 dB. The negative effect of vegetation was less noticeable beyond 160
m (Figure 4).
Figure 4. Propagation of broadcasted captive records of maned wolves roar-barks at 4 sites at Serra da
Canastra National Park, MG/Brazil. Re-recordings made with autonomous recorders (Song Meter SM2+;
Wildlife Acoustics, Inc., Concord, Massachusetts). The intensity loss is relative to the re-recording at
1.25m.
83
Regarding the Speaker Position effect (Table 3), Tukey contrasts revealed that
the there was no difference in the sound broadcasted in the Inclined position compared
to the Straight position. However, different from the other factors tested, the Speaker
position caused an intensity difference already in the 1.25m reading (Figure 5 shows
the dB loss relative to the 1.25m Straight), with the Inclined position being on average
2.50 dB (± 2.22 dB) lower in intensity than the Straight position at this initial distance.
This difference is maintained at greater distances (-1.44 ± 3.36 dB x the Straight). As
our measure of intensity loss is relative to the 1.25m re-recording of each roar-bark, no
difference in the Speaker Position was detected. This indicates that once the initial
lower intensity of the Inclined position is taken into account, both positions propagate
similarly.
The general lower intensity of the Inclined broadcast contradicted out theory that
the head/muzzle elevation enhanced propagation. At the site Low to High, and specially
at the site Vegetation (Figure 5), the mean intensity difference between Inclined and
Straight positions was smaller, and at 160 m even positive (i.e. the Inclined position was
better than the Straight). This can also be seen on the model on the interaction between
site and position (Table 2). The interaction effect is negative for the position Straight on
the site Vegetation and larger than the other combinations. This suggests that on this site
the Straight speaker position had a much smaller effect than on other combination.
84
Figure 5. Propagation of broadcasted captive records of maned wolves roar-barks at 4 sites at Serra da
Canastra National Park, MG/Brazil. We conducted broadcasts with the speaker box positioned straight
forward (Straight) and with the speaker box inclined 45o upward (Inclined) to simulate the inclination of
the head/muzzle seen when animals roar-bark. Re-recordings made with autonomous recorders (Song
Meter SM2+; Wildlife Acoustics, Inc., Concord, Massachusetts). The intensity loss is relative to the
straight re-recording at 1.25m.
Finally, regarding the effect of the broadcasting time (Table 3 and Figure 6),
Tukey contrasts revealed that the propagation at 17h, 18h and 05h were more efficient
than at 11h, 23h and 06h, and indistinguishable among them. Those results were
partially in accordance with our prediction of the beginning of the night being the
period with the most efficient propagation of roar-barks, even though our results show
the end of the night as an equally efficient period to call. Propagation at 23h and 11h
were less efficient than all other times and not significantly different. An impaired
propagation at midday was expected, but not at midnight, which was not expected.
85
The secondary model revealed fewer differences between broadcasting times
(Supplementary material 4), suggesting temperature and its inverse correlate humidity
are the main influence on propagation differences. Some differences remained, as
between the 05h and 06h broadcasts and 18h and 23h broadcasts, indicating other
characteristics of those periods are driving the differences, potentially the direction of
the air temperature gradient and wind masses.
Figure 6. Propagation of broadcasted captive records of maned wolves roar-barks at 6 time intervals at
Serra da Canastra National Park, MG/Brazil. The time shown is the beginning of a 1 hour interval in
which broadcasts were made. Re-recordings made with autonomous recorders (Song Meter SM2+;
Wildlife Acoustics, Inc., Concord, Massachusetts). The intensity loss is relative to the re-recording at
1.25m.
86
Discussion
In this work we investigated how roar-barks, the long-range maned wolf call,
propagate through the species’ natural environment. Roar-barks broadcasted and re-
recorded from a higher to a lower altitude (down-slope) lost more intensity than when
propagated from a lower to a higher altitude (up-slope). At the site with more vegetation
sound attenuated more than at all other sites, which was in accordance with our
prediction. We found that inclining the speaker 45o upward to simulate the head/muzzle
position during vocalization had a negative effect on sound intensity instead of
enhancing transmission. However, this position did counteract partially the negative
effects of vegetation. Finally, we found that the maned wolf roar-bark calling time
choice is partially correlated with sound propagation properties of its habitat, since
transmission loss is lower in the beginning of the night. However, attenuation was also
minimal between 05 and 06 AM and these are times when the species rarely vocalizes.
Better sound propagation (measured as a smaller intensity loss) when
broadcasting from a lower to a higher position was consistent with our first predicted
scenario (and opposite to our second) which is likely due to positive interference from
sound waves reflected from the surface of the bare soil road (Bradbury & Vehrencamp
1998).
On the site with vegetation, although it was also a site where sound was
propagated from a lower to a higher position, any interference is less likely as the
uneven leaf surface would reflect sound in many different directions. This fact,
combined with the presence of more obstacles to attenuate the sound, resulted in a
worse sounds transmission at this site compared to all others. It is worth to note that the
effect of this site was the larger on the model (larger estimate except for distance) even
87
considering that the grass and bushes were just below the speaker. Beyond 160 m the
vegetation tended to get lower, and the difference between this and other sites became
less evident.
Captivity studies describe that maned wolves usually emit roar-barks with the
nose pointed upward (Sábato 2011; Balieiro & Monticelli 2019; see Figure B in the
General Introduction). However, here we found that broadcasting those calls mimicking
this position with the speaker lead to lower registered intensities. At least for the Flat
site and where the broadcast was done from a higher to a lower place, this difference
was very consistent, appearing in the first recorder (1.25m) and maintaining itself
through distance. We can only assume in general situations there is a tradeoff between
signal intensity and other positive effects of this posture.
One possibility is the potential of this posture to counteract the negative effects
of the vegetation, as seen in this work. On the site with vegetation the upward inclined
speaker position could have reduced the amount of obstacles between the sound source
and the recorders, in some distances resulting in a smaller loss of intensity. It is worth
noting this site probably represents the normal vocalization scenario for maned wolves
on their natural environment. That would indicate a behavioral adaptation rather than an
acoustic signal adaptation for better propagation on the species’ natural habitat
(Acoustic Adaptation Hypothesis; Morton 1975).
Some other possible advantages of the nose/muzzle elevation could be that
emitting the sound upward makes it more omnidirectional, provides a larger body
impression through maximization of the vocal tract length, or that it enhances the
maintenance of signal integrity despite the intensity loss. Another related hypothesis is
that this posture can make it easier for the animal to produce the call, being a
physiological requirement.
88
Regarding the time of the day, the times when propagation was substantially
impaired was between 11h and 12h, which was expected and reported elsewhere
(dingos’ long-range vocalizations; Déaux et al. 2016c), and between 23h and 00h,
which was unexpected. Different from the day, during the night the colder surface and
hotter air would refract the sound downward enhancing propagation (Embleton 1996). It
is possible, however, that the air at the Serra da Canastra National Park at that period
had already lost its heat and there was no temperature gradient.
Propagation of the broadcasted roar-barks was more efficient near dusk (17h to
19:40h) and before dawn (between 05h and 06h). The efficient sound propagation
before dawn is one of the most cited theories explaining the existence of the dawn
chorus (Brenowitz 1982; Brown & Handford 2003). Those studies are in general
conducted in temperate zones and sometimes this effect is not found in the tropics
where other factors may be prevalent (Berg et al. 2006). In this study we did find a
better sound propagation before dawn and this could mean the Serra da Canastra
National Park may behave more like a temperate than a tropical area in relation to sound
transmission.
Maned wolves are crepuscular, being more active during twilight (Melo et al.
2007; Jácomo et al. 2004). However, they rarely vocalize around dawn (Rocha et al.,
2016; Ferreira et al., unpublished; Chapter 3). That indicates that propagation properties
alone are not the main reason for the species calling period, or they would vocalize
equally on both twilights instead of concentrating the vocal activity in the beginning of
the night.
Again, there may be other acoustic factors not tested that could explain why
maned wolves prefer to vocalize in the beginning of the night, as signal integrity and
transmission consistency (Brown & Handford 2003). There may be also social factors
89
involved. Rocha et al. (2016) proposes that the beginning of the activity would be the
most important period to acoustically announce/ defend territories, as vagrant
individuals would also be starting their activities and the probability of invading a
territory while searching for food would be highest.
Finally, there may be a weather influence. Wind speed has been previously
found to affect roar-bark detection and to increase throughout the night for the same
area between April and July (Rocha et al. 2016). Although the wind speed registered
during this experiment were very low and apparently had little influence on the model,
general year-round patterns may be influencing the evolution of the maned wolf time
choice for long distance calling.
We expected more intense backgrounds to impair the signal, but the background
noise level fluctuated with the signal intensity, having a small positive effect on the
model. This could mean our measure was mainly related to the equipment sensitivity. In
accordance, we did not include on the selection boxes wind noise or any noticeable
zoophony. As the recorders were not calibrated together, our measure probably acted on
the model as a control for equipment variance, which was a positive feature. It could
also act as a control for differences in the background noise levels between sites.
In conclusion, maned wolves have to deal with dynamic propagation scenarios
that will dramatically influence their vocalization active space, which explains why the
signal must be redundant. Their simultaneous listening might enable behavioral
adjustments to local propagation scenarios, i.e., moving their head. Finally, we can also
expect receiver maned wolves to have evolved characteristics and behaviors that
enhance signal reception, and further studies will clarify other aspects of this
communication system.
90
References
BALIEIRO, F.S. 2015. The voice of the not so lonely maned wolf: evidence of
individual discrimination via playback of the long distance extended bark. M.S. Thesis,
Universidade de São Paulo, Ribeirão Preto, São Paulo, Brazil.
BALIEIRO, F., & MONTICELLI, P.F. 2019. Evidence of individual discrimination in
the maned wolf long-distance extended-bark. Behavioural processes, 158, 219-227.
BERG, K.S., BRUMFIELD, R.T., & APANIUS, V. 2006. Phylogenetic and ecological
determinants of the neotropical dawn chorus. Proceedings of the Royal Society of
London B: Biological Sciences, 273(1589), 999-1005.
BESTELMEYER, S.V. 2000. Solitary, reproductive, and parental behavior of maned
wolves (Chrysocyon brachyurus). PhD Thesis, Colorado State University, Fort Collins,
U.S.A.
BRADBURY, J.W., & VEHRENCAMP, S.L. 1998. Principles of animal
communication.
BRADY, C.A. 1981. The vocal repertoires of the bush dog (Speothos venaticus), crab-
eating fox (Cerdocyon thous), and maned wolf (Chrysocyon brachyurus). Animal
Behaviour, 29(3), 649-669.
BRENOWITZ, E.A. 1982. The active space of red-winged blackbird song. Journal of
Comparative Physiology, 147(4), 511-522.
BROWN, T.J., & HANDFORD, P. 2000. Sound design for vocalizations: quality in the
woods, consistency in the fields. The Condor, 102(1), 81-92.
BROWN, T.J., & HANDFORD, P. 2003. Why birds sing at dawn: the role of consistent
song transmission. Ibis, 145(1), 120-129.
BURT, J.M., & VEHRENCAMP, S.L. 2005. Dawn chorus as an interactive
communication network. Animal communication networks, 320-343.
CATO, D.H. 1978. Marine biological choruses observed in tropical waters near
Australia. The Journal of the Acoustical Society of America, 64(3), 736-743.
CHARLTON, B.D., REBY, D., & MCCOMB, K. 2007. Female perception of size-
related formant shifts in red deer, Cervus elaphus. Animal Behaviour, 74(4), 707-714.
COELHO, C.M., DE MELO, L.F.B., SÁBATO, M.A.L., VAZ MAGNI, E.M.,
HIRSCH, A., & YOUNG, R.J. 2008. Habitat use by wild maned wolves (Chrysocyon
brachyurus) in a transition zone environment. Journal of Mammalogy, 89(1), 97-104.
CORBETT, L. 2001, October. The conservation status of the dingo Canis lupus dingo
in Australia, with particular reference to New South Wales: threats to pure dingoes and
potential solutions. In A Symposium on the Dingo’.(Eds C.R. Dickman and D. Lunney.)
pp (pp. 10-19).
91
DÉAUX, E.C., CHARRIER, I., & CLARKE, J.A. 2016c. The bark, the howl and the
bark-howl: Identity cues in dingoes’ multicomponent calls. Behavioural Processes, 129,
94-100.
DIETZ, J.M. 1984. Ecology and social organization of the maned wolf (Chrysocyon
brachyurus). Smithsonian Contribution to Zoology. 1-51.
EMBLETON, T.F. 1996. Tutorial on sound propagation outdoors. The journal of the
Acoustical Society of America. 100(1), 31-48.
EY, E., & FISCHER, J. 2009. The “acoustic adaptation hypothesis”—a review of the
evidence from birds, anurans and mammals. Bioacoustics, 19(1-2), 21-48.
FITCH, W.T. 1997. Vocal tract length and formant frequency dispersion correlate with
body size in rhesus macaques. The Journal of the Acoustical Society of
America, 102(2), 1213-1222.
FOX, M. W. 1975. The wild canids: their systematics, behavioral ecology and
evolution. R.E. Krieger.
HARRINGTON, F.H., & MECH, L. D. 1979. Wolf howling and its role in territory
maintenance. Behaviour, 207-249.
JÁCOMO, A.T., SILVEIRA, L., & DINIZ-FILHO, J.A.F. 2004. Niche separation
between the maned wolf (Chrysocyon brachyurus), the crab-eating fox (Dusicyon
thous) and the hoary fox (Dusicyon vetulus) in central Brazil. Journal of
Zoology, 262(1), 99-106.
JÁCOMO, A.T., KASHIVAKURA, C.K., FERRO, C., FURTADO, M.M., ASTETE,
S.P., TÔRRES, N.M., ... & SILVEIRA, L. 2009. Home range and spatial organization
of maned wolves in the Brazilian grasslands. Journal of Mammalogy, 90(1), 150-157.
KIME, N.M., TURNER, W.R., & RYAN, M.J. 2000. The transmission of
advertisement calls in Central American frogs. Behavioral Ecology, 11(1), 71-83.
MARTEN, K., & MARLER, P. 1977. Sound transmission and its significance for
animal vocalization. Behavioral ecology and sociobiology, 2(3), 271-290.
MELO, L.F., LIMA SÁBATO, M.A., VAZ MAGNI, E.M., YOUNG, R.J., &
COELHO, C.M. 2007. Secret lives of maned wolves (Chrysocyon brachyurus Illiger
1815): as revealed by GPS tracking collars. Journal of Zoology, 271(1), 27–36.
MMA/IBAMA, 2005. Plano de manejo – Parque Nacional da Serra da Canastra.
Brasília: Ibama. 799p.
MORTON, E.S. 1975. Ecological sources of selection on avian sounds. The American
Naturalist, 109(965), 17-34.
NAGUIB, M. & WILEY, H. 2001. Estimating the distance to a source of sound:
mechanisms and adaptations for long-range communication. Animal
Behaviour. 62, 825-837.
92
PISANSKI, K., & RENDALL, D. 2011. The prioritization of voice fundamental
frequency or formants in listeners’ assessments of speaker size, masculinity, and
attractiveness. The Journal of the Acoustical Society of America, 129(4), 2201-2212.
ROCHA, L.H., FERREIRA, L.S., PAULA, B.C., RODRIGUES, F.H., & SOUSA-
LIMA, R.S. 2015. An evaluation of manual and automated methods for detecting
sounds of maned wolves (Chrysocyon brachyurus Illiger 1815). Bioacoustics, 24(2),
185-198.
ROCHA, L.H., FERREIRA, L.S., VENTICINQUE, E.M., RODRIGUES, F.H., &
SOUSA-LIMA, R.S. 2016. Temporal and environmental influences on long-distance
calling by free-ranging maned wolves (Chrysocyon brachyurus). Journal of
Mammalogy, 98(1), 302-311.
RODDEN, M., RODRIGUES F., & BESTELMEYER, S. 2004. Maned wolf
(Chrysocyon brachyurus). Canids: Foxes, Wolves, Jackals and Dogs, Status Survey and
Conservation Action Plan. 1st ed. IUCN Canid Specialist Group, Cambridge, United
Kingdom 38–43.
SÁBATO, V. 2011. Aspectos do comportamento acústico do lobo-guará Chrysocyon
brachyurus (Illiger 1815).
SCHEL, A.M., & ZUBERBÜHLER, K. 2012. Dawn chorusing in guereza colobus
monkeys. Behavioral Ecology and Sociobiology, 66(3), 361-373.
SCHWARTZ, J.J., HUNCE, R., LENTINE, B., & POWERS, K. 2016. Calling site
choice and its impact on call degradation and call attractiveness in the gray treefrog,
Hyla versicolor. Behavioral ecology and sociobiology, 70(1), 1-19.
SEDGWICK, E. 1940. The Dawn Chorus. Emu, 40(5), 422-422.
SILVEIRA, L. 1999. Ecologia e conservação da comunidade de carnívoros do Parque
Nacional das Emas, GO. M.S. Thesis, Universidade Federal de Goiás, Goiânia, Goiás,
Brazil.
STAICER, C.A., SPECTOR, D.A., & HORN, A.G. (1996). The dawn chorus and other
diel patterns in acoustic signaling. Ecology and evolution of acoustic communication in
birds, 426-453.
WALSH, P.B., & INGLIS, J.M. 1989. Seasonal and diel rate of spontaneous
vocalization in coyotes in south Texas. Journal of mammalogy, 70(1), 169-171.
WIENER, F.M., & KEAST, D.N. 1959. Experimental study of the propagation of sound
over ground. The Journal of the Acoustical Society of America, 31(6), 724-733.
WILEY, R.H., RICHARDS, J. 1982. Adaptations for acoustic communication in
birds: sound transmission and signal detection. In: Kroodsma, D.H., Miller, E.H. (eds)
Ecology and evolution of acoustic communication in birds. Academic Press, New
York, pp 131–181
93
WILEY, R.H., & RICHARDS, D.G. 1978. Physical constraints on acoustic
communication in the atmosphere: implications for the evolution of animal
vocalizations. Behavioral Ecology and Sociobiology, 3(1), 69-94.
94
Supplementary material 1 – Preliminary data exploration
95
2 – Normality and homogeneity of residuals
96
3 – Predicted x observed values of the model
97
4 – Secondary model (including temperature):
Table S2. ANOVA test for the fixed factors of the secondary model for the intensity propagation of
maned wolves roar-barks broadcasted on their natural environment.
numDF denDF F-value p-value
Intercept 1 3735 41460.96 <.0001
Distance 5 3735 3347.82 <.0001
Site 3 3735 1148.00 <.0001
Speaker position 1 3735 85.66 <.0001
Site:Speaker.position 3 3735 5.63 0.0008
Time 5 3735 125.37 <.0001
Wind speed 1 3735 14.87 0.0001
Background Noise 1 3735 11095.45 <.0001
Temperature 1 3735 103.66 <.0001
Table S3. Secondary model: simultaneous tests for general linear hypotheses using Tukey contrasts for
multiple comparisons of means. The reported p values are adjusted by single-step method. Significance
codes: 0 '***', 0.001 '**', 0.01 '*', 0.05 '.', 0.1 ' '. Only com comparisons of consecutive distances are
shown.
Hypothesis Estimate Std. Error z value Pr(>|z|)
D2(40m) - D1(20m)= 0 -6.497 0.245 -26.480 <2e-16 ***
D3(80m) - D2(40m) = 0 -5.851 0.252 -23.193 <2e-16 ***
D4(160m) - D3(80m) = 0 -15.047 0.239 -62.879 <2e-16 ***
D5(320m) - D4(160m) = 0 -6.873 0.318 -21.600 <2e-16 ***
D6(640m) - D5(320m) = 0 -4.049 0.470 -8.608 <2e-16 ***
Low to High - Flat = 0 -5.888 0.679 -8.675 <0.001 ***
Vegetation - Flat = 0 -10.860 0.428 -25.368 <0.001 ***
High to Low - Flat = 0 -6.558 0.487 -13.479 <0.001 ***
Vegetation - Low to High = 0 -4.972 0.560 -8.884 <0.001 ***
High to Low - Low to High = 0 -0.670 0.480 -1.396 0.485
High to Low - Vegetation = 0 4.302 0.431 9.974 <0.001 ***
Straight - Inclined = 0 -0.176 0.358 -0.491 0.623
98
06h - 05h = 0 -1.110 0.363 -3.062 0.019 *
11h - 05h = 0 1.134 1.146 0.989 0.8729
17h - 05h = 0 2.303 0.792 2.908 0.0301 *
18h - 05h = 0 1.081 0.429 2.521 0.0853
23h - 05h = 0 -1.517 0.352 -4.316 <0.001 ***
11h - 06h = 0 2.244 1.045 2.148 0.1974
17h - 06h = 0 3.413 0.668 5.107 <0.001 ***
18h - 06h = 0 2.191 0.443 4.945 <0.001 ***
23h - 06h = 0 -0.407 0.389 -1.047 0.8453
17h - 11h = 0 1.169 0.664 1.759 0.3962
18h - 11h = 0 -0.054 0.817 -0.066 1
23h - 11h = 0 -2.651 0.901 -2.944 0.027 *
18h - 17h = 0 -1.222 0.590 -2.072 0.2298
23h - 17h = 0 -3.820 0.625 -6.114 <0.001 ***
23h - 18h = 0 -2.598 0.250 -10.382 <0.001 ***
99
Chapter 3
Temporal and spatial patterns of the long-range calls
of maned wolves
100
Temporal and spatial patterns of the long-range calls of maned wolves
Luane Ferreira, Luciana Rocha, Danielly Duarte, Edvaldo Neto, Júlio Baumgarten, Flávio
Rodrigues, Renata Sousa-Lima
Intended for submission on: Biota neotropica
Abstract
Passive acoustic monitoring has a great potential for aiding conservation efforts
and elucidating the behavior and ecology of nocturnal/crepuscular secretive species, like
the maned wolf. Here we characterize the seasonal, lunar, and diel patterns in the long-
range vocalizations (roar-barks) of free ranging maned wolves at Serra da Canastra
National Park (Brazil) throughout eight months of recordings over two years with a grid
of 12/13 autonomous recorders. We found it is possible to identify the mating season
and probably the circa-parturition period through an increase in vocal activity of the
species. Those peaks in vocal activity indicate a role of roar-barks in partner attraction
and mate guarding, and also in intra-familiar-group communication. Additionally, vocal
activity happened throughout all recorded period and was much higher at some sites
than at others, corroborating that the species uses roar-barks to announce territorial
ownership and defense. Maned wolves vocalize more around the waxing gibbous lunar
phase, and after dusk until mid-night, following the seasonal variation in sunset time.
Moonlight likely reduces foraging time, resulting in more time available to invest in
acoustic signaling for communication, while vocalizations early on the onset of activity
suggest a territorial announcement function similar to bird dawn chorus. Group
101
vocalizations follow the seasonal variation of the general vocal activity, but not always
the lunar or nightly vocal activity pattern. This suggests that social events may require:
response independent of illumination and hour, as territorial contests; and/or
simultaneous location of animals, as mate guarding and joint territorial defense. Based
on spatial patterns, we estimate between 6 and 11 individuals contributed to the
recordings, and found the vocal activity varies among sites and nights as results of a
spatial-temporal dynamic that still needs to be further explored.
Resumo
O monitoramento acústico passivo tem grande potencial para ajudar esforços
para conservação e elucidar o comportamento e ecologia de espécies evasivas noturno-
crepusculares, como o lobo-guará. Aqui nós caracterizamos os padrões sazonais, lunares
e nictemerais nas vocalizações de longo alcance (aulidos) de lobos-guará de vida livre
no Parque Nacional da Serra da Canastra (MG, Brasil) através de oito meses de
gravações ao logo de dois anos com uma rede de 12/13 gravadores autônomos. Nós
descobrimos que é possível identificar a estação de acasalamento, e possivelmente o
período em torno do parto, através de um aumento na atividade vocal. Esses picos de
atividade vocal sugerem que os aulidos têm um papel na atração e guarda de parceiros e
também na comunicação intra grupo familiar. Contudo, houve atividade vocal durante
todo o período amostral e muito mais em alguns locais do que em outros, corroborando
a hipótese de que a espécie usa aulidos para anúncio e defesa territorial. Lobos-guará
vocalizam mais na Lua crescente gibosa e depois do anoitecer até meia-noite, seguindo
a variação sazonal no pôr do Sol. A luz da Lua provavelmente reduz o tempo de
forrageio, resultando em mais tempo disponível para investir na sinalização acústica
102
para comunicação, enquanto vocalizações logo no início da atividade sugerem uma
função de anúncio territorial. Vocalizações de grupo seguiram a variação sazonal da
atividade vocal geral, mas nem sempre o padrão lunar ou noturno de atividade vocal.
Essas vocalizações em grupo podem ocorrer devido a eventos sociais que requeiram:
resposta independente de iluminação ou hora, como disputas territoriais; e/ou
localização simultânea dos animais, como para guarda de parceiro e defesa territorial
conjunta. Baseado nos padrões espaciais, nós estimamos entre 6 e 11 animais
contribuíram para as gravações, e descobrimos que a atividade vocal varia entre os
locais e noites como resultado de uma dinâmica espaço-temporal que ainda precisa ser
explorada.
Key-words: maned wolf, Chrysocyon brachyurus, passive acoustic, vocalization,
seasonal, moon, diel.
103
Introduction
The majority of mammals are nocturnal, crepuscular or cathemeral (Jones et al.
2009), with nocturnality assumed to be the ancestral condition (Gerkema at al. 2013).
Nocturnality favors communication modalities other than vision, such as chemical and
acoustic (Fox 1975). Most mammals are also solitary (Lukas & Clutton-Brock 2013),
which implies that an important part of their social interaction is mediated by long range
signals to maintain spacing among individuals (Kleiman 1972; Morton 1977).
For those reasons, passive acoustic monitoring of terrestrial mammals has great
potential in aiding conservation efforts and in elucidating their behavior and ecology
(Blumstein et al. 2011). In fact, many mammals are more easily heard than seen at
distance (Marques et al. 2013). Besides allowing monitoring of species hard to visually
follow, passive acoustic monitoring has the advantage of enabling behavioral sampling
over large temporal and spatial scales (Van Parijs et al. 2009; Blumstein et al. 2011).
This feature is crucial for investigating seasonal patterns in animal behavior. For
instance, some species can have the breeding season tracked by increased number of
vocalizations, as most cervids, e.g. the red deer (Clutton-Brock & Albon 1979; Bocci &
Laiolo 2013), and in birds, as the critically endangered Araripe Manakin (Girão &
Souto 2005).
The lunar cycle is also an important driver of the activity of nocturnal animals
(Kronfeld-Schor et al. 2013), and variations over this cycle can also be acoustically
investigated. For instance, coyotes emit more group vocalizations on new moon nights,
when territorial pressure is higher and/or pack coordination for hunting large prey is
required (Bender et al. 1996). As another example, eagle owls are more active and
vocalize more on bright moonlit nights and during twilight (Penteriani et al. 2009),
104
although they also have less success in capturing prey during those periods (Penteriani
et al. 2011). A meta-analysis found that the majority of terrestrial mammals reduce their
activity around full moon nights, probably as a predator avoidance strategy and as an
energy conserving response to lower prey availability (Prugh & Golden 2014).
Information on the relationship between terrestrial mammals’ locomotion/physical
activity and vocal activity over the moon cycles is lacking (except for the echolocation
of bats, which sometimes is used as a proxy for activity per se; Hecker & Brigham
1999; Kronfeld-Schor et al. 2013).
Finally, the diel vocalization pattern is also frequently investigated and can
reveal times better suited for acoustic communication. One of the most prominent
examples of such times are the dawn and dusk choruses, the latter being especially
relevant for crepuscular avian species. There are over 12 non-exclusive hypotheses for
this phenomenon (Staicer et al. 1996), including that this is a period of better acoustic
propagation (Brown & Handford 2003). Another explanation states that those would be
moments when individual quality would be honestly advertised due to the energetic
constraint of having passed through a period without food (Zahavi 1975; Montgomerie
1985; Cuthill & Macdonald 1990). Accordingly, nocturnal birds vocalize more during
the dusk chorus than in the dawn chorus (Hardouin et al. 2008). Yet another explanation
points that dawn and dusk are not ideal for foraging but are good to communicate while
avoiding predators (Berg et al. 2006). For nocturnal animals those are the more
illuminated periods of their activity, improving visual acuity (Prugh & Golden 2014).
This is the case of the afore mentioned eagle owl, which couples visual and vocal
displays on dim light (Penteriani et al. 2009).
The South American maned wolf (Chrysocyon brachyurus, Illiger 1815) is a
promising species for acoustic monitoring. They are large (70-90 cm, 20-30 kg; Silveira
105
1999), but very shy and cryptic, crepuscular/nocturnal canids (Mello et al. 2007).
Monogamous breeding pairs share and defend the same extensive home range (15-115
km2; Rodrigues 2002; Azevedo 2008), yet rarely meet, living largely solitary lives
(Dietz 1984). They forage for small vertebrates and fruits (Rodden et al. 2004), and thus
conspecific presence may interfere (Jácomo et al. 2009). All of which makes them hard
to observe and visually follow in the wild (personal [non] observation), even when
tagged with GPS collars (Emmons, 2012). Hence, many aspects of their behavior are
poorly understood (Rodden et al. 2004). To our benefit, however, they communicate
throughout the year with a long-range explosive call (the roar-bark, Kleiman 1972;
same as “extended bark”, Balieiro & Monticelli 2019). Roar-barks are emitted in
sequences (bouts) of 5-15 roar-barks separated by 2-4 seconds (see spectrograms in next
session) and are proposed to function as: territorial announcement, especially for intra-
sexual spacing; partner attraction and guarding; and intra-group (partner and offspring)
communication (Kleiman 1972; Brady 1981; Dietz 1984; Bestelmeyer 2000; Sábato
2011; Emmons 2012; Rocha et al. 2016; Balieiro & Monticelli 2019). Animals from the
same group or from adjacent territories can exchange vocalizations, creating a group
sequence of alternating roar-barks (Dietz 1984; Emmons 2012).
Our goal was to characterize the maned wolf seasonal, lunar, and diel long-range
acoustic communication pattern at a protected Brazilian area with passive audio
recordings of eight months over two years. Our first specific objective was to test if the
vocalization increase registered in captivity during the mating season (Sábato 2011)
could be identified on natural recordings. Our second specific objective was to confirm
the lunar and diel patterns suggested by our previous small sample record (32 nights)
that suggested that maned wolves vocalize more on the waxing gibbous phase and
between 18h and 19h (Rocha et al. 2016). Our third specific objective was to test if
106
group vocalizations followed the seasonal, lunar, and diel patterns of solo vocalizations
or if they varied independently as in coyotes (Bender et al. 1996). Finally, we wanted to
detect spatio-temporal patterns that could contribute evidence of roar-bark functions,
and to help in estimating the number of animals recorded.
107
Material and Methods
1. Study área
The study was conducted at Serra da Canastra National Park, Minas Gerais state,
Brazil (Figure 1). The park is mainly composed of Cerrado open savannas with a cold,
dry season (April-September) and a hot, rainy season (October-March; Queirolo &
Motta-Junior 2007). Maned wolf density at the park is considered high (0.08
individuals/km2; Paula et al. 2013).
Figure 1. Study region at the Serra da Canastra National Park, MG/Brazil. Yellow squares indicate
autonomous recorder (SongMeter SM2+) sites used only in 2014, pink triangles sites used only in 2016,
and white circles sites used in both years.
2. Recordings
Recordings were made with autonomous recorders (SongMeter SM2+; Wildlife
Acoustics, Inc., Concord, Massachusetts) coupled with a single SMX-II weatherproof
microphone each (Wildlife Acoustics, Inc.). Autonomous recorders were programed to
108
record continuously for 12 hours each night, partitioning samples in 30 minutes files,
with an +36dB gain, 8 kHz sample rate, and 16-bit wav coding. Recorders were
attached to 1.4m high wooden stakes and distributed in areas where tracks, scats, and
reported observations of maned wolves had been made. We aimed to sample high
elevation sites (1.373±56.65 m; all measures are reported in mean±SD, unless noted
otherwise) and the broadest distribution possible, but accessibility was a major
constraint.
In 2014 we deployed 12 autonomous recorders, with the linear distance between
them of 2.27 km (±0.72 km). Recorders remained active between April 05 to August 08,
recording from 18h to 06h. In 2016, we deployed 13 autonomous recorders (8 were the
same from 2014), with the linear distance between them of 3.03 km (±0.78 km).
Recorders remained active between March 09 to July 01, recording from 17h to 05h.
Due to a technical problem the equipment failed to record between March 29 and April
03 2016.
3. Audio processing and measures
The most recent vocal repertoire of the maned wolf (Sábato 2011) describes 10
different vocalization types (only for adult wolves, see Brady 1981 for vocalizations
during development). From those, the long-range types described are the roar-bark and
the similar, but slightly shorter and higher in frequency, single bark. During our pilot
study we manually searched for any maned wolf vocalization and the only type found
was the roar-bark. Therefore, the present study was focused in this vocalization type.
Roar-barks were detected automatically using XBAT (Extensible Bioacoustic
Tool; Figueroa 2007) extension for Matlab (R2010a version; MathWorks, Inc., Natick,
MA, USA) following the methodology detailed by Rocha et al. (2015). In summary,
spectrograms are scanned with a mobile cross correlation using 4 roar-barks templates.
109
Matches above a threshold (0.21) are then manually verified for false positives and
undetected roar-barks within 24 seconds of the detected ones (Figure 2a). We used 4
different templates, including different frequency portions, and a very low threshold to
guarantee that even very faint, partially masked, or uncommon frequency shaped roar-
barks were detected, at the cost of increasing the number of false positives. In our test
data, this methodology resulted in 100% of roar-bark sequences being detected in half
the processing time. This method yielded even more detections than found by manually
scanning spectrograms (93%; Rocha et al. 2015).
We considered a single sequence when roar-barks were not separated by more
than 10 seconds (based in Bender et al. 1996, and preliminary observation of the data).
Sequences with a single roar-bark are possible. For each roar-bark sequence we noted
the recording site, date, absolute start time (17h to 06h), start time in relation to sunset
(± 0-12h), number of roar-barks, and the number of animals vocalizing at the same
moment (solo x group sequences, detailed below).
Sunset times used to calculate the start time in relation to sunset were extracted
from https://www.sunearthtools.com/pt/solar/sunrise-sunset-calendar.php, which allows
for GPS location specification.
The number of vocalizing animals was detected inspecting Raven’s pro 1.5
spectrograms (Bioacoustics Research Program, 2014. Ithaca, NY: The Cornell Lab of
Ornithology. http://www.birds.cornell.edu/raven. Configurations: grayscale, 50%
brightness and contrast, 50% overlap, 512 points Hann window, smoothing active;
Figure 2). The presence of a second, and very rarely a third, animal could be verified by
differences in the spectral shape of roar-barks, cadence, intensity, and eventual overlaps.
When two or more animals intercalated roar-barks on the same sequence we termed it a
group vocalization (Figure 2b). When only one animal could be detected on the
110
sequence, we termed it a solo vocalization (Figure 2a). General vocal activity refers to
both number of sequences and roar-barks, with no separation between solo and group
vocalizations.
In some cases, the same sequence was recorded on more than one Song Meter.
This was verified by temporal proximity and comparison of inter roar-bark intervals,
guarantying that it was in fact the same sequence in two different sensors and not two
animals vocalizing at the same moment. Only the most intense sequence (measured with
Raven’s peak power function) was counted for the analysis. The autonomous recorders’
clocks were not exactly synchronized, therefore triangulation of the emitter position
based on the time difference in the signal arrival at different sensors was not possible.
111
Fig
ure 2
. Wild
man
ed w
olf ro
ar-bark
sequ
ences reco
rded
passiv
ely w
ith a g
rid o
f 12/1
3 au
tono
mou
s record
ers (So
ng
Meter S
M2
+) at S
erra da C
anastra N
ation
al Park
, MG
/Brazil. a) so
lo
roar-b
ark seq
uen
ce with
detectio
n b
ox
es from
XB
AT
(Fig
uero
a 200
7) ex
tensio
n fo
r Matlab
(Math
Wo
rks, In
c.). Sp
ectrog
ram p
arameters: 5
12
win
do
w size, H
ann
win
dow
, 100
% b
righ
tness,
43%
con
trast, 50%
ov
erlap o
n an
8 k
Hz sam
plin
g rate w
av file
. b) g
roup
vo
calization
consistin
g o
f two
anim
als alternatin
g ro
ar-bark
s. Sp
ectrog
ram m
ade o
n R
aven
pro
1.5
(Co
rnell
Bio
acoustics L
ab, Ith
aca, NY
, US
A), H
ann
win
do
w, 5
12
win
do
w size, 5
0%
brig
htn
ess and
contrast, 5
0%
ov
erlap, sm
ooth
ing
on
, on
an 8
kH
z samplin
g rate w
av file
112
The total area recorded by Song Meters – in the sense of roar-bark detectability
– was estimated in two ways: first a ‘lower estimate’ by using half the mean distance
between recorders as the radius for each recorder (we used 2014 mean recorders’
distance for both year so values would be comparable); and second a ‘maximum
estimate’ by using half the distance between the most distant recorders that ever
registered the same sequence as the radius for each recorder (overlapping areas were
discounted).
We mapped the vocal activity distribution over the months according to recorder
sites, identifying regions with concentrated activity. This was done to verify if space
and roar-bark use were associated, which would indicate a territorial or resource defense
function for this vocalization. We also wanted to see if the vocal activity during the
mating season occurred at the same sites as during other periods.
We also used the spatial information to estimate the number of animals recorded.
We calculated the radius (5.36 km) of the mean home range (90.29 km2) from a recent
study in the area (Paula 2016) and looked for roar-bark sequences on sites farther away
than this value within 1-3 consecutive nights. Those sequences were considered having
been emitted by different animals, or group of animals (in the cases of group
vocalizations). We estimated the number of animals in two other ways: by dividing the
mean number of roar-bark sequences per night by the mean number of sequences
emitted per individual per night in captivity (Sábato 2011; 0.68 proestrous, 0.28
anestrous); and by multiplying our estimate of total area recorded by the maned wolf
density in the studied park (0.08/km2; Paula et al. 2013).
113
4. Statistical analyses:
The monthly vocal activity was compared with ANOVAs followed by Tukey
contrasts on R software (R version 3.5.1 [2018-07-02], The R Foundation for Statistical
Computing). For the circular data, moon phases and hours of the day, we used
Rayleigh’s test on Oriana software (version 4; Kovach Computing Services, Anglesey,
Wales; https://www.kovcomp.co.uk/oriana/). In an effort to eliminate the seasonal
variation as a confounding effect, we standardized the vocal activity by each lunar
month and then compared moon phases with ANOVAs.
114
Results
1. Roar-bark calling general characterization
We detected a total of 13180 roar-barks distributed in 1210 sequences over the
233 nights of recording (2014 and 2016). There were fewer sequences and roar-barks in
2014 than in 2016 (Table 1 shows only the vocal activity of entire months). Few
sequences were emitted by more than one animal (12%), and there were more group
vocalizations in 2014 than 2016 (Table 1). On five occasions the group vocalization
involved three animals (3.3% of group vocalizations). All other group vocalizations
involved only two animals (96.7%, 145 sequences).
Overall the mean number of roar-barks per sequence was 10.89 (± 8.07 SD). The
sequences were shorter in 2014 than in 2016 (Table 1 and Figure 3). Sequences with a
single roar-bark in general preceded or followed other sequences (60% within 1
minute). The longest sequence (highest number of roar-barks) of 2014 involved two
animals and had 91 roar-barks, followed 10 seconds later by a 56 roar-bark sequence
and right after that 6 more sequences summing 31 roar-barks. In 2016 the longest
sequence had 50 roar-barks uttered by a single animal.
Table 1. Summary of maned wolf’s vocal activity recorded passively with a grid of 12/13 autonomous
recorders (Song Meter SM2+) at Serra da Canastra National Park, MG/Brazil. Year Total March April May June July
Sequences of roar-barks 2014 478 - 193 86 118 81
2016 708 229 184 117 178 -
Number of roar-barks 2014 4748 - 1950 819 1253 726
2016 8177 2655 2246 1306 1970 -
Group vocalizations 2014 79 - 26 16 21 16
2016 65 25 14 11 15 -
% of group vocalizations 2014 16.5 - 13.5 18.6 17.8 19.8
2016 9.2 10.9 7.6 9.4 8.4 -
115
Figure 3. Histogram of the number of roar-barks on each sequence of maned wolves’ vocalizations
recorded passively with a grid of 12/13 autonomous recorders (Song Meter SM2+) at Serra da Canastra
National Park, MG/Brazil. One sequence was defined by one or a bout of roar-barks not separated by
more than 10 seconds.
In 2014, 23.1% of roar-bark sequences were detected by more than one
autonomous recorder, with 8% of those being detected by three or more (up to five)
recorders. In 2016 we obtained the same percentage of sequences detected by more than
one recorder (23.9%), but only 3.2% of those were detected by three recorders and none
in four or more.
In 32 cases, group sequences were registered in more than two recorders. In all
but two of those events, the time difference between the same vocalization of each
individual differed by 0.5-8.8 seconds between sensors. Therefore, the relative position
of each emitter suggest animals were between 86 m and 1517 m away from each other
(considering a sound speed of 343 m/s). Those were the only situations in which we
could estimate distances (and not from the recorder).
2. Seasonal variation
The seasonal fluctuation in maned wolf vocal activity was very similar in both
years (Figure 4). During March and April, the vocal activity was at its maximum,
dropping in mid-April and remaining low until the beginning of June. There was a
second smaller increase in vocal activity in June, dropping to the previous levels again
116
at the end of the month. There was no visible seasonal pattern on the number of roar-
barks by sequence (the sequence size), neither on the percentage of group vocalizations.
The number of group vocalization in general followed the vocal activity, except for a
marked peak in the middle of June on 2014 (Figure 4, bottom right).
Figure 4. Seasonal variation in the maned wolf vocal activity recorded passively with a grid of 12/13
autonomous recorders (Song Meter SM2+) at Serra da Canastra National Park, MG/Brazil. Each point is
a sum of 5 nights. Photos: Endangered Wolf Center, St.Louis, and Adriano Gambarini.
In 2014 there was a significant monthly variation on the mean number of
sequences by night (F=4.4978, df=3, p=0.005; Table 2). There were more sequences by
night in April than in other months (p<0.05), except for June (p=0.0954). The result was
0
200
400
600
800
1000
1200
09
/03
19
/03
29
/03
10
/04
20
/04
30
/04
10
/05
20
/05
30
/05
09
/06
19
/06
29
/06
09
/07
19
/07
29
/07
Ro
ar-b
arks
date (5 nights sum)
0
2
4
6
8
10
12
14
09
/03
19
/03
29
/03
10
/04
20
/04
30
/04
10
/05
20
/05
30
/05
09
/06
19
/06
29
/06
09
/07
19
/07
29
/07G
rou
p v
oca
liza
tio
ns
date (5 nights sum)
117
the same for the monthly mean number of roar-barks by night (F=7.5311, df=3,
p=0.0001), except that there were also more roar-barks by night in April than June
(p=0.0317). No significant differences were found for the size of the sequences
(F=0.9681, df=3, p=0.4115), number of group vocalizations (F=1.0522, p=0.3724), and
percentage of group vocalizations (F=0.2698, df=3, p=0.847) by night.
In 2016 there were also a significant monthly variation on the mean number of
sequences by night (F=5.4306, df=3, p=0.0016; Table 2). There were more sequences
by night in March than in other months (p<0.05), except for April (p=0.0953). The
results were the same for the monthly number of roar-barks (F=4.9486, df=3, p=0.003).
There was a marginal trend in the monthly number of group vocalizations by night
(F=2.6757, df=3, p=0.051), with March mean higher than May’s (p=0.0461). No
significant difference was found in the size of the sequences (F=0.8459, df=3,
p=0.4725), and percentage of group vocalizations (F=0.2068, df=3, p=0.8915) by night.
Table 2. Maned wolf vocal activity recorded passively with a grid of 12/13 autonomous recorders (Song
Meter SM2+) at Serra da Canastra National Park, MG/Brazil. Values reported are mean by night ± SD. Year Total March April May June July
Sequences of
roar-barks
2014 4.0±5.7 - 7.4±8.7 2.8±3.9 4.0±4.0 2.6±4.6
2016 6.6±7.1 11.5±8.6 6.8±8.9 3.8±4.0 5.9±5.0 -
Number of
roar-barks
2014 39.9±49.0 - 75.0±71.3 26.1±29.5 41.3±39.6 23.6±34.0
2016 73.8±83.4 127.6±100.5 80.0±111.2 42.0±39.1 65.0±56.5 -
Roar-barks /
sequences
2014 10.8±5.7 - 12.7±5.6 10.1±5.4 10.9±4.4 10.1±7.1
2016 11.9±4.7 11.5±3.6 12.0±4.8 13.0±5.9 10.9±3.9 -
Group
vocalizations
2014 0.7±1.3 - 1.0±1.3 0.5±0.8 0.7±1.6 0.5±1.0
2016 0.6±1.2 1.3±1.9 0.5±1.1 0.4±0.8 0.5±0.9 -
% of group
vocalizations
2014 18.5±26.8 - 15.0±18.4 20.9±27.3 17.5±27.0 21.1±3.7
2016 9.4±20.5 8.9±13.1 7.2±15.1 11.9±28.1 9.5±21.6 -
3. Lunar pattern
Overall (2014 and 2016) vocal activity was concentrated on the waxing gibbous
phase (Figure 5). This was true for the number of sequences (122.9±98.4° [mean angle
± Circular Standard Deviation]; Z=63.125, p<0.0001, N=1206), for total number of
118
roar-barks (124.7±99.3°; Z=623.101, p<0.0001, N=12556), and also for the number of
group vocalizations (114.2°±111.8°; Z=3.325, p=0.036, N=150).
In 2014 the roar-bark sequences were significantly concentrated in the waxing
gibbous phase (Figure 5; Z=31.225, p<0.0001, N=501), with a mean angle of 158.9° (±
95.5°). The total number of roar-barks followed the same pattern (152.6±101.1°
[waxing gibbous]; Z=221.87, p<0.0001, N=4989). The number of group vocalization
was not significantly concentrated in any moon phase (Z=2.336, p=0.097, N=87).
2014
New
1st quarter
Full
3rd quarter60 60
60
60
50 50
50
50
40 40
40
40
30 30
30
30
20 20
20
20
10 10
10
10
2016
New
1st quarter
Full
3rd quarter60 60
60
60
50 50
50
50
40 40
40
40
30 30
30
30
20 20
20
20
10 10
10
10
data
New
1st quarter
Full
3rd quarter100 100
100
100
80 80
80
80
60 60
60
60
40 40
40
40
20 20
20
20
119
Figure 5. Maned wolf roar-bark sequences distribution over the lunar phases (gray = total). Records were
made from April to July on 2014 (blue) and from March to June on 2016 (red) with a grid of 12/13
autonomous recorders (Song Meter SM2+) at Serra da Canastra National Park, MG/Brazil. The radial line
represents the mean angle and the concentric bar at the end of the line the 95% confidence interval.
In 2016 there were significantly more roar-bark sequences during the waxing
crescent and waxing gibbous phases (Figure 5), with a mean angle of 95.3° (±90.9° [1st
quarter]; Z=57.097, p<0.0001, N=707). The total number of roar-barks followed a
similar pattern, with a slightly greater angle (mean 104.6±93.4° [waxing gibbous];
Z=530.156, p<0.0001, N=7578). The number of group vocalizations was concentrated
during the waxing crescent phase (mean 74.4±92.6°; Z=530.156, p=0.01, N=63).
Higher vocal activity during March and April could have biased the moon
concentration results, therefore we tested each moon cycle separately (Table 3). Of
eight cycles recorded on both years, for three the number of roar-bark sequences was
concentrated during the waxing gibbous phase, two during the waxing crescent, one
during the waning gibbous, and two were not concentrated on any moon phase. Results
were similar for the number of roar-barks, and only in March and April 2016 the
number of group vocalizations was concentrated on the waxing gibbous and waxing
crescent phases, respectively (Table 3).
After standardizing vocal activity by each lunar month, the mean values seemed
to decrease from the waxing crescent (e.g. number of roar-barks: 0.732) to the new
moon phase (e.g. number of roar-barks: -0.512). However, the difference was not
significant for any variable (ANOVAs; number of sequences: p=0.3714; number of
roar-barks: p=0.0809; group vocalizations: p=0.4055).
120
Table 3. Concentration of maned wolf vocal activity on each of eight moon cycles recorded passively
with a grid of 12/13 autonomous recorders (Song Meter SM2+) at Serra da Canastra National Park,
MG/Brazil. When the concentration is significant, the mean moon phase, mean angle ± circular standard
deviation, and Rayleigh test statistics are reported.
Lunar cycle Sequences Roar-barks Group vocalizations
2014 April
4/5 – 5/4
Waxing gibbous
147.3±74.6°
Z=36.72, df=199, p<0.0001
Waxing gibbous
145.9±77.7°
Z=321.103, df=2019, p<0.0001
-
2014 May
(5/5 – 6/2) - - -
2014 June
(6/3 – 7/2)
Waning gibbous
193.0±108.3°
Z=5.564, df=122, p=0.004
Full
178.9±110.8°
Z=30.225, df=1269, p<0.0001
-
2014 July
(7/3 – 7/31) -
Waning crescent
313.7±94.0°
Z=45.021, df=665, p<0.0001
-
2016 March
(3/9 – 4/7)
Waxing gibbous
116.3±77.7°
Z=41.109, df=258, p<0.0001
Waxing gibbous
125.9±76.1°
Z=499.409, df=2915, p<0.0001
Waxing gibbous
106.9±47.8°
Z=12.958, df=25, p<0.0001
2016 April
(4/8 – 5/6)
Waxing crescent
44.3±77.5°
Z=30.472, df=189, p<0.0001
Waxing crescent
43.8±76.9°
Z=298.245, df=1809, p<0.0001
Waxing crescent
26.6±62.8°
Z=4.811, df=15, p=0.006
2016 May
(5/7 – 6/5)
Waxing crescent
43.1±102.4°
Z=3.978, df=96, p=0.019
Waxing crescent
33.6±108.6°
Z=30.738, df=1116, p<0.0001
-
2016 June
(6/6 – 6/30)
Waxing gibbous
135.3±81.6°
Z=21.146, df=160, p<0.0001
Waxing gibbous
144.1±83.7°
Z=205.843, df=1734, p<0.0001
-
4. Nightly pattern
Vocal activity during the first two recording hours (17-19h) revealed a seasonal
variation correlated with sunset (Figure 6). During March and April, when sunset was
after or close to 18h, the vocal activity was lower between 17-18h and higher between
18-19h. The inverse pattern is seen when sunset was before 18h, in May and June. In
July the sunset starts to get later again and the vocal activity between 18-19h starts to
increase again. Because of this variation we decided to report the nightly vocalization
pattern as time relative to sunset (Figure 7).
121
Figure 6. Maned Wolf roar-barks registered between 17-19h on passive audio recordings made with a
grid of 12/13 autonomous recorders (Song Meter SM2+) at Serra da Canastra National Park, MG/Brazil.
In 2014 there was no recordings in March, and in 2016 no recordings in July.
In both years the vocal activity was concentrated in the first half of the night
(Figure 7). In 2014 the first hour after sunset alone comprised 18% of roar-barks, and
the first 3 hours 44%. In 2016 there was a moderate vocal activity already during the
hour preceding sunset, and this level doubled during the hour after. Although there was
a tendency for vocal activity to decrease throughout the night, on 2016 it continued
relatively high until 5-6h after sunset.
Group vocalizations did not always follow vocal activity (number of sequences
and roar-barks). In 2014 there was an increase in group vocalizations after 7 hours after
sunset despite the overall low vocal activity on this period. That resulted in a higher
percentage of group vocalization between 7 and 12 hours after sunset in 2014 (Figure
7). In 2016, while the vocal activity was high between 2 and 4 hours after sunset, the
number of group vocalizations decreased markedly during this period (Figure 8). On 3
occasions a peak in the vocal activity was followed by a peak in group vocalizations in
the next hour (Figure 7): in 2014 6-7 hours (sequences) and 7-8 hours (group
vocalizations), in 2016 0-1 hours and 1-2 hours, and on 2016 3-4 hours and 4-5 hours.
0
100
200
300
400
500
March[18:17:14]
April[17:53:06]
May[17:37:14]
June[17:34:31]
July[17:43:01]
Ro
ar-b
arks
Month [mean sunset time]
17-18h 2016
18-19h 2016
18-19h 2014
122
Figure 7. Maned wolf nightly vocal activity relative to sunset. Recordings were made with a grid of
12/13 autonomous recorders (Song Meter SM2+) at Serra da Canastra National Park, MG/Brazil.
The mean hour of roar bark sequence start for the two combined years was 22:02
(±03:16 [mean hour ± Circular Standard Deviation]). In 2014 the mean hour of
sequence emission was 22:14 (±03:27; Z=220.97, p<0.0001, N=503), and in 2016 21:51
(±03:05; Z=317.911, p<0.0001, N=614). Some caution with the Rayleigh test is needed
in this case, as the recording period of the two years differed in one hour and on both
years half of the 24h cycle was not sampled (only night records). Despite this bias, the
result showing a concentration of vocal activity in the first half of the night is robust.
0
200
400
600
800
1000
1200
-1h 0h
0h +1h
+1h +2h
+2h +3h
+3h +4h
+4h +5h
+5h +6h
+6h +7h
+7h +8h
+8h +9h
+9h +10h
+10h +11h
+11h +12h
Ro
ar-b
arks
2014
2016
02468
10121416
-1h
0
h
0h
+1
h
+1h
+2
h
+2h
+3
h
+3h
+4
h
+4h
+5
h
+5h
+6
h
+6h
+7
h
+7h
+8
h
+8h
+9
h
+9h
+1
0h
+10
h
+11
h
+11
h
+12
h
Gro
up
vo
cali
zati
on
s
Time relative to sunset
0%
5%
10%
15%
20%
25%
30%
-1h
0
h
0h
+1
h
+1h
+2
h
+2h
+3
h
+3h
+4
h
+4h
+5
h
+5h
+6
h
+6h
+7
h
+7h
+8
h
+8h
+9
h
+9h
+1
0h
+10
h
+11
h
+11
h
+12
h
% o
f gr
ou
p v
oca
liza
tio
n
Time relative to sunset
123
5. Spatial patterns and number of recorded animals
The mean distance between Song Meters in 2014 was 2.27 km, giving a 1.14 km
radius for the total area recorded ‘low estimate’. The most distant recorders to register
the same sequence were 4.86 km away (K and N), giving a radius for the total area
recorded ‘maximun estimate’ of 2.43 km. That resulted in a estimated area covered of
48.97 to 139.90 km2 (low to maximun) for 2014 and of 53.05 to 195.25 km2 for 2016.
The spatial distribution of vocal activity revealed 4 main regions with
concetrated activity in 2014: C/D sites, H/J sites, K/L/M sites, and O/P sites (Figure 8
and supplementary Figure S1). There was high vocal activity at the sites between C/D
and H/I, making it unclear if this larger area was the home range of two animals that
moved around, three animals with one moving between the two areas, or four (or
more?) different animals. There were many group vocalizations in all those regions
(Figure 8).
In April 2014 there was high and widespread vocal activity. In May the vocal
activity reduced and becomes more concetrated at the eastern portion of the park. In
June 2014 the vocal activity increased again and was concentrated at 3 points (C, H, and
O). Finally, on July 2014 the vocal activity was low and concentrated on the western
and eastern sides.
124
Figure 8. Maned wolf roar-bark sequences recorded passively with 12 autonomous recorders (Song
Meter SM2+) at Serra da Canastra National Park, MG/Brazil (gray contour). Circles have approximately
0.5 km radius with the center point being the recorder site. Heat colors represent the intensity of vocal
activity (number of sequences). Letters indicates the site name and numbers following them on the circles
indicates the amount of group vocalizations. * Indicates at least one sequence involved 3 animals
(otherwise group vocalizations involve 2 animals).
Based on the vocal activity of distant (> 5.36 km) sites in consecutive 1-3 nights,
we estimate tha between 6 and 9 different animals were recorded during 2014 (example
in Figue S1 top). Based on Sábato (2011) captivity report of mean sequences by animal
by night (0.68 proestrous, 0.28 anestrous) and our mean sequences by night (Table 1
and 2), our estimative would be 10.88 (April) and 11.17 (May-July) recorded animals in
2014. Based on our estimmates of recorded area and the reported maned wolf density in
this park (0.08/km2; Paula et al. 2013) the number of animals would range from 3.92 to
11.19 individuals (low and maximum area estimate).
125
The spatial distribution of vocal activity revealed 5 not well separated regions of
concentrated activity in 2016: A/B sites, C/D/F sites, H/I sites, J site, and M/N sites
(Figure 9 and supplementary Figure S1b). Compared to 2014, the K/L/M focus seem
to have shifted south and the O/P focus seem to have become much less active (only
recorder P is present on 2016). It was not clear if the activity on sites J and G were from
animals of near by sites (I,K and E,H), or from different animals. Although this year had
fewer group vocalizations than 2014, there were still group vocalizations on most
regions (Figure 9).
Figure 9. Maned wolf roar-bark sequences recorded passively with 13 autonomous recorders (Song
Meter SM2+) at Serra da Canastra National Park, MG/Brazil (gray contour). Circles have approximately
0.5 km radius with the center point being the recorder site. Heat colors represent the intensity of vocal
activity (number of sequences). Letters indicates the site name and numbers following them indicates the
amount of group vocalizations (sequences involving 2 animals).
126
In 2016 the vocal activity on sites C, F, H, and J was high during all recorded
months. This included a single night (from 17:06 to 00:24) in April with 41 sequences,
totaling 510 roar-barks, involving at least two animals. From the vocal activity on this
night, it seems there were an individual near site C and one or two near site J. After
some hours of the J animal/s vocalizing the animal from C comes between H and J and
they utter group vocalizations (only after 23:45).
Based on the vocal activity of distant sites in consecutive 1-3 nights, we estimate
that between 7 and 11 different animals were recorded during 2016 (example in Figue
S1 bottom). Based on the captivity report of mean sequences by animal by night and our
mean sequences by night (Table 1 and 2), our estimative would be 13.20 (March-April)
and 17.28 (May-June) recorded animals in 2016. Based on our estimmates of recorded
area and the reported maned wolf density in this park the number of animals would
range from 4.24 to 15.62 individuals (low and maximum area estimate).
127
Discussion
We passively recorded spontaneous long-distance calls (roar-barks) of maned
wolves during eight months over two years in a protected park with the aim of
elucidating the maned wolf’s long-range acoustic behavior. We found that there is an
increase in vocal activity during March and April, indicating that it is possible to
identify the species mating period in the wild by the higher number of roar-barks. We
also found a second smaller increase in June, coinciding with the peak period of
parturition, which suggest that this reproduction period can also be monitored
acoustically. We found that maned wolves vocalize more on more moon illuminated
nights, especially during the waxing gibbous lunar phase. They concentrate their
vocalizations from dusk to mid-night, adjusting to the seasonal variation in sunset time.
Group vocalizations (detected vocal activity coming from 2 or more animals) did follow
the seasonal variation of the general vocal activity, but not always the lunar or nightly
vocal activity pattern, suggesting that environmental cues are less important than social
interactions. Vocal activity was much higher at some recording sites than at others over
all months, corroborating that the species uses roar-barks as territory announcement and
defense besides reproductive related purposes. Based on the vocal activity and group
vocalizations on consecutive nights, we that estimate between 6 and 11 different
animals contributed to recordings.
The peak in vocal activity in March and April coincides with the reported mating
season for the species (Carvalho & Vasconcellos 1995; Rodden et al. 2004). Female
maned wolves are monoestral and stay fertile for only 5 days (Rodden et al. 2004). In
this scenario, vocalizations could play an important role for normally solitary wolves to
quickly locate their potential mate or partner on an extensive home range. Reports from
captivity describe the frequency of vocalizations increase weeks prior to estrous up to
128
the end of the mating season (Brady 1981; Dietz 1984; Sabato 2011). A report from the
wild describes that pairs stay together during one or two nights, copulating often, and
foraging close to each other (Rodden et al. 2004). This report showed that animals
emitted roar-barks whenever the partner was out of sight (Rodden et al. 2004). Our
findings are in accordance with those reports and reinforce the role of roar-barks in
partner attraction and possibly mate guarding.
We also found an increase in vocal activity in June, approximately two months
after the mating period, which coincides with the species’ gestation period (65 days;
Carvalho & Vasconcellos 1995), and the reported peak in births both in captivity (Maia
& Gouveia 2002; Rodden et al. 2004) and for the Serra da Canastra park (Dietz 1984;
de Mello et al. 2007, 2009). We have confirmation that at least one female on the area
was lactating in July 2014 and 2016 (R. C. de Paula, personal communication; Annex
I). The increase in vocalizations was smaller and, although clear in the 5-night sum of
both years (Figure 4), was not detected on the statistics of monthly rate of vocalization
by night (Table 2). Probably the time scale used, an entire month, was too long for the
event to capture these sudden and more isolated increases in calling behavior.
Mating has been reported up to June (Dietz 1984), which raises the possibility
that the second peak observed in June was related to other pairs (different than the ones
of the first peak) vocalizing near estrous, instead of related to vocal activity around
parturition as proposed above. However, this second vocal activity peak happened at
similar sites from those during the first vocal peak in March and April, which suggests
it involves the same individuals, unless the first breeding pair has lost its territory (e.g.,
due to reproductive failure). Maned wolves enter estrous only once per year, pairs do
not mate twice in the same year. Thus, we argue the second increase in vocal activity is
likely related to the birth of pups.
129
GPS tracking results indicate that males reduce their activity and stay closer to
females on the days around parturition (de Mello et al 2007; Emmons 2012). The male
may rest all daylight hours with the female and pup, and at night visit the den regularly
(de Mello et al. 2007, 2009). Extensively in captivity and occasionally in the wild,
males maned wolves have been observed to guard, defend, regurgitate, and carry food to
the mother and pups, and also groom, play with, and accompany their young (Dietz
1984; Carvalho & Vasconcellos 1995; Bestelmeyer 2000; Rodrigues 2002; Jácomo et
al. 2009).
Emmons (2012) reports maned wolves vocalize more when pups are present and
suggest an intra group (pair and offspring) communication function. However, for the
reported cases pups/juveniles were older (3-15 months) and had already left the den
(Emmons 2012). Only Dietz (1984) reports roar-barks around parturition, which
suggests a more prominent role of this vocalization in intra-pair communication. The
use of roar-barks for intra-pair vocal communication is poorly discussed, but we
speculate that around parturition while pups are still dependent of parental care acoustic
communication may mediate family coordination, e.g. signal the location of females
and their den to males, since females are known to shift pups’ location often (Dietz
1984; Bestelmeyer 2000), or mediate negotiation/manipulation of parental care
(Wachtmeister 2001). Another possible cause would be an increased urge for
announcing and defending the territory, as other wolves, pumas, and feral dogs could be
a threat to the pups (Dietz 1984; de Mello et al 2007).
While maned wolves have been shown to decrease movement around full moon
nights (Sábato et al. 2009), here we found they vocalize more on bright nights. This
highlights the importance of coupling the investigation of activity levels and
vocalization levels, as they may not fluctuate together. The most accepted explanation
130
of decreased activity for mammal predators is a strategy to minimize energy loss due to
reduced prey availability (Prugh & Golden 2014). Alternatively, there may be an
increase in foraging efficiency if: the increase in the ability of maned wolves to visually
detect preys surpass both the low prey availability and the increase in the ability of
preys to detect the maned wolves (Prugh & Golden 2014); and/or they have a gain in
fruit detectability. It should be noted that periodicity of spatial movement was not
correlated with the lunar cycle on maned wolves of the same region (Péron et al. 2017),
indicating they do not change habits. In either case, i.e. energy saving or increased
foraging efficiency, would result in less time foraging and therefore more time to
announce territory and interact with conspecifics, increasing vocalization levels.
Other explanation for the lunar pattern observed is that roar-barks reveal the
emitter position to prey and other predators. This could be particularly detrimental in
darker nights compared to brighter nights, when the emitter is already more easily
detected independent of sound and can itself detect threats more easily (as poorwills:
Woods & Brigham 2008). The problem with this line of thought is that maned wolves
vocalize too rarely to seriously impair hunting or expose themselves to risks (mean 0.68
sequences/individual/night: Sábato 2011; 4-7 sequences/night/12-13 recorders: Table
2). Yet another hypothesis would be that coupled visual and vocal communication is
important for the species, and they would take advantage of better iluminated periods to
increase those displays (as eagle owls: Penteriani et al. 2011). However, maned wolves
show only a small concentration of vocalizations on the minutes of dusk twilight
(30.6% x 18.3% chance on the ±1h sunset; data not shown), and not at all on the dawn
twilight minutes (8.9% x 18.3% chance -1h sunrise; data not shown). Besides, in 2014,
vocalizations during the moonlit versus dark portion of the night were not different than
expected by chance (Duarte et al. unpublished; Appendix I). Finally, the wolves’ typical
131
habitat has many tall grass/bushes and most of the time wolves are far enough that they
would not see each other even in bright daylight (Jácomo et al. 2009).
On the other hand, we found more vocalizations during the waxing gibbous
phase, but also during the waxing crescent and the waning gibbous (with an angle near
full). For those moon phases the first half of the night is illuminated, and this period is
when they vocalize more. That fact corroborates the idea that light has a direct influence
on the species vocalization, and that deserves further investigation. It also indicates that
it is best to describe maned wolves vocalize less on darker nights than the other way
around.
To announce the territory ownership right at the start of the activity may be
especially important as other wolves will also start moving around at this time and may
decide to trespass or not (also suggested by Rocha et al. 2016). This is also the most
constraining hour for communication, as they haven’t feed over the day. In this situation
the advertisement of the individual quality should be honest (Zahavi 1975; Cuthill &
Macdonald 1990). This will influence even monogamous territorial species, as, beside
mate guarding, territorial intrusions can be presumably avoided by advertising body
condition (Cuthill & Macdonald 1990; Hardouin et al. 2008). Territory pressures may
be high, and there are several reports of resident maned wolves extending their home
range shortly after a neighbor death/disappearance (Dietz 1984; Rodrigues 2002;
Jácomo et al. 2009).
Conversely, the peak in vocal activity on the first hour of the night in 2016 was
not accompanied by an increase in group vocalizations. Those interactions only appear
latter, sometimes one hour after a previous increase in vocal activity (Figure 7). This
can indicate the first roar-bark display is used to announce territory, before others can
trespass. Later the partner can respond, or another wolf may dispute the area, creating
132
group vocalizations. Also corroborating this idea, maned wolves only responded to
playbacks between 17-19:40h (Ferreira et al. unpublished; Chapter 1).
Although peaks in activity (Jácomo et al. 2004; Melo et al. 2007) and
vocalizations (Brady 1981; Balieiro & Monticelli 2019) have been reported on both
dusk and dawn, here we registered very few roar-bark sequences preceding dawn (05-
06h of 2014). Maybe there is a second peak in vocalizations after dawn that we did not
record. However, 24h recordings in the same park, including periods when roar-bark
playbacks were conducted before and after dusk, did not reveal any vocalization around
dawn or morning (8 days of continuous recordings and playbacks: Chapter 1; 5
recorders, 13 days of continuous recordings in 2015: Ferreira et al. unpublished).
Group vocalizations occur in a variety of contexts, including interactions
between neighbors, breeding pairs, and parent offspring (Kleiman 1972; Brady 1981;
Dietz 1984; Emmons 2012; Balieiro & Monticelli 2019). In captivity group
vocalizations are common and roar-barks of one wolf often induce vocalizations of
other wolves (Sábato 2011). Emmons (2012) reports a particularly familiar group of
maned wolves (5) that responded sequences to each other 31% of times on average.
In our study, there were more group vocalizations on 2014 than in 2016, when
recorders were spaced over a larger area (49-140 km2 x 53-195 km2 estimated recorded
area). This suggests we could have failed to detect the second animal on some
sequences classified as solo in 2016. However, considering Emmons and our own
personal observations of the data, we still think those are low percentages to believe all
sequences heard by another maned wolf are actually answered. For instance: on several
occasions, sequences appearing to come from different individuals were separated by 5-
20 minutes, a time interval short enough to assume the second wolf heard the first wolf
but choose not to intercalate roar-barks with it. Secondly: on a playback experiment
133
(Chapter 1), wolves never intercalated the response with the broadcasted roar-barks,
even during an interactive playback (but intercalated entire sequences; Ferreira et al.
unpublished). This suggest wolves intentionally choose to intercalate roar-barks on
some occasions but not on others. Our results suggest solo and group vocalizations may
have slightly different functions as they do not always fluctuate together.
We hypothesize group vocalizations happen more often when the response to a
conspecific vocalization must be immediate, as territorial defense (versus passive
territory ownership announcement), and/or when pair members must be located
simultaneously, as for mate guard and joint territorial/offspring defense. Accordingly,
group vocalizations increased in the mating period, and in 2014 near parturition, when
those situations would happen more often. On the other hand, group vocalizations did
not follow the lunar cycle (except for, and probably biased by, the mating season of
2016), neither the diel solo vocalization pattern (especially on 2016). Those facts
indicate events of social interaction not related to breeding occur at any time and must
be addressed independent of illumination and hour. As a final remark on this topic,
animals emitting group vocalizations were still several meters apart, that is, vocal
interactions do not mean close physical interaction.
Further studies on solo and group vocalizations are needed to clarify if they have
indeed distinct functions, which was not the focus of this work. As a recommendation,
clearly separating sequences with and without intercalating roar-barks would help in
this matter, something that has not been done previously (Kleiman 1972; Brady 1981;
Dietz 1984; Balieiro & Monticelli 2019).
Our estimates of number of animals that contributed to the recordings ranged
from 4 to 17. While we are certain that at least 4 different animals were recorded (in
March 20, 2016 there were group vocalizations on both extremes of the park), we think
134
17 animals is an overestimate. Data from telemetry tracking show that long straight
trajectories that are bigger than their territories are uncommon (Paula 2016), thus it is
likely that we have recorded 6 individuals because on March 22, 2016 (two days after
detection of group vocalizations on the limits of the park) there was another group
vocalization in the middle of the park.
The higher estimate of 17 individuals was obtained based on captivity data of
mean roar-bark sequences by individual by night (Sábato 2011). Interestingly, we
imagined animals in captivity would vocalize more than those in the wild, as they are
closer and roar-barks from one individual stimulate responses of other individuals,
which would result in lower estimative for the wild. Therefore, maned wolves appear to
vocalize less in captivity, maybe as a combined result of being habituated to each
other’s roar-bark sequences, favoring close-range communication types as they are in
proximity, and/or a reduced pressure to defend a territory/mate.
Nevertheless, based on telemetry data obtained in our study area (Paula 2016)
we estimate that 5 ranges could have been recorded, considering generally ranges do not
change drastically (Dietz 1984; Emmons 2012). Each range would have a breeding pair,
and occasionally up to 3 juveniles or adult offspring (maximum group size reported:
Azevedo 2008, Emmons 2012). Additionally, transient animals may also vocalize and
contribute to the maximum estimate, so it would be possible that 17 different animals
were recorded.
As conclusion, maned wolves use roar-barks in many contexts, and more in the
mating season, on better moonlit nights (or less on darker ones), at the onset of their
activity in the first part of the night. All of these findings support the role of this
vocalization in territorial announcement and defense, partner attraction and guarding,
and intra-group communication. This multi functionality attests that maned wolves
135
interact in complex ways and more frequently than previously thought. Additionally, we
were able to estimate the number of vocalizing individuals in a way useful to monitor
populations and detect large scale fluctuations in the number of individuals at a low cost
(passive acoustic monitoring). Those fluctuations would indicate serious environmental
problems requiring urgent decision making and action.
136
References
AZEVEDO, F. C. 2008. Home range and spatial organization of maned wolf
(Chrysocyon brachyurus) in the Serra da Canastra National Park region, Minas Gerais,
Brazil. M.S. thesis, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil.
BALIEIRO, F., & MONTICELLI, P.F. 2019. Evidence of individual discrimination in
the maned wolf long-distance extended-bark. Behavioural processes, 158, 219-227.
BENDER, D.J., BAYNE, E.M., & BRIGHAM, R.M. 1996. Lunar condition influences
coyote (Canis latrans) howling. American Midland Naturalist, 413-417.
BERG, K.S., BRUMFIELD, R.T., & APANIUS, V. 2006. Phylogenetic and ecological
determinants of the neotropical dawn chorus. Proceedings of the Royal Society of
London B: Biological Sciences, 273(1589), 999-1005.
BESTELMEYER, S.V. 2000. Solitary, reproductive, and parental behavior of maned
wolves (Chrysocyon brachyurus). PhD Thesis, Colorado State University, Fort Collins,
U.S.A.
BLUMSTEIN, D.T., MENNILL, D.J., CLEMINS, P., GIROD, L., YAO, K.,
PATRICELLI, G., ... & HANSER, S.F. 2011. Acoustic monitoring in terrestrial
environments using microphone arrays: applications, technological considerations and
prospectus. Journal of Applied Ecology, 48(3), 758-767.
BOCCI, A., TELFORD, M., & LAIOLO, P. 2013. Determinants of the acoustic
behaviour of red deer during breeding in a wild alpine population, and implications for
species survey. Ethology Ecology & Evolution, 25(1), 52-69.
BRADY, C.A. 1981. The vocal repertoires of the bush dog (Speothos venaticus), crab-
eating fox (Cerdocyon thous), and maned wolf (Chrysocyon brachyurus). Animal
Behaviour, 29(3), 649-669.
BROWN, T.J., & HANDFORD, P. 2003. Why birds sing at dawn: the role of consistent
song transmission. Ibis, 145(1), 120-129.
CARVALHO, C.T.D., & VASCONCELLOS, L.E. 1995. Disease, food and
reproduction of the maned wolf: Chrysocyon brachyurus (Illiger)(Carnivora, Canidae)
in southeast Brazil. Revista Brasileira de Zoologia, 12(3), 627-640.
CLUTTON-BROCK, T.H., & ALBON, S.D. 1979. The roaring of red deer and the
evolution of honest advertisement. Behaviour, 145-170.
CUTHILL, I.C., & MACDONALD, W.A. 1990. Experimental manipulation of the
dawn and dusk chorus in the blackbird Turdus merula. Behavioral Ecology and
Sociobiology, 26(3), 209-216.
DIETZ, J.M. 1984. Ecology and social organization of the maned wolf (Chrysocyon
brachyurus). Smithsonian Contribution to Zoology. 1-51.
EMMONS, L.H. 2012. The maned wolves of Noel Kempff Mercado National Park.
Smithsonian Contributions to Zoology 639:1–135.
137
FIGUEROA, H. 2007. Extensible BioAcoustic Tool. XBAT R7. Cornell Bioacoustics
Research Program, Ithaca, New York.
FOX, M. W. 1975. The wild canids: their systematics, behavioral ecology and
evolution. R.E. Krieger.
GERKEMA, M.P., DAVIES, W.I., FOSTER, R.G., MENAKER, M., & HUT, R.A.
2013. The nocturnal bottleneck and the evolution of activity patterns in
mammals. Proceedings of the Royal Society B: Biological Sciences, 280(1765),
20130508.
GIRÃO, W., & SOUTO, A. 2005. Breeding period of Araripe Manakin Antilophia
bokermanni inferred from vocalization activity. Cotinga, 24, 35-37.
HARDOUIN, L.A., ROBERT, D., & BRETAGNOLLE, V. 2008. A dusk chorus effect
in a nocturnal bird: support for mate and rival assessment functions. Behavioral ecology
and sociobiology, 62(12), 1909.
HECKER, K.R., & BRIGHAM, R.M. 1999. Does moonlight change vertical
stratification of activity by forest-dwelling insectivorous bats?. Journal of
Mammalogy, 80(4), 1196-1201.
JÁCOMO, A.T., KASHIVAKURA, C.K., FERRO, C., FURTADO, M.M., ASTETE,
S.P., TÔRRES, N.M., ... & SILVEIRA, L. 2009. Home range and spatial organization
of maned wolves in the Brazilian grasslands. Journal of Mammalogy, 90(1), 150-157.
KLEIMAN, D.G. 1972. Social behavior of the maned wolf (Chrysocyon brachyurus)
and bush dog (Speothos venaticus): a study in contrast. Journal of Mammalogy, 53(4),
791-806.
KRONFELD-SCHOR, N., DOMINONI, D., IGLESIA, H., LEVY, O., HERZOG, E.D.,
DAYAN, T., & HELFRICH-FORSTER, C. 2013. Chronobiology by
moonlight. Proceedings of the Royal Society B: Biological Sciences, 280(1765),
20123088.
LUKAS, D., & CLUTTON-BROCK, T.H. 2013. The evolution of social monogamy in
mammals. Science, 341(6145), 526-530.
MAIA, O.B., & GOUVEIA, A. M.G. 2002. Birth and mortality of maned wolves
Chrysocyon brachyurus (Illiger, 1811) in captivity. Brazilian Journal of Biology, 62(1),
25-32.
MARQUES, T.A., THOMAS, L., MARTIN, S.W., MELLINGER, D.K., WARD, J.A.,
MORETTI, D.J., ... & TYACK, P.L. 2013. Estimating animal population density using
passive acoustics. Biological Reviews, 88(2), 287-309.
MELO, L.F., LIMA SÁBATO, M.A., VAZ MAGNI, E.M., YOUNG, R.J., &
COELHO, C.M. 2007. Secret lives of maned wolves (Chrysocyon brachyurus Illiger
1815): as revealed by GPS tracking collars. Journal of Zoology, 271(1), 27–36.
MELO, L.F.B., SÁBATO, M.A.L., VAZ MAGNI, E.M., YOUNG, R.J., & COELHO,
C.M. 2009. First observations of nest attendance behavior by wild maned wolves,
138
Chrysocyon brachyurus. Zoo Biology: Published in affiliation with the American Zoo
and Aquarium Association, 28(1), 69-74.
MONTGOMERIE, R.D. 1985. Why do birds sing at dawn. spoken papers, 19th
International Ethology Congress, 1, 242.
MORTON, E.S. 1977. On the occurrence and significance of motivation-structural rules
in some bird and mammal sounds. The American Naturalist, 111(981), 855-869.
PAULA, R. C., RODRIGUES, F. H. G., QUEIROLO, D., JORGE, R. P. S., LEMOS, F.
G., & DE ALMEIDA RODRIGUES, L. 2013. Avaliação do risco de extinção do lobo-
guará Chrysocyon brachyurus (Illiger, 1815) no Brasil. Biodiversidade Brasileira, (1),
146-159.
PAULA, R. C. D. 2016. Adequabilidade ambiental dos biomas brasileiros à ocorrência
do lobo-guará (Chrysocyon brachyurus) e efeitos da composição da paisagem em sua
ecologia espacial, atividade e movimentação. Doctoral dissertation, Universidade de
São Paulo, São Paulo, Brazil.
PENTERIANI, V., & DEL MAR DELGADO, M. 2009. The dusk chorus from an owl
perspective: eagle owls vocalize when their white throat badge contrasts most. PLoS
One, 4(4), e4960.
PENTERIANI, V., FERRER, M., & DELGADO, M.D. M. 2011. Floater strategies and
dynamics in birds, and their importance in conservation biology: towards an
understanding of nonbreeders in avian populations. Animal Conservation, 14(3), 233-
241.
PÉRON, G., FLEMING, C.H., DE PAULA, R.C., & CALABRESE, J.M. 2016.
Uncovering periodic patterns of space use in animal tracking data with periodograms,
including a new algorithm for the Lomb-Scargle periodogram and improved
randomization tests. Movement ecology, 4(1), 19.
PRUGH, L.R., & GOLDEN, C.D. 2014. Does moonlight increase predation risk?
Meta‐analysis reveals divergent responses of nocturnal mammals to lunar
cycles. Journal of Animal Ecology, 83(2), 504-514.
QUEIROLO, D., AND J. C. MOTTA-JUNIOR. 2007. Prey availability and diet of
maned wolf in Serra da Canastra National Park, southeastern Brazil. Acta Theriologica
52:391–402.
ROCHA, L.H., FERREIRA, L.S., PAULA, B.C., RODRIGUES, F.H., & SOUSA-
LIMA, R.S. 2015. An evaluation of manual and automated methods for detecting
sounds of maned wolves (Chrysocyon brachyurus Illiger 1815). Bioacoustics, 24(2),
185-198.
ROCHA, L.H., FERREIRA, L.S., VENTICINQUE, E.M., RODRIGUES, F.H., &
SOUSA-LIMA, R.S. 2016. Temporal and environmental influences on long-distance
calling by free-ranging maned wolves (Chrysocyon brachyurus). Journal of
Mammalogy, 98(1), 302-311.
139
RODDEN, M., RODRIGUES F., & BESTELMEYER, S. 2004. Maned wolf
(Chrysocyon brachyurus). Canids: Foxes, Wolves, Jackals and Dogs, Status Survey and
Conservation Action Plan. 1st ed. IUCN Canid Specialist Group, Cambridge, United
Kingdom 38–43.
RODRIGUES, F.H.G. 2002. Biologia e conservação do lobo-guará na Estação
Ecológica de Águas Emendadas, DF. Ph.D Thesis, Universidade Estadual de
Campinas, Campinas, Brazil.
SÁBATO, M.A.L., MELO, L.F.B., MAGNI, E.M.V., YOUNG, R.J., & COELHO,
C.M. 2006. A note on the effect of the full moon on the activity of wild maned wolves,
Chrysocyon brachyurus. Behavioural processes, 73(2), 228-230.
SÁBATO, V. 2011. Aspectos do comportamento acústico do lobo-guará Chrysocyon
brachyurus (Illiger 1815).
SILVEIRA L. 1999. Ecologia e conservação da comunidade de carnívoros do Parque
Nacional das Emas, GO. M.S. Thesis, Universidade Federal de Goiás, Goiânia, Goiás,
Brazil.
STAICER, C.A., SPECTOR, D.A., & HORN, A.G. (1996). The dawn chorus and other
diel patterns in acoustic signaling. Ecology and evolution of acoustic communication in
birds, 426-453.
VAN PARIJS, S. M., CLARK, C. W., SOUSA-LIMA, R. S., PARKS, S. E., RANKIN,
S., RISCH, D., & VAN OPZEELAND, I. C. 2009. Management and research
applications of real-time and archival passive acoustic sensors over varying temporal
and spatial scales. Marine Ecology Progress Series, 395, 21-36.
WACHTMEISTER, C.A. 2001. Display in monogamous pairs: a review of empirical
data and evolutionary explanations. Animal Behaviour, 61(5), 861-868.
WOODS, C.P., & BRIGHAM, R.M. 2008. Common poorwill activity and calling
behavior in relation to moonlight and predation. The Wilson Journal of
Ornithology, 120(3), 505-512.
ZAHAVI, A. 1975. Mate selection—a selection for a handicap. Journal of theoretical
Biology, 53(1), 205-214.
140
Supplementary material
Figure S1. Nightly maned wolf number of roar-barks recorded passively with 12/13 autonomous
recorders (Song Meter SM2+) at Serra da Canastra National Park, MG/Brazil (gray contour), in
2014/2016. From top-left to bottom-right: April 05/06/07/08 2014, April 15/16/17/18 2014, March
19/20/21/22/23 2016, and May 01/02/03/04 2016. **indicates at least one sequence of roar-barks
involved two animals. *** indicates at least one sequence of roar-barks involved three animals.
141
Chapter 4
Identity and sex discrimination of roar-barks for
captive and free-ranging maned wolves
142
Identity and sex discrimination of roar-barks for captive and free-
ranging maned wolves
Ferreira, L.S., Sábato V.R., Baumgarten, J., Rodrigues, F.H., Sousa-Lima, R.S.
Intended for submission on: Biological Conservation
Abstract
Although vocal identity is a widespread trait, it is rarely tested for application in
natural conditions. Such methodology could aid in the understanding and conservation
of elusive nocturnal/crepuscular species, especially those whose populations are
predicted to decline, as are maned wolf populations. Here we recorded captive maned
wolves and found that their long-range call, the roar-bark, is individually and sexually
distinct. Roar-barks could be correctly assigned to individuals 72.6% of times, and to
sex 78.6%. The roar-bark duration and the concentration of energy in lower frequencies
were the most important parameters. However, when roar-barks were experimentally
broadcasted on natural habitat most parameters were not stable, even at distances as
short as 160 m. The few stable parameters were mediocre in discriminating among
individuals (42.5% success x 13.0% chance). Site characteristics, as vegetation and
relief, and individual differences, were more influential than distance, suggesting some
identity information may be ranging far given the appropriate conditions. Unfortunately,
we also found that in passive records of spontaneous roar-barks sequences of free-
ranging maned wolves the variation in parameters due to propagation is larger than the
individual differences, which seriously compromises the applicability of vocal
identification of individuals of the species on natural conditions.
143
Resumo
Apesar da identidade vocal ser um traço bastante difundido em animais, ela é
raramente testada para aplicação em condições naturais. Tal metodologia poderia ajudar
na compreensão e conservação de espécies elusivas noturno-crepusculares,
especialmente aquelas em que há previsão de declínio populacional, como o lobo-guará.
Neste trabalho nós gravamos lobos-guará em cativeiro e descobrimos que seus
chamados de longa distância, os aulidos, são individual e sexualmente distintos. Aulidos
puderam ser corretamente designados aos indivíduos em 72,6% dos casos, e aos sexos
78,6%. A duração do aulido e a concentração de energia em frequências baixas foram os
parâmetros mais importantes. Entretanto, quando os aulidos foram experimentalmente
propagados em ambiente natural, a maioria dos parâmetros variaram, mesmo em
distâncias curtas (160 m). Os poucos parâmetros que se mantiveram estáveis foram
capazes de discriminar entre indivíduos, mas com sucesso medíocre (42,5% x 13,0% ao
acaso). Características locais, como vegetação, e diferenças individuais, foram mais
influentes do que a distância, sugerindo que alguma informação de identidade pode ter
longo alcance em condições propícias. Infelizmente, nós também descobrimos que em
gravações passivas de sequências de aulidos espontâneas de lobos-guará em vida livre a
variação nos parâmetros causada pela propagação é maior do que as diferenças
individuais identificadas, o que compromete seriamente a aplicação da identificação
vocal individual da espécie em condições naturais.
Key-words: vocal identity, Chrysocyon brachyrurus, propagation, passive monitoring
144
Introduction
Maned wolves (Chrysocyon brachiurus, Illiguer 1815) are an exception among
large canids (70-90cm, 20-30kg; Silveira 1999). They forage alone for fruits and small
vertebrates and rarely interact with conspecifics (Dietz 1984), despite the similar sized
family members being highly social (Kleiman & Eisenberg 1973; Mohelman 1987,
1989). The monogamous mated pair share the same extensive home range (15-115 km2;
Rodrigues 2002; Azevedo 2008), being seen together almost exclusively in the breeding
season (Dietz 1984). Females are monoestral and the litter size is small (captivity
average is 3; Maia & Gouveia 2002). Both sexes provide parental care (Bestelmeyer
2000; Mello et al. 2009) and the young stay in their natal range for 1 or more years
before dispersing (Rodden et al. 2004; Emmons 2012). The species has mainly
nocturnal/crepuscular habits (4pm-8am; Paula 2016) and delimit/announce their
territory through urine, scats (Bestelmeyer 2000; Rodden et al. 2004), and long-range
calls (the roar-bark; Kleiman 1972; Rocha et al. 2016).
More extensive studies on the behavior ecology of this peculiar species could
help the understanding of evolutionary patterns on the Canidae family (Mohelman
1987, 1989). Additionally, the species’ numbers are predicted to decline 30% over 3
generations (21 years) due to habitat loss alone (Paula et al. 2008) and additional
populational estimates are needed (Paula et al. 2013). Unfortunately, many of the
characteristics cited above, coupled with the fact that maned wolves are shy, occur in
low densities, in difficult terrains (Melo et al. 2007; Trolle et al.2007), makes finding
and monitoring the species a challenge. This is a common scenario for many terrestrial
mammals, and therefore, methods for identifying individuals in the wild without need of
visual confirmation would be an extremely useful tool and would allow for less invasive
studies.
145
For many species long range vocalizations are ideal for use in identification
(Terry et al. 2005). The presence of identity information in vocalizations, referred to as
vocal signature, is a well-documented phenomenon in birds (e.g. Falls 1982), primates
(e.g. Marler & Hobbett 1975), ungulates (e.g. Reby et al. 2001), cetaceans (e.g. Sayighet
al.1998), proboscids (e.g. McComb et al. 2003), rodents (e.g. Kober et al. 2008), canids
(e.g. Durbin 1998; Darden et al. 2003; Mitchell et al. 2006), and many others, indicating
this is a widespread trait (Falls 1982). Some studies have additionally investigated sex
and age cues on vocalizations, for instance in manatees (Sousa-Lima et al. 2002, 2008)
and dingos (Déaux et al. 2016a).
While vocal signatures are a widespread trait, it does not mean all species show
them, or are capable of detecting or use those clues to recognize conspecifics (Terry et
al. 2005). Vocal discrimination in general evolves in situations in which failing to
identify the emitter – offspring, partner, competitor – generates a loss in the receptor
fitness, and, simultaneously, not being correctly identified generates a loss in the emitter
fitness (Bradbury & Vehrencamp 1998). Those situations normally involve: the risk of
misdirecting parental care, common in colonial species, as penguins (Jouventinet al.
1999), bats (Fanis & Jones 1996), and pinnipeds (Charrier & Harcourt 2006) and in
species that invest highly in parental care, such as manatees (Sousa-Lima et al. 2001,
2009). The importance of discriminating established neighbors from trespassing
vagrants to optimize the energetical investment in territorial defense is also dependent
on individual recognition, as the observed in artic foxes (Frommolt et al. 2003). Besides
artic foxes, vocal discrimination by canids has been demonstrated in gray wolves
(Palacios et al. 2015), domestic dogs (Molnáret al. 2009), and dingos (Déaux et al.
2016b).
146
Therefore, vocal signatures and individual discrimination and recognition in
maned wolves may have evolved to maintain pups close to parents and hidden from
strangers, juveniles inside their parents’ range, to optimize the energetic investment in
territorial defense, or to find a mate or reunite a previously established pair during the
breading season. Long ago, Brady (1981) reported that a human would be able to
discriminate individual maned wolves from 1 kilometer, and, recently, it has been
demonstrated that captive maned wolves discriminate between roar-barks of familiar
versus unfamiliar individuals (Balieiro & Monticelli 2019). Although neither work
quantified individual characteristics of the roar-bark (but see Sabato 2011), they suggest
maned wolves have vocal signatures that could be used by researchers to identify
individuals.
Despite the potential for the use of vocal signature on animal research and
conservation, very few studies in fact apply the vocal identity after its description
(Gilbert et al. 2002; Delport et al. 2002), a critic highlighted by Terry et al. (2005).
From all studies of vocal signature in canids, only 3 uses data from natural habitat
besides or instead of captivity (Frommolt et al. 2003; Hartwig 2005; Root-Guteridge et
al. 2014b), and none apply identification protocols in vocalizations of unknown
emitters.
The main problem in the classification of vocalizations recorded from natural
habitat is the degradation of acoustic parameters used in the discrimination when the
sound propagates through long distances (Mitchell et al. 2006). Frommolt et al. (2003) e
Hartwig (2005) results stem from data acquired at small distances, between 20 and 100
meters. However, in the work of Root-Guteridge et al. (2014b) recordings were of
unknown distance, with low signal-to-noise ratios, and, still, discriminant functions
attained 100% of correct classification of individual grey wolves.
147
Here we used a combination of maned wolf captivity recordings, field sound
propagation experiment re-recordings, and recordings from wild animals, to (i) test the
individual distinctiveness of roar-barks, (ii) test the sexual distinctiveness of roar-barks,
(iii) test the acoustic distinctive parameters’ stability over distance, and (iv) test the
applicability of vocal identification on natural habitat recordings.
We previously used a grid of autonomous recorders to passively monitor maned
wolves, detecting their presence by roar-barks (Rocha et al. 2016; see Chapter 3).
Therefore, we were interested in knowing if a vocal identification of maned wolves
could be applied in natural habitat recordings of this kind. Roar-barks recorded this way
rarely have a high signal-to-noise ratio and the emitter identity, distance, and location in
the habitat are unknown. Thus, for an acoustic parameter to be reliably used to
distinguish between individuals, its value must remain unchanged (or change less than
within individuals) at distance and in different habitats. If parameter values change the
identity classification will change for the same individual depending on where it is
calling from, making long-term monitoring impossible. Also, vocal identification must
apply to typical low signal-to-noise ratio recordings, otherwise it will only be useful for
a small amount of fortuitous high-quality recordings. For those reasons we conducted
the roar-bark propagation experiment so we could test how much the acoustic parameter
values of known individuals would change over controlled distances and habitats. The
final goal was to test the resulting vocal identification method on a 8-month passive
acoustic monitoring dataset at the Serra da Canastra National Park (MG/BR).
148
Material and Methods
1. Data acquisition in Captivity
We recorded 10 sexually mature maned wolves (Table 1, Figure 1a), from two
facilities in Minas Gerais state: Criadouro Científico de Fauna Silvestre para Fins de
Conservação da Companhia Brasileira de Metalurgia e Mineração (CC-CBMM), and
Zoológico da Associação Esportiva e Recreativa dos Funcionários das Usinas
Siderúrgicas de Minas Gerais (ZOO-USIPA). Recordings on CC-CBMM were made
from April 17 to June 03 2010, during the species breeding period (proestrous was
confirmed by swabs), and on ZOO-USIPA from November 12 to 27 2010, which
correspond to the non-breeding period (anestrous). Enclosures had between 1500 and
6000 m2 on CC-CBMM, and between 175 and 185m2 on ZOO-USIPA. All recordings
were made between 17:30h and 7:00h, from 5 to 200 meters away from the animals.
Table 1. Maned wolves recorded on 2010 at two facilities in Minas Gerais, Brazil. *estimated age. m# are
non-participant males. m3 is GA/GI half-brother.
ID Sex Age Facility Origin Relatedness Roar-barks
Recorded Selected
SH male 3y9m* CC-CBMM Nature ? 827 20
FI female 3y6m* CC-CBMM Nature ? 354 20
RO female 13y8m* CC-CBMM Nature JU/NE mother 148 20
JU female 9y6m CC-CBMM Captivity m1+RO offspring 353 20
NE male 5y6m CC-CBMM Captivity m2+RO offspring 297 20
SA female 6y4m CC-CBMM Captivity m3+JU offspring 33 20
GA male 5y6m ZOO-USIPA Captivity GI litter mate 137 20
GI male 5y6m ZOO-USIPA Captivity GA litter mate 29 20
LU female 12y ZOO-USIPA Captivity ? 73 20
BA female 4y* ZOO-USIPA Nature ? 70 20
Total 2321 200
Roar-barks were recorded with a unidirectional microphone Sennheiser K6
coupled with a Sennheiser ME-66 module (40-20000 Hz ± 2.5 dB flat response
frequency), protected by a windscreen, and connected to a Marantz PMD-661 solid state
149
recorder using 96 kHz sampling rate and 24-bit wav coding format. Recordings were
monitored with a headset Sony MDR-7506 and the gain was manually set to maximize
the signal-to-noise ratio while avoiding clipping.
Using R program (R version 3.5.1 [2018-07-02], The R Foundation for
Statistical Computing) we randomly selected 20 good quality non-overlapping roar-
barks for each of the 10 animals, totaling 200 roar-barks from captivity.
2. Propagation experiment
We selected 5 good quality roar-barks of each of 4 individuals (GA, SH, JU, SA)
that were recorded close (5-8 m) and that we judged encompassed the variation
observed in captivity (Figure 1b). Broadcasts were done at the Serra da Canastra
National park, representing a typical maned wolf habitat (Dietz 1984), between 18:44
and 19:41h, the period of highest vocal activity for the species (Rocha et al. 2016), from
March 4 to 9, 2017. All 20 roar-barks were broadcasted in 4 different sites at the park,
resulting in a total of 80 playbacks. The most prominent features of sites were: flat
terrain on open bare soil road (“Flat” site; -20.22401 -46.55861 [decimal degrees
WGS84]); broadcast from a lower to a higher position on open bare soil road (“Low-to-
high” site; -20.26104 -46.42372); 20 m from Low-to-high but on a tall grass and shrubs
area (“Vegetation” site; see General Introduction Figure C); and broadcast from a higher
to a lower position on a sinuous bare soil road (“High-to-low” site; -20.25403 -
46.4188).
We used an Acer AspireOne notebook to broadcast the sounds using Raven Pro
1.5 software (Bioacoustics Research Program, 2014. Ithaca, NY: The Cornell Lab of
Ornithology; http://www.birds.cornell.edu/raven) and a Pioneer S-DJ50X speaker (class
A/B Bi-amp, 80 W output, 50-20000 Hz frequency range) 86 cm above the ground to
150
simulate the height of a maned wolf. The sound level for the playback was calibrated to
be similar to a natural emission based on captivity recordings of known distance.
Broadcasted roar-barks were re-recorded simultaneously by 7 autonomous
recorders (Song Meter SM2+; Wildlife Acoustics, Inc., Concord, Massachusetts) with
one omnidirectional weatherproof microphone each (SMX-II; Wildlife Acoustics, Inc.;
sensitivity -36±4dB [0dB=1V/pa@1kHz]; 20-20000 Hz flat response frequency).
Recorders were set on the road side in a single direction from the speaker positioned at
distances of 1.25 m, 20 m, 40 m, 80 m, 160 m, 320 m, and 640 m. Distances were
measured using a measuring tape (1.25 to 80 m) and a GPS (Garmin GPSMAP® 76S;
accuracy < 15 m). The autonomous recorders were attached to stakes of the same height
of the speaker (86 cm) with the omnidirectional microphone in a perpendicular position
in relation to the speaker. Recordings were made continuously, with an +36dB gain, 8
kHz sample rate, and 16-bit wav files (same configurations used for recordings of
spontaneous vocalizations in the wild).
At some sites the re-recorded roar-bark quality sharply decreased after 160 m or
there were no detectable roar-bark. We used all re-recordings of the 80 broadcasted
roar-barks in which we could distinguish the two first bands (around 250-1000 Hz,
detailed below).
3. Data acquisition in maned wolves’ natural habitat
We recorded for a total of 233 nights, from April 05 to August 08 2014 and from
March 09 to July 01 2016, with a grid of autonomous recorders (SongMeters SM2+) at
the Serra da Canastra National Park, MG/Brazil. Autonomous recorders were
programed to record continuously for 12 hours each night, partitioning samples in 30
minutes files, with an +36dB gain, 8 kHz sample rate, and 16-bit wav coding (same
configurations as Rocha et al. 2015).
151
In 2014, 12 recorders were active from 18:00h to 06:00h with a linear distance
between them of 2.27 ± 0.72 km, and in 2016, 13 recorders were active from 17:00h to
05:00h with a linear distance between then of 3.03 ±0.78 km.
Roar-barks were detected automatically using XBAT (Extensible Bioacoustic
Tool; Figueroa 2007) extension for Matlab (R2010a version; MathWorks, Inc., Natick,
MA, USA) following the methodology detailed by Rocha et al. (2015). In summary,
spectrograms are scanned with a mobile cross correlation of 4 roar-barks templates.
Matches above a threshold (0.21) are then manually verified for false positives and
undetected roar-barks within 24 seconds of the detected ones.
We inspected the detected roar-bark sequences on Raven’s spectrograms and
noted if there was more than one animal emitting roar-barks on the same sequence
(Figure 1c top), and if the same sequence was recorded by more than one autonomous
recorder (Figure 1c bottom). The presence of a second animal could be verified by
differences in the spectral shape of roar-barks, cadence, intensity, and eventual overlaps.
Sequences recorded by more than one recorder were verified by comparing inter roar-
bark intervals, besides emission time and spectral similarity, to ensure that it was in fact
the same sequence and not two different animals vocalizing at the same moment in two
different recorders.
For the prospect of applying vocal identity we used only sequences with two
animals vocalizing together and those recorded simultaneously by two recorders. We
randomly selected sequences among good quality ones (with the two first frequency
bands visible, see below) and selected up to 5, minimum 2, roar-barks until we had at
least 80 roar-barks of each of the two animals and two recorders.
152
Figure 1. Maned wolves roar-barks recorded in Minas Gerais, Brazil. a. One example of roar-bark from
each of 10 individuals (letters) recorded with unidirectional microphone and a hand recorder in two
captivity facilities. b. Roar-barks of GA and JU broadcasted and re-recorded with autonomous recorders
at 7 different distances from the speaker on site “Flat” at the Serra da Canastra National Park. c. Free-
ranging animals spontaneous roar-bark sequences recorded with autonomous recorders at the same park:
top spectrogram shows some roar-barks (numbers) from a sequence involving two animals (letters);
bottom spectrogram shows the same sequence recorded by another autonomous recorder 2.41 km away,
roar-barks from “b” reach the recorders on different times because animals are at different positions.
Spectrogram parameters: a. 96 kHz sample rate, 3080 window size, Hann window, 55% brightness, 60%
contrast, 24-bit wav; b. and c. 8 kHz sample rate, 512 window size, Hann window, 50% brightness, 60%
contrast, 16-bit wav.
4. Acoustic measurements
Although some individuals’ roar-barks have harmonic-like structure (Figure 1a
SH), often roar-barks are very harsh, noisy, broad-banded (Figure 1a GI), sometimes
varying in structure for the same individual (Figure 1a JU x 1b JU). Sub-harmonics
153
may also appear (Figure 1a LU), generally less intense and incomplete, i.e. not lasting
the entire roar-bark duration. The presence of non-linear phenomenon (Fitch et al. 2002)
also occurred, as SH (Figure 1a) abrupt loss of sub-harmonics and JU (Figure 1b)
abrupt transition from noisy to a more tonal signal. Finally, formants (Fitch 1997)
appear to be present in at least some roar-barks (Figure 1a SA), making the distinction
between harmonics and sub-harmonics unclear.
Those characteristics made it difficult to identify harmonic correspondences and
we opted to use “frequency bands” instead. Using the natural habitat roar-barks as base
(Figure 1c), we chose not to use signal energy below 150 Hz and above 2000 Hz since
these portions of the signal are rarely conserved at distance. Still based on the natural
habitat roar-barks, we identified two frequency bands that frequently were the only
visible portion of the signal at distance (Figure 1c a7), the lower around 250-620 Hz
and the next one above it around 620-1000 Hz. We termed them 1st and 2nd Bands and
made separated selection boxes on Raven Pro 1.5 spectrograms for each, in addition to a
selection box of the entire roar-bark (150-2000 Hz), referred as “Full”.
In those 3 spectrogram selection boxes we measured 41 parameters that we
considered biologically relevant and with potential for coding identity and sex on the
roar-barks (Table S1). The choice was made based on literature on canid that looked for
vocal identity (Darden et al. 2003; Hartwig 2005) and on previous work on maned
wolves (Sabato 2011). We also included some parameters considered “robust
measurements” on Raven’s manual (Charif et al. 2010). For the propagation experiment
we measured the signal-to-noise ratio of re-recorded roar-barks and roar-barks recorded
in the maned wolves’ natural habitat. We used the Full roar-bark in-band power (given
in dB in Raven pro 1.5) as the signal level and the in-band power of an equal dimension
box immediately before the roar-bark as the background noise level. As the autonomous
154
recorders were not calibrated, this signal-to-noise ratio should be regarded as a relative
measure of the signal quality instead of an absolute intensity measure. Detailed
description of the variables can be found on the Raven’s manual (Charif et al. 2010).
For the captivity roar-barks, we made selection boxes on spectrograms with the
following parameters: Hann window, 4096 window size, 50% overlap, 50% brightness,
75% contrast, smoothing “off”, which resulted in pixels of 23.4 Hz x 0.022 s. For the
propagation experiment and natural habitat roar-barks we used the following
parameters: Hann window, 512 window size, 50% overlap, 45-55% brightness, 60%
contrast, smoothing “off”, which resulted in pixels of 15.6 Hz x 0.032 s. For the
propagation experiment and natural habitat, if the Full 1st frequency quartile or Full
peak frequency were on a noise (mostly wind) on lower frequencies, we elevated the
Full selection box inferior limit (150 Hz) until the measured frequency parameters were
placed on the roar-bark.
Before starting to create the selection boxes, three analysts preliminary observed
roar-barks and agreed on the general position of frequency bands and standardized the
scales and size of the spectrogram while making measures.
5. Procedure and statistical analyses
The two analysts (LSF, VS) each created measuring boxes for all 200 captivity-
selected roar-barks. We calculated for all variables the difference between analysts’
measurements, using the Spearman’s correlation (in R program), and the Potential for
Identity Coding (PIC; as in Robisson et al. 1993). Beside the absolute difference
between analysts’ measurements, we also compared the difference with the parameter
overall standard deviation as a way to predict the potential impact on the discrimination
analysis. We looked for parameters with low difference between analysts, low
correlation among them, and high PIC values, as well as biological relevant
155
combinations in our opinion, to select among the original 41 parameters those that
would be included in the identity and sexual discriminant analyses (Table S1). The
number of variables included to create discriminant functions should be smaller than the
number of samples in the smaller group (Tabachnick & Fidell 2001), in our case a
maximum of 9 parameters for identity and 3 for sex.
We choose 8 parameters among the 41 (Table 2) and tested their distribution
with the Shapiro-Wilk test, transforming the non-parametric to normal-like distribution
with the Yeo-Johnson transformation, and them centralized and scaled them with the R
function preprocess{caret}.
We used the permuted Discriminant Function Analysis (pDFA; Mundry &
Sommer 2007) which balances the tendency of traditional DFA to overestimate
discriminability by permuting elements between groups and comparing the
discriminability of random versus real grouping (Mundry & Sommer 2007). We set the
R function written by Roger Mundry (2015 version) to randomly select 100 of the 200
roar-barks, balancing individuals (10 of each), and ran the discriminant analysis for
1000 permutations. For the identity pDFA we used sex as a control factor for
permutation.
We conducted a separate pDFA for the data collected by each analyst, and, as a
further measure of difference between analysts, conducted a pDFA on 100 random
combinations of captivity roar-barks measured by both analysts and took the average
percentage of correctly cross classified individuals. We further used a homogeneous
combination of data from both analyst (i.e. equal number of roar-barks measured by
each analyst for each animal), which is referred here as the “combined dataset”.
156
To evaluate the importance of functions and each parameter on the model we
used the combined dataset to conduct a normal DFA, and observed the cumulative
variance explained by functions and standardized coefficients.
For the sexual discrimination, we used the combined dataset and tested all
combinations of 3 parameters among the 8 we had chosen and analyzed combinations
that were statistically significant on the pDFA (p<0.05). We used Welch’s t-test (for
unequal sample sizes and/or variances) to test if sexes differed on the most recurrent
parameters.
For the propagation experiment each analyst measured half of the data, with
each re-recording file (containing all 20 roar-barks on a specific site and distance) being
randomly assigned to each of the 2 analysts. We conducted a MANOVA, followed by
Bonferroni p-corrected ANOVAs, for all 41 parameters (α=0.0012). We then observed
if the effect size (F) of the distance on the parameters was at least 2 times smaller than
the effect size of the individual (as in Mitchell et al. 2006). The parameters attending
this requisite were termed “Stable parameters”. As we used only 4 animals on the
propagation experiment, we choose 3 parameters among stable ones from the 8 selected
for captivity discrimination. As we had only 2 males and 2 females and based on the
captivity sexual discrimination results (see below), we decided not to conduct further
sexual analysis.
With those 3 parameters we trained a DFA for identity calls using the measures
from 1.25 m of each site and them predicted the identity of the roar-barks for the other
distances. Additionally, we re-conducted the pDFA on the captivity dataset including
only the 3 selected stable parameters to compare the success of discrimination.
157
For the natural habitat data, on sequences involving two animals, we measured
the absolute difference between roar-barks of different animals of the same sequence.
We then tested if the animals’ parameters differed with a one-way t-test of the absolute
difference. We also tested, with Pearson’s correlation, if the size of the difference was
correlated with the absolute difference in signal-to-noise ratio between animals.
For sequences simultaneously registered by more than one recorder, we
measured the difference between parameters from the sound with the higher signal-to-
noise ratio to the one with the lower signal-to-noise ratio. We then tested if the
difference was significant with a paired t-test. We also tested, using Pearson’s
correlation, if the absolute difference was correlated with the absolute difference in the
signal-to-noise ratio among roar-barks.
Finally, to test if the absolute difference between animals was greater than the
absolute difference caused by sound degradation, we compared the two differences with
Welch’s t-test.
158
Results
1. Captivity
All parameters had PIC values above 1, indicating they could be used for
identity discrimination, but none had high values, i.e. above 2 (Robisson et al. 1993;
Table 2 and S1). The parameters that varied most between analysts were related to the
box limits, such as the 2nd Band higher frequency (70.14% relative to the SD), or more
subjective, such as the 2nd Band “true” lower frequency (120.59% relative to the SD)
which we measured manually. Raven’s robust parameters were in fact very similar
between analysts. Among the 8 parameters chosen, none had correlation coefficients
higher than 0.660 (1st and 2nd Band peak frequency).
Table 2. Selected parameters on maned wolves roar-barks (LSF analyst). “Full” measures refer to the
roar-bark from 150 Hz to 2000 Hz, while “First” and “Second” bands refer to portions from lower to
higher frequencies. Means±SD are for all 10 individuals, 20 roar-barks each. PIC = Potential for Identity
Coding (Robisson et al. 1993). (A) = unitless: proportion relative to entire duration. Parameters detailing
can be found on Raven’s manual (Charif et al. 2010).
Parameter Mean±SD Dif. to VSR analyst PIC
Full duration (s) 0.591±0.112 -0.032±0.051 1.817
Full position of 1st time quartile (A) 0.375±0.056 0.013±0.039 1.039
Full position of 3rd time quartile (A) 0.635±0.070 0.025±0.044 1.154
Full 1st frequency quartile (Hz) 609.8±169.7 1.1±7.2 1.997
Full 3rd frequency quartile (Hz) 905.2±220.2 1.4±7.7 1.657
Full average entropy (bits) 3.765±0.584 -0.015±0.056 1.691
1st band peak frequency (Hz) 528.2±60.9 -8.4±59.5 1.627
2nd band peak frequency (Hz) 848.4±139.0 -35.5±113.6 1.641
The pDFA for identity discrimination resulted, for each analyst (LSF/VS), in
74.5%/71.80% of correctly cross classified roar-barks, which is significantly (p=0.001)
higher than expected by chance on the cross classification (17.24%/15.98%). The 100
random combination of both datasets resulted in a mean 72.58% (±0.98%) of correctly
cross classified roar-barks, being always significant above chance (p=0.001).
159
The LDA resulted in 8 discriminant functions, with the first 3 explaining 89% of
the variance (LD1 44.22%, LD2 29.08%, LD3 15.59%; Table S2). The most important
parameters were: for the first function Full duration and Full 1st frequency quartile; for
the second, Full average entropy; and for the third, Full 1st frequency quartile and Full
duration. PIC measure predicted accurately the best discriminant parameters (Table 2).
Figure 2 shows the prediction of identity for the first 3 discriminant functions.
Figure 2. First 3 linear discriminant functions for identity discrimination of 10 captive maned wolves
(colors) roar-barks recorded from two facilities at Minas Gerais, Brazil.
Table 3 shows the mean classification results for all individuals (confusion
matrix). We could not identify any relationship of the errors with the individuals’
relatedness or origin (nature/captivity; see Table 1). All mistakes above 1 roar-bark
(From 10. Values are means) involve non-kin animals (e.g. FI and GI).
160
Table 3. Confusion matrix for the cross-validation classification of 10 captive maned wolves by their
roar-barks. Average from the classification results of two analysts (LSF, VS), each constructed by means
of the results from 1000 randomizations of 100 roar-barks (10 for each individual) from a total of 200.
Extracted by the pDFA R function written by Roger Mundry (2015 version).
Individual BA FI JU LU RO SA GA GI NE SH
BA 8.09 0.33 0.83 0.27 0.42 0.05 0.00 0.00 0.03 0.00
FI 0.05 5.18 0.01 0.00 2.74 0.00 0.12 1.86 0.06 0.00
JU 0.01 0.00 9.30 0.43 0.03 0.02 0.00 0.00 0.23 0.00
LU 0.23 0.00 0.55 9.22 0.00 0.00 0.00 0.00 0.00 0.00
RO 0.41 1.29 0.93 0.00 6.32 0.05 0.11 0.36 0.37 0.18
SA 0.27 0.00 0.08 0.01 0.75 8.48 0.16 0.00 0.18 0.09
GA 0.18 0.29 0.00 0.00 0.23 0.03 6.16 0.11 0.12 2.90
GI 0.17 2.69 0.00 0.00 0.11 0.00 0.00 6.70 0.03 0.31
NE 0.46 0.42 0.14 0.24 0.65 0.21 0.11 0.00 7.71 0.09
SH 0.22 0.65 0.00 0.00 0.18 0.12 2.30 0.07 0.45 6.03
For the sex classification, the only combination of parameters that significantly
discriminated sex better than chance (p=0.048) was Full duration, Full 1st frequency
quartile, and 1st Band peak frequency. This combination yielded 81.93%/75.27%
(LSF/VS) of correctly cross classification of sexes, compared to the expected
63.37%/62.48%. Other two combinations yielded marginally significant results: Full
duration, Full 1st frequency quartile, and 2nd Band peak frequency (p=0.063); and Full
duration, Full 1st frequency quartile, and Full position of 1st time quartile (p=0.066).
Increasing the permutations to 10,000 (as suggested by Mundry & Sommer 2007) did
not changed results (p=0.049, p=0.061, p=0.058, respectively).
Of those 5 different parameters, only Full duration was significantly different
between sexes (t=-9.108, df=172.38, p<0.0001), with the male roar-bark being longer
(difference: 0.122±0.022 s; Figure 3). Full 1st frequency quartile (t= -2.612, df=117.44,
p=0.0102) and the 2nd Band peak frequency (t=2.484, df=184.94, p=0.0139) were
marginally significant (α=0.01; Bonferroni correction).
161
Figure 3. Differences in the roar-barks parameters between females and males maned wolves recorded in
two facilities in Minas Gerais, Brazil.
2. Propagation experiment
Only roar-barks broadcasted on the Flat site could be detected and measured up
to 640 m, while on the other sites roar-barks were too degraded or completely
undetectable at 320 m and beyond. Therefore, to balance the MANOVA and subsequent
ANOVAs we used only the data up to 160 m.
The repeated measures MANOVA for the 41 original parameters showed
significant individual (F=40.35, Pillai=2.9047, df=3, p<0.0001), distance (F=5.1061,
Pillai=1.5514, df=4, p<0.0001), and interaction (F=2.9636, Pillai=3.1590, df=14,
p<0.0001) effects. Individual effects for each parameter are reported on Table 4 and S3.
From the 8 selected parameters, only Full position of 3rd time quartile, Full 1st frequency
quartile, and 2nd Band peak frequency were significantly affected by the individual but
not by distance (Table 4). The other 5 parameters did not have individual effects two
times greater than the distance effect. An equal number of Raven’s robust measures
were stable (Full position of 3rd time quartile, Full 1st frequency quartile) but others so
called were not (Full position of 1st time quartile, Full 3rd frequency quartile). The
parameters variation with distance are shown in Figure 4.
162
Table 4. ANOVA effect sizes (F) for the 8 selected parameters of20 roar-barks from 4 maned wolves
broadcasted and re-recorded at 5 distances (1.25-160m) in 4 sites. (A) = unitless: proportion relative to
entire duration. α=0.0012.
Parameter Fwolf /
Fdist
Wolf Distance Interaction
F P F p F p
Full duration (s) 0.32 39.57 <0.0001 124.60 <0.0001 13.03 <0.0001
Full position of 1st time quartile (A) 0.56 20.00 <0.0001 35.69 <0.0001 2.96 0.0325
Full position of 3rd time quartile (A) 11.60 39.48 <0.0001 3.40 0.0660 6.70 0.0002
Full 1st frequency quartile (Hz) 66.94 93.07 <0.0001 1.39 0.2392 0.70 0.5501
Full 3rd frequency quartile (Hz) 1.07 21.60 <0.0001 20.23 <0.0001 2.66 0.0484
Full average entropy (bits) 0.02 6.42 0.0006 272.20 <0.0001 0.52 0.6672
1st band peak frequency (Hz) 1.00 18.73 <0.0001 18.81 <0.0001 3.81 0.0105
2nd band peak frequency (Hz) 1459 149.44 <0.0001 0.10 0.7492 1.49 0.2168
163
Figure 4. Variation of 8 selected parameters of broadcasted maned wolves roar-barks re-recorded at 7
distances (1.25-640m) at the Serra da Canastra National Park, MG/Brazil.
LDA for re-recorded roar-barks with the 3 stable parameters had an overall
mean percentage of correct identity classification of 73.96% (±8.96%, not including the
1.25 m training set, Figure 5). On site Vegetation the classification success dropped
rapidly, while on site Flat it remained relatively high across all distances. There was a
great difference among individuals for the classification success, with the animal JU
164
achieving the lowest rates, while the other 3 achieved 100% of correct classification at
320 m, and GA and SA remaining at this level at 640 m (Figure 5).
Figure 5. LDA percentage of correctly identity classification of broadcasted roar-barks of 4 maned
wolves (bottom) re-recorded at 7 distances (1.25-640m) in 4 sites (top) at the Serra da Canastra National
Park, MG/Brazil.
The pDFA for the captivity dataset with all 10 animals and only the 3 stable
parameters achieved 43.68%/41.39% (LSF/VS) of correctly cross validated
classification. Although much lower than the results using all 8 selected parameters,
discrimination was significantly (p=0.001) above the expected by chance on the cross
classification (13.75%/12.22%).
165
Signal-to-noise ratio decayed over distance as predicted (Figure 6). Although
not so clear on the LDA results, 160 m seem to be a limit that affects most parameters
(Figure 4). Mean signal-to-noise ratio on 160 m was 11.73 (±6.06 dB; Figure 6), with
all sites having significantly different means (F=139.74, df=3, p<0.0001), except
Vegetation and High-to-low sites (Tukey contrasts: p=0.4140). Despite the great
individual difference in classification success at 160 m, there was no difference in
signal-to-noise ratio among them (F=2.22, df=3, p=0.0927).
Figure 6. Signal-to-noise ratio of broadcasted roar-barks of maned wolves re-recorded at 7 distances
(1.25-640m) in 4 sites at the Serra da Canastra National Park, MG/Brazil. The signal-to-noise ratio was
calculated subtracting from the in-band power of each roar-bark (150-2000 Hz) and the same
measurement taken from an equal sized spectrogram portion immediately before the vocalization
(measure of the background noise level).
3. Natural habitat
We detected a total of 1210 roar-bark sequences over the 233 nights of
recordings. Of those, 150 involved two (very rarely three) different animals exchanging
roar-barks and 284 were registered by two (or more) recorders.
We measured the parameters of 92 pairs of roar-barks from different animals
vocalizing together (184 roar-barks total, from 21 sequences). Mean signal-to-noise
166
ratio of those roar-bark records was 11.58 (±6.93 dB). Animals were significantly
different from each other for all parameters (Table 5). The absolute difference between
animals was not generally correlated with the absolute signal-to-noise ratio difference
(Table 5), with maximum coefficients being 0.329, for the 2nd Band peak frequency,
and 0.485, for the Full average entropy.
We measured the parameters of 83 unique roar-barks registered by two recorders
(166 records total, from 17 unique sequences). Mean signal-to-noise ratio of roar-barks
from the most intense record was 11.37 (±7.77 dB), and for the last intense record 4.62
(±2.84 dB). Full duration, 1st and 2nd Band peak frequencies, and Full 3rd quartile of
frequency were all not significantly different between recorders (Table 5), although the
last presented a tendency for difference (α=0.0012, p=0.0128). The absolute difference
in parameters was also not generally correlated with the absolute difference of signal-to-
noise ratio (Table 5), except for Full 1st frequency quartile (0.594) and Full average
entropy (0.528).
The absolute parameter value difference between animals was not greater than
the difference between roar-barks registered in different recorders (Table 6), except for
Full duration. Figure 7 shows an example of two different sequences involving
apparently the same animals recorded with 4 nights of difference in recorders
approximately 5 km apart. The more stable parameters, on the propagation experiment
and the natural records, of both animals and sequences are compared in the graphs
below the spectrogram (Figure 7).
167
Table 5. Effect size (t) for the absolute difference between parameters of roar-barks from two different
maned wolves vocalizing together (2 wolves) and roar-barks simultaneously recorded by two different
autonomous recorders (2 recorders; SongMeter SM2+). 2 wolves: one sample t-test, df=91. 2 recorders:
paired t-test, df=82. α=0.0012. Cor. SNR: Pearson’s correlation coefficient with the signal-to-noise ratio
difference. (A) = unitless: proportion relative to entire duration. The vocalizations were recorded
passively with a grid of 12/13 autonomous recorders at Serra da Canastra National Park, MG/Brazil.
Parameter t2wolves
/ t2rec.
Dif. 2 wolves Dif. 2 recorders
t p Cor.
SNR t p
Cor.
SNR
Full duration (s) 45.14 11.826 <0.0001 0.095 -0.262 0.7941 0.095
Full position of 1st time quartile (A) 1.94 10.618 <0.0001 0.007 5.483 <0.0001 0.037
Full position of 3rd time quartile (A) 2.75 13.411 <0.0001 0.039 -4.883 <0.0001 0.195
Full 1st frequency quartile (Hz) 2.27 11.869 <0.0001 0.066 5.228 <0.0001 0.594
Full 3rd frequency quartile (Hz) 4.17 10.610 <0.0001 0.118 -2.547 0.0128 -0.115
Full average entropy (bits) 1.64 14.425 <0.0001 0.485 -8.815 <0.0001 0.528
1st band peak frequency (Hz) 15.20 13.616 <0.0001 0.216 0.896 0.3730 0.145
2nd band peak frequency (Hz) 17.40 10.090 <0.0001 0.329 -0.580 0.5636 0.160
Table 6. Mean±SD differenceof the absolute difference between parameters of roar-barks from two
different maned wolves vocalizing together (2 wolves) and roar-barks simultaneously recorded by two
different autonomous recorders (2 recorders; SongMeter SM2+). Those differences were compared with a
Welch two sample t-test.α=0.0012.(A) = unitless: proportion relative to entire duration. The vocalizations
were recorded passively with a grid of 12/13 autonomous recorders at Serra da Canastra National Park,
MG/Brazil.
Parameter Mean±SD difference Comparison of differences
2 wolves 2 recorders t df p
Full duration (s) 0.108±0.088 0.002±0.063 5.729 126.5 <0.0001
Full position of 1st time quartile (A) 0.067±0.060 -0.037±0.061 1.255 165.5 0.2113
Full position of 3rd time quartile (A) 0.082±0.059 0.044±0.082 0.702 173 0.4835
Full 1st frequency quartile (Hz) 123.0±99.4 -98.1±170.9 -0.171 139.4 0.8641
Full 3rd frequency quartile (Hz) 161.2±145.7 87.7±313.8 -0.861 124.6 0.3910
Full average entropy (bits) 0.819±0.545 0.610±0.631 1.399 172.5 0.1637
1st band peak frequency (Hz) 40.9±28.8 -4.7±47.8 1.968 155.8 0.0508
2nd band peak frequency (Hz) 79.1±75.2 5.8±91.7 2.309 171.3 0.0222
168
Figure 7. Two different roar-bark sequences involving the same two maned wolves each (top), and their
roar-bark parameters (bottom). Recordings made passively by two different autonomous recorders
(SongMeter SM2+) at the Serra da Canastra National Park, MG/Brazil.
169
Discussion
Here we found that the maned wolf long range call, the roar-bark, is individually
distinct and that captive animals can be discriminated from each other by their roar-
barks with a mean 72.6% correct classification. We also found that roar-barks carry sex
information, with sexes being correctly discriminated 78.6% of times. The roar-bark
duration and the concentration of energy in lower frequencies were the most important
parameters both for identity and sexual discrimination. The majority of acoustic
parameters were not stable when experimentally broadcasted over distance, even
through distances as short as 160 m. The few stable parameters were capable of
discriminating among individuals, although classification success rate dropped
substantially (42.54%). Interestingly, site characteristics (i.e., presence of vegetation),
and individual differences, were more important for discrimination success than
distance. Unfortunately, we found that in natural habitat the variation on parameters due
to propagation is larger than the individual difference, which seriously compromises the
applicability of vocal identification of maned wolves in the wild.
One of the most important parameters for individual discrimination was the first
frequency quartile, which sometimes can be viewed as a correlate of the fundamental
frequency. Several other works on canids (Tooze et al. 1990; Frommolt et al. 2003;
Hartwig 2005; Mitchell at al. 2006), and other mammals (Rendall et al. 1998; Sousa-
Lima et al. 2002, 2008), have similarly found frequency related parameters, especially
the fundamental frequency, to be important for individual discrimination. This was
expected, as the fundamental frequency and the energy distribution over frequencies are
highly related to the individual morphology of the vocal tract, affecting both source and
filter (Fitch 1997; Sousa-Lima et al. 2002).
170
On the other hand, the importance of the roar-bark duration for identity
discrimination was unexpected. Other studies in canids have found temporal patterns to
be important for identity discrimination (Durbin 1998; Darden et al. 2003), but in those
cases the parameter was the time between the start of two consecutive elements. While
in dholes and swift fox this interval is short (0.6 s: Durbin 1998; 0.2 s: Darden et al.
2003), for maned wolves it is much longer and unrelated to vocal identity (4.0-5.4 s;
Ferreira et al. unpublished; Appendix II).
The vocalization duration can be related to lung capacity (Fitch 2002), and
therefore indirectly related to body size. Body size may be important to distinguish
young from adults or females from males (in sexually dimorphic species) but seems to
provide very limited information for discriminant functions to separate among many
individuals. Thus, roar-bark duration may be at least partially under the emitter control
and reflect the animals’ internal state and/or preference.
Roar-bark duration was important to discriminate sexes, as also the first
frequency quartile. Both characteristics may provide body size cues: duration because of
the lung capacity (as cited above), and the first frequency quartile because larger vocal
folds are capable of producing lower frequencies (Morton 1977), although in several
species vocal tract structures vary independent of body size (Fitch 1997). Maned wolves
do not have a pronounced sexual dimorphism. Males are on average 2 kg and 2.5 cm
larger than females (N=79; Jácomo et al. 2009). Vocal differences may be less related to
morphology and more related to the animals’ motivation to appear bigger.
In captivity males vocalize more than females during the breeding season
(Sábato 2011) and produce longer roar-barks, which require higher energy investment.
This may be used as an indicator of the emitter motivation to fight (Vehrencamp 2000;
Linhart et al. 2012), for instance to defend their partner during the breeding season.
171
Independent of being more related to morphology or motivation, the sexual
difference in the roar-barks seems to be small, as the body dimorphism. Only one
combination of parameters (from 56) discriminated sexes better than expected by
chance (78.6% obtained x 62.9% chance). Also, the difference in frequency parameters
between sexes was only marginally significant (p=0.0102 and p=0.0139, α=0.01),
despite being on the expected direction (males with lower frequencies).
Randomly testing combinations of discriminating variables can generate
spurious results (Whittingham et al. 2006). We emphasize then that the 3 parameters
were drawn from a small pool of variables that were considered biologically relevant
and that the two first parameters in the 3 significant or marginally significant
combinations were the same as the most important for identity discrimination (Full
duration and Full 1st frequency quartile). All of which suggests that our result reflects a
real difference and not spurious characteristics of our data.
The LDA on the propagation experiment had a 73.96% identity classification
success. However, this value is overestimated as we did not used the permuted approach
(pDFA; Mundry & Sommer 2007), because we needed different training and test sets,
and that was not feasible due to the small sample size (included only 4 individuals).
Probably the value attained using the 3 stable parameters on the captivity data is closer
to reality (42.54%). Nevertheless, one interesting observation was that, in most cases,
the classification success did not drop progressively with distance. The two cases in
which success clearly dropped with distance was on the site with vegetation and for
individual JU. This result suggests that obstacles for sound propagation in their habitat
is more relevant in discriminating individuals than degradation. It also suggests the
identity information in some individuals’ roar-bark is more likely to travel far than
others, revealing the importance of the particular roar-bark acoustic structure for
172
discrimination at long distances. In the end this suggests some identity information may
travel far, given that environment and the vocalization spectrographic structure is
appropriate.
One thing worth noting is that while in the propagation experiment few roar-
barks even reached 640 m (less than 5%), with a mean signal-to-noise ratio of 5.92
(±2.14 dB) at this distance, on natural recordings 23.5% of sequences were registered by
more than one recorder. Based on the distance between recorders of the measured pairs
of roar-barks, sound would have to travel a mean of 1.19 km (±0.39 km) before
reaching recorders (in the minimum distance situation, in which the emitter is exactly in
the middle), and the signal-to-noise ratio of the less intense signal recorded was 4.62
(±2.84 dB). Thus, our propagation experiment, while perfectly valid, underestimates
roar-barks reach. This was probably a matter of the recorder height, which was lower on
the experiment than in the natural recordings (0.86 m x 1.4 m). It could also be that
captive maned wolves vocalize with a different intensity level than wild ones, which
would compromise the level we calibrated for. Captive maned wolves, for instance, are
maintained in much closer proximity with each other and would need a lower level to
communicate efficiently among themselves.
Unfortunately, despite some parameters being stable in the propagation
experiment (3 of 8), in natural conditions they were significantly altered in the
comparison of same roar-barks on different recorders (Full position of 3rd time quartile,
Full 1st frequency quartile) or the difference between individuals was correlated with
signal-to-noise ratio (2nd Band peak frequency). Furthermore, for all parameters (except
Full duration, see below) the variation magnitude among the same roar-bark on different
recorders was equal to the magnitude of the individual difference. The consequence is
173
that none of those parameters are reliable to apply vocal identity classification in free-
ranging maned wolves.
The duration of the roar-bark, one of the most important discriminatory
parameters in captivity, presented a paradoxical propagation behavior. While it
decreased steadily on the propagation experiment measures, it was the only parameter
not altered between different records of the same roar-bark in the natural recordings.
Maybe after some distance (>1km) the degradation of this parameter stabilizes. At this
moment we cannot state this is a reliable discriminatory parameter for application in the
wild, and further research will be needed to clarify this topic.
One last curious observation on the matter is that the mean duration difference
between roar-barks of animals vocalizing at the same moment (108±88 ms) was similar
to the mean difference for males and females in captivity (122±22 ms). This could
indicate that, for this sample of roar-barks (21 of 150 group sequences), this kind of
sequence is normally emitted by breeding pairs. Therefore, group vocalizations could be
a duet-like behavior, as common in some birds (Marshall-Ball et al. 2006) and primates
(Méndez-Cárdenas et al. 2009), and may have the same proposed functions: pair
bonding/coordination, joint territorial defense, mate guarding, and parental care
manipulation (Wachtmeister 2001). However, we do not know if duration is a reliable
parameter, as stated above, and a previous work from our laboratory found exactly the
opposite result (i.e. roar-barks in group vocalizations generally have similar duration,
N=15; Frigo 2016). Reports from captivity and the wild indicates that group
vocalizations occur between same and opposite sex maned wolves (Dietz 1984;
Emmons 2012).
For the captivity data, the presence of identity information in the roar-barks was
very consistent, as revealed by the pDFA, which considers the chance of random false
174
groups attaining discriminability because of information noise unrelated to the tested
factor (in our case identity). Yet, our success of vocal identity classification rate
(72.6%) was low compared to other canids, as artic foxes (91.7%, Frommolt et al.
2003), swift foxes (99%, Darden et al. 2003), and gray wolves (100%, Root-Guteridge
et al. 2014b). Also, all acoustic parameters degraded over distance, in the propagation
experiment or in the natural records, or both.
The high degradation of identity and sexual cues is probably related to the noisy
(harsh, broadband) nature of roar-barks. By contrast, long (several seconds), tonal, and
frequency modulated vocalizations are the ideal for transmitting identity information
over long distances (Wiley & Richards 1978, 1982; Fitch 1997).
In fact, studies on canids recurrently find that vocalizations like howls and
whistles (Durbin 1998; Root-Guteridge et al. 2014a), that are longer and more tonal,
attain better identity discrimination rates than barks and similarly noisy vocalizations
(Yin & McCowan 2004; Hartwig 2005). Domestic dogs’ barks are very noisy and have
the lowest identity discrimination rates among canid studies (40%), but this rate
improves in contexts in which dogs make more tonal and modulated barks (Yin &
McCowan 2004). Dingoes produce “bark-howls” concatenating a noisy and a tonal part,
with that last part attaining higher identity classification success than the first (Déaux et
al. 2016c). For coyotes, howls not only present best discrimination rates than barks
(83% x 69%), but also maintain high discrimination levels at 1000 m (81%, Mitchell et
al. 2006).
Those results raise the question of why free-ranging maned wolves do not
produce more tonal roar-barks in addition to the noisy ones, since captivity animals
demonstrate the species is capable of such. One hypothesis, that would also explain the
only moderate levels of identity and sexual discrimination obtained here, would be that
175
the selective pressure for identity discrimination by the species is low. Maned wolves
live solitary, in low densities (Trolle et al. 2007), and share their extensive home range
with very few conspecifics (the breeding partner and eventual offspring; Rodden et al.
2004). In this scenario, one animal would only need to distinguish between group-
members and non-group members, which does not demand a very fine identity coding.
Besides, maned wolves use other communication modalities, as visual (postures,
piloerection) and chemical (scent marks, urine, feces), to signalize its presence and
status over short and long distances and duration (Rodden et al. 2004). The combination
of modalities as well as redundancy (e.g. sequences of roar-barks) maximizes the
information transmission efficiency and prevent interpretation errors (Bradbury
&Vehrencamp1998).
The moderate discrimination rates, the high degradation of identity and sexual
cues, and the rarity of more tonal elements on free-ranging maned wolves’ roar-barks
could also mean that for the species the estimation of the emitter location is more
important than precise identification of sex and identity. Broadband vocalizations are
ideal to estimate the emitter distance, as they degrade predictably over distance (Wiley
& Richards 1978; Naguib & Wiley 2001). Accordingly, the roar-bark entropy steadily
increased on the propagation experiment. That would reinforce the idea that roar-barks
are vocalizations better suited for intra-group communication, for which identifying the
emitter position is important to coordinate activities, rather than for territorial
announcement, in which propagation over long distances is desired but precise location
is not.
Despite very rarely participating in cooperative activities close to conspecifics,
e.g. hunting ground birds (Jácomo et al. 2009), knowing the location of group members
may be most important for maned wolves. For instance, the breeding pair must reunite
176
on the short 5 days of estrous (Rodden et al. 2004), and the male must know the location
of the female and pups to defend and provide for them (Bestelmeyer 2000). Breeding
pairs are reported to use the same paths, fruit plants, and hunting areas, but not
simultaneously (Dietz 1984; Carvalho & Vasconcellos 1995), and the species present
periodicity on the use of space (Péron et al. 2017). Thus, maybe a more adaptive
function of roar-barking would be for each individual to plan their movement, according
to the position of others, as other maned wolves may interfere in their foraging and
deplete foraging areas.
Finally, one plausible alternative explanation for moderate discrimination and
low propagation of identity cues is that maned wolves are using parameters others than
the ones tested here to discriminate the identity and sex of roar-bark emitters (we
measured 41 but only tested 8 considered more relevant). Besides, maned wolves
potentially hear roar-barks at much further distances that the ones recorded by our
autonomous recorders (which was also true for the field researchers; personal
observation). When used to the captivity data, analysts could generally discriminate
with ease the emitter identity on the propagation experiment by the shape of the roar-
bark on spectrograms, no matter the distance (personal observation). In Figure 7, it is
also possible to discriminate individuals by the shape of their roar-barks. Although the
animal’s points in the graph do not overlay, they are visually (to us) more distinct than
the graphs suggest. That indicates the chosen parameters may not reliably reflect roar-
bark characteristics, and maybe other approaches, such as frequency contour tracking or
spectrogram cross correlation, will result in better discriminability increasing the
applicability of vocal identification of free-ranging maned wolves.
177
References
AZEVEDO, F. C. 2008. Home range and spatial organization of maned wolf
(Chrysocyon brachyurus) in the Serra da Canastra National Park region, Minas Gerais,
Brazil. M.S. thesis, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil.
BALIEIRO, F., & MONTICELLI, P.F. 2019. Evidence of individual discrimination in
the maned wolf long-distance extended-bark. Behavioural processes, 158, 219-227.
BESTELMEYER, S. V. 2000. Solitary, reproductive, and parental behavior of maned
wolves (Chrysocyon brachyurus). PhD Thesis, Colorado State University, Fort Collins,
U.S.A.
BRADBURY, J.W., & VEHRENCAMP, S.L. 1998. Principles of animal
communication.
BRADY, C.A. 1981. The vocal repertoires of the bush dog (Speothos venaticus), crab-
eating fox (Cerdocyon thous), and maned wolf (Chrysocyon brachyurus). Animal
Behaviour, 29(3), 649-669.
CARVALHO, C.T.D., & VASCONCELLOS, L.E. 1995. Disease, food and
reproduction of the maned wolf: Chrysocyon brachyurus (Illiger)(Carnivora, Canidae)
in southeast Brazil. Revista Brasileira de Zoologia, 12(3), 627-640.
CHARIF, R.A., WAACK, A.M., & STRICKMAN, L.M. 2010. Raven Pro 1.4 user’s
manual. Cornell Lab of Ornithology, Ithaca, NY, 25506974.
CHARRIER, I. & HARCOURT, R.G. 2006. Individual vocal identity in mother and
pup Australian sea lions. Journal of Mammalogy, 87(5), 929-938.
DARDEN, S.K., DABELSTEEN, T. & PEDERSEN, S.B. 2003. A potential tool for
swift fox (Vulpes velox) conservation: individuality of long-range barking sequences.
Journal of Mammalogy. 84(4), 1417-1427.
DÉAUX, E.C., CLARKE, J.A., & CHARRIER, I. 2016a. Dingo Howls: The Content
and Efficacy of a Long‐Range Vocal Signal. Ethology, 122(8), 649-659.
DÉAUX, E.C., ALLEN, A.P., CLARKE, J.A., & CHARRIER, I. 2016b. Concatenation
of ‘alert’ and ‘identity’ segments in dingoes’ alarm calls. Scientific Reports, 6.
DÉAUX, E.C., CHARRIER, I., & CLARKE, J.A. 2016c. The bark, the howl and the
bark-howl: Identity cues in dingoes’ multicomponent calls. Behavioural Processes, 129,
94-100.
DELPORT, W., KEMP, A.C., & FERGUSON, J.W.H. 2002. Vocal identification of
individual African Wood Owls Strix woodfordii: a technique to monitor long‐term adult
turnover and residency. Ibis, 144(1), 30-39.
DIETZ, J.M. 1984. Ecology and social organization of the maned wolf (Chrysocyon
brachyurus). Smithsonian Contribution to Zoology. 1-51.
178
DURBIN, L.S. 1998. Individuality in the whistle call of the Asiatic dog Cuon alpinus.
Bioacoustics, 9, 197-206.
EMMONS, L.H. 2012. The maned wolves of Noel Kempff Mercado National Park.
Smithsonian Contributions to Zoology 639:1–135.
FALLS, J.B. 1982. Individual recognition by sound in birds. In: Acoustic
communication in birds. D.E. Kroodsma, E. H. Miller. Academic Press, NY.
237-238.
FANIS, E. & JONES, G. 1996. Allomaternal care and recognition between mothers and
young in pipistrelle bats (Pipistrellus pipistrellus). Journal of Zoology, 240,
781-787.
FIGUEROA, H. 2007. Extensible BioAcoustic Tool. XBAT R7. Cornell Bioacoustics
Research Program, Ithaca, New York.
FITCH, W.T. 1997. Vocal tract length and formant frequency dispersion correlate with
body size in rhesus macaques. The Journal of the Acoustical Society of
America, 102(2), 1213-1222.
FITCH W. T. 2002. Primate vocal production and its implications for auditory research.
In: Ghazanfar, A. (Eds). Primate audition: ethology and neurobiology. Boca
Raton CRC Press. 87-108.
FRIGO, R.G. 2016. Interações acústicas entre lobos-guará (Chrysocyon brachyurus,
illiger 1815) alteram seus aulidos? Undergraduate thesis, Universidade Federal do Rio
Grande do Norte, Natal, RN, Brazil.
FROMMOLT, K.H., GOLTSMAN, M.E., & MACDONALD, D.W. 2003. Barking
foxes, Alopex lagopus: field experiments in individual recognition in a territorial
mammal. Animal Behaviour, 65(3), 509-518.
GILBERT, G., TYLER, G.A., & SMITH, K.W. 2002. Local annual survival of
booming male Great Bittern Botaurus stellaris in Britain, in the period 1990–1999. Ibis,
144(1), 51-61.
HARTWIG, S. 2005. Individual acoustic identification as a non-invasive conservation
tool: an approach to the conservation of the African wild dog an approach to
the conservation of the African wild dog Lycaon pictus (Temminck, 1820).
Bioacoustics, 15, 35-50.
JÁCOMO, A.T., KASHIVAKURA, C.K., FERRO, C., FURTADO, M.M., ASTETE,
S.P., TÔRRES, N.M., ... & SILVEIRA, L. (2009). Home range and spatial organization
of maned wolves in the Brazilian grasslands. Journal of Mammalogy, 90(1), 150-157.
JOUVENTIN, P., AUBIN, T. & LENGAGNE, T. 1999. Finding a parent in a king
penguin colony: the acoustic system of individual recognition. Animal Behaviour, 57,
1175-1183.
KLEIMAN, D.G., & EISENBERG, J.F. 1973. Comparisons of canid and felid social
systems from an evolutionary perspective. Animal behaviour, 21(4), 637-659.
179
KLEIMAN, D.G. 1972. Social behavior of the maned wolf (Chrysocyon brachyurus)
and bush dog (Speothos venaticus): a study in contrast. Journal of Mammalogy, 53(4),
791-806.
KOBER, M., TRILLMICH, F. & NAGUIB, M. 2008. Vocal mother-offspring
communication in guinea pigs: females adjust maternal responsiveness to litter size.
Frontiers in Zoology, 5(13)
LINHART, P., SLABBEKOORN, H., & FUCHS, R. 2012. The communicative
significance of song frequency and song length in territorial chiffchaffs. Behavioral
Ecology, 23(6), 1338-1347.
MAIA, O.B., & GOUVEIA, A. M.G. 2002. Birth and mortality of maned wolves
Chrysocyon brachyurus (Illiger, 1811) in captivity. Brazilian Journal of Biology, 62(1),
25-32.
MARLER, P. & HOBBETT, L. 1975. Individuality in long-range vocalization of wild
chimpanzees. Zeitschrift für Tierpsychologie. 38, 97-109.
MARSHALL-BALL, L., MANN, N., & SLATER, P. J. B. 2006. Multiple functions to
duet singing: hidden conflicts and apparent cooperation. Animal Behaviour, 71(4), 823-
831.
MCCOMB, K., REBY, D., BAKER, L., MOSS, C., & SAYIALEL, S. 2003. Long-
distance communication of acoustic cues to social identity in African elephants. Animal
Behaviour, 65, 317-329.
MELO, L.F., LIMA SÁBATO, M.A., VAZ MAGNI, E.M., YOUNG, R.J., &
COELHO, C.M. 2007. Secret lives of maned wolves (Chrysocyon brachyurus Illiger
1815): as revealed by GPS tracking collars. Journal of Zoology, 271(1), 27–36.
MELO, L.F.B., SÁBATO, M.A.L., VAZ MAGNI, E.M., YOUNG, R.J., & COELHO,
C.M. 2009. First observations of nest attendance behavior by wild maned wolves,
Chrysocyon brachyurus. Zoo Biology: Published in affiliation with the American Zoo
and Aquarium Association, 28(1), 69-74.
MÉNDEZ‐CÁRDENAS, M. G., & ZIMMERMANN, E. 2009. Duetting—A
mechanism to strengthen pair bonds in a dispersed pair‐living primate (Lepilemur
edwardsi)?. American Journal of Physical Anthropology: The Official Publication of
the American Association of Physical Anthropologists, 139(4), 523-532.
MITCHELL, B.R., MAKAGON, M.M., JAERGER, M.M. & BARRETT, R.H. 2006.
Information content of coyote barks. Bioacoustics, 15, 289-314.
MOEHLMAN, P.D. 1987. Social organization in jackals: the complex social system of
jackals allows the successful rearing of very dependent young. American Scientist,
75(4), 366-375.
MOEHLMAN, P.D. 1989. Intraspecific Variation in Canid Social Systems. In J.L.
Gittleman (Org.), Carnivore Behavior, Ecology, and Evolution (p. 143-163). Springer
US.
180
MOLNÁR, C., PONGRÁCZ, P., FARAGÓ, T., DÓKA, A., & MIKLÓSI, Á. 2009.
Dogs discriminate between barks: the effect of context and identity of the caller.
Behavioural processes, 82(2), 198-201.
MORTON, E.S. 1977. On the occurrence and significance of motivation-structural rules
in some bird and mammal sounds. The American Naturalist, 111(981), 855-869.
MUNDRY, R., & SOMMER, C. 2007. Discriminant function analysis with
nonindependent data: consequences and an alternative. Animal Behaviour, 74(4), 965-
976.
NAGUIB, M. & WILEY, H. 2001. Estimating the distance to a source of sound:
mechanisms and adaptations for long-range communication. Animal
Behaviour. 62, 825-837.
PALACIOS, V., FONT, E., MÁRQUEZ, R., & CARAZO, P. 2015. Recognition of
familiarity on the basis of howls: a playback experiment in a captive group of
wolves. Behaviour, 152(5), 593-614.
PAULA, R. D., MEDICI, P., & MORATO, R. G. 2008. Plano de Ação para a
Conservação do Lobo-Guará - Análise de Viabilidade Populacional e de
Habitat. Brasília: Centro Nacional de Pesquisas para Conservação dos Predadores
Naturais-CENAP/ICMBio, 158p.
PAULA, R. C., RODRIGUES, F. H. G., QUEIROLO, D., JORGE, R. P. S., LEMOS, F.
G., & DE ALMEIDA RODRIGUES, L. 2013. Avaliação do risco de extinção do lobo-
guará Chrysocyon brachyurus (Illiger, 1815) no Brasil. Biodiversidade Brasileira, (1),
146-159.
PAULA, R. C. D. 2016. Adequabilidade ambiental dos biomas brasileiros à ocorrência
do lobo-guará (Chrysocyon brachyurus) e efeitos da composição da paisagem em sua
ecologia espacial, atividade e movimentação. Doctoral dissertation, Universidade de
São Paulo, São Paulo, Brazil.
PÉRON, G., FLEMING, C.H., DE PAULA, R.C., & CALABRESE, J.M. 2016.
Uncovering periodic patterns of space use in animal tracking data with periodograms,
including a new algorithm for the Lomb-Scargle periodogram and improved
randomization tests. Movement ecology, 4(1), 19.
REBY, D., HEWISON, M., IZQUIERDO, M. & PÉPIN, D. 2001. Red deer (Cervus
elaphus) hinds discriminate between the roars of their current harem-holder stag and
those of neighboring stags. Ethology, 107, 951-959.
RENDALL, D., OWREN, M.J., & RODMAN, P.S. 1998. The role of vocal tract
filtering in identity cueing in rhesus monkey (Macaca mulatta) vocalizations. Journal of
the Acoustical Society of America, 103, 602-614.
ROBISSON, P., AUBIN, T., & BREMOND, J.C. 1993. Individuality in the voice of the
emperor penguin Aptenodytes forsteri: adaptation to a noisy
environment. Ethology, 94(4), 279-290.
181
ROCHA, L.H., FERREIRA, L.S., PAULA, B.C., RODRIGUES, F.H., & SOUSA-
LIMA, R.S. 2015. An evaluation of manual and automated methods for detecting
sounds of maned wolves (Chrysocyon brachyurus Illiger 1815). Bioacoustics, 24(2),
185-198.
ROCHA, L.H., FERREIRA, L.S., VENTICINQUE, E.M., RODRIGUES, F.H., &
SOUSA-LIMA, R.S. 2016. Temporal and environmental influences on long-distance
calling by free-ranging maned wolves (Chrysocyon brachyurus). Journal of
Mammalogy, 98(1), 302-311.
RODDEN, M., RODRIGUES F., & BESTELMEYER, S. 2004. Maned wolf
(Chrysocyon brachyurus). Canids: Foxes, Wolves, Jackals and Dogs, Status Survey and
Conservation Action Plan. 1st ed. IUCN Canid Specialist Group, Cambridge, United
Kingdom 38-43.
RODRIGUES, F.H.G. 2002. Biologia e conservação do lobo-guará na Estação
Ecológica de Águas Emendadas, DF. Ph.D Thesis, Universidade Estadual de
Campinas, Campinas, Brazil.
ROOT-GUTTERIDGE, H., BENCSIK, M., CHEBLI, M., GENTLE, L.K., TERRELL-
NIELD, C., BOURIT, A., & YARNELL, R.W. 2014. Improving individual
identification in captive Eastern grey wolves (Canis lupus lycaon) using the time course
of howl amplitudes. Bioacoustics, 23(1), 39-53.
ROOT-GUTTERIDGE, H., BENCSIK, M., CHEBLI, M., GENTLE, L.K., TERRELL-
NIELD, C., BOURIT, A., & YARNELL, R.W. 2014. Identifying individual wild
Eastern grey wolves (Canis lupus lycaon) using fundamental frequency and amplitude
of howls. Bioacoustics, 23(1), 55-66.
SÁBATO, V. 2011. Aspectos do comportamento acústico do lobo-guará Chrysocyon
brachyurus (Illiger 1815).
SAYIGH, L. S., TYACK, P. L., WELLS, R. S., SOLOW, A. R., SCOTT, M.D., &
IRVINE, A.B. 1998. Individual recognition in wild bottlenose dolphins: a field test
using playback experiments. Animal Behaviour, 57, 41-50.
SILVEIRA L. 1999. Ecologia e conservação da comunidade de carnívoros do Parque
Nacional das Emas, GO. M.S. thesis, Universidade Federal de Goiás, Goiânia, Goiás,
Brazil.
SOUSA-LIMA, R.S., PAGLIA, A.P. & FONSECA, G.A. D.B. 2002. Signature
information and individual recognition in the isolation calls of Amazonian
manatees, Trichechus inunguis (Mammalia: Sirenia). Animal Behaviour, 63(2),
301-310.
SOUSA-LIMA, R.S., PAGLIA, A.P. & FONSECA, G.A. D.B. 2008. Gender, age and
identity in the isolation calls of Antillean manatees (Trichechus manatus
manatus). Aquatic Mammals, 34(1), 109-122.
TABACHNICK, B.G. & FIDELL, L.S. 2001. Using Multivariate Statistics.
4th edn. Boston, Massachusetts: Allyn and Bacon.
182
TERRY, A.M., PEAKE, T.M., & MCGREGOR, P.K. 2005. The role of vocal
individuality in conservation. Frontiers in Zoology, 2(1), 10.
TOOZE, Z.J., HARRINGTON, F.H. & FENTRESS, J.C. 1990. Individually distinct
vocalizations in timber wolves, Canis lupus. Animal Behaviour, 40, 723-730.
TROLLE, M., NOSS, A.J., LIMA, E.S., & DALPONTE, J.C. 2007. Camera-trap
studies of maned wolf density in the Cerrado and Pantanal of Brazil. Biodiversity and
Conservation, 16, 1197-1204.
VEHRENCAMP, S.L. 2000. Handicap, index, and conventional signal elements of bird
song. Animal signals: signalling and signal design in animal communication.
WACHTMEISTER, C.A. 2001. Display in monogamous pairs: a review of empirical
data and evolutionary explanations. Animal Behaviour, 61(5), 861-868.
WHITTINGHAM, M. J., STEPHENS, P. A., BRADBURY, R. B., & FRECKLETON,
R. P. 2006. Why do we still use stepwise modelling in ecology and behaviour?. Journal
of animal ecology, 75(5), 1182-1189.
WILEY, R.H., RICHARDS, J. 1982. Adaptations for acoustic communication in
birds: sound transmission and signal detection. In: Kroodsma, D.H., Miller, E.H. (eds)
Ecology and evolution of acoustic communication in birds. Academic Press, New
York, pp 131-181
WILEY, R.H., & RICHARDS, D.G. 1978. Physical constraints on acoustic
communication in the atmosphere: implications for the evolution of animal
vocalizations. Behavioral Ecology and Sociobiology, 3(1), 69-94.
YIN, S. & MCCOWAN, B. 2004. Barking in domestic dogs: context specificity and
individual recognition. Animal behavior, 68, 343-355.
183
Supplementary material
Table S1. All measured parameters on maned wolves roar-barks (LSF analyst). “Full” measures refer to
the roar-bark from 150 Hz to 2000 Hz, while “First” and “Second” bands refer to portions from lower to
higher frequencies. Means±SD are for all 10 individuals, 20 roar-barks each. PIC = Potential for Identity
Coding (Robisson et al. 1993). # selected for the identity classification. (A) = unitless: proportion relative
to entire duration. *the selection box of the 2nd band is limited below by the selection box of the 1st band,
therefore we manually measured the 2nd band true lower frequency. Parameters detailing can be found on
Raven’s manual (Charif et al. 2010).
Parameter Mean±SD Dif. to VSR analyst PIC
# Full duration (s) 0.591±0.112 -0.032±0.051 1.817
Full duration of 1st quartile (ms) 74±25 -1±6 1.212
Full duration of 3rd quartile (ms) 78±26 -2±8 1.138
Full inter quartile duration (ms) 152±41 -3±9 1.259
# Full position of 1st time quartile (A) 0.375±0.056 0.013±0.039 1.039
Full position of 2nd time quartile (A) 0.502±0.064 0.019±0.042 1.077
# Full position of 3rd time quartile (A) 0.635±0.070 0.025±0.044 1.154
# Full 1st frequency quartile (Hz) 609.8±169.7 1.1±7.2 1.997
Full 2nd frequency quartile (Hz) 731.3±196.4 0.4±4.4 1.649
# Full 3rd frequency quartile (Hz) 905.2±220.2 1.4±7.7 1.657
Full peak frequency (Hz) 716.4±242.6 0. 1.535
# Full average entropy (bits) 3.765±0.584 -0.015±0.056 1.691
1st band duration (s) 0.551±0.112 -0.029±0.092 1.787
1st band duration of 1st quartile (ms) 65±25 -1±19 1.043
1st band duration of 3rd quartile (ms) 77±41 -2±23 1.252
1st band inter quartile duration (ms) 142±55 -3±30 1.204
1st band position of 1st time quartile (A) 0.327±0.096 0.022±0.070 1.265
1st band position of 2nd time quartile (A) 0.448±0.105 0.030±0.089 1.230
1st band position of 3rd time quartile (A) 0.588±0.112 0.037±0.091 1.245
1st band lower frequency (Hz) 255.6±86.6 18.9±70.9 1.514
1st band higher frequency (Hz) 618.1±53.9 -9.7±63.5 1.390
1st band 1st frequency quartile (Hz) 493.1±55.4 -3.2±52.9 1.647
1st band 2nd frequency quartile (Hz) 520.9±54.7 -6.1±52.3 1.822
1st band 3rd frequency quartile (Hz) 546.2±54.7 -7.4±57.9 1.726
# 1st band peak frequency (Hz) 528.2±60.9 -8.4±59.5 1.627
1st band average entropy 2.447±0.355 -0.104±0.310 1.386
2nd band duration (s) 0.520±0.092 0.004±0.089 1.665
2nd band duration of 1st quartile (ms) 76±33 -5±27 1.072
2nd band duration of 3rd quartile (ms) 78±32 -8±29 1.161
2nd band inter quartile duration (ms) 154±50 -13±46 1.192
2nd band position of 1st time quartile (A) 0.375±0.079 0.026±0.059 1.232
2nd band position of 2nd time quartile (A) 0.522±0.092 0.014±0.067 1.193
2nd band position of 3rd time quartile (A) 0.671±0.091 -0.002±0.073 1.177
2nd band lower frequency (Hz)* 623.9±53.3 -3.0±73.1 1.350
2nd band true lower frequency (Hz)* 467.5±100.7 117.0±135.5 1.193
2nd band higher frequency (Hz) 1036.3±137.8 -75.6±136.6 1.706
2nd band 1st frequency quartile (Hz) 794.2±107.1 -23.8±82.5 1.798
2nd band 2nd frequency quartile (Hz) 844.7±127.0 -37.6±106.0 1.942
2nd band 3rd frequency quartile (Hz) 891.4±139.0 -48.4±118.7 1.920
# 2nd band peak frequency (Hz) 848.4±139.0 -35.5±113.6 1.641
2nd band average entropy (bits) 2.753±0.354 -0.113±0.328 1.311
184
Table S2. Explained variance and coefficients of each linear discriminant function for identity
discrimination of 10 captive maned wolves roar-barks. Non-normal parameters were transformed (Yeo
Johnson). All parameters were centralized and scaled.
Functions LD1 LD2 LD3 LD4 LD5 LD6 LD7 LD8
Proportion of trace: 44.22% 29.08% 15.59% 5.56% 3.65% 1.28% 0.61% 0.01%
Coefficients of linear
discriminants:
Full average entropy -0.276 1.283 0.544 -0.977 0.755 0.019 0.722 0.819
Full duration -1.310 -0.190 1.221 0.210 0.527 -0.226 -0.108 -0.090
Full position of the 1st
time quartile -0.327 0.177 -0.363 -0.522 0.205 0.053 0.348 -1.005
Full position of the 3rd
time quartile 0.048 -0.158 0.446 1.200 -0.126 0.027 0.503 0.357
Full 1st frequency quartile 0.909 -0.110 1.464 -0.915 -0.335 0.020 0.716 0.357
Full 3rd frequency quartile 0.421 0.418 -0.154 1.267 -0.053 -0.415 -1.136 -0.947
1st Band peak freq 0.337 -0.514 0.042 0.094 0.777 1.094 -0.372 -0.057
2nd Band peak freq 0.040 -0.261 -0.515 -0.102 0.546 -1.159 0.587 0.026
185
Table S3. ANOVA effect sizes (F) for all measured parameters on maned wolves roar-barks broadcasted
and re-recorded at 5 distances (1.25-160m). “Full” measures refer to the entire roar-bark (from 150 Hz to
2000 Hz), while “First” and “Second” bands refer to portions from lower to higher frequencies. (A) =
unitless: proportion relative to entire duration. α=0.0012.
Parameter FID /
FDist
Individual Distance Interaction
F p F p F p
Full duration (s) 0.32 39.57 <0.0001 124.60 <0.0001 13.03 <0.0001
Full inter quartile duration (ms) 149.99 10.38 <0.0001 0.07 0.7926 4.06 0.0075
Full position of 1st time quartile (A) 0.56 20.00 <0.0001 35.69 <0.0001 2.96 0.0325
Full position of 2nd time quartile (A) 2.88 25.43 <0.0001 8.82 0.0032 6.39 0.0003
Full position of 3rd time quartile (A) 11.60 39.48 <0.0001 3.40 0.0660 6.70 0.0002
Full 1st frequency quartile (Hz) 66.94 93.07 <0.0001 1.39 0.2392 0.70 0.5501
Full 2nd frequency quartile (Hz) 2.46 48.07 <0.0001 19.51 <0.0001 2.45 0.0636
Full 3rd frequency quartile (Hz) 1.07 21.60 <0.0001 20.23 <0.0001 2.66 0.0484
Full peak frequency (Hz) 29.88 64.30 <0.0001 2.15 0.1434 0.26 0.8536
Full average entropy (bits) 0.02 6.42 0.0006 272.20 <0.0001 0.52 0.6672
1st band duration (s) 0.02 2.05 0.1133 103.70 <0.0001 10.89 <0.0001
1st band inter quartile duration (ms) 0.12 7.00 0.0003 57.70 <0.0001 11.14 <0.0001
1st band position of 1st time quartile (A) 1.26 15.96 <0.0001 12.67 0.0004 17.03 <0.0001
1st band position of 2nd time quartile (A) 2.25 22.51 <0.0001 10.00 0.0017 20.49 <0.0001
1st band position of 3rd time quartile (A) 0.45 25.62 <0.0001 56.82 <0.0001 19.21 <0.0001
1st band lower frequency (Hz) 5.26 146.64 <0.0001 27.89 <0.0001 8.42 <0.0001
1st band higher frequency (Hz) 0.70 27.00 <0.0001 38.61 <0.0001 6.65 0.0002
1st band 1st frequency quartile (Hz) 0.19 15.34 <0.0001 80.54 <0.0001 8.79 <0.0001
1st band 2nd frequency quartile (Hz) 0.78 21.36 <0.0001 27.26 <0.0001 3.54 0.0150
1st band 3rd frequency quartile (Hz) 5.86 27.87 <0.0001 4.75 0.0300 3.94 0.0088
1st band peak frequency (Hz) 1.00 18.73 <0.0001 18.81 <0.0001 3.81 0.0105
1st band average entropy 0.24 27.20 <0.0001 115.43 <0.0001 1.92 0.1255
2nd band duration (s) 0.15 23.49 <0.0001 157.65 <0.0001 11.67 <0.0001
2nd band inter quartile duration (ms) 89.67 22.39 <0.0001 0.25 0.6177 9.94 <0.0001
2nd band position of 1st time quartile (A) 0.16 1.74 0.1655 11.19 0.0009 5.10 0.0018
2nd band position of 2nd time quartile (A) 103.82 28.49 <0.0001 0.27 0.6008 0.83 0.4786
2nd band position of 3rd time quartile (A) 8.61 56.84 <0.0001 6.60 0.0107 6.51 0.0003
2nd band lower frequency (Hz) 0.62 21.11 <0.0001 33.90 <0.0001 6.73 0.0002
2nd band higher frequency (Hz) 5.63 106.63 <0.0001 18.92 <0.0001 3.02 0.0301
2nd band 1st frequency quartile (Hz) 12995 291.08 <0.0001 0.02 0.8812 5.77 0.0008
2nd band 2nd frequency quartile (Hz) 130.49 302.48 <0.0001 2.32 0.1289 5.40 0.0012
2nd band 3rd frequency quartile (Hz) 15802 207.00 <0.0001 0.01 0.9090 1.86 0.1363
2nd band peak frequency (Hz) 1459 149.44 <0.0001 0.10 0.7492 1.49 0.2168
2nd band average entropy (bits) 0.11 15.75 <0.0001 137.04 <0.0001 7.89 <0.0001
186
Final remarks
Through a combination of approaches, including recordings in captivity, in the
wild, and a propagation experiment, the present work supports the hypothesis that the
maned wolf roar-bark is a multifunctional vocalization. Roak-barks are used for
territorial announcement and defense, as indicated by the observation that: wild maned
wolves vocally responded and approached roar-bark playbacks of unknown individuals
(of both sexes); and roar-barks are emitted all year, not only in the breeding season.
Maned wolves emit more roar-barks at the beginning of the night, but not near dawn,
despite both these times presenting efficient propagation. Vocalizing at the on-set of the
activity period can provide an honest advertisement of quality as it is more restrictive to
vocalize after a period without food. They also do not respond to playbacks after 20h,
which can indicate, as in some birds, this is the most important period to announce
territory occupancy and maned wolves will pay less attention to it outside this period.
On two occasions two different maned wolves responded to the playback, one
time with an apparent coordinated approach. Maybe it was a mated pair (or same group
members), which if true would show maned wolves also use roar-barks to jointly defend
their territory. The roar-bark structure favors location of the emitter instead of long-
range identity information transmission, in accordance with the idea that group-
members must coordinate their activities, even if it is only to avoid interference in
foraging. During the mating season maned wolves use roar-barks to attract new partners
or reunite previously established pairs and mate guard, as indicated by the increase in
both solo and group vocalizations. Captive males emitted longer, and thus costlier, roar-
barks, which reinforce the role of roar-barks in announcing motivation to defend sexual
partners. Around parturition solo and group vocal activity also increased, as the pair
must coordinate the parental care and/or territorial defense becomes more important.
187
Besides finding support for the previously proposed functions of roar-barks, the
present work highlights some new aspects of the species vocal behavior, as the: low
vocal activity and responsiveness near dawn, despite the species being more active
around twilight; the increase in solo vocal activity during better moonlit nights; and the
possibility that group vocalizations, i.e., when animals alternate roar-barks on the same
sequence, have a differentiated function, as no animal intercalated roar-barks during the
playback, and group vocalizations were in general independent of moon illumination or
time of the night. Those issues require further investigation to clarify their underlying
cause and provide new exciting research paths in the study of maned wolf acoustic
ecology.
On very good conditions a roar-bark can reach over 3 km, but on others, less
than 160 m, which shows how drastically their environment influences their active
space. Even pointing the muzzle up and finding higher grounds to vocalize do not
guarantee better transmission. This instability was reflected in the vocal discrimination
of individuals at longer distances. The unfortunate conclusion of this work is that, at the
moment, this method is not applicable for natural recordings with the parameters used.
Yet, visual differences in roar-barks are perceivable on spectrograms, which, despite
being a subjective impression, encourages new approaches on the matter.
Nevertheless, even without vocal identification, exploring the long-range
acoustic communication of maned wolves is proving highly valuable to monitor the
species and reveal their secretive behavior.
188
Extra-textual references
BALIEIRO, F., & MONTICELLI, P.F. 2019. Evidence of individual discrimination in
the maned wolf long-distance extended-bark. Behavioural processes, 158, 219-227.
BEISIEGEL, B. D. M., & ADES, C. 2002. The behavior of the bush dog (Speothos
venaticus Lund, 1842) in the field: a review. Revista de Etologia, 4(1), 17-23.
BESTELMEYER, S.V. 2000. Solitary, reproductive, and parental behavior of maned
wolves (Chrysocyon brachyurus). PhD Thesis, Colorado State University, Fort Collins,
U.S.A.
BRADY, C.A. 1981. The vocal repertoires of the bush dog (Speothos venaticus), crab-
eating fox (Cerdocyon thous), and maned wolf (Chrysocyon brachyurus). Animal
Behaviour, 29(3), 649-669.
COURTENAY, O., & MAFFEI, L., 2004. Crab-eating fox (Cerdocyon thous). Canids:
Foxes, Wolves, Jackals and Dogs, Status Survey and Conservation Action Plan. 1st ed.
IUCN Canid Specialist Group, Cambridge, United Kingdom 32–38.
DIETZ, J.M. 1984. Ecology and social organization of the maned wolf (Chrysocyon
brachyurus). Smithsonian Contribution to Zoology. 1-51.
EMMONS, L.H. 2012. The maned wolves of Noel Kempff Mercado National Park.
Smithsonian Contributions to Zoology 639:1–135.
FOX, M. W. 1975. The wild canids: their systematics, behavioral ecology and
evolution. R.E. Krieger.
FREEBERG, T. M. 2006. Social complexity can drive vocal complexity: group size
influences vocal information in Carolina chickadees. Psychological Science, 17(7), 557-
561.
FROMMOLT, K.H., GOLTSMAN, M.E., & MACDONALD, D.W. 2003. Barking
foxes, Alopex lagopus: field experiments in individual recognition in a territorial
mammal. Animal Behaviour, 65(3), 509-518.
IUCN 2015: PAULA, R.C. & DEMATTEO, K. 2015. Chrysocyon brachyurus (errata
version published in 2016). The IUCN Red List of Threatened Species 2015.
http://dx.doi.org/10.2305/IUCN.UK.2015-4.RLTS.T4819A82316878.en. Accessed 29
January 2019.
JÁCOMO, A.T., KASHIVAKURA, C.K., FERRO, C., FURTADO, M.M., ASTETE,
S.P., TÔRRES, N.M., ... & SILVEIRA, L. 2009. Home range and spatial organization
of maned wolves in the Brazilian grasslands. Journal of Mammalogy, 90(1), 150-157.
KLEIMAN, D.G., & EISENBERG, J.F. 1973. Comparisons of canid and felid social
systems from an evolutionary perspective. Animal behaviour, 21(4), 637-659.
KLEIMAN, D.G. 1972. Social behavior of the maned wolf (Chrysocyon brachyurus)
and bush dog (Speothos venaticus): a study in contrast. Journal of Mammalogy, 53(4),
791-806.
189
MAIA, O.B., & GOUVEIA, A. M.G. 2002. Birth and mortality of maned wolves
Chrysocyon brachyurus (Illiger, 1811) in captivity. Brazilian Journal of Biology, 62(1),
25-32.
MCNAB, B. K. 1986. The Influence of Food Habits on the Energetics of Eutherian
Mammals. Ecological Monographs, 56(1), 1–19. doi:10.2307/2937268
MELO, L.F., LIMA SÁBATO, M.A., VAZ MAGNI, E.M., YOUNG, R.J., &
COELHO, C.M. 2007. Secret lives of maned wolves (Chrysocyon brachyurus Illiger
1815): as revealed by GPS tracking collars. Journal of Zoology, 271(1), 27–36.
MELO, L.F.B., SÁBATO, M.A.L., VAZ MAGNI, E.M., YOUNG, R.J., & COELHO,
C.M. 2009. First observations of nest attendance behavior by wild maned wolves,
Chrysocyon brachyurus. Zoo Biology: Published in affiliation with the American Zoo
and Aquarium Association, 28(1), 69-74.
MMA 2008: Livro vermelho da fauna brasileira ameaçada de extinção / editores Angelo
Barbosa Monteiro Machado, Gláucia Moreira Drummond, Adriano Pereira Paglia. -
1.ed. - Brasília, DF : MMA; Belo Horizonte, MG : Fundação Biodiversitas, 2008.
MOEHLMAN, P.D. 1987. Social organization in jackals: the complex social system of
jackals allows the successful rearing of very dependent young. American Scientist,
75(4), 366-375.
MOEHLMAN, P.D. 1989. Intraspecific Variation in Canid Social Systems. In J.L.
Gittleman (Org.), Carnivore Behavior, Ecology, and Evolution (p. 143–163). Springer
US.
MYERS, N., MITTERMEIER, R. A., MITTERMEIER, C. G., DA FONSECA, G. A.
B., & KENT, J. 2000. Biodiversity hotspots for conservation priorities. Nature,
403(6772), 853–858.
PAULA, R. D., MEDICI, P., & MORATO, R. G. 2008. Plano de Ação para a
Conservação do Lobo-Guará - Análise de Viabilidade Populacional e de
Habitat. Brasília: Centro Nacional de Pesquisas para Conservação dos Predadores
Naturais-CENAP/ICMBio, 158p.
PAULA, R. C., & GAMBARINI, A. 2013. Histórias de um Lobo. Stories of a Golden
Wolf. Avisbrasilis. São Paulo: Editora Vinhedo.
QUEIROLO, D., AND MOTTA-JUNIOR, J. C. 2007. Prey availability and diet of
maned wolf in Serra da Canastra National Park, southeastern Brazil. Acta Theriologica
52:391–402.
RASMUSSEN, J. L., & TILSON, R. L. 1984. Food provisioning by adult maned
wolves (Chrysocyon brachyurus). Zeitschrift für Tierpsychologie, 65(4), 346-352.
ROCHA, L.H., FERREIRA, L.S., PAULA, B.C., RODRIGUES, F.H., & SOUSA-
LIMA, R.S. 2015. An evaluation of manual and automated methods for detecting
sounds of maned wolves (Chrysocyon brachyurus Illiger 1815). Bioacoustics, 24(2),
185-198.
190
ROCHA, L.H., FERREIRA, L.S., VENTICINQUE, E.M., RODRIGUES, F.H., &
SOUSA-LIMA, R.S. 2016. Temporal and environmental influences on long-distance
calling by free-ranging maned wolves (Chrysocyon brachyurus). Journal of
Mammalogy, 98(1), 302-311.
RODDEN, M., RODRIGUES F., & BESTELMEYER, S. 2004. Maned wolf
(Chrysocyon brachyurus). Canids: Foxes, Wolves, Jackals and Dogs, Status Survey and
Conservation Action Plan. 1st ed. IUCN Canid Specialist Group, Cambridge, United
Kingdom 38–43.
RODRIGUES, F.H.G. 2002. Biologia e conservação do lobo-guará na Estação
Ecológica de Águas Emendadas, DF. Ph.D Thesis, Universidade Estadual de
Campinas, Campinas, Brazil.
SÁBATO, V. 2011. Aspectos do comportamento acústico do lobo-guará Chrysocyon
brachyurus (Illiger 1815).
SILVA, J. M. C. D., & BATES, J. M. 2002. Biogeographic Patterns and Conservation
in the South American Cerrado: A Tropical Savanna Hotspot. BioScience, 52(3), 225–
234.
SILVEIRA L. 1999. Ecologia e conservação da comunidade de carnívoros do Parque
Nacional das Emas, GO. M.S. Thesis, Universidade Federal de Goiás, Goiânia, Goiás,
Brazil.
SLATER, G. J., THALMANN, O., LEONARD, J. A., SCHWEIZER, R. M., KOEPFLI,
K. P., POLLINGER, J. P., ... & WAYNE, R. K. 2009. Evolutionary history of the
Falklands wolf. Current biology, 19(20), R937-R938.
TEMBROCK, G. 1976. Canid vocalizations. Behavioural Processes, 1(1), 57-75.
191
Appendix I
Maned wolves do not emit more roar-barks than expected by chance during the
illuminated versus the non-illuminated portion of the night, except from new to waxing
crescent phase. In this phase only the first part of the night is illuminated, and thus the
difference may be caused by the species preference to vocalize on this time. *t=2.906,
df=33, p=0.006. Graph extracted and translated from: ÁRAUJO, D.D., FERREIRA,
L.S., ROCHA, L.H.S., & SOUSA-LIMA, R.S. 2016. Influência do ciclo lunar nas
vocalizações de lobo guará. Abstract and poster presentation at the III Conferência e
VIII Simpósio de Psicobiologia, UFRN, Natal, Rio Grande do Norte, Brazil.
192
Appendix II
Individual variation of the time interval between the start of one maned wolf
roar-bark to the next one in the sequence. Potential for Identity Coding (PIC; as in
Robisson et al. 1993) for this parameter is 1.04. Data from 10 captive maned wolves,
24-124 roar-barks by individual, 897 roar-barks in total, recorded in 2010 by V. Sábato,
at 2 facilities in Minas Gerais, Brazil.
193
Annex I
A sedated lactating maned wolf being examined by the Lobos da Canastra team.
This female (known as “Rose”) was captured on the night between July 12-13 2016 at
the Serra da Canastra National Park, around site F (Figure 1, Chapter 3). At this year
she was already without a VHS collar, but her data shows she lived on the present study
area since at least 2014, when she was first captured, also lactating. Photo: Rogério
Cunha de Paula. Used with permission.
194
Annex II
Justificativa ética
A pesquisa de Luane Maria Stamatto Ferreira, CPF 049.973.494-70, RG
001850911, enquanto aluna do Programa de Pós-Graduação em Psicobiologia, nível
doutorado, e orientada pela Profa. Dra. Renata Santoro Sousa-Lima, foi autorizada pelo
Instituto Chico Mendes de Conservação da Biodiversidade (ICMBio; licença SISBIO
número 41329-2). O ICMBio é o órgão federal com finalidade de executar e fazer
executar a política e diretrizes governamentais fixadas para o meio ambiente (lei nº
6.938/1981), sendo competência deste, através do SISBIO, conceder autorização para a
coleta de material biológico de espécimes da fauna silvestre em território brasileiro e
realização de pesquisa em unidade de conservação, como consta nas leis nº 5.197/1967,
nº 9.605/1998, nº 9.985/2000, no decreto nº 4.340/2002, lei nº 12.651/2012, e IN
ICMBio nº 03/2014. A observação e gravação de imagens e som para fins científicos
não requer autorização pelo ICMBio, como consta no IN ICMBio nº 03/2014,
entendendo-se, portanto, que sejam atividades não intrusivas e de impacto
negligenciável. Entretanto, a autorização é necessária quando as atividades de coleta são
feitas em Unidades de Conservação, como foi o caso desta pesquisa (UC: Parque
Nacional da Serra da Canastra). Logo, o projeto foi submetido e aprovado pelo ICMBio,
sob a referida licença número 41329-2, que dá autorização para Atividades com
Finalidades Científicas em UC federal (anexada a seguir).
Ministério do Meio Ambiente - MMAInstituto Chico Mendes de Conservação da Biodiversidade - ICMBioSistema de Autorização e Informação em Biodiversidade - SISBIO
Autorização para atividades com finalidade científicaNúmero: 41329-7 Data da Emissão: 17/02/2017 09:50 Data para Revalidação*: 19/03/2018
* De acordo com o art. 28 da IN 03/2014, esta autorização tem prazo de validade equivalente ao previsto no cronograma de atividades do projeto,mas deverá ser revalidada anualmente mediante a apresentação do relatório de atividades a ser enviado por meio do Sisbio no prazo de até 30 diasa contar da data do aniversário de sua emissão.
SISBIODados do titular
Nome: RENATA SANTORO DE SOUSA LIMA MOBLEY CPF: 705.712.446-53
Título do Projeto: Ecologia acústica do lobo-guará (Chrysocyon brachyurus) no Parque Nacional da Serra da Canastra ? MG
Nome da Instituição : UFRN - UNIVERSIDADE FEDERAL DO RIO G. NORTE CNPJ: 24.365.710/0001-83
Cronograma de atividades# Descrição da atividade Início (mês/ano) Fim (mês/ano)
1 Piloto no campo 12/2013 12/20132 Coleta de dados 01/2014 10/20173 Levantamento bibliográfico 01/2014 12/20174 Análise de dados 02/2014 12/20175 Elaboração de documentos cientificos 06/2014 12/2017
Observações e ressalvas
1As atividades de campo exercidas por pessoa natural ou jurídica estrangeira, em todo o território nacional, que impliquem o deslocamento de recursos humanos emateriais, tendo por objeto coletar dados, materiais, espécimes biológicos e minerais, peças integrantes da cultura nativa e cultura popular, presente e passada,obtidos por meio de recursos e técnicas que se destinem ao estudo, à difusão ou à pesquisa, estão sujeitas a autorização do Ministério de Ciência e Tecnologia.
2
Esta autorização NÃO exime o pesquisador titular e os membros de sua equipe da necessidade de obter as anuências previstas em outros instrumentos legais, bemcomo do consentimento do responsável pela área, pública ou privada, onde será realizada a atividade, inclusive do órgão gestor de terra indígena (FUNAI), daunidade de conservação estadual, distrital ou municipal, ou do proprietário, arrendatário, posseiro ou morador de área dentro dos limites de unidade de conservaçãofederal cujo processo de regularização fundiária encontra-se em curso.
3Este documento somente poderá ser utilizado para os fins previstos na Instrução Normativa ICMBio n° 03/2014 ou na Instrução Normativa ICMBio n° 10/2010, no queespecifica esta Autorização, não podendo ser utilizado para fins comerciais, industriais ou esportivos. O material biológico coletado deverá ser utilizado para atividadescientíficas ou didáticas no âmbito do ensino superior.
4O titular de licença ou autorização e os membros da sua equipe deverão optar por métodos de coleta e instrumentos de captura direcionados, sempre que possível,ao grupo taxonômico de interesse, evitando a morte ou dano significativo a outros grupos; e empregar esforço de coleta ou captura que não comprometa a viabilidadede populações do grupo taxonômico de interesse em condição in situ.
5O titular de autorização ou de licença permanente, assim como os membros de sua equipe, quando da violação da legislação vigente, ou quando da inadequação,omissão ou falsa descrição de informações relevantes que subsidiaram a expedição do ato, poderá, mediante decisão motivada, ter a autorização ou licençasuspensa ou revogada pelo ICMBio, nos termos da legislação brasileira em vigor.
6Este documento não dispensa o cumprimento da legislação que dispõe sobre acesso a componente do patrimônio genético existente no território nacional, naplataforma continental e na zona econômica exclusiva, ou ao conhecimento tradicional associado ao patrimônio genético, para fins de pesquisa científica,bioprospecção e desenvolvimento tecnológico. Veja maiores informações em www.mma.gov.br/cgen.
7Em caso de pesquisa em UNIDADE DE CONSERVAÇÃO, o pesquisador titular desta autorização deverá contactar a administração da unidade a fim de CONFIRMARAS DATAS das expedições, as condições para realização das coletas e de uso da infra-estrutura da unidade.
Outras ressalvas
1 Não pode haver novas capturas além das previstas no projeto citado no parecer ora emitido. Todas as citações referentes ao PNSC devem tercomo base o Plano de Manejo de 2005.
Equipe# Nome Função CPF Doc. Identidade Nacionalidade1 Jean Pierre Santos assistente de campo 013.190.686-00 M7928211 SSP-MG Brasileira2 FLAVIO HENRIQUE GUIMARÃES RODRIGUES Colaborador 536.695.171-20 1028375 SSP-DF Brasileira3 Victor Sábato Rocha Colaborador 081.762.096-60 MG13000879 SSP-MG Brasileira4 Luciana Helena Silva Rocha Responsável de campo 047.443.344-74 1800413 ITEP-RN Brasileira5 Julio Baumgarten Colaborador 473.370.951-04 - Brasileira6 Marcello Montagno do Valle assistente de campo 115.682.788-45 6895163 SSP-SP Brasileira7 Luane Maria Stamatto Ferreira Responsável de campo 049.973.494-70 001850911 ITEP-RN Brasileira8 Julia Simões Damo assistente de campo 119.633.486-21 19157461 PCMG-MG Brasileira
SISBIOEste documento (Autorização para atividades com finalidade científica) foi expedido com base na Instrução Normativa nº 03/2014. Através do código
de autenticação abaixo, qualquer cidadão poderá verificar a autenticidade ou regularidade deste documento, por meio da página do Sisbio/ICMBio na
Internet (www.icmbio.gov.br/sisbio).
Código de autenticação: 48814279Página 1/3
Ministério do Meio Ambiente - MMAInstituto Chico Mendes de Conservação da Biodiversidade - ICMBioSistema de Autorização e Informação em Biodiversidade - SISBIO
Autorização para atividades com finalidade científicaNúmero: 41329-7 Data da Emissão: 17/02/2017 09:50 Data para Revalidação*: 19/03/2018
* De acordo com o art. 28 da IN 03/2014, esta autorização tem prazo de validade equivalente ao previsto no cronograma de atividades do projeto,mas deverá ser revalidada anualmente mediante a apresentação do relatório de atividades a ser enviado por meio do Sisbio no prazo de até 30 diasa contar da data do aniversário de sua emissão.
SISBIODados do titular
Nome: RENATA SANTORO DE SOUSA LIMA MOBLEY CPF: 705.712.446-53
Título do Projeto: Ecologia acústica do lobo-guará (Chrysocyon brachyurus) no Parque Nacional da Serra da Canastra ? MG
Nome da Instituição : UFRN - UNIVERSIDADE FEDERAL DO RIO G. NORTE CNPJ: 24.365.710/0001-83
Locais onde as atividades de campo serão executadas# Município UF Descrição do local Tipo
1 MG PARQUE NACIONAL DA SERRA DA CANASTRA UC Federal
Atividades X Táxons# Atividade Táxons
1Observação e gravação de imagem ou som de taxon em UCfederal
Canidae
Material e métodos1 Método de captura/coleta (Carnívoros) Bioacústica
Destino do material biológico coletado# Nome local destino Tipo Destino
1 UFRN - UNIVERSIDADE FEDERAL DO RIO G. NORTE
SISBIOEste documento (Autorização para atividades com finalidade científica) foi expedido com base na Instrução Normativa nº 03/2014. Através do código
de autenticação abaixo, qualquer cidadão poderá verificar a autenticidade ou regularidade deste documento, por meio da página do Sisbio/ICMBio na
Internet (www.icmbio.gov.br/sisbio).
Código de autenticação: 48814279Página 2/3
Ministério do Meio Ambiente - MMAInstituto Chico Mendes de Conservação da Biodiversidade - ICMBioSistema de Autorização e Informação em Biodiversidade - SISBIO
Autorização para atividades com finalidade científicaNúmero: 41329-7 Data da Emissão: 17/02/2017 09:50 Data para Revalidação*: 19/03/2018
* De acordo com o art. 28 da IN 03/2014, esta autorização tem prazo de validade equivalente ao previsto no cronograma de atividades do projeto,mas deverá ser revalidada anualmente mediante a apresentação do relatório de atividades a ser enviado por meio do Sisbio no prazo de até 30 diasa contar da data do aniversário de sua emissão.
SISBIODados do titular
Nome: RENATA SANTORO DE SOUSA LIMA MOBLEY CPF: 705.712.446-53
Título do Projeto: Ecologia acústica do lobo-guará (Chrysocyon brachyurus) no Parque Nacional da Serra da Canastra ? MG
Nome da Instituição : UFRN - UNIVERSIDADE FEDERAL DO RIO G. NORTE CNPJ: 24.365.710/0001-83
Registro de coleta imprevista de material biológicoDe acordo com a Instrução Normativa nº 03/2014, a coleta imprevista de material biológico ou de substrato nãocontemplado na autorização ou na licença permanente deverá ser anotada na mesma, em campo específico, porocasião da coleta, devendo esta coleta imprevista ser comunicada por meio do relatório de atividades. O transporte domaterial biológico ou do substrato deverá ser acompanhado da autorização ou da licença permanente com a devidaanotação. O material biológico coletado de forma imprevista, deverá ser destinado à instituição científica e, depositado,preferencialmente, em coleção biológica científica registrada no Cadastro Nacional de Coleções Biológicas (CCBIO).
Táxon* Qtde. Tipo de amostra Qtde. Data
* Identificar o espécime no nível taxonômico possível.
SISBIOEste documento (Autorização para atividades com finalidade científica) foi expedido com base na Instrução Normativa nº 03/2014. Através do código
de autenticação abaixo, qualquer cidadão poderá verificar a autenticidade ou regularidade deste documento, por meio da página do Sisbio/ICMBio na
Internet (www.icmbio.gov.br/sisbio).
Código de autenticação: 48814279Página 3/3