Model vs. playback experiments: The impact of sensory mode ...

Ethology. 2020;126:563–575.  |  563wileyonlinelibrary.com/journal/eth

Received: 16 March 2019  |  Revised: 30 November 2019  |  Accepted: 7 January 2020

DOI: 10.1111/eth.13008

R E S E A R C H P A P E R

Model vs. playback experiments: The impact of sensory mode on predator-specific escape responses in saki monkeys

Dara B. Adams1,2  | Dawn M. Kitchen1,3

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.© 2020 The Authors. Ethology published by Blackwell Verlag GmbH.

1Department of Anthropology, The Ohio State University, Columbus, OH, USA2Department of Anthropology, Humboldt State University, Arcata, CA, USA3Department of Anthropology, The Ohio State University, Mansfield, OH, USA

CorrespondenceDara B. Adams, Department of Anthropology, The Ohio State University, 4034 Smith Laboratory, 174 W. 18th Ave., Columbus, OH 43210, USA.Email: [email protected]

Funding informationDepartment of Anthropology, Texas State University; Division of Behavioral and Cognitive Sciences, Grant/Award Number: 1341174; Animal Behavior Society; Department of Anthropology, The Ohio State University; Tinker Foundation; Society for Integrative and Comparative Biology

Editor: Redouan Bshary

AbstractAlthough experimentally simulating predator presence helps improve sample sizes in studies of free-ranging animals, few studies have examined whether auditory play-backs and visual models produce similar results. Additionally, it is unclear if anti-pred-ator strategies are specific to predator hunting styles in understudied Neotropical pitheciid primates, limiting what we can generalize about this phenomenon across this taxonomic order. We conducted predator simulation experiments to assess whether wild Rylands' bald-faced saki monkeys (Pithecia rylandsi) recognize preda-tors based solely on acoustic cues, exhibit predator-specific responses to different predator types, and vary responses to presentations in different sensory modes. In our playback experiments, sakis had weak responses to non-predator control vo-calizations compared to jaguar growls and harpy eagle shrieks. In most predator playbacks, subjects' first glance corresponded to the direction from which simulated predators would typically attack (above vs. below). However, although sakis exhib-ited appropriate movement responses to harpy playbacks (i.e., descending canopy), they exhibited no clear movement patterns when presented with jaguar playbacks. In contrast, jaguar model experiments consistently elicited fast approaches, mobbing-style responses, and long alarm calling bouts. Thus, if we had relied on playbacks alone, we might have concluded that sakis have only generalized responses to ter-restrial ambush predators. In fact, in all variables measured (e.g., latency, number of calls, and response duration), models of both predator species elicited stronger reac-tions than playbacks. Results indicate that bald-faced sakis can identify predators based solely on vocalizations, but do not exhibit predator-specific escape responses to terrestrial predators based on acoustic cues alone. The differential response to playbacks and models calls into question the reliability of using acoustic-only stimuli to assess the specificity of anti-predator behavior to predator hunting styles in some primate species.

K E Y W O R D S

alarm calls, anti-predator behavior, harpy eagle, jaguar, Pithecia rylandsi, predation

564  |     ADAMS AnD KITCHEn

1  | INTRODUC TION

Animals face the threat of predation on a daily basis (Caro, 2005; Lima & Dill, 1990). However, anti-predator behaviors, such as alarm calling, mobbing, and fleeing, are energetically costly. To increase odds of survival and avoid wasting energy, prey should readily dis-tinguish predators from non-threatening animals, as well as respond appropriately based on the hunting tactics employed by different types of predators (Robinson, 1980), such as being cryptic when faced with a pursuit-style predator versus loud calling, approach-ing, and even mobbing when encountering ambush predators (e.g., Zuberbühler, Noë, & Seyfarth, 1997).

Predator encounters are rare, particularly when human observers are present. Thus, predator recognition has traditionally been studied in primates through the use of acoustic playback experiments in which prey are exposed to auditory broadcasts of predator vocalizations (e.g., Hauser & Wrangham, 1990; Macedonia & Young, 1991; Searcy & Caine, 2003; Zuberbühler, 2000). Even though most predators do not vocal-ize when hunting, meta-reviews of the playback literature demonstrate that many species of birds and mammals respond with appropriate es-cape responses to the sounds of their predators (Barrera, Chong, Judy, & Blumstein, 2011; Blumstein, Cooley, Winternitz, & Daniel, 2008; Hettena, Munoz, & Blumstein, 2014). For example, red-tailed sportive lemurs (Lepilemur ruficaudatus) behave cryptically and exhibit freezing behavior when exposed to sounds of aerial predators but climb up the tree and scan the ground after hearing vocalizations of their main ter-restrial predators (Fichtel, 2007).

Although acoustic playbacks are considered a reliable method of simulating predator presence (e.g., Arnold, Pohlner, & Zuberbühler, 2008), it is important to note that the mode of presentation may have a significant impact on behavioral responses based on which sensory modality dominates. Captive research has demonstrated that pri-mates learn more quickly and retain information longer in visual rather than acoustic modes (e.g., D'Amato & Colombo, 1985; Fritz, Mishkin, & Saunders, 2005; Munoz-Lopez, Mohedano-Moriano, & Insausti, 2010; Wright, 2007). Additionally, visual cues provide direct informa-tion about a predator's identity, location, movements, and behavior, whereas acoustic cues are more ambiguous, leaving prey unable to confirm the predator's location or behavior (Billings, Greene, & Lucia Jensen, 2015). As a result, primates may exhibit weaker anti-predator responses in the auditory than visual sensory domain. For example, Old World putty-nosed monkeys (Cercopithecus nictitans martini) emit significantly more alarm calls and longer calling durations to leopard models than to leopard growls (Arnold et al., 2008). Although a num-ber of studies have used models (e.g., Rahlfs & Fichtel, 2010; Wheeler, 2008, 2010; Wich & Sterk, 2003), only a few studies of free-ranging primates have compared the differing effects of acoustic and visual cues on primate anti-predator behavior (e.g., Cercopithecus nictitans martini, Arnold et al., 2008; Cercopithecus campbelli campbelli, Ouattara, Lemasson, & Zuberbühler, 2009; Colobus guereza, Schel, 2009; see also captive study on Callithrix jacchus, Kemp & Kaplan, 2011).

Experimental evidence of the function and specificity of alarm calling and other anti-predator responses is emerging

slowly in New World monkeys (e.g., Berthet, Neumann, Mesbahi, Cäsar, & Zuberbühler, 2018; Cäsar, Byrne, Hoppitt, Young, & Zuberbühler, 2012a; Fichtel, Perry, & Gros-Louis, 2005; Kirchhof & Hammerschmidt, 2006; Wheeler, 2008, 2010) compared to the well-studied Old World primates (reviewed in: Zuberbühler, 2009), limiting what can be said about the taxonomic breadth of these phenomena (Cäsar & Zuberbühler, 2012). The Pitheciidae (titi monkeys, Callicebus spp., Cheracebus spp., and Plecturocebus spp.; saki monkeys, Pithecia spp.; cuxiú, Chiropotes spp.; and uaca-ris, Cacajao spp.) are the least well studied of the three families of New World monkeys (Wildman, Jameson, Opazo, & Soojin, 2009). Anecdotal field reports indicate pitheciids exhibit a com-bination of anti-predator strategies including crypsis (i.e., silence and hiding), alarm calling, and mobbing (e.g., Barnett et al., 2018; Barnett et al., 2017; Ferrari, 2009; Gleason & Norconk, 2002; de Luna, Sanmiguel, Difiore, & Fernandez-Duque, 2010; Mourthé & Barnett, 2014). For example, white-faced sakis (P. pithecia) re-spond to small terrestrial threats (small cats and some snakes) with alarm calls and mobbing behaviors but move lower in the canopy and remain motionless for long periods of time (i.e., “freezing”) in response to large raptors (Buchanan, Mittermeier, & Roosmalen, 1981; Gleason & Norconk, 2002). In the only ex-perimental study on pitheciids to date, black-fronted titi monkeys (Callicebus nigrifrons) have been shown to respond to terrestrial and aerial predator models with alarm call sequences that convey information about predator type and location in the forest matrix (Berthet et al., 2019; Cäsar, Byrne, Young, & Zuberbühler, 2012b; Cäsar, Zuberbühler, Young, & Byrne, 2013), and listeners can dis-tinguish among conspecific alarm calls (Cäsar, Byrne, Hoppitt, et al., 2012a). Thus, pitheciid monkeys may exhibit anti-predator strategies adapted to the predator's hunting style and the level of perceived threat, but more systematic data are needed to assess the consistency of these responses across taxa.

Our study examines anti-predator strategies of a little known pitheciid primate, Rylands' bald-faced saki monkeys (Pithecia ry-landsi), in Amazonia Peru. Our objectives were to determine whether bald-faced sakis could discriminate between different types of predators (i.e., aerial-pursuit vs. terrestrial ambush preda-tors) and whether presentation mode (i.e., acoustic vs. visual cues) impacted the strength or specificity of their responses. To test this, we conducted acoustic playback experiments on sakis using jaguar (Panthera onca) growls, harpy eagle (Harpia harpyja) shrieks, and vo-calizations from non-predators. We compared playback results to those generated in predator model experiments using life-like jaguar and harpy eagle decoys.

2  | METHODS

2.1 | Study species

Rylands' bald-faced sakis are found in northwestern Bolivia, south-eastern Peru, and in the Brazilian states of Rondônia and Mato

     |  565ADAMS AnD KITCHEn

Grosso (Marsh, 2014; Orsini, Nunes, & Marsh, 2017). The species was formerly included in Gray's bald-faced sakis (Pithecia irrorata) prior to Marsh's (2014) taxonomic revision, which introduced P. ry-landsi as one of five newly described saki species. However, there remains debate regarding the classification of sakis at our field site, with some researchers recognizing P. rylandsi as a junior synonym of P. irrorata (see Serrano-Villavicencio, Hurtado, Vendramel, & do Nascimento, F. O., 2019). Information on the behavioral ecology of the species is lacking due to challenges in accessing their remote and dense Amazonian habitats, as well as difficulties in locating and ha-bituating such cryptic primates.

To our knowledge, our study site contains the only well-habitu-ated groups of P. rylandsi. A prior 3-year study on five groups at our field site documented group sizes averaging 4.7 ± 1.5 SD individuals, consisting of 1 adult male, 1–3 adult females, and related juveniles (as P. irrorata, Palminteri & Peres, 2012). Mean home range size for those five groups was 35.9 ± 16.3 ha (Palminteri & Peres, 2012). Sakis at our study site show a strong preference for terra firme for-ests but are also found in seasonally flooded forests (Palminteri & Peres, 2012). Fruit comprises over 95% of their monthly diet, with seeds alone accounting for 75% (Palminteri, Powell, & Peres, 2012). While P. rylandsi spend most of their time in the mid- to high canopy, our previous research has shown that they will descend to the low canopy to feed on soils from arboreal termite mounds and occasion-ally descend to the forest floor to feed on army ants for long periods, making them potentially susceptible to terrestrial predators (Adams, Rehg, & Watsa, 2017).

2.2 | Study site and subjects

Playback experiments were conducted at La Estación Biológica Los Amigos in southeastern Peru from May to August 2008. Predator model experiments were conducted at the same site from June to December 2014. The field station (12°34′07″S 70°05′57″W) is located on a high terrace above the Madre de Dios River near the confluence of the Los Amigos River. The conservation area protects over 140,000 hectares of Amazonian lowland forests and encom-passes a variety of habitats, including terra firme primary forests, palm swamps, dense bamboo patches, floodplain forests, and dis-turbed forests (Pitman, 2010).

We conducted experiments on 10 individuals belonging to three saki groups in 2008 and 6 individuals belonging to the same three groups in 2014. Groups were identified each year based on home range, recognition of some individuals, and group census data from ongoing behavioral studies (Table 1). Two of the focal groups have been habituated since 2004 as part of a separate long-term study on their habitat use and feeding ecology (Palminteri & Peres, 2012). The third group was not previously habituated but was included in the study due to accessibility, frequency of sight-ings, and behavior in observer presence that suggested they were not sensitive to observers (e.g., lack of alarm calling/heightened vigilance/flight).

2.3 | Predators

Weighing only 1,770 g on average (range 1,300–3,000 g), saki mon-keys have a wide array of known and potential predators (Gleason & Norconk, 2002; Norconk, 2007). The study region is home to over 40 raptor species, including several species known to prey on sakis such as crested eagles (Morphnus guianensis), harpy ea-gles (Harpia harpyja), and black-and-white hawk-eagles (Spizaetus melanoleucus) (Adams & Williams, 2017; Aguiar-Silva, Sanaiotti, & Luz, 2014; Barnett et al., 2011; Gilbert, 2000; Martins, Lima, & Silva, 2005; Peres, 1990; Rettig, 1978). In addition, several carni-vore species known to hunt primates, and thus capable of preying on adult or infant sakis, are found at the site (e.g., Puma concolor: Matsuda & Izawa, 2008; Panthera onca: Peetz, Norconk, & Kinzey, 1992; Leopardus pardalis: Bianchi & Mendes, 2007; Leopardus wiedii: Mondolfi, 1986; Puma yaguaroundi: Ximenes, 1982; Eira barbara: Bezerra, Barnett, Souto, & Jones, 2009). Several snakes documented to be predators of Neotropical primates also occur in the study area, including anacondas (Eunectes murinus, Heymann, 1987) and red-tailed boas (Boa constrictor, Quintino & Bicca-Marques, 2013). Of these species, only boa constrictors are known to consume sakis (Ferrari, Pereira, Santos, & Veiga, 2004). The two species used as predator stimuli in our study, the harpy eagle and jaguar, have both been sighted by our research team within our focal monkey groups' home ranges (Adams.D., & Kitchen, D. M. unpubl. data). While there are no published data on harpy eagle densities at our study site, densities found for jaguars in the study area are among the highest documented, suggesting that this re-gion of the Amazon is a core area for the species (Tobler, Carrillo-Percastegui, Zuñiga Hartley, & Powell, 2013).

2.4 | Acoustic predator stimuli

Playback stimuli of predators included jaguar growls (Figure 1a) and harpy eagle shrieks (Figure 1b). To ensure sakis did not simply respond to any startling call (e.g., pseudo-predators: Mourthé & Barnett, 2014), we used control stimuli of loud vocalizations that were not associated with threats and were produced by common birds and mammals that are found in the study area yet harmless to sakis (e.g., screaming piha, Lipaugus vociferas: n = 4; warbling antbird, Hypocnemis peruviana: n = 4; mealy parrot, Amazona farinosa: n = 4; and brown agouti, Dasyprocta variegata: n = 4). To avoid pseudor-eplication, we used several exemplars from each playback category (jaguar: n = 4; harpy: n = 5; and controls: n = 7) so that each group only heard a particular stimulus once (Kroodsma, 1989).

All calls used as playback stimuli were recorded in the study area, with the exception of jaguar growls obtained from The Cornell Laboratory of Ornithology, Macaulay Library. The context of jaguar calls used to create playback stimuli was not known; however, it is unlikely that the calls were produced by hunting animals because felids are known to rely on stealth when hunting prey. The calls used to create harpy eagle playback stimuli consisted of vocalizations

566  |     ADAMS AnD KITCHEn

emitted by eagles that were perched within the canopy (i.e., not flying overhead). Although most predators do not vocalize while hunting, previous studies have used recordings of felid or eagle vo-calizations to successfully elicit anti-predator behavior in sakis (Neal, 2009; Stenzel, 2017) and other Neotropical primate species (Gil-da-Costa, Palleroni, Hauser, Touchton, & Kelley, 2003; Neal, 2009; Wheeler, 2008).

Playbacks of predator and control vocalizations consisted of two bouts of calling, each approximately 15 s in duration, with a 20 s interval of silence in between. Each audio playback file began with 5 min of silence, which allowed us to move out of the sakis' visual range prior to the onset of playbacks, thereby avoiding association between our presence and predator calls (following Papworth, Böse, Barker, Schel, & Zuberbühler, 2008). The order

in which each playback stimulus was administered was random-ized and balanced across subjects, with subjects from two groups hearing the jaguar first and subjects from one group hearing the harpy first. Each group was subjected to no more than one pred-ator playback trial per week to prevent habituation. Our goal was to test each group with all possible exemplars of each playback category.

2.5 | Visual predator stimuli

We used commercially produced models of a harpy eagle and jag-uar for visual stimuli (Figure 2). The eagle decoy was custom-de-signed (Ure-a-Duck Decoys, North Carolina) and painted to match

TA B L E 1   Composition of saki monkey focal groups during the data collection period in 2008 and 2014

Group Year Total individuals Adult males Adult females Subadults/Large juveniles Infants/Small juveniles

A 2008 4 1 1 2 0

A 2014 4 1 1 1 1

B 2008 6 1 2 2 1

B 2014 6 1 2 2 1

C 2008 4/5a 1 1 2 0/1

C 2014 4 1 1 2 0

aAn infant was born in this group during the data collection period.

F I G U R E 1   Spectrograms of typical (a) jaguar (Panthera onca) growls and (b) harpy eagle (Harpia harpyja) shrieks used as stimuli in predator playback experiments

F I G U R E 2   Predator models presented to sakis: (a) jaguar in sitting position and (b) harpy eagle with claws protracted and wings expanded [Colour figure can be viewed at wileyonlinelibrary.com]

(a) (b)

     |  567ADAMS AnD KITCHEn

harpy eagle coloration patterns based on photographs of harpies from our study site. The eagle's posture resembled an individual in attack mode, with claws protracted and the wings (1.68 m wing-span) expanded as if in flight. The model was attached at its base to a removable rod that allowed for easier handling and mounting in the field. For the visual jaguar model, we used a commercially produced leopard replica (Viahart LLC), which we modified with paint to match jaguar spot patterns based on camera trap im-ages of jaguars at our study site. The model was over 1 m in body length (not including tail) and a head that reached ~0.6 m in height while in a sitting position. Models were life-like in size, form, and coloration patterns. To avoid habituation, model types were pre-sented to groups in a randomized order and each specific model was used only once within a two-week period; in most instances, each model was used only once with each group in a given 30-day period (following Wheeler, 2008).

2.6 | Experimental protocol

Observers located and followed the focal group for at least 30 min prior to experiments. Trials were conducted opportunistically when certain conditions were met: Alarm calls had not been heard in the previous half-hour period and members of the focal group were feeding or resting in low to mid-canopy and located in close proxim-ity to one another, but not interacting. We then estimated the path the group would likely take following the feeding or resting bout, which was determined based on prior knowledge of their typical day ranges and feeding or sleeping tree locations. One observer then moved outside the group's visual range to place the playback speaker or visual model along the predicted path at a distance be-tween 15 and 30 m from the focal group. One or more observers remained with the group to ensure that they did not approach the experiment area before the speaker or model was in place (following Arnold et al., 2008; Wheeler, 2008).

Playbacks of jaguar growls and harpy eagle shrieks were broad-cast using an Apple iPod and Saul Mineroff Electronics AFS field speaker, which was hidden behind brush or trees. Stimuli ampli-tudes were measured prior to trials using a sound pressure meter (Galaxy Audio CM-130) within 3 m from the playback speaker in similar environments in which playbacks were conducted to en-sure they were broadcast at comparable amplitudes and sounded natural to experienced human listeners. The speaker was posi-tioned on the ground except when sakis were located in ravines during harpy playbacks (7 of 15 harpy trials). Here, we attempted to position the speaker at the top of hillsides to emulate the nat-ural direction of attack. Visual jaguar models were placed in open areas on the ground, whereas harpy models were placed 3–5 m off the ground along ridgelines when possible. Once the speaker or model was set up, observers moved away from the immediate experiment area to a covert location to begin recording data. The speaker or model was removed only after sakis retreated from the area (>15 m) and was no longer in visual range.

2.7 | Behavioral data collection

Prior to each experiment, we noted general information, such as the date, time, stimulus type (i.e., control/harpy eagle/jaguar), presenta-tion mode (i.e., audio/visual), exemplar number (e.g., jaguar record-ing 3), location of the experiment, activity of group majority (e.g., feeding, foraging, and resting), and presence of infants or young ju-veniles. We then randomly chose an adult or subadult focal subject that was located close to the experiment area and in clear site of observers. We documented the focal animal's latency to respond to acoustic or visual stimuli. For acoustic experiments, this was meas-ured as time between playback onset and the focal subject's first response, such as rapid movement or alarm calling. For visual experi-ments, this was measured as time between the focal subject making eye contact with the model and their first rapid movement or vocal response. This measure was not taken in 4 visual trials because we could not clearly see the immediate focus of the focal monkey. We also recorded the focal animals first direction of glance, which was scored as “up” or “down” with head movements of >45 degrees from the pre-trial direction, “toward” when individuals turned their gaze directly at the speaker area or model, and “no look” when there were no head movements or changes were <45 degrees. Additionally, we noted the focal subject's first direction of vertical and horizontal movement of >1 m following playback onset or visual detection of models, which was scored as “up” or “down,” “toward” or “away,” or “no movement.”

We also recorded data on the focal subject's vocal behavior following experiment onset, including whether or not alarm calls were produced (scored as “yes” or “no”) and the number of vocal-izations emitted within 5 min of detection. Vocalizations produced during experiments were recorded using either a Sony PCM-D50 or Marantz PMD660 solid state audio recorder (sampling rate 44.1 kHz) and a Sennheiser ME66 directional microphone with a K6 power module.

Finally, we recorded the subject's overall reaction duration, which was measured as the length of time from the focal subject's first response until the majority of the group returned to baseline be-haviors including feeding or engaging in non-vigilant behaviors such as grooming. This measure was not taken in 3 playback trials because we lost the group before they returned to non-vigilant behaviors.

While one observer collected data on the focal animal (de-scribed above), a second observer used continuous sampling to record descriptions of overall group response during experiments. Specifically, we noted all approaches to the experiment area, any in-stances of hiding (remaining concealed in thick vegetation) or fleeing (rapid departure from the area), obvious scanning behaviors (rapid back and forth movement of the head), and whether group members emitted alarm calls. For visual model experiments, we also noted any instances of inspection (i.e., staring intently at the model) and mobbing (i.e., approaching, surrounding, and harassing the model; see Crofoot, 2012). If animals engaged in mobbing, we documented the total mobbing duration and noted detailed descriptions of the behavior immediately after the experiment ended.

568  |     ADAMS AnD KITCHEn

2.8 | Data analysis

We excluded four playback trials (two harpy and two jaguar trials) from data analyses due to unexpected occurrences during experiments that could have affected the monkeys' responses (predator in the area: n = 1; loud humans entering area: n = 2; and capuchins harassing focal group: n = 1). We used Fisher's exact tests to examine whether sub-jects categorical behavioral responses (i.e., direction of first glance, direction of movement, and propensity to alarm call) differed between control and experimental trials, as well as between predator type (i.e., jaguar and harpy eagle), and modes of presentation (i.e., audio and vis-ual). We used the Holm-Bonferroni correction (Holm, 1979) to adjust p-values for multiple testing where appropriate.

Because numeric response variables (i.e., latency to move, num-ber of calls produced, and total reaction duration) were correlated (Spearman correlation between all responses: p < .022), we com-bined all three of these variables into a “response strength index” using a principal component analysis (PCA; McGregor, 1992). Following Kitchen, Cortés-Ortiz, Dias, Canales-Espinosa, and Bergman (2018), we focused on the component that explained the most variance, hereafter referred to as PC1. Since data were approx-imately normally distributed, PC1 was analyzed using a linear mixed model (LMM) with subject identity nested within group identity as a random factor, and presentation mode (audio and visual), predator type (jaguar and harpy eagle), and the interaction of presentation mode and predator type as fixed factors. To test for the effect of potentially confounding variables, we also included trial order (first and subsequent trials) and presence of young (i.e., infants or small juveniles) as fixed factors in the full model. Trial order was included to account for the possibility that subject response may differ be-tween first and subsequent exposure to stimuli. Presence of young was included because previous research indicates that sakis respond to predators more strongly when infants are present (de Luna et al., 2010).

Likelihood ratio tests (LRT) showed that our full model was a significant improvement over a reduced model that did not control for the interaction of presentation mode and predator type (LRT: 8.255, p < .01). The full model was also an improvement over the null model, which included only the intercept and random effects (LRT: 31.142, p = .00); therefore, we present a full model that includes the

interaction effect in the results. We report the estimated marginal means (EMM) as well as standard error (SE) and 95% confidence in-tervals (CI). All tests were two-tailed and performed in SPSS v. 22.0 (SPSS) with significance levels set at .05.

2.9 | Ethical note

All applicable institutional and/or national guidelines for the care and use of animals were followed, and all methods complied with the ABS/ASAB guidelines for ethical treatment of animals. All proto-cols adhered to Peruvian legal requirements and were authorized by Peru's Ministry of Agriculture (permits No. 49-2008-INRENA-IFFS-DCB and 210-2014-MINAGRI-DGFFS/DGEFFS).

3  | RESULTS

3.1 | Does presentation modality affect saki monkey responses to predators?

After excluding four trials (see Methods), the final playback dataset included 16 control trials, 15 harpy eagle trials, and 11 jaguar trials. Our dataset for visual predator model experiments included 6 harpy eagle trials and 6 jaguar trials.

The PCA of the three numeric response variables (i.e., move latency, number of calls produced, and total reaction duration) re-sulted in only one principal component score with an eigenvalue >1.0 (PC1: eigenvalue = 1.970) that explained 65.7% of the variance in the data. All variables had moderate to high loading scores on PC1 (move latency = −0.532; number of calls = 0.901; and reaction du-ration = 0.935). Thus, high PC1 values indicated strong overall re-sponses (e.g., short latencies, high number of calls, and long reaction durations; Table 2).

As shown in Figure 3, presentation mode had a strong effect on PC1 (F1,21.078 = 34.447, p < .001); sakis were more likely to have a stron-ger response (i.e., shorter latencies, more calls, and longer reaction du-rations; Table 2) to visual models (EMM ± SE: 1.161 ± 0.243, 95% CI: 0.654 to 1.668) than to audio playbacks of predator stimuli (EMM ± SE: −0.450 ± 0.165, 95% CI: −0.813 to −0.087).

TA B L E 2   The mean and standard errors of three numeric response variablesa used to measure response strength in the principal component analysis

Experiment type Response latency (s) Number of calls Reaction duration (s) PC1

Audio playback

Harpy eagle 24.36 ± 4.67 9.13 ± 4.31 188.77 ± 34.00 −0.500 ± 0.077

Jaguar 21.27 ± 8.52 4.09 ± 0.94 429.30 ± 125.96 −0.342 ± 0.143

Visual model

Harpy eagle 6.25 ± 0.63 24.00 ± 10.65 1554.67 ± 519.30 0.403 ± 0.168

Jaguar 3.50 ± 1.04 138.83 ± 45.62 1865.50 ± 486.47 1.953 ± 0.705

aSee text for description of measurements.

     |  569ADAMS AnD KITCHEn

Examination of categorical response data also revealed significant differences in subjects' responses based on presentation mode. There were significant differences in subjects' direction of first movement between visual model and acoustic playback experiments (Fisher's exact test with Bonferroni correction: p < .001, n = 38). While sakis moved toward predator models in 100% of visual trials, they only ap-proached the speaker in 23.1% of acoustic playbacks (6 of 23 trials). Sakis also exhibited significant differences in alarm call response based on presentation mode (Fisher's exact test with Bonferroni correction:

p < .001, n = 38); regardless of model type, subjects emitted alarm calls in 100% of visual trials but in only 23.1% of acoustic trials.

3.2 | Do saki monkeys recognize the vocalizations of their predators?

Although weaker than in visual model experiments, responses to playback experiments were stronger than to control sounds. For example, individuals never emitted alarm calls following control tri-als, and this difference compared to 23.1% of experimental trials ap-proached statistical significance (Fisher's exact test with Bonferroni correction: p = .067, n = 42; Table 3). Additionally, focal subjects were more likely to maintain the same direction of glance (i.e., not look up, down, or toward the speaker; Table 3) following exposure to control trials compared to experimental trials (Fisher's exact test with Bonferroni correction: p < .001, n = 42). Similarly, subjects were more likely to maintain their vertical or horizontal positions in the canopy following control trials (moving in only 2 of 16 control tri-als, Table 3) compared to experimental trials (Fisher's exact test with Bonferroni correction for vertical movement: p < .001, n = 42; for horizontal movement: p = .002, n = 42).

3.3 | Which predator type elicited overall stronger responses in saki monkeys?

There was a strong effect of predator type on PC1 (F1,20.392 = 14.585, p = .001; Figure 3); regardless of presentation mode, subjects had an overall stronger response to jaguar stimuli (EMM ± SE: 0.813 ± 0.201, 95% CI: 0.385 to 1.241) than to harpy eagle stimuli (EMM ± SE: −0.102 ± 0.192, 95% CI: −0.510 to 0.306). However, there was also a

F I G U R E 3   Boxplot illustrating the estimated marginal means (±SE) for PC1 scores based on predator type and presentation modality. The bar in each boxplot indicates the median value. The data point plotted with an open circle is an outlier

TA B L E 3   Contingency tables from Fisher's exact tests of categorical responses to acoustic playbacks

Responses Harpy Eagle Jaguar Control

Direction of glance

Up 12 0 3

Down 0 9 1

Toward 2 2 2

No change 1 0 10

Horizontal movement

Toward 2 4 0

Away 4 4 0

No movement 9 3 16

Vertical movement

Up 1 4 0

Down 14 5 2

No movement 0 2 14

Alarm calling

Present 4 2 0

Absent 11 9 16

570  |     ADAMS AnD KITCHEn

significant interaction effect of presentation mode by predator type (F1,20.465 = 9.784, p = .005; Figure 3); whereas sakis responded more strongly to visual jaguar models (EMM ± SE: 1.991 ± 0.325, 95% CI: 1.320 to 2.662) than to harpy eagle models (EMM ± SE: 0.332 ± 0.311, 95% CI: −0.310 to 0.973), there was no difference in subjects' overall response strength between predator types in audio playbacks (jaguar stimuli: EMM ± SE: −0.364 ± 0.211, 95% CI: −0.809 to 0.081; harpy eagle stimuli: EMM ± SE: −0.535 ± 0.200, 95% CI: −0.953 to −0.118). The lack of a difference in audio playbacks was driven by the fact that sakis responded with more vocalizations to harpy eagles but with slightly shorter latencies and longer durations to jaguars (Table 2).

3.4 | Do saki monkeys have predator-specific responses in both presentation modes?

Examination of categorical response data revealed significant differ-ences in subjects' responses between predator types in playback ex-periments. The direction of the focal subjects' first glance following onset of playback stimuli significantly differed (Fisher's exact test with Bonferroni correction: p < .001, n = 26); sakis typically looked down and never looked up immediately following jaguar playbacks. Conversely, they never looked down and typically looked up after harpy playbacks (Table 3). This difference in direction continued after completion of the trial, with one or more group members seen scanning out and down following five of 11 jaguar trials (45.4%) versus at least one group mem-ber scanning upward following seven of 15 harpy trials (46.7%).

Although horizontal response movement did not significantly differ between playback types (Fisher's exact test with Bonferroni correction: p = .352, n = 26; Table 3), there was a significant differ-ence in subjects vertical response movement to each predator play-back type (Fisher's exact test with Bonferroni correction: p = .045, n = 26). Subjects moved down in the canopy immediately following the onset of harpy playbacks on all but one occasion (when the sub-ject moved up), whereas there was no clear pattern in direction of vertical movement following jaguar playbacks (Table 3).

Likelihood of emitting alarm calls did not significantly differ based on predator playback types (Fisher's exact test with Bonferroni cor-rection: p = 1.000, n = 26); alarm calls were emitted by at least one group member in four out of 15 (26.7%) harpy playback trials and two out of 11 (18.2%) jaguar playback trials (Table 3). However, the overall group behavior at the end of playback trials did differ based on predator type. In the majority of jaguar playback trials (7 of 11 tri-als, 63.6%), the interaction ended with the group departing the area with rapid movement and fleeing more than 50 m away. In contrast, almost all interactions (14 of 15 trials, 93.3%) with harpy playbacks ended with the group quietly beginning to feed in the same area as the experiment.

Similar to playbacks, the overall behavior of the group differed based on visual predator type; however, responses were generally more consistent within trial types in model experiments. In all jaguar model trials, group members (adult males, adult females, and occa-sionally juveniles) approached decoys and responded with mobbing

behaviors, including moving in a circular motion in the area directly above the model (typically about 10–15 m away but an adult male descended to within 2 to 3 m on one occasion) and sometimes lunging toward the model (3 of 6 trials, 50.0%). Sakis also exhibited piloerection, back arching, and body and tail shaking during all mob-bing bouts. The duration of mobbing bouts ranged from 325.2 to 1,819.8 s (x ± SE: 1,310.4 ± 208.8 s). Mobbing ended abruptly with all group members rapidly fleeing the area.

Unlike jaguar model experiments, sakis exhibited mobbing-style behavior in only one of the six harpy model experiments (Fisher's exact test: p = .015, n = 12). In the remaining five experiments, in-dividuals approached the harpy model (although never closer than 15 m), but after spending a couple of minutes (x ± SE: 143.6 ± 22.2 s) visually inspecting the model, group members either moved a few meters away into thicker foliage and remained motionless (2 of 6 trials, 33.3%) or quickly fled the area (3 of 6 trials, 50.0%). Overall response times including approaches and hiding in vegetation near the model lasted from 180.0 to 3,360.0 s (x ± SE: 1,034.4 ± 500.4 s) and ended with sakis abruptly fleeing the area in all cases.

Although sakis alarm called in all visual model experiments, the types of calls produced differed based on predator types. In trials with jaguar models, alarm calls consisted of a continuous series of noisy grunts, chucks, twitters, growls, and occasional shrieks (Adams, 2009). Whereas during harpy eagle trials, sakis emitted oc-casional high-pitched whistle-like calls in brief bursts (5–10 s), rather than constant calling.

3.5 | Effect of other factors?

There was no significant effect of trial order on PC1 (F1,23.072 = 0.044, p = .836). In other words, subjects hearing or seeing stimuli for the first time were not more likely to produce a stronger response than those that had been exposed previously to a particular stimuli type (first trial: EMM ± SE: 0.329 ± 0.223, 95% CI: −0.158 to 0.817; subsequent trials: 0.382 ± 0.175, 95% CI: 0.008 to 0.756). Similarly, the presence of infants or young juveniles in the group had no significant effect on PC1 (F1,22.030 = 0.128, p = .724). Subjects with young offspring pre-sent did not produce stronger responses than those without young offspring present (young present: EMM ± SE: 0.309 ± 0.181, 95% CI: −0.085 to 0.702; no young present: EMM ± SE: 0.403 ± 0.224, 95% CI: −0.084 to 0.889).

4  | DISCUSSION

4.1 | Does presentation modality affect saki monkey responses to predators?

Using both model and playback experiments to simulate predator presence, we found that wild Rylands’ bald-faced saki monkeys in Amazonia Peru had overall stronger responses (an index combining our numeric measures; Table 2) to visual models than to acoustic

     |  571ADAMS AnD KITCHEn

playbacks. Regardless of predator type, subjects had shorter laten-cies to respond, produced more vocalizations, and had longer reac-tion durations when presented with a model than a playback trial. While subjects approached and alarm called in all visual model trials, they approached the experiment area and alarm called in less than a quarter of playback trials. Similarly, mobbing (multiple group mem-bers approaching, circling, and lunging at the stimulus while shaking their bodies, tails, and branches, and alarm calling for extended pe-riods of time) occurred in model trials (all jaguar and one harpy trial) but never in playback trials.

In similar tests, Campbell's monkeys (Cercopithecus camp-belli: Ouattara et al., 2009) and predator-naïve common marmo-sets (Callithrix jacchus: Kemp & Kaplan, 2011) exhibited stronger responses to visual predator stimuli than to auditory playbacks. Likewise, Old World putty-nosed monkeys (Cercopithecus nictitans martini) have stronger responses to terrestrial predator models than to playbacks, although the two sensory modes elicited similar re-sponses to aerial predators (Arnold et al., 2008).

As a proximate explanation for the stronger responses to mod-els than to playbacks in our study, sound can be difficult to localize and thus, particularly in the case of a predator like harpy eagles that frequently rely on pursuit-hunting strategies (Gil-da-Costa, 2007; Gil-da-Costa et al., 2003; Robinson, 1994), the best option to avoid an unseen threat would be to drop into the lower canopy and hide. This is in fact what saki subjects did in all but one of our playback trials. The sakis also concealed themselves in thick vegetation for the majority of their response time during harpy model experiments, but only after first approaching and visually inspecting the model. Perhaps this was because they could see the predator's location and behavior so that evasive action would be possible if the predator attacked.

In addition to alerting conspecifics, alarm calls may also con-vey information to predators by informing them that their presence has been detected (Caro, 1995; Hasson, 1991; Woodland, Jaafar, & Knight, 1980; Zuberbühler, Jenny, & Bshary, 1999). This pursuit-de-terrent function of alarm calls may be particularly important in the presence of predators like jaguars that rely solely on ambush hunting strategies. By eliminating the surprise aspect of the attack, the preda-tor's hunting strategy is no longer effective, and they are likely to flee the area (Curio, 1978; Gursky, 2006). Indeed, our previous research data related to this paper are available in the figshare repository at https ://doi.org/10.6084/m9.figsh are.8969864 has demonstrated that ocelots (Leopardus pardalis) are deterred by saki alarm calls, which may explain why sakis exhibited more consistent approach responses and mobbing reactions to jaguar models than to jaguar playbacks given that the specific location of the ambush predator was still unknown in the playback trials..

4.2 | Do saki monkeys recognize the vocalizations of their predators?

Our control experiments demonstrated that saki monkeys can distin-guish predator vocalizations from those of non-predators, consistent

with other primate playback studies (e.g., Fichtel & Kappeler, 2002; Searcy & Caine, 2003; Seiler, Schwitzer, & Holderied, 2013). In con-trast to the strong look and movement responses to experimental playbacks, saki subjects maintained their position and continued for-aging or resting without changing their gaze in the majority (62.5%) of control (i.e., non-predator) playbacks. Thus, there was no indica-tion that sakis perceived non-predator calls, at least of the species we used as controls, as threatening as they might a pseudopredator (see Barnett et al., 2018).

4.3 | Do saki monkeys have predator-specific responses in both presentation modes?

Although saki monkeys had a stronger overall response to jaguar than to harpy eagle stimuli, the difference was driven entirely by visual experiments. In the acoustic mode, sakis had the same overall strength of response to both predator stimuli. Thus, had we relied on this strength of response data alone, we might have concluded that sakis cannot differentiate among predator classes (e.g., felids vs. eagles) based on sound alone.

However, the lack of a difference in overall strength of response to audio playbacks was because of predator-specific differences in the three continuous response variables used to develop the re-sponse strength index; whereas sakis responded with more vocal-izations to harpy eagles, they had shorter latencies to respond and longer response durations to jaguars (Table 2). Other details on saki monkey behavioral responses to playbacks demonstrate that they indeed associate different types of predator vocalizations with spe-cific types of threats. For example, direction of subject's first glance (i.e., up to harpy eagles and down to jaguars) corresponded to the di-rection from which the simulated predator would typically attack in almost all occasions. Additionally, sakis almost invariably responded with immediate movement into the lower canopy after harpy eagle playbacks, whereas their response movements to jaguar playbacks were highly varied, with subjects approaching the experiment area at times and remaining motionless or moving away on other occasions. Thus, sakis appear to exhibit more specific movement responses to harpy eagle shrieks compared to jaguar growls.

One possible explanation for the more specific response move-ments to harpy eagle shrieks is that aerial predators appear to prey on arboreal platyrrhine primates more frequently than terrestrial predators (Ferrari, 2009; Gleason & Norconk, 2002; Hart, 2007). If aerial predators indeed pose a greater threat to sakis, then selec-tion should favor more specific escape responses. However, there is increasing evidence that felids, such as jaguars, pumas, and oce-lots, consume primates at relatively high rates in some Neotropical locations (Bianchi & Mendes, 2007; Santos, Paschoal, Massara, & Chiarello, 2014). For example, Peetz et al. (1992) report on a jaguar killing nearly an entire howler monkey (Alouatta seniculus) group on an island in Venezuela. Interestingly, evidence suggested that the jaguar hunted the monkeys while they were sleeping in defoliated trees. Therefore, while jaguars are primarily terrestrial (Tewes &

572  |     ADAMS AnD KITCHEn

Schmidly, 1987), they are certainly capable of climbing trees and may attack arboreal prey from the canopy (Peetz et al., 1992). This ability to attack from both the ground and trees might explain why sakis in our study exhibited varied response movements upon hearing jaguar growls, particularly because they were not able to visually locate the predator in the forest matrix.

The distinctive hunting strategies of harpy eagles and jaguars might also help explain why sakis exhibited differential response movements to our playbacks depending on predator type. Gil-da-Costa et al. (2003) found that harpy eagles often pursue howler monkeys (Alouatta palliata) by perching in nearby trees and watching them while emitting a series of calls known as “predator-assessment” calls prior to attack. In contrast, jaguars are stealthy hunters that rely on surprise attack (Tewes & Schmidly, 1987). Therefore, it is possible that sakis do not as strongly associate jaguar vocalizations with the threat of being attacked because jaguars do not typically advertise their presence when hunting. Although sakis clearly recognize the jaguar growl sound as a terrestrial predator (e.g., based on look di-rection), they may not associate it with a specific escape response.

The detailed behavioral responses of groups to the two types of predator models also suggest escape specificity based on pred-ator hunting style. Although sakis approached all models, spent approximately the same amount of time in the experiment area in both types of model trials, and all trials with a model ended in the monkeys fleeing the area, sakis spent the majority of the time in the harpy model trials concealed in vegetation. On the other hand, al-though sakis mobbed the harpy eagle model once, they mobbed the jaguar model in all trials. Mobbing is a risky behavior for these small primates (e.g., Tórrez, Robles, González, & Crofoot, 2012) and may function more as a deterrent to ambush predators than pursuit pred-ators (Crofoot, 2012). Finally, although sakis called at both model types, they produced loud vocalizations at a continuous rate to the jaguar models, whereas they produced high-pitched, whistle-like calls in brief bursts to the harpy models. High-pitched calls in other species have been determined to be less locatable and more appro-priate for pursuit-style predators like raptors (Caro, 2005), and thus, we plan to examine context differences in alarm calls in bald-faced sakis in the future.

5  | GENER AL CONCLUSIONS

In sum, the consistency of responses between groups suggest that sakis recognize predators based on both acoustic and visual cues and generally respond differentially to predator types based on predator hunting strategies and behavior. The scarcity of predation studies on pitheciids clearly indicates the need for additional research to see whether more members of this taxonomic group respond with pred-ator-specific escape strategies (e.g., Cäsar, Byrne, Hoppitt, et al., 2012a; Cäsar, Byrne, Young, et al., 2012b). However, the vastly dif-ferent responses to visual and acoustic simulations suggest that re-searchers need to consider the explanatory power of the traditional approach of acoustic-only experiments. Even when focusing on

more nuanced behavioral data rather than overall response strength, jaguar playbacks elicited a variable movement response in subjects, whereas jaguar models produced a consistent approach and mob-bing response. Thus, if we had relied on playbacks alone, we might have concluded that sakis have a generalized response to terrestrial ambush predators. Our future work will explore whether varying both predator class and threat levels (by placing models in differ-ent habitats and at different heights) will change saki responses to models, including the types and the patterns of alarm calls produced (e.g., Cäsar et al., 2013).

ACKNOWLEDG MENTSWe are grateful to the Asociación para la Conservación de la Cuenca Amazónica for allowing us to conduct research at La Estación Biológica Los Amigos. We thank Los Amigos staff for their support and friendship during our study. In addition, we thank Beth Erhart, Kerrie Lewis Graham, Joseph Macedonia, and Nigel Pitman for their guidance and support during early stages of this fieldwork. This pro-ject was supported by Texas State University in 2008 and by The Ohio State University (Columbus and Mansfield campuses), Animal Behavior Society, The Society for Integrative and Comparative Biology, Tinker Foundation, and National Science Foundation (BCS 1341174) in 2014.

CONFLIC T OF INTERE S TThe authors declare that they have no conflict of interest.

DATA AVAIL ABILIT Y S TATEMENTData are available in the figshare repository (https ://doi.org/10.6084/m9.figsh are.8969864; Data related to this paper are available in the figshare repository at https ://doi.org/10.6084/m9.figsh are.8969864.

ORCIDDara B. Adams https://orcid.org/0000-0002-3607-3370 Dawn M. Kitchen https://orcid.org/0000-0002-1590-004X

R E FE R E N C E SAdams, D. B. (2009). A preliminary study on vocal communication in the

Gray’s bald-faced saki monkey, Pithecia irrorata. Master’s Thesis. San Marcos, TX: Texas State University.

Adams, D. B., Rehg, J. A., & Watsa, M. (2017). Observations of termi-tarium geophagy by Rylands’ bald-faced saki monkeys (Pithecia ry-landsi) in Madre de Dios, Peru. Primates, 58, 449–459. https ://doi.org/10.1007/s10329-017-0609-8

Adams, D. B., & Williams, S. M. (2017). Fatal attack on a Rylands’ bald-faced saki monkey (Pithecia rylandsi) by a black-and-white hawk-ea-gle (Spizaetus melanoleucus). Primates, 58, 361–365. https ://doi.org/10.1007/s10329-017-0598-7

Aguiar-Silva, F. H., Sanaiotti, T. M., & Luz, B. B. (2014). Food habits of the harpy eagle, a top predator from the Amazonian rainforest can-opy. Journal of Raptor Research, 48, 24–35. https ://doi.org/10.3356/JRR-13-00017.1

Arnold, K., Pohlner, Y., & Zuberbühler, K. (2008). A forest monkey’s alarm call series to predator models. Behavioral Ecology and Sociobiology, 62, 549–559. https ://doi.org/10.1007/s00265-007-0479-y

     |  573ADAMS AnD KITCHEn

Barnett, A. A., Oliveira, T., Soares da Silva, R. F., Albuquerque Teixeira, S., Tomanek, P., Todd, L. M., & Boyle, S. A. (2018). Honest error, precaution or alertness advertisement? Reactions to vertebrate pseudopredators in red-nosed cuxiús (Chiropotes albinasus), a high-canopy neotropical primate. Ethology, 124, 177–187. https ://doi.org/10.1111/eth.12721

Barnett, A. A., Schiel, V., Deveny, A., Valsko, J., Spironello, W. R., & Ross, C. (2011). Predation on Cacajao melanocephalus ouakary and Cebus albifrons (Primates: Platyrrhini) by harpy eagles. Mammalia, 75, 169–172. https ://doi.org/10.1515/mamm.2011.004

Barnett, A. A., Silla, J. M., de Oliveira, T., Boyle, S. A., Bezerra, B. M., Spironello, W. R., … Pinto, L. P. (2017). Run, hide, or fight: Anti-predation strategies in endangered red-nosed cuxiú (Chiropotes albi-nasus, Pitheciidae) in southeastern Amazonia. Primates, 58, 353–360. https ://doi.org/10.1007/s10329-017-0596-9

Barrera, J. P., Chong, L., Judy, K. N., & Blumstein, D. T. (2011). Reliability of public information: Predators provide more information than con-specifics. Animal Behaviour, 81, 779–787. https ://doi.org/10.1016/j.anbeh av.2011.01.010

Berthet, M., Mesbahi, G., Pajot, A., Cäsar, C., Neumann, C., & Zuberbühler, K. (2019). Titi monkeys combine alarm calls to create probabilistic meaning. Science Advances, 5, eaav3991. https ://doi.org/10.1126/sciadv.aav3991

Berthet, M., Neumann, C., Mesbahi, G., Cäsar, C., & Zuberbühler, K. (2018). Contextual encoding in titi monkey alarm call sequences. Behavioral Ecology and Sociobiology, 72, 8. https ://doi.org/10.1007/s00265-017-2424-z

Bezerra, B. M., Barnett, A. A., Souto, A., & Jones, G. (2009). Predation by the tayra on the common marmoset and the pale-throated three-toed sloth. Journal of Ethology, 27, 91–96. https ://doi.org/10.1007/s10164-008-0090-3

Bianchi, R. C., & Mendes, S. L. (2007). Ocelot (Leopardus pardalis) pre-dation on primates in Caratinga Biological Station, southeast Brazil. American Journal of Primatology, 69, 1173–1178. https ://doi.org/10.1002/ajp.20415

Billings, A. C., Greene, E., & De La Lucia Jensen, S. M. (2015). Are chick-adees good listeners? Antipredator responses to raptor vocaliza-tions. Animal Behaviour, 110, 1–8. https ://doi.org/10.1016/j.anbeh av.2015.09.004

Blumstein, D. T., Cooley, L., Winternitz, J., & Daniel, J. C. (2008). Do yel-low-bellied marmots respond to predator vocalizations? Behavioral Ecology and Sociobiology, 62, 457–468. https ://doi.org/10.1007/s00265-007-0473-4

Buchanan, D. B., Mittermeier, R. A., & Van Roosmalen, M. G. M. (1981). The saki monkeys, genus Pithecia. In A. F. Coimbra-Filho, & R. A. Mittermeier (Eds.), Ecology and behavior of neotropical primates, Vol. I (pp. 391–417). Rio de Janeiro, Brazil: Academia Brasileira de Ciencias.

Caro, T. M. (1995). Pursuit-deterrence revisited. Trends in Ecology and Evolution, 10, 500–503. https://doi.org/10.1016/S0169-5347(00)89207-1

Caro, T. M. (2005). Antipredator defenses in birds and mammals. Chicago, IL: University of Chicago Press.

Cäsar, C., Byrne, R. W., Hoppitt, W., Young, R. J., & Zuberbühler, K. (2012a). Evidence for semantic communication in titi monkey alarm calls. Animal Behaviour, 84, 405–411. https ://doi.org/10.1016/j.anbeh av.2012.05.010

Cäsar, C., Byrne, R., Young, R. J., & Zuberbühler, K. (2012b). The alarm call system of wild black-fronted titi monkeys, Callicebus nigri-frons. Behavioral Ecology and Sociobiology, 66, 653–667. https ://doi.org/10.1007/s00265-011-1313-0

Cäsar, C., & Zuberbühler, K. (2012). Referential alarm calling behaviour in New World primates. Current Zoology, 58, 680–697. https ://doi.org/10.1093/czool o/58.5.680

Cäsar, C., Zuberbühler, K., Young, R. J., & Byrne, R. W. (2013). Titi mon-key calls sequences vary with predator location and type. Biology Letters, 9, 20130535. https ://doi.org/10.1098/rsbl.2013.0535

Crofoot, M. C. (2012). Why Mob? Reassessing the costs and benefits of primate predator harassment. Folia Primatologica, 83, 252–273. https ://doi.org/10.1159/00034 3072

Curio, E. (1978). The adaptive significance of avian mobbing. I. Teleonomic hypotheses and predictions. Ethology, 48, 175–183. https ://doi.org/10.1111/j.1439-0310.1978.tb00254

D'Amato, M. R., & Colombo, M. (1985). Auditory matching-to sample in monkeys (Cebus apella). Animal Learning and Behavior, 13, 375–382. https ://doi.org/10.3758/BF032 08013

de Luna, A. G., Sanmiguel, R., Difiore, A., & Fernandez-Duque, E. (2010). Predation and predation attempts on red titi monkeys (Callicebus discolor) and equatorial sakis (Pithecia aequatorialis) in Amazonian Ecuador. Folia Primatologica, 81, 86–95. https ://doi.org/10.1159/00031 4948

Ferrari, S. F. (2009). Predation risk and anti-predator strategies. In P. A. Garber, & A. Estrada (Eds.), South American primates: Comparative perspectives in the study of behavior, ecology, and conservation (pp. 251–278). New York, NY: Springer.

Ferrari, S. F., Pereira, W. L. A., Santos, R. R., & Veiga, L. M. (2004). Fatal at-tack of a Boa constrictor on a bearded saki (Chiropotes satanas utahicki). Folia Primatologica, 75, 111–113. https ://doi.org/10.1159/00007 6272

Fichtel, C. (2007). Avoiding predators at night: Anti-predator strategies in red-tailed sportive lemurs (Lepilemur ruficaudatus). American Journal of Primatology, 69, 611–624. https ://doi.org/10.1002/ajp.20363

Fichtel, C., & Kappeler, P. M. (2002). Anti-predator behavior of group-liv-ing Malagasy primates: Mixed evidence for a referential alarm call system. Behavioral Ecology and Sociobiology, 51, 262–275. https ://doi.org/10.1007/s00265-001-0436-0

Fichtel, C., Perry, S., & Gros-Louis, J. (2005). Alarm calls of white-faced capuchin monkeys: An acoustic analysis. Animal Behavior, 70, 165–176. https ://doi.org/10.1016/j.anbeh av.2004.09.020

Fritz, J., Mishkin, M., & Saunders, R. C. (2005). In search of an audi-tory engram. Proceedings of the National Academy of Sciences of United States of America, 102, 9359–9364. https ://doi.org/10.1073/pnas.05039 98102

Gilbert, K. A. (2000). Attempted predation on a white-faced saki in the Central Amazon. Neotropical Primates, 8, 103–104.

Gil-da-Costa, R. (2007). Howler monkeys and harpy eagles: A commu-nication arms race. In S. L. Gursky, & K. A. I. Nekaris (Eds.), Primate anti-predator strategies (pp. 289–307). Boston, MA: Springer.

Gil-da-Costa, R., Palleroni, A., Hauser, M. D., Touchton, J., & Kelley, J. P. (2003). Rapid acquisition of an alarm response by a neotropical pri-mate to a newly introduced avian predator. Proceedings of the Royal Society of London. Series B: Biological Sciences, 270, 605–610. https ://doi.org/10.1098/rspb.2002.2281

Gleason, T. M., & Norconk, M. A. (2002). Predation risk and anti-preda-tor adaptations in white-faced sakis, Pithecia pithecia. In L. A. Miller (Ed.), Eat or be eaten: Predator sensitive foraging among primates (pp. 169–186). Cambridge, UK: Cambridge University Press.

Gursky, S. (2006). Function of snake mobbing in spectral tarsiers. American Journal of Physical Anthropology, 129, 601–608. https ://doi.org/10.1002/ajpa.20364

Hart, D. (2007). Predation on primates: A biogeographic analysis. In S. Gursky, & K. A. I. Nekaris (Eds.), Primate anti-predator strategies (pp. 27–59). New York, NY: Springer.

Hasson, O. (1991). Pursuit-deterrent signals: Communication between prey and predator. Trends in Ecology and Evolution, 6, 325–329. https ://doi.org/10.1016/0169-5347(91)90040-5

Hauser, M. D., & Wrangham, R. W. (1990). Recognition of predator and competitor calls in nonhuman primates and birds: A preliminary report. Ethology, 86, 116–130. https ://doi.org/10.1111/j.1439-0310.1990.tb004 23.x

Hettena, A. M., Munoz, N., & Blumstein, D. T. (2014). Prey responses to predator’s sounds: A review and empirical study. Ethology, 120, 427–452. https ://doi.org/10.1111/eth.12219

574  |     ADAMS AnD KITCHEn

Heymann, E. W. (1987). A field observation of predation on a mous-tached tamarin (Saguinus mystax) by an anaconda. International Journal of Primatology, 8, 193–195. https ://doi.org/10.1007/BF027 35163

Holm, S. (1979). A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics, 6, 65–70.

Kemp, C., & Kaplan, G. (2011). Individual modulation of anti-predator re-sponses in common marmosets. International Journal of Comparative Psychology, 24, 112–136.

Kirchhof, J., & Hammerschmidt, K. (2006). Functionally referential alarm calls in tamarins (Saguinus fuscicollis and Saguinus mystax)–evidence from playback experiments. Ethology, 112, 346–354. https ://doi.org/10.1111/j.1439-0310.2006.01165.x

Kitchen, D. M., Cortés-Ortiz, L., Dias, P. A. D., Canales-Espinosa, D., & Bergman, T. J. (2018). Alouatta pigra males ignore A. palliata loud calls: A case of failed rival recognition? American Journal of Physical Anthropology, 166, 433–441. https ://doi.org/10.1002/ajpa.23443

Kroodsma, D. E. (1989). Suggested experimental designs for song playbacks. Animal Behaviour, 37, 600–609. https ://doi.org/10.1016/0003-3472(89)90039-0

Lima, S. L., & Dill, L. M. (1990). Behavioral decisions made under the risk of predation: A review and prospectus. Canadian Journal of Zoology, 68, 619–640. https ://doi.org/10.1139/z90-092

Macedonia, J. M., & Young, P. L. (1991). Auditory assessment of avian predator threat in semi-captive ringtailed lemurs (Lemur catta). Primates, 32, 169–182. https ://doi.org/10.1007/BF023 81174

Marsh, L. K. (2014). A taxonomic revision of the saki monkeys, Pithecia Desmarest, 1804. Neotropical Primates, 21, 1–163. https ://doi.org/10.1896/044.021.0101

Martins, S. S., Lima, E. M., & Silva, J. S. Jr (2005). Predation of a bearded saki (Chiropotes utahicki) by a harpy eagle (Harpia harpyja). Neotropical Primates, 13, 7–10. https ://doi.org/10.1896/1413-4705.13.1.7

Matsuda, I., & Izawa, K. (2008). Predation of wild spider monkeys at La Macarena, Columbia. Primates, 49, 65–68. https ://doi.org/10.1007/s10329-007-0042-5

McGregor, P. K. (1992). Quantifying responses to playback: One, many, or composite multivariate measures? In P. K. McGregor (Ed.), Playback and studies of animal communication (pp. 79–96). New York, NY: Plenum Press.

Mondolfi, E. (1986). Notes on the biology and status of the small wild cats in Venezuela. In S. D. Miller, & D. D. Everett (Eds.), Cats of the world: Biology, conservation, and management (pp. 125–146). Washington, DC: National Wildlife Federation.

Mourthé, Í., & Barnett, A. A. (2014). Crying tapir: The functionality of errors and accuracy in predator recognition in two Neotropical high-canopy primates. Folia Primatologica, 85, 379–398. https ://doi.org/10.1159/00037 1634

Munoz-Lopez, M., Mohedano-Moriano, A., & Insausti, R. (2010). Anatomical pathways for auditory memory in primates. Frontiers in Neuroanatomy, 4, 129. https ://doi.org/10.3389/fnana.2010.00129

Neal, O. (2009). Responses to the audio broadcasts of predator vocaliza-tions by eight sympatric primates in Suriname, South America. Master’s Thesis. Kent State University, Ohio.

Norconk, M. A. (2007). Sakis, uakaris, and titi monkeys. In C. J. Campbell, A. Fuentes, K. MacKinnon, M. Panger, & S. Bearder (Eds.), Primates in Perspective (pp. 123–138). New York, NY: Oxford University Press.

Orsini, V. S., Nunes, A. V., & Marsh, L. K. (2017). New distribution records of Pithecia rylandsi and Pithecia mittermeieri (Primates, Pitheciidae) and an updated distribution map. Check List, 13, 2123. https ://doi.org/10.15560/ 13.3.2123

Ouattara, K., Lemasson, A., & Zuberbühler, K. (2009). Anti-predator strategies of free-ranging Campbell’s monkeys. Behaviour, 146, 1687–1708. https ://doi.org/10.1163/00057 9509X 12469 53372 5585

Palminteri, S., & Peres, C. (2012). Habitat selection and use of space by bald-faced sakis (Pithecia irrorata) in Southwestern Amazonia: Lessons

from a multiyear, multigroup study. International Journal of Primatology, 33, 401–417. https ://doi.org/10.1007/s10764-011-9573-0

Palminteri, S., Powell, G. V., & Peres, C. A. (2012). Advantages of granivory in seasonal environments: Feeding ecology of an arboreal seed predator in Amazonian forests. Oikos, 121, 1896–1904. https ://doi.org/10.1111/j.1600-0706.2012.20456.x

Papworth, S., Böse, A., Barker, J., Schel, A. M., & Zuberbühler, K. (2008). Male blue monkeys alarm call in response to danger experienced by others. Biology Letters, 4, 472–475. https ://doi.org/10.1098/rsbl.2008.0299

Peetz, A., Norconk, M., & Kinzey, W. (1992). Predation by jaguar on howler monkeys (Alouatta seniculus) in Venezuela. American Journal of Primatology, 28, 223–228. https ://doi.org/10.1002/ajp.13502 80307

Peres, C. A. (1990). A harpy eagle successfully captures an adult male red howler monkey. Wilson Bulletin, 102, 560–561.

Pitman, N. C. A. (2010). An overview of the Los Amigos watershed, Madre de Dios, southeastern Perú. Retrieved from https ://www.resea rchga te.net/publi catio n/27535 0970_An_overv iew_of_the_Los_Amigo swate rshed_Madre_de_Dios_south easte rn_Peru

Quintino, E. P., & Bicca-Marques, J. C. (2013). Predation of Alouatta puruensis by Boa constrictor. Primates, 54(4), 325–330. https ://doi.org/10.1007/s10329-013-0377-z

Rahlfs, M., & Fichtel, C. (2010). Anti-predator behaviour in a nocturnal primate, the grey mouse lemur (Microcebus murinus). Ethology, 116, 429–439. https ://doi.org/10.1111/j.1439-0310.2010.01756.x

Rettig, N. L. (1978). Breeding behavior of the harpy eagle (Harpia harpyja). The Auk, 95, 629–643.

Robinson, S. R. (1980). Antipredator behaviour and predator recognition in Belding’s ground squirrels. Animal Behaviour, 28, 840–852. https ://doi.org/10.1016/S0003-3472(80)80144-8

Robinson, S. K. (1994). Habitat selection and foraging ecology of rap-tors in Amazonian Peru. Biotropica, 26, 443–458. https ://doi.org/10.2307/2389239

Santos, J. L., Paschoal, A. M. O., Massara, R. L., & Chiarello, A. G. (2014). High consumption of primates by pumas and ocelots in a remnant of the Brazilian Atlantic Forest. Brazilian Journal of Biology, 74, 632–641. https ://doi.org/10.1590/bjb.2014.0094

Schel, A. M. (2009). Anti-predator behavior of Guereza colobus monkeys (Colobus guereza). Ph.D. Dissertation. University of St. Andrews, St. Andrews, Fife, Scotland.

Searcy, Y. M., & Caine, N. G. (2003). Hawk calls elicit alarm and defensive reactions in captive Geoffroy’s marmosets (Callithrix geoffroyi). Folia Primatologica, 74, 115–125. https ://doi.org/10.1159/00007 0645

Seiler, M., Schwitzer, C., & Holderied, M. (2013). Anti-predator behaviour of Sahamalaza sportive lemurs, Lepilemur sahamalazensis, at diurnal sleeping sites. Contributions to Zoology, 82, 131–143. https ://doi.org/10.1163/18759 866-08203003

Serrano-Villavicencio, J. E., Hurtado, C. M., Vendramel, R. L., & do Nascimento, F. O. (2019). Reconsidering the taxonomy of the Pithecia irrorata species group (Primates: Pitheciidae). Journal of Mammalogy, 100, 130–141. https ://doi.org/10.1093/jmamm al/gyy167

Stenzel, C. (2017). Anti-predator behavior and discriminative abilities: Playback experiments with free-ranging equatorial saki monkeys (Pithecia aequatorialis) in the Peruvian Amazon. Master’s Thesis. Winthrop University, Rock Hill, South Carolina.

Tewes, M. E., & Schmidly, D. J. (1987). The neotropical felids: Jaguar, oce-lot, margay, and jaguarundi. In M. Novak, J. Baker, M. E. Obbard, & B. Malloch (Eds.), Wild furbearer management and conservation in North America (pp. 703–705). Ontario, Canada: Ministry of Natural Resources.

Tobler, M. W., Carrillo-Percastegui, S. E., Zuñiga Hartley, A., & Powell, G. V. N. (2013). High jaguar densities and large population sizes in the core habitat of the southwestern Amazon. Biological Conservation, 159, 375–381. https ://doi.org/10.1016/j.biocon.2012.12.012

     |  575ADAMS AnD KITCHEn

Tórrez, L., Robles, N., González, A., & Crofoot, M. C. (2012). Risky business? Lethal attack by a jaguar sheds light on the costs of predator mobbing for capuchins (Cebus capucinus). International Journal of Primatology, 33, 440–446. https ://doi.org/10.1007/s10764-012-9588-1

Wheeler, B. C. (2008). Selfish or altruistic? An analysis of alarm call func-tion in wild capuchin monkeys, Cebus apella nigritus. Animal Behaviour, 76, 1465–1475. https ://doi.org/10.1016/j.anbeh av.2008.06.023

Wheeler, B. C. (2010). Production and perception of situationally vari-able alarm calls in wild tufted capuchin monkeys (Cebus apella nigri-tus). Behavioral Ecology and Sociobiology, 64, 989–1000. https ://doi.org/10.1007/s00265-010-0914-3

Wich, S. A., & Sterk, E. H. M. (2003). Possible audience effect in thomas langurs (primates; Presbytis thomasi): An experimental study on male loud calls in response to a tiger model. American Journal of Primatology, 60, 155–159. https ://doi.org/10.1002/ajp.10102

Wildman, D. E., Jameson, N. M., Opazo, J. C., & Soojin, V. Y. (2009). A fully resolved genus level phylogeny of neotropical primates (Platyrrhini). Molecular Phylogenetics andEvolution, 53, 694–702. https ://doi.org/10.1016/j.ympev.2009.07.019

Woodland, D. J., Jaafar, Z., & Knight, M. L. (1980). The ‘pursuit deterrent’ function of alarm signals. The American Naturalist, 115, 748–753. https ://doi.org/10.1086/283596

Wright, A. A. (2007). An experimental analysis of memory processing. Journal of Experimental Analysis of Behavior, 88, 405–433. https ://doi.org/10.1901/jeab.2007.88-405

Ximenes, A. (1982). Notas sobre félidos neotropicales VIII: Observaciones sobre el contenido estomal y el comportamiento alimentar de diver-sas especies de felinos. Revista Nordestina De Biologia, 5, 89–91.

Zuberbühler, K. (2000). Referential labeling in Diana monkeys. Animal Behaviour, 59, 917–927. https ://doi.org/10.1006/anbe.1999.1317

Zuberbühler, K. (2009). Survivor signals: The biology and psychology of animal alarm calling. Advances in the Study of Behavior, 40, 277–322. https ://doi.org/10.1016/S0065-3454(09)40008-1

Zuberbühler, K., Jenny, D., & Bshary, R. (1999). The predator deterrence function of primate alarm calls. Ethology, 105, 477–490. https ://doi.org/10.1046/j.1439-0310.1999.00396.x

Zuberbühler, K., Noë, R., & Seyfarth, R. M. (1997). Diana monkey long-distance calls: Messages for conspecifics and predators. Animal Behaviour, 53, 589–604. https ://doi.org/10.1006/anbe.1996.0334

How to cite this article: Adams DB, Kitchen DM. Model vs. playback experiments: The impact of sensory mode on predator-specific escape responses in saki monkeys. Ethology. 2020;126:563–575. https ://doi.org/10.1111/eth.13008