UNIVERSIDADE DE SÃO PAULO
FACULDADE DE FILOSOFIA, CIÊNCIAS E LETRAS DE RIBEIRÃO PRETO
FLORA SILVEIRA BALIEIRO
The voice of the not so lonely maned wolf: evidence of individual
discrimination via playback of the longdistance extendedbark
FLORA SILVEIRA BALIEIRO
The voice of the not so lonely maned wolf: evidence of individual
discrimination via playback of the longdistance extendedbark
Dissertação apresentada à Faculdade de
Filosofia, Ciências e Letras de Ribeirão Preto
da Universidade de São Paulo, como parte dos
requisitos para obtenção do grau de Mestre
em Ciências.
Área de concentração: Psicobiologia
Orientadora: Profª. Dra. Patrícia Ferreira
Monticelli
Autorizo a reprodução e divulgação total ou parcial deste trabalho, por qualquer meio convencional ou eletrônico, para fins de estudo e pesquisa, desde que citada a fonte.
Balieiro, Flora Silveira
The voice of the not so lonely maned wolf: evidence of individual discrimination via playback of the longdistance extendedbark . Ribeirão Preto, 2015.
60p. : il. ; 30 cm
Dissertação de Mestrado, apresentada à Faculdade de Filosofia Ciências e Letras, Ribeirão Preto/USP. Área de concentração: Psicobiologia.
Orientadora: Ferreira Monticelli, Patrícia.
1. Comportamento Animal. 2. Comunicação Acústica.
Nome: Balieiro, Flora Silveira
Titulo: The voice of the not so lonely maned wolf: evidence of individual discrimination via playback of the longdistance extendedbark
Dissertação apresentada à Faculdade de Filosofia, Ciências e Letras de Ribeirão Preto da Universidade de São Paulo, como parte dos requisitos para obtenção do grau de Mestre em Ciências.
Aprovado em:
Banca Examinadora
Prof. Dr. ___________________________________________________________________ Instituição: _________________________________ Assinatura: ______________________ Prof. Dr. ___________________________________________________________________ Instituição: _________________________________ Assinatura: ______________________ Prof. Dr. ___________________________________________________________________ Instituição: _________________________________ Assinatura: ______________________
To the Maned Wolf, this amazing canid, that deserves to be understood, respected and protected.
ACKNOWLEDGMENTS
I am thankful to the ones who contributed to my “unique history” as a scientist.
À minha orientadora Patrícia pela oportunidade de trabalho e pelo apoio durante esses primeiros passos rumo à minha carreira acadêmica. A toda a equipe do Laboratório de Etologia e Bioacústica, em especial à Juliana pelas gravuras impecáveis e às companheiras de campo, Lilian e Laís, cujas presenças durante as campanhas sem dúvida tornaram muito mais curtas as longas horas de espera na madrugada. À minha família, em especial à minha irmã Lara pela primorosa revisão e consultoria linguística durante o processo árduo da redação final; à minha irmã Lia pela paciência e compreensão durante minhas ausências justificadas pelo trabalho; ao meu marido Rafael por me apoiar sempre e facilitar meu cotidiano, bem como todas as outras dimensões da minha vida; aos meus pais Ari e Valéria, que me ensinaram a pensar como cientista e a agir com justiça.
Também agradeço aos estagiários de campo, de póscampo e aos juízes voluntários da minha “gestalt”. Aos Professores Luiz Carlos Vulcano e Carlos Roberto Teixeira pelas imagens de tomografia computadorizada do loboguará e do cão doméstico. Aos zoológicos: Parque Ecológico de São Carlos, Parque Zoológico Municipal Quinzinho de Barros, Zoo Bosque Ribeirão Preto, Zoológico Municipal de Americana, Zoológico Municipal de Catanduva, Zoológico Municipal de Paulínia, Zoológico Municipal de Piracicaba, Zoológico Municipal de São José do Rio Preto, pela permissão de acesso aos lobos. À CAPES, pela concessão da bolsa de mestrado, à Universidade de São Paulo e ao Programa de PósGraduação em Psicobiologia pelo apoio na realização deste trabalho.
“The scientist, like any organism, is the product of a unique history” (Skinner, 1956)
ABSTRACT
The acoustic channel is an efficient longdistance signaling system that may be especially effective for animals with crepuscular/nocturnal habits. The maned wolf is a threatened canid with crepuscular/nocturnal habits that is thought to be a solitary species in common sense. As a matter of fact, it would be better defined as a gregarious species, since male and female share the same wide territory and the spatial distance between them varies according to the female’s reproductive period. The maned wolf’s extendedbark is a longdistance vocalization that functions as a mechanism to increase spatial distance among conspecifics as well as to enable pairmates to find each other. Individual variations in this vocalization have been reported, but the possibility that they can be perceived and used by the species has never been tested. One should expect these individual variations to be perceived by the species, since only in this scenario it would be plausible for the extendedbark to have the dual function stated above. If this individual variability is not perceived by conspecifics, the efficiency of this vocalization at long distances, at least for the hypothesized functions, would be compromised, as the hearer would not be able to identify if the sender is its reproductive partner or a possible rival. In our study we used playbacks to test if these individual variations can be perceived by captive wolves and have concluded that they can. To our knowledge, this is the first time it has been demonstrated that the maned wolf is capable of discriminating among extended barks of different individuals.
Key Words: habituationdiscrimination; Chrysocyon brachyurus ; individual discrimination; vocal communication; acoustic structure; acoustic variability.
SUMMARY
ABSTRACT
1 INTRODUCTION
2 METHODS
2.1 Study Sites and Animals
2.2 Animal ethics and Welfare
2.3 Recordings
2.4 Acoustic Analysis
2.4.1 Measurements
2.4.2 Potential for Individuality Coding (PIC)
2.5 Pilot Phase
2.6 Description of the Maned Wolf's vocal tract
2.7 Hypothesis Test
2.7.1 Habituation–discrimination paradigm
2.7.2 Playback Procedure
2.7.3 Standard Playback Protocol
2.7.3.1 Criteria of Response
2.7.3.2 Procedure
2.7.4 Ethogram
2.7.5 Complementary Playback Protocol
2.7.5.1 Criteria of Response
2.7.6 Other Procedures
2.7.6.1 Reproductive pair discrimination test
2.7.6.2 Background noise effect evaluation
2.7.6.3 Stimulus control test
2.7.6.4 Playback tracks preparation
2.7.6.5 Equipment Quality Procedure Test
3 RESULTS
3.1 Hypothesis Test
3.1.1 Standard Playback Protocol
3.1.2 Complementary Playback Protocol
3.1.2.1 Response to stimuli
3.1.2.2 Validity of the “gestalt” Disturbance
3.1.3 Other Results
3.1.3.1 Reproductive pair discrimination test
3.1.3.2 Background noise effect evaluation
3.1.3.3 Stimulus control test
3.1.3.4 Equipment Quality Procedure Test
3.1.3.5 Recordings
3.1.3.6 Acoustic Analysis
3.1.3.7 Pilot Phase
3.1.3.8 Description of the Maned Wolf's vocal tract
4 DISCUSSION
5 REFERENCES
1 INTRODUCTION
The maned wolf ( Chrysocyon brachyurus , Illiger, 1815) is the largest South American canid inhabiting open vegetation areas, mainly savanna’s phytophysiognomy (Batalha, 2011) of Cerrado (Dietz, 1985; Bueno, 2002). It occurs in Argentina, Brazil, Bolivia, Paraguay, Peru and Uruguay in a distribution range estimated in 5 million square kilometers (Dietz, 1985; Bueno, 2002). The species is listed as “Nearly Threatened” by the International Union for Conservation of Nature (IUCN) red list of threatened species (Rodden et al., 2008) and as “Vulnerable” in the Species List of the Brazilian Fauna Threatened with Extinction Species List ( MMA, 2014 ). In Brazil, the main threats to the species are the loss of the original Cerrado areas (Myers et al ., 2000) due to the expansion of agriculture and livestock (Fonseca et al. , 1994; Klink & Moreira, 2002), the roadkills on highways, diseases spread by domestic dogs and illegal hunting committed by farmers to protect their livestock (Dietz, 1984; Rodden et al., 2008).
The maned wolf is thought to be a solitary species in common sense. It is a monogamous species and male and female usually share the same wide territory (Kleiman, 1972; Dietz, 1984), ranging from 20 to 115 km² (Dietz, 1984, 1985; Carvalho & Vasconcellos, 1995; Melo et al. , 2007). As a matter of fact, the species would be better defined as gregarious species with level two of sociality ( i.e. aggregation for at least one activity, sensu Lee, 1994), since the members of a reproductive pair are scarcely found together during activity periods (Dietz, 1984), but still may share resting places and sleep together in natural areas (Melo et al. , 2007). The spatial distance between male and female varies according to the female’s reproductive period, as the proestrus is a social oncoming period and anestrous is characterized by mutual avoidance (Dietz, 1984).
signals, including the loud "singlebark" and the "roarbark" (called "singlebark" by Kleiman, 1972; and "extendedbark" by Dietz, 1984, which is the term we chose to adopt in our work). The extendedbark resembles an extended variation of the bark of a large domestic dog (Dietz, 1984; see Figure 6 in Methods) and is emitted in sequences both by males and females. Kleiman (1972) has suggested that the extendedbarks function as a spacing mechanism among individuals, such as the roars of the great cats. It has also been noticed to be produced in duets between couples (Dietz, 1984; personal observations). Since its emission rate increases during the breeding season, it has been suggested that it may also have a role in aiding the localization between pairmates (Dietz, 1984).
The maned wolf's body is large and requires a lot of energy supply. As its food is scarce and dispersed (mostly fruits and small vertebrates such as rodents and birds Bueno et al. , 2002) the best strategy for this species is to hunt alone and use mutually exclusive foraging areas. It has been suggested that the lack of exposure to factors promoting sociality has made this social structure relatively inflexible and turned to be a genetic constraint to the extent of cooperative social interaction among individuals (Dietz, 1984). If a strict territoriality has increased the survival of individuals, then natural selection must have favoured characteristics which emphasize the property of an area (Dietz, 1984). In fact, the maned wolf exhibits scentmarking behaviours with urine and faeces and produces a long distance vocalization (the extendedbark), which are related to territorial defense in close proximity and at longer range, respectively (Dietz, 1984; Kleiman, 1972).
2002). It has been demonstrated that individuals of some species are capable of distinguishing familiar vocalizations from unfamiliar ones, even when they are far away from the sender ( e.g. Gray wolf Canis lupus : Tooze et al ., 1990; Rhesus monkey Macaca mulatta : Rendall et al ., 1996; Lion Panthera leo : Grinnell & McComb, 2001; African Elephant Loxodonta africana : McComb et al ., 2000). It has even been reported foraging group signature in the social and cooperative spearnosed bat Phyllostomus hastatus (Boughman, 1997), favoring membership pertencement signaling and pair recognition.
The extendedbark of the maned wolf is a loud vocalization whose abrupt start and broadband frequency favor its perception and its location source (McGregor, 2005). Thereby, we hypothesized that the extendedbark of the maned wolf might have sufficient individual variations if involved in provision of clues about the sender’s identity, either to permit reproductive partner recognition or to signal territory property. In fact, Brady (1981) has reported individual variations in maned wolf extendedbarks and stated that a human would distinguish individual differences in extendedbarks at a distance of 1 km. Nevertheless, the possibility that this variability can be used by wolves for individual recognition has never been tested. One should expect these individual variations to be perceived by the species, for only in this scenario it would be plausible for the extendedbark to function as a mechanism to increase spatial distance among conspecifics as well as to enable pairmates to find each other. If this individual variability is not perceived by conspecifics, the efficiency of this vocalization at long distances would be compromised, as the hearer would not be able to identify if the sender is its reproductive partner or a possible rival. We used playbacks to test if these individual variations can be perceived by the species. To our knowledge, this is the first time it has been demonstrated that the maned wolf is capable of discriminating among extended barks of different individuals.
2 METHODS
2.1 Study sites and animals
total number of wolves visited in different zoos of São Paulo state, Brazil, were 16 (Table 1). As not all of them produced extendedbarks, our results are related to extendedbarks produced by 6 adult wolves (four males and two females) and playback trials were conducted with 10 (one couple and eight solitary males). The enclosures in the zoos were often open for visitors, hence the wolves were accustomed to human presence (exceptions are presented in footnotes of Table 6).
2.2 Animal ethics and welfare
The maned wolf is catalogued as “Nearly Threatened” by the IUCN red list of threatened species (2015), and we obtained official research license from the Brazilian government (Sisbio, permit number 410283). The research was approved by the Ethics Commission for the Use of Animals in research (CEUA) of University of São Paulo (Protocol number 14.1.153.53.8), and permissions were obtained from the person in charge of each zoo. Both recording and playback procedures were conducted in accordance with the Guidelines of the American Society of Mammalogists (Sikes et al. , 2011).
Table 1. Wolves that participated in this study. They were held in captivity in different zoos of São Paulo state, Brazil, isolated or in pairs. Sexes are presented as M = male, F = female. The contribution of each animal in the different phases of the research is presented in the last column.
CITY / STATE HEADQUARTERS OF
DATA GATHERING
SUBJECT (INITIALS) SEX
PHASE OF STUDY IN WHICH PARTICIPATED AS
SUBJECT
Botucatu/ SP Wolverine (WO) M Playback
Botucatu/ SP Logan (LO) M Playback
Catanduva/ SP Mel (ME) F Recording
Ribeirão Preto/ SP Orfeu (OR) M Recording
São Carlos/ SP Frutti (FR) F Recording
São Carlos/ SP Tutti (TU) M Pilot, Recording and Playback
São José do Rio Preto/ SP Mostarda (MO) M Recording
Sorocaba/ SP Fantasma (FA)*** M Pilot, Recording and Playback
Sorocaba/ SP Nymeria (NY)*** F Pilot, Recording and Playback
Sorocaba/ SP Perônio (PE)*** M* Playback
Sorocaba/ SP Tíbio (TÍ)*** M* Playback
Americana/SP Colombo (CO) M Recording and Playback
Paulínia/SP Pluto (PL) M Playback
Piracicaba/SP Lupin (LU) M Recording and Playback
Itirapina/SP Mogli (MOG) Unknown Pilot
*All the subjects, except Perônio and Tíbio, marked with a star, were more than 2 years old and, therefore, are considered adults (Rodden et al. , 2004). **Subject Hortelã was a freeliving male (unofficial information from Zoo’s employees and personal verification) at the cerrado reserve bordering the Zoo of São Carlos that occasionally visited Tutti and Frutti's enclosure. For further details, see topic on pilot phase. *** The only wolves that shared the same enclosure were: (1) FA and NY; (2) TÍ and PE. TU and FR were kept apart most of the time, but shared the same enclosure during part of the day. The rest of the wolves did not share enclosures.
2.3 Recordings
Cesar Ades (FOCA), the neotropical mammals sound library of our institution, for future reference.
Field work took place from 2013 to 2015 in 13 short field trips of approximately two weeks each. Recording sessions were taken in each zoo continuously from 17:00 to 07:00 (UTC3) within the period of greatest activity of the species, as described in literature (Dietz, 1984). Daylight saving time was disregarded. Vocalizations of interest were recorded ad libitum at a minimum distance of 7 to no more than 15 meters, except for FR that was impossible to be approached to less than 40 meters. Whenever possible, we made sound pressure level measurements (of vocalizations and environmental noise in dB) using a sound level meter (Instrutemp, ITDEC4080; range 30dB to 130dB). At the end of each recording I dictated the following information: (1) date and time; (2) distance from the subject; (3) place and geographic coordinates; (4) environment (captivity); (5) temperature and relative humidity (when possible); (6) scientific and common name of the species ( Chrysocyon brachyurus , Maned Wolf); (7) subject identity; (8) gender; (9) female’s estrous stage; (10) age (adult or juvenile); (11) detailed behavioural description of the individual that vocalized (before, after and during vocalization), and a description of the behaviors presented by enclosure mates (when applicable); (12) equipments used.
In order to complement information about vocal activity of the subjects, we also made remote sound recordings using automated digital recorders (Wildlife Acoustics, SongMeter 2+). Each mobile unit can be programmed to record automatically on a schedule, so it does not require the researcher's presence for gathering data. The recorders were settled inside the enclosures and adjusted to work continuously from 17:00 to 07:00. Each twochannel unit holds two omnidirectional microphones (Wildlife Acoustics, SMXII) whose gains can be set independently. To protect recordings from distortion of the signal through clipping , we 1 always let a difference of at least 6 dB between the gain of both microphones. Other recording parameters were the same as used in manual recordings (see above). The amount of data collected was then scanned manually in Song Scope software (Wildlife Acoustics). We also used some of these vocalizations automatically acquired as stimuli for some of the playback tests (see “Standard Playback Protocol” for details).
1 Clipping is a distortion of the signal due to severe over loading during sound recording or further analysis and it can be easily detected in the oscillogram. (Wilden et al. , 1998).
2.4 Acoustic Analysis
Brady (1981) stated that the acoustic structure of the extendedbark notes vary between wolves, but is fairly constant at individual level. He has reported sharp and consistent differences in the number of emphasized frequencies among individuals’ extendedbarks, but has not deeply investigated what other possible acoustic variables can contribute to the vocal individuality. We explored the potential to promote vocal individuality of four acoustic variables selected in both temporal and frequency domains (Table 2).
Table 2 Description of the acoustic variables for measured to investigate the potential for individual coding (PIC). All measurements were made in Raven Pro version 1.4. Delta Time and 3rd Quartile Frequency are considered “robust signal measurements”, which means that they do not vary much based on the exact bounds of the selection created by the user (Cornell Lab of Ornithology, 2014).
Variable Description
Average Entropy The degree of disorder ( i.e. chaos) in the note.
Delta time (s) Duration of the note ( the end of the note). i.e. the temporal difference between the beginning and
3rd Quartile Frequency (Hz) The frequency that divides the note into two frequency intervals containingrespectively 75% and 25 % of the energy distribution.
Max Frequency (Hz) The frequency at which the maximum energy occurs ( frequency or F0). i.e. the fundamental
2.4.1 Measurements
21.5 Hz (frequency grid spacing); 70% (brightness); 30% (contrast). The accuracy of the automated measurements extracted by Raven Pro was confirmed by visual assessment (Fischer et al. , 2013) and all note selections ( i.e. the rectangle displayed in the software that delimits the area to be measured) were done by the same researcher. For each note Raven Pro generated: (1) an audio file containing the note alone; (2) a table in “txt” format containing all measures and the identity of the wolf that vocalized. These tables were automatically unified with a software made at the Ethology and Bioacoustics Research Laboratory (EBAC) for this purpose. Taking the advice of Fischer and colleagues (2013) we used a hierarchical naming system to the files. Such system allows every note to be tracked back to the sequence and the original file it came from. Only notes with a good signaltonoiseratio entered the analysis and, to avoid pseudoreplication, we picked notes from different recording sessions whenever possible. We filtered two frequency bands that appeared in a recording from a duet between FR and TU using Raven Pro (Bandstop Filters: Lower Limit = 2250 Hz; Upper Limit = 2550 Hz; Lower Limit = 4000 Hz; Upper Limit = 4650 Hz) none of these sequences of notes were used in the playback phase.
2.4.2 Potential for Individuality Coding (PIC)
For each acoustic parameter measured, the betweenindividual and withinindividual coefficients of variation (CVb and CVw) were calculated according to the following formula:
CV = 100
×
(
1+14N)
×
(
X SD
)
For each parameter, CVw was evaluated as the mean value of individual CVs. Then we defined the potential for individuality coding (PIC) as the CVb/CVw ratio. For a given parameter, a PIC value greater than 1 indicates that this parameter has potential to be used in individual recognition, this is because its intraindividual variability is smaller than the interindividual variability (de la Torre et al. , 2015). Calculations were made on R program version 3.1.3 (see Appendix 1 for script details).
2.5 Pilot Phase
The first step of our ethological analysis started with extensive observation of maned wolf behaviour. For logistical reasons we could not make preliminary observations of every wolf that participated as subject in the playback phase. Despite that, we gathered enough hours of observation (400 hours approximately) to accurately describe behaviour patterns of the species. The observations were made mostly at night (together with the recording phase), including a photic period from 17:00 to 19:00 and another one from 06:00 to 07:00 in some occasions the field work had to be extended to later hours in the morning, since the wolves were still active. The subjects on which we made the preliminary behavioural observations were the reproductive pair FA and NY, from Sorocaba’s Zoo (which had artificial lighting at night) and TU from São Carlos’ Zoo (in which we used a flashlight and counted on full moon illumination). Sampling rules chosen to preliminary observation were “ad libitum” (Altmann, 1974), together with behaviour sampling for vocalization events. We used “all occurrences” as recording rule in both cases (Martin et al., 1993).
In addition to the captivity wolves, we also had the chance to observe two wild individuals. The first (HOR) is a free animal that lives in the cerrado reserve bordering the Zoo of São Carlos and often entered the Zoo through a hole in the outer fence. We assume it is a male from observation and unofficial information from Zoo’s employees. The second wolf (MOG) inhabits the Ecological Station of Itirapina and its surroundings.
2.6 Description of the Maned Wolf's vocal tract
2.7 Hypothesis Test
For the maned wolf to be capable of distinguishing its reproductive partner from a possible competitor through the extendedbark, there might be individual variations in this vocalization that provide clues about the sender’s identity.
H 1 : In case our hypothesis is valid, one can expect that the wolves will be able to distinguish
among different conspecifics’ extendedbarks.
H 0 : The wolves will not be able to distinguish among different conspecifics’ extendedbarks.
2.7.1 Habituation–discrimination paradigm
In order to test the prediction that maned wolves are able to discriminate among extendedbarks of different individuals, we used the habituationdiscrimination paradigm (Friedman, 1972; Cheney & Seyfarth, 1988; Rendall et al ., 1996; “habituationdishabituation” Hauser, 1998; Reby & McComb, 2003; “habituationdishabituation” Mumm et al. , 2014; Palacios et al. , 2015). Habituation is defined as a gradual decrease in the magnitude and/or frequency of the response due to repeated presentations of a stimulus (Catania, 1999; Rankin et al. , 2009 ). When the response is extinct or, at least, decreased enough in magnitude to consider the subject habituated in the presence of a stimulus and maintained or recovered in the presence of another stimulus, one can say that discrimination occurred (Holland & Skinner, 1975).
The typical habituationdiscrimination paradigm consists on presenting repeatedly a given stimulus ( i.e. habituation stimulus; e.g. extendedbark from individual ‘A’) at fixed time intervals until the subject habituates to it, followed by the presentation of a different stimulus ( i.e. test stimulus; e.g. extendedbark from individual ‘B’). If this subsequent exposure to the test stimulus immediately increases the magnitude and/or frequency of the response, discrimination between stimuli A and B occurred.
2.7.2 Playback Procedure
Sessions started within the period of activity of the species (Dietz, 1984), in early morning (6:00), to take advantage of photoperiod for video documentation. We waited approximately 10 minutes in silence before starting each test, to familiarize the subjects to the equipments and to our presence. Acoustic stimuli were played back using a power speaker (Boxer M10; output power 100 W; impedance 48 Ω; frequency range 0,05 kHz 20 kHz). The speaker was positioned 30 m away from the subjects and out of their sight. Volume level was adjusted to ensure the stimuli were audible by the subjects and simulated the vocalization contexts we witnessed during the recording phase. In order to prevent sound attenuation by atmospheric absorption (Wiley & Richards, 1978), we avoided days with high humidity. And to prevent changes in the wavesound caused by ground attenuation (Wiley & Richards, 1978), we placed the powered speakers at a distance of approximately 0.5 m from the ground (simulating a typical context of vocalization). During the tests, we documented the wolves’ behaviours with a Sony Handycam for couple FA and NY we used two cameras, one for each wolf. After each test we made sound pressure level measurements of the playback sound and without it (Sound Level Meter Instrutemp, ITDEC4080; range: 30dB to 130dB).
2.7.3 Standard Playback Protocol
To test whether the maned wolf is able to discriminate among extendedbarks of different individuals (A and B) we adopted the following playback protocol (modified from Rendall et al. , 1996):
(1) Habituation phase: we used a sequence of 6 extendedbarks (A 1 , A 2 , A 3 , A 4 , A 5 , A 6 ) from wolf A as the habituation stimulus. The sequence was presented to the subject repeatedly until it was habituated (for details see Criteria of Response). We chose to use a sequence as stimulus, rather than a single extendedbark, to expose the subject to six different samples of stimulus from class A. Thus, the subject was habituated to the individual content of wolf A's extendedbarks, not to the particular acoustic characteristics of a single note, which is more faithful to the natural situation of emission.
subject does not recover from habituation, we assume it is habituated to the acoustic features virtually present in all stimuli from class A. Otherwise, our postulate is discarded and we conclude that the subject has only habituated to the habituation series ( i.e. A 1 , A 2 , A 3 , A 4 , A 5 , A 6 ). We chose to use three new stimuli rather than a single one. This way we avoid the possibility of the subject not having heard the new stimulus. (3) Discrimination phase: after the habituation control phase, we broadcast a single
stimulus from wolf B. Stimulus B is also new, it only does not share the acoustic characteristics to which the subject is habituated ( i.e. those virtually present in class A stimuli). Therefore, if stimulus B elicits a response similar to that obtained in the beginning of the habituation phase ( i.e. before the wolf was habituated), we assume it has discriminated between extendedbarks from wolves A and B.
(4) Rehabituation phase: at last we present a new stimulus from class A ( i.e. A 10 ). If the subject rehabituates ( i.e. we observe a diminution of response magnitude or no response anywise), we endorse the statement that the subject is able to discriminate between A and B. Albeit we could reexposure the subject to a stimulus of type A that had already been used, we preferred to broadcast a new one (A 10 ) as a way to increase experimental control.
To avoid violating the intervals found between extendedbarks from a natural sequence or duet, the intervals between two stimuli were no longer than 10 seconds (Figure 1).
difference among the types of background noise each track contained, we inserted background noise extracted from track B (background noise B) in the habituation soundtrack and just before stimulus B. This strategy allowed us to know if the subjects’ behaviour altered in response to the change of stimulus ( i.e. from A to B) or due to the background noise change. If the subject did not react to the inserted fragments of background noise, but to the stimulus B, we assumed there had been discrimination between A and B. Only soundtracks played to the subjects TU, FA and NY contained stimuli recorded with different equipments. The spectrogram was generated in Raven Pro 1.4. ( Window type = Hann; Window Size = 900 samples; 3dB filter Bandwidth = 70.5 Hz; Overlap = 99.7; Hop Size = 3 samples; Frequency Grid DFT size = 2048 samples; Grid Spacing 21.5 Hz; Screen brightness = 50; Screen Contrast = 50) and the marks were made with Photoshop CS6.
2.7.3.1 Criteria of Response
During the habituation phase of each playback trial we studied the responses shown by the subjects in order to judge which could be elected as good estimators of the wolves’ reaction to the stimuli. We adopted the following criteria:
(1) the response must have followed an habituation pattern ( i.e. gradually decreased in magnitude and/or frequency after repeated presentations of a stimulus);
(2) the occurrence of the response must have been easy to identify by an untrained observer; (3) the duration of the response must be brief (less than 10 seconds), since the interval between the discrimination stimulus and the rehabituation stimulus was of about 10 seconds.
The response that best fulfilled all requirements was “Oriented Attention” (see Ethogram for description).
2.7.3.2 Procedure
In order to comply with the habituationdiscrimination paradigm we only analyzed tests in which the subjects displayed the response Oriented Attention in the beginning of the habituation phase. However, any response to the test stimulus ( i.e. B) might have been influenced by the fact that the subject had already heard extendedbarks. Thus, to consider the subject habituated during playback procedures, both the frequency of the Oriented Attention response and the magnitude of the subject’s overall “disturbance state” should decrease.
S ince we used Oriented Attention as a measure of the response, we only played the test soundtrack ( i.e. A 7 , A 8 , A 9 , B, A 10 ) when the subject was not facing the speaker and when no other noticeable acoustic stimuli was present.
the audio’s waveforms on the same screen, which allowed us to accurately visualize the timing between stimulus (sound) and response (image). To measure the subjects’ responses to the stimuli we used a single measure: if the behaviour Oriented Attention occurred or not. Thus, discrimination between A and B happened if the subject displayed Oriented Attention only in the beginning of the habituation phase and during/immediately after B. We could also have measured latency and/or duration of Oriented Attention or even included more behaviours as responses. However, since experimental conditions widely vary among subjects and trials, it would only give us extra information about the subject and/or individual trials. Since we are interested in understanding if the subject responds more than how it responds to the stimuli, other measures did not reach significance.
The magnitude of the wolves overall “disturbance state” were examined under the concept of “gestalten”, proposed by Konrad Lorenz (1962, 1960 apud Hutt & Hutt, 1974), which is the notion of certain responses seeming to follow a pattern that can be easily recognized. The practical notion of a “gestalt” comes with experience on behavioural observation. The experience leads to the ability of identifying natural units and grouping them according to some regularity. Thus, “gestalten” are groups whose natural units are highly redundant in some way. In our work one “gestalt” corresponds to a set of responses that characterizes a state of general perturbation in the animal’s regular state, labelled here “Disturbance”. Therefore, each animal presents a particular set of responses that characterizes its Disturbance (see Ethogram). During habituation, the magnitude of the “gestalt” Disturbance must decrease.
To prevent “experimenter bias” we elected five untrained observers to judge the
validity of our “gestalt” Disturbance. Each judge watched video excerpts from CO’s playback tests. Four of the excerpts corresponded to moments in which we considered the subject was “very perturbed” ( i.e. high magnitude level of Disturbance). The other four excerpts were from moments we considered it “less perturbed” ( i.e. lower magnitude level of Disturbance). We randomly mixed the video excerpts and, as a result, we obtained four pairs of video containing excerpts we judged as different “gestalten”, one of them corresponding to the transition from stimuli A to stimuli B.
mute mode the original audio contained the researcher’s narration of the test, which could induce their judgement. To determine interobserver agreement, we used Fleiss’ Kappa statistic calculated with and without author’s scores. Fleiss’ Kappa statistic (k) varies from 0 (no agreement or agreement exactly as expected by chance) to 1 (full agreement), while equal to 0 indicates no agreement or that the agreement was exactly as expected by chance. To determine the strength of agreement, we adopted Landis & Koch (1977) benchmarks (Table 3).
Table 3 Strength of agreement of Fleiss’ Kappa statistic (k) by Landis & Koch (1977). The second column gives the interpretation for the k value.
Kappa Statistic Interpretation
< 0.00 Poor agreement
0.000.20 Slight agreement
0.210.40 Fair agreement
0.410.60 Moderate agreement
0.610.80 Substantial agreement
0.811.00 Almost Perfect agreement
2.7.4 Ethogram
We built the ethogram based on observations from pilot and playback phases (Table 4).
Table 4 Behaviours and their assumed disturbance value as defined during preliminary observations of a captive wolf and following the literature. The third column indicates the contribution of the response to the “gestalt” disturbance, as evaluated by 5 untrained judges, being positive when it increases and negative when decreases in its magnitude.
Natural Unit Description Disturbance
Piloerection When body hair is spiky noticeable in video (Figure 2). +
Relaxed Pelage When body hair is not spiky noticeable in video.
Oriented Attention
Wolf turns its head towards the speaker and remains with ears raised, eyes wide open and outer ears facing the speaker during at least 1 s. The reaction can occur just after the stimulus or have a latency of maximum 4 seconds. (Adapted from Palacios et al ., 2015) (Figure 3).
+
Attention Wolf turns its head towards any direction but the speaker’s and
remains with ears raised and eyes wide open during at least 1 s. +
Ear scan Wolf turns one or both ears towards the speaker (Figure 4). +
Fast Walk
Wolf suddenly increases speed of walk. It differs from running or trote since it is a locomotion without an inair phase, in which hindlegs do not extend forward past the midline of the body (Adapted from Fletchall et al. , 1995).
+
Pacing
Wolf comes and goes compulsively in a small space describing a predictable route, with no apparent function. Mouth is usually half open and tail is relaxed (Vasconcellos, 2009).
Perambulate The animal walks through an apparently random route. Mouth is
usually half open and tail is relaxed.
Approach speaker Wolf walks towards speaker. +
Move away from
speaker Walks away from speaker. +
Marking Urinates, defecates, or rubs itself on the enclosure’s vegetation
(Coelho et al ., 2011). +
Lick the nose Passes tongue over the nose. +
Sniff the air Draws air into the nose supposedly to smell something, nose is up and
ears are usually facing forwards. +
Sniff enclosure Draws air into the nose supposedly to smell a specific place of the
enclosure.
Eat Selfexplanatory.
Yawn
Involuntarily opens its mouth widely and inhales deeply, ears are usually back. Could happen on playback context supposedly due to anxiety.
+
Rest Lies down with raised body or not, occasionally moving the ears ,
with closed or semienclosed eyes.
Rest alert Lies down with raised body, ears turning forwards or moving laterally
and eyes wide open. +
Charge/lunge* Advances towards another wolf, piloerection, stiff forelegs, ears back
( Fletchall et al. , 1995 ). +
Gape* Open mouth, ears back, oriented towards other wolf; often
accompanies charge ( Fletchall et al. , 1995 ). +
Growl* Growls at conspecific ( Fletchall et al. , 1995 ). +
Hunt
Jumps over something on the ground with both forelegs, ears towards the ground. Considered as a ludic activity and assigned as negative value for our purposes.
Selfgroom Scratches or licks its own body (Adapted from Fletchall et al. , 1995 ).
Play ** Manipulates objects with front paw or mouth.
Refuge Runs or walks fast towards the enclosure’s den. +
Harassed Posture in which the hip is slightly lowered relatively to the line of the
back, leading the tail to be partially between the legs. +
Bark (Extendedbark)
Wolf emits a sequence of extendedbarks. The wolf is usually standing up with its tail relaxed. Between one note and another its ears rotate in a back and forth movement as if searching for sounds (Figure 5).
+
Whine Wolf whines. Ears can be raised or flattened and turned backwards. If
the wolf is standing up its tail is usually up. +
* Valid only for the wolves FA and NY that were held in the same enclosure. ** Valid only for TU session, in which there was a tire as enrichment object.
Figure 2 Piloerection. The spiky hair is specially pronounced in the wolf’s mane, lower back and tail (Drawing made by Juliana Takata based on video excerpts of our playback tests).
Figure 4 Ear Scan. Wolf turns one or both ears towards the speaker. The ear movement is indicated by the arrow in the drawing (Drawing made by Juliana Takata based on video excerpts of our playback tests).
Figure 5 Bark. Wolf emits a sequence of extendedbarks. The wolf is usually standing up with its tail relaxed. Between one note and another its ears rotate in a back and forth movement as if searching for sounds (Drawing made by Juliana Takata based on author’s verbal description).
2.7.5 Complementary Playback Protocol
The first playback session of the subject CO was harmed by confounding variables: loud background noise, frequent human interference, dog barking and others. Under such adverse conditions the subject did not show clear response to the stimuli change in the standard protocol. One of the possible reasons to the absence of response was that the subject did not perceive the B stimulus along the track, since there were too many distractions during the test. Another possibility was that the subject did not have enough time to react to B stimulus, since there was about 5 seconds between it and the next stimulus ( i.e. A 8 ). Thus, even if it had started to react to B, the following stimulus ( i.e. A 8 ) could have interrupted it. We could not elongate the interval between notes, since the silence itself could be a confounding factor. Thus, in order to see if the subject could perceive the stimuli change, we ran a new test: (1) we played a series of stimulus from wolf A (A 1 , A 2 , A 3 , A 4 , A 5 , A 6 ), and then (2) a series of stimulus from wolf B (B 1 , B 2 , B 3 , B 4 , B 5 , B 6 ). This approach gave the subject time to perceive the stimulus change and react.
2.7.5.1 Criteria of Response
During the playback tests we noticed that each subject responded to the stimuli in a particular way. For already stated reasons, the wolf CO did not display Oriented Attention in response to the stimuli change on the first playback session. To solve the difficulty of objectively evaluate its responses, we adopted again the “gestalten”. Discrimination happened when the “gestalt” Disturbance reappears in the presence of the new stimulus.
2.7.5.2 Procedure
We watched the playback videos (Camtasia Studio software version 8.0; TechSmith Corporation) and evaluated the subject’s responses to the stimuli.
2.7.6 Other Procedures
2.7.6.1 Reproductive pair discrimination test
During FA and NY vocalization recordings it was observed that the female reacted with a “whine” most of the times in which the male emitted an extendedbark sequence (Kleiman, 1972; Brady, 1981). With that in mind, we ran a playback test to verify if the female NY could distinguish its pair’s extendedbarks from unknown ones. The procedure adopted was: (1) we isolated the female in the enclosure den and kept the male out of her visual range, in the open enclosure; (2) we then played back an extendedbark sequence from FA. To simulate a natural context of vocalization, we adjusted the sound intensity and positioned the power speaker out of the subject's sight; (3) a few minutes later, we played a male’s extendedbark sequence which had been previously recorded in another zoo, being unfamiliar to NY. The purpose of this test was to verify if the female would react to the FA’s playback emitting whines and would not have the same reaction when extendedbarks of an unknown male were played. If the female whined only to FA’s extendedbarks, it would be possible to consider that discrimination occurred between stimuli.
2.7.6.2 Background noise effect evaluation
Since recordings were made in uncontrolled environments, the background noise can differ from one recording to another. That difference is especially pronounced between the recordings automatically obtained by the Song Meter remote recorder and the ones acquired manually with Sennheiser microphone attached to Marantz recorder (Figure 1). The recordings obtained by Song Meter are much noisier ( i.e. more background noise is captured with the signal of interest), which is mainly due to the nature of its microphones (see Recordings for details) and this difference cannot be eliminated. To guarantee that the subjects’ responses occurred due to the change in the stimuli and not to the difference among the types of background noise contained in each track, we used two forms of control:
the background noise change. If the subject did not react to the inserted fragments of background noise, we assumed there had been discrimination between A and B. The duration of each background noise fragment was of approximately 5 seconds.
(2) We also held a playback trial with subjects NY and FA using only noise as stimuli to test their responses to the difference in background noise alone. The protocol we used was simpler than the standard one: we played a track which contained an excerpt of noise extracted from the stimulus A recording (background noise A), followed by an excerpt of noise extracted from the stimulus B recording (background noise B). Our expectation was that the subject would not react to any of the noises or, if there was a reaction, the behaviours presented would differ from the ones that occurred in the presence of the extendedbarks. For this test we used only background excerpts that evidently differed from each other, such as background from Song Meter recordings against background from Marantz recordings.
2.7.6.3 Stimulus control test
With one of the subjects (male TU), we had the opportunity to play an audio track comprising 15 minutes of surround sound which had been previously recorded before initiating the extendedbarks playback tests. This controltrack allowed us to evaluate the animal’s reaction to artificially played sounds. If the subject responded in some way to the playback of surround sound it would have been harder to guarantee that its reaction to the other playback tests (all artificially played) were due to the stimuli.
We also had the opportunity to test the reaction of the subjects TU, FA and NY to a domestic dog’s barks playback to ascertain if they would respond differently from the way they did during the extendedbarks presentation phases. The dog barks were played after the subject was habituated to the extendedbarks. A large dog (Pit Bull American breed of Canis lupus familiaris ) bark that resembled extendedbark notes (Figure 6) was chosen to test if the subjects would notice any difference.
Figure 6 Spectrogram showing a maned wolf’s bark to the right ( i.e. single note of an extendedbark sequence) and a domestic dog’s bark to the left. The image was made on Raven Pro 1.4 using the following spectrograph parameters: Window type = Hann; Window Size = 900 samples; 3dB filter Bandwidth = 70.5; Overlap = 99.7; Hop Size = 3 samples; Frequency Grid DFT size = 2048 samples; Grid Spacing = 21.5 Hz; Brightness = 70; Contrast = 50.
2.7.6.4 Playback tracks preparation
To avoid any possible influences of the females’ estrous stage in the vocalizations, and as we do not know if there is a perceptible difference between the vocalizations of both sexes, we decided to use only male vocalizations as stimuli. Thus, our soundtrack stimuli were restricted to three wolves: FA, TU and OR, which were the only males of which we had vocalizations at the time. We avoided using the same test stimuli ( i.e. A 7 , A 8 , A 9 , B, A 10 ) in more than one test soundtrack and for each playback session we used new test soundtracks. That approach increases the consistency of our results by preventing simple pseudoreplication ( i.e. when a simple stimulus is used to represent a class of stimuli Kroodsma et al., 2001), since we used as many test stimuli as possible.
2.7.6.5 Equipment Quality Procedure Test
In one of the tests we recorded a playback to attest the quality of the sound played by
our equipment. The recording was made analogously to the recordings made during the field trips of acoustic data gathering. We stood near the same spot where the subject was during the playback to record the sound from the animal’s perspective.
3 RESULTS
3.1 Hypothesis Test
Our experimental results showed that the wolves are able to distinguish among different conspecifics’ extendedbarks, leading us to reject H 0 .
3.1.1 Standard Playback Protocol
We ran 19 playback sessions with 10 different subjects from six different zoos. Sessions on a single zoo were separated by at least 1 week from one another except the ones carried out in Sorocaba’s Zoo which were separated by only 1 day from each other for logistical reasons. We ran the test more than once in several sessions to test habituation to B. However, we only considered the first trial of each session as valid, since the habituation control phase stimuli (A 7 , A 8 , A 9 ) and the rehabituation stimuli (A 7 ) were no longer new after the first trial. Therefore, the response to B, or even its absence, would not be reliable as a result, since there would have been a lack in experimental control.
We discarded 9 of the sessions, which did not fulfill minimal experimental conditions (Table 6). In two of the 10 valid trials, the subjects (FA and NY) did not respond as expected.
Table 5 List of the valid playback trials. We considered that the subject discriminated between stimulus B and stimuli A only if they displayed “Oriented Attention” only in the beginning of the habituation phase and during/immediately after B. We considered the results from the second playback session of the couple FA and NY as “Inconclusive”, since the subjects failed discrimination in the second trial, but responded as expected in the first trial (see Discussion for details).
DID IT DISPLAY ORIENTED ATTENTION?
SUBJECT TRIAL A 1 A 6 (BEGINNING)
A 1 A 6 (END)
A 7 A 9 B A 10 RESULT
CO 2 nd Yes No No Yes No Discriminated
FA 1 st Yes No No Yes No Discriminated
FA 2 nd Yes No No No No Inconclusive
LO 1 st Yes No No Yes No Discriminated
LU 1 st Yes No No Yes No Discriminated
NY 1 st Yes No No Yes No Discriminated
NY 2 nd Yes No No No No Inconclusive
TU 1 st Yes No No Yes No Discriminated
TU 2 nd Yes No No Yes No Discriminated
TU 3 rd Yes No No Yes No Discriminated
Table 6 List of the invalidated playback trials. We considered as invalid every trial in that: the subject did not displayed Oriented Attention in the beginning of the habituation phase; was harmed by confounding variables; the magnitude of the subject’s overall disturbance state did not decrease (see “gestalt” Disturbance).
SUBJECT TRIAL REASON FOR INVALIDATION
CO 1 st Did not habituate. Test harmed by confounding variables. 1, 2
FA 3 rd Did not habituate. Test harmed by confounding variables. 1
LU 2 nd Did not habituate. Test harmed by confounding variables. 1, 3
PE 1 st Did not habituate. Test harmed by confounding variables. 1, 4
PL 1 st Did not habituate. Test harmed by confounding variables. 1, 5, 6, 7
PL 2 nd Did not habituate. Test harmed by confounding variables. 1, 5, 7
TI 1 st Did not habituate. Test harmed by confounding variables. 3
WO 1 st Did not habituate. 5, 8
1 Loud background noise, frequent human interference; 2 Dog barking for this session we ran the Complementary Playback Protocol (see Methods); 3 Presence of caretaker during part of the test; 4 Test suspended before the end by the Zoo’s biologist; 5The wolf had recently undergone surgery and was limping; 6 Presence of veterinary during part of the test; 7The subject was not used to unfamiliar humans, since the zoo had been closed to visitors for more than one year before the test; 8 Subject had been recently rescued from nature and was not used to human presence.
3.1.2 Complementary Playback Protocol
3.1.2.1 Response to stimuli
The first playback session with CO was harmed by confounding variables (see Methods). When we ran the standard protocol to see if any reaction would appear, the wolf did not respond to the test stimulus in the first try and responded with “attention” (i.e. turning its head, but not towards the sound source) in the second one. Since the response was not clear, we ran the Complementary Playback Protocol (see Methods). This time the response was clear: the magnitude of the “gestalt” Disturbance suddenly increased in the presence of the B stimuli (excerpts from VideoB).
3.1.2.2 Validity of the “gestalt” Disturbance
The judgment on validity of our “gestalt” Disturbance indicated moderate agreement without the author as judge (Fleiss’ kappa statistic: k = 0.56; n judges = 5; Pvalue 0.001) and substantial agreement with the author as judge (Fleiss’ kappa statistic: k = 0.625; n judges = 6; Pvalue 0.001). The agreement in the increase of Disturbance magnitude from stimuli A to stimuli B was unanimous in both cases (Table 7).
Table 7 Evaluation of the “gestalt” Disturbance based on video excerpts of the subject CO by 5 naive adult judges of both sexes. The author’s ratings were judged by five untrained observers.
Author Judge 1 Judge 2 Judge 3 Judge 4 Judge 5
VideoA Increases Increases Increases Increases Decreases Increases
VideoB* Increases Increases Increases Increases Increases Increases
VideoC Increases Increases Increases Increases Decreases Increases
VideoD Decreases Decreases Decreases Decreases Decreases Decreases *VideoB corresponds to the transition from stimuli A to stimuli B and the agreement among judgments was unanimous.
3.1.3 Other Results
3.1.3.1 Reproductive pair discrimination test
The female NY did not “whine” in response to either of the playbacks. During the test NY was lying down inside the den with a raised body and its ears were moving forwards and backwards (see “Ear Scan” and “Rest” in Ethogram). Moreover, its ears moved many times in the direction of the enclosure where the male FA was which was not the same direction of the sound source. The male responded to the playbacks (including its own extendedbarks) with “Piloerection” and “Oriented Attention”.
3.1.3.2 Background noise effect evaluation
The difference of background noise did not produce any response in the subjects. We used the first type of control in four playback soundtracks (two with TU; two with FA and NY) and the subjects did not show any response to the fragment of background noise preceding the test stimulus (see Methods for details). Also, the subjects FA and NY did not react to the second type of control (see Methods for details).
3.1.3.3 Stimulus control test
The subject TU did not show any response to the playback of surround sound. It remained resting (see “Rest” in Ethogram) before, during and after the test, which increases the credibility of the other playback tests’ results. The responses of the subjects (TU, FA and NY) to the playback of a domestic dog barks varied. The wolf TU, that was resting alert (see “Rest Alert” in Ethogram) when the extendedbark sequence was replaced by the domestic dog’s barks, got up and walked towards the speaker. The male FA did not show any clear response to the stimulus and NY responded with “Attention” (see Ethogram).
3.1.3.4 Equipment Quality Procedure Test
The spectrogram analysis of the recorded playback showed no distortion or anomaly in the notes (Figure 7).
Figure 7 Spectrogram (below) and oscillogram (above) of extendedbarks. The audio file is the result of the recording of a playback (see Methods for details). The image was made on Raven Pro 1.4 using the following spectrograph parameters: Window type = Hann; Window Size = 900 samples; 3dB filter Bandwidth = 70.5; Overlap = 99.7; Hop Size = 3 samples; Frequency Grid DFT size = 2048 samples; Grid Spacing = 21.5 Hz; Brightness = 70; Contrast = 50.
3.1.3.5 Recordings
registered 5 duet events between the couples FANY (n = 3) and FRTU (n = 2) (Figure 8). Male and female were out of each other’s sight just before and during the events (Table 9).
During field work in São Carlos’ Zoo, we registered two very long sequences with an automatic recorder (Wildlife Acoustics, SM2+). The first sequence contained 712 extendedbarks emitted in a row and the second one contained 208. Both sequences were emitted in the same day 20 minutes apart from each other. There are no records of such long sequences in the literature neither in captivity nor in the wild. Unfortunately, we could not include the recorded notes in the acoustic analysis, since the sampling frequency was adjusted to only 16 kHz (an error that went unnoticed at the time). The animal that emitted these sequences is also uncertain, since we were not present during the event all we can infer is that it was not TU, since it was locked in its enclosure, far away from the recorder. This problem also happened to the sequences automatically recorded in the enclosure of FA and NY.
All notes emitted by CO (n = 34) and most of the notes emitted by LU (n = 42) were ruined by aliasing.
Table 8 Information about the recording samples acquired during 13 field trips. Recording sessions were taken in each zoo continuously from 17:00 to 07:00 (UTC3) within the period of greatest activity of the species. The 4 th column indicates the recording procedure: automatic (recorded with SongMeter 2+) or manual (recorded with Marantz).
TOTAL NUMBER OF
NOTES
NUMBER OF SEQUENCES
AVERAGE OF NOTES PER SEQUENCE
RECORDING PROCEDURE
DATA GATHERING
EFFORT
ZOO’S
LOCATION SUBJECT
337 8 42,1 Automatic 37:52 h R. Preto OR
0 0 0 Automatic 111:40 h Catanduva ME
0 0 0 Automatic 181:58 h S. J. R. Preto MO
111 12 9,2 Manual 221:00 h Sorocaba NY
145 33 4,4 Manual 221:00 h Sorocaba FA
54 7 7,7 Manual 120:00 h São Carlos FR
32 3 10,7 Manual 120:00 h São Carlos TU