Behavioural effects of noise on Linnaeus’s two-toed sloth (Choloepus didactylus) in a walk-through enclosure

Anthropogenic noise has been related to stress in captive animals; despite this there have been few studies on animal welfare assessment in walk-through zoo enclosures. We aimed to investigate the behavioural effects of noise on a male-female pair of two-toed sloths (Choloepus didactylus), housed in a walk-through enclosure in a zoo in the UK. The animals were filmed for 24 h per day, during three days per week, including days with potential low and high flow of visitors, for three weeks. Sound pressure measurement was performed four times each collection day (twice in the morning, once at noon and once in the afternoon), for 15 min per session, using a sound level meter. The number of visitors passing the enclosure during each session was also recorded. The videos were analysed using focal sampling, with continuous recording of behaviour. Correlations between noise and the behaviours expressed during, and in the 24 h after the acoustic recording, were investigated. The number of visitors correlated with the acoustic parameters. At the moment of exposure, higher levels of noise correlated with decreased inactivity, and longer expression of locomotion and maintenance behaviours for the male; the female spent more time inside a box in these moments. During the 24 h hours after exposure to loud noise, the female showed no behavioural changes while the male tended to reduce foraging. The behavioural changes observed in both individuals have already been reported in other species, in response to stressful events. Our study indicates the need for a good acoustic management in walk-through zoo enclosures where sloths are housed.


Introduction
Animals housed in zoos are exposed to various stimuli that can impinge upon their welfare (Birke 2002;Cooke & Schillaci 2007;Clark et al. 2012;Maia et al. 2012).Among these stimuli there is exposure to visitors (Carder & Semple 2008;Clark et al. 2012;Farrand et al. 2014), which has been connected to the 'zoo-visitor effect' (e.g.Davey 2007).Such effect may be assessed through changes in behaviour and/or physiological responses of the animals, when exposed to zoo visitors (Davey 2006).However, research on visitor effect may lack scientific rigorousness as a result of constraints regarding the control of variables related to visitor presence (Farrand et al. 2014).Animals may perceive the presence of visitors via a variety of different perception channels: visual, olfactory and auditory (Young 2003).While visual and olfactory stimuli are difficult to measure, accurate quantification of auditory stimuli is feasible.Zoo visitor-noise pollution has been referred to as having a negative effect on animal welfare (Owen et al. 2004;Powell et al. 2006).Besides the effects of noise on the stress-response system (e.g.Bowles & Eckert 1997;Ward et al. 1999;Owen et al. 2004;Wysocki et al. 2006), noise has been reported to cause DNA damage, alterations in gene expression and numerous cellular processes with effects on neural, developmental, immunological and physiological functioning (Kight & Swaddle 2011).
Zoo-animal responses to visitors may depend upon species-and individual-characteristics, the nature of visitor-animal interactions, and enclosure design (Woolway & Goodenough 2017;Learmonth et al. 2018).Compared to traditional facilities (in which visitors remain outside), walk-through enclosures have greater potential to affect the welfare of animals, since contact with visitors (auditory, visual) is magnified, due to greater physical proximity and/or the absence of physical barriers between animals and visitors (Learmonth et al. 2018).However, such enclosures are increasingly popular, and their impacts on animal welfare are understudied (Sherwen et al. 2015b).The few studies addressing this type of housing report that visitors have an effect on animals: quokkas (Setonix brachyurus) spent more time hidden when the enclosure was open to visitors (Learmonth et al. 2018) and squirrels (Sciurus vulgaris: Woolway & Goodenough 2017) moved more and fed less.In contrast, Jones et al. (2016) pointed to a positive effect on crowned lemurs (Eulemur coronatus) via a decrease in aggression among conspecifics with an increase in numbers of visitors.The scarcity of data indicates the need for investigations into the effects of such enclosures on the welfare of animals.
The two-toed sloth (Choloepus didactylus) is a mid-sized, nocturnal tree mammal found in the rainforests of South America, which spends most of its time in the treetops (Eisenberg & Redford 1999;Nowak & Walker 1999;Bezerra 2008;Peery & Pauli 2012).Having very low metabolic rates (roughly half those of other placental mammals) and feeding on a low-energy diet, they require up to 14 h of daily inactivity, and locomotion occurs slowly (Montgomery & Sunquist 1975;Nagy & Montgomery 1980).In this study, we investigated the possible effects of visitor noise on the behaviour of two-toed sloths, housed in a walk-through zoo enclosure.

Ethical permission
All procedures here were evaluated and approved by the Ethics Committee for Animal Use from the Pontifical Catholic University of Minas Gerais (Permit n 007/2014).

Study protocol
One male and one female two-toed sloth, housed in a walk-through enclosure in a zoo in the UK were the subject of this study.Visitors were unable to touch the sloths in the enclosure,with animals remaining above them (at a height of approximately 6 m) for the majority of the time, i.e. hanging from ropes, without any barriers to isolate the animals from visitor noise.The enclosure was indoors, made of concrete and round in shape (approximately 11 m in diameter).The animals were fed in the mornings, prior to the admittance of visitors.Food was made available through environmental enrichment: feeding items were spread over the floor, and within the branches of trees in the enclosure.The animals were filmed for 24 h per day, for three consecutive days a week (Fridays, Saturdays and Sundays), during the first three weeks of July 2017.The videos were analysed through focal sampling with continuous recording of behaviour, using the Solomon Coder programme (Copyright 2006-2017 by András Péter).Behavioural observations were based on an ethogram (Table 1) adapted from Hayssen (2011), Silva et al. (2013), and Clark and Melfi (2012).Sound measurement was performed four times per day using the SVAN 977A sound level meter (SVAN 977A, SVANTEK, Poland).The equipment was fixed on a tripod, in the visitor space, facing the enclosure, 1.5 m above the floor and a good distance clear from the boundaries of the enclosure.Each session lasted 15 min, distributed between morning (0930-0945h; with the zoo still closed to visitors), mid-morning (1045-1100h), midday (1200-1215h) and afternoon (1500-1515h).
During each session, the number of visitors passing through the enclosure was also assessed.Since there have been no studies to date assessing which aspects of sound have the greatest effect on animals, we chose to evaluate three parameters, using the A-weighting filter: Equivalent Continuous Noise Level (LA eq ), Maximum Sound Pressure Level (LA max ), and Peak Value (LA peak ).The LA eq is used to measure the average sound pressure levels and calculate the average noise level (energy) to which an environment is exposed (Duarte et al. 2011), LA max is the maximum level of noise (mean square root) during a certain period and LA peak is the highest point of raw sound pressure, without considering time.A-weighting was chosen because this filter has been tested previously, and was shown to correlate to sloths' behaviour more than filter Z.This correlation suggests their acoustic sensitivity may be similar to ours (Queiróz 2018).

Statistical analysis
For statistical analysis, we grouped the observed behaviours into the following categories: Inactivity; Locomotion; Maintenance; Foraging; Sexual behaviour; Affiliative interaction; Agonistic interaction; and Non-visible.Every behavioural category was correlated with each acoustic parameter (LA eq , LA max , LA peak ) using Pearson's correlation test for parametric data or Spearman's correlation test for non-parametric data (Zar 2010).These analyses were performed for the behaviours exhibited during the 15-min of acoustic collection, and for the behaviours recorded during the subsequent 24 h, to assess a possible lasting effect of sound pressure.In the case of data correlation within 24 h of noise exposure, the behaviours recorded on video were correlated with the LA eq recorded on the previous day.The number of visitors passing in the enclosure during the 15-min sessions were also correlated to the acoustic parameters.Considering all behaviour data were correlated with three acoustic parameters, we applied Bonferroni correction, and considered P-values which were not greater than 0.017 as significant.

Results
In total, 216 h of behavioural recording, and 36 × 15-min sessions of acoustic recording, were produced.The activity budget of male and female animals can be seen in the Supplementary material.The noise levels recorded ranged from 51.7 to 122.2 dBA (LA peak ), with an LA eq of 85.5 dBA.The acoustic parameters from the mornings (first acoustic collections, without visitors) were: LA eq ranging from 58.2 to 74.7 dBA; LA max ranging from 56.5 to 62.4 dBA; LA peak from 67.5 to 73.3 dBA (Figure 1[a]-[c]).The number of visitors inside the enclosure during each 15-min session (ranging from 2 to 322 people) correlated strongly and positively with the three acoustic parameters (LA peak P < 0.0001, Pearson = 0.92; LA max P < 0.0001, Pearson = 0.92; LA eq P < 0.0001, Spearman = 0.87).During exposure to increasing noise levels, we recorded significant reduction in inactivity (P = 0.0016, r = -0.5141LA peak ; P = 0.0015, r = -0.5158LA max ; P = 0.0011, r = -0.5273LA eq ), and increased locomotion (P = 0.0004, r = 0.5625 LA peak ; P = 0.0004, r = 0.5626 LA max ; P = 0.0019, r = 0.5071 LA eq ; Figure 2) and maintenance behaviours (P = 0.0081, r = 0.4403 LA peak ; P = 0.0083, r = 0.4389 LA max ; P = 0.0149, r = 0.4082 LA eq ; Figure 3) for the male.The female, when noise was higher, spent significantly more time inside a box (P = 0.0170, r = 0.4010 LA peak ; P = 0.0170, r = 0.4006 LA max ; P = 0.0149; Figure 4).Within 24 h of noise exposure, the female showed no behavioural changes but the male tended to spend less time foraging when LA max was higher the previous day (P = 0.0479, r = -0.6709; Figure 5).For all other behaviours recorded, the correlations with acoustic parameters were not statistically significant (for more details, please see Tables S1 and S2, in the Supplementary material).

Discussion
In this study, we evaluated the behaviours exhibited by a pair of two-toed sloths housed in a walk-through enclosure, on days with different levels of sound pressuremeasured as LA peak , LA max and LA eqdue to human visitation.During exposure to higher sound pressure, the male moved more and spent longer time in maintenance behaviours while the female spent more time out of view.
Within 24 h of experiencing intense noise exposure, the male displayed a tendency to reduce foraging.An increase in locomotion, or in general activity in zoo animals as observed in the male sloth in response to increased noise in this studyhas been interpreted as indicative of improved welfare for certain species (e.Marques 2008).However, studies with other species have shown a correlation between increased activity/locomotion and acoustic stress; in pumas (Maia 2009;Maia et al. 2012), gorillas (Clark et al. 2012), chimpanzees and spectacled bears (Noga 2010).In these cases, such increased activity was interpreted as an attempt by the animal to mitigate stress (Mitchell et al. 1992;Boere 2001;Hosey 2005), or simply as restlessness, caused by noise (Davey 2006;Quadros et al. 2014;Hashmi & Sullivan 2020).These apparently contradictory interpretations point to the need for a careful study of data, based on species characteristics (Queiroz & Young 2018) and, preferably, also on a joint evaluation based on technical measurements of sound pressure levels using appropriate equipment and protocols for a more accurate analysis (Quadros et al. 2014;Jakob-Hoff et al. 2019;Hashmi & Sullivan 2020).
The same debate pertaining to the interpretation of increased activity in zoo animals also applies regarding maintenance behaviours; the correlation of such behaviours with stress is often uncertain.Giant pandas increased locomotion, maintenance, scentmarking, and stereotypies during noisy periods but, according to the authors, there was no sign of a decline in welfare (Powell et al. 2006).However, white-handed gibbons (Hylobates lar) increased frequency and duration of scratching behaviour on days with more zoo visitors, especially children (Ribeiro 2016): in this case, scratching may be interpreted as a displacement behaviour (Roth & Cords 2020), suggestive of stress.
Here, we collected behavioural and acoustic data to carefully evaluate possible connections between noise and behaviour.Taking the species' characteristics into account (low metabolic rates, low-energy diet, inactivity during most of the day, and slow locomotion; Montgomery & Sunquist 1975;Nagy & Montgomery 1980), the increased locomotion/activity and maintenance might be connected to acoustic stress.Another factor contributing to this interpretation is that the average noise level in the rainforest, Animal Welfare natural habitat of the study species, is usually around 38 dB(A), considerably lower than the noise levels recorded in this study (Santos 2012).Besides, a study on visitor effect in zoos found that herbivorous species and those from closed habitats, such as ours here, were more negatively impacted by visitors than species from open habitats (Queiroz & Young 2018).The fact that the male has increased locomotion in the moments the noise was more intense during the day, and the records during the subsequent 24 h did not point to such an increase suggests the animal adjusted its activity budget, by relocating locomotion from their natural time (at night) to day-time.Such an adjustment may have unpredictable impacts on the welfare of this individual.Although we have found no data on the hearing capacities of C. didactylus, studies on the ancestors of the sloth extant species point to a possible need for great sound sensitivity in the species (Blanco & Rinderknecht 2012;Blanco & Jones 2016).Those authors base their argument on the short snouts of certain sloth species, which suggest that they possess more focused and specialised sight, and as a consequence, a need for good sound acuity.This greater hearing sensitivity could make the species more prone to developing stress reactions when exposed to high levels of noisesuch as those recorded in this study (e.g. up to 122.2 dBA).Apart from that, the potential (non-measured) high levels of reverberation can be also a factor affecting the sloths' welfare.Both the circular shape of the enclosure and the material with which it was constructed favoured reverberation.Reverberation has been associated with impaired cognitive processes in humans (Kjellberg 2004), and interferences in animal communication (Padgham 2004).The disturbing effects of reverberation may also contribute to animal stress.Different responses to stress in males compared to females have previously been reported (e.g.Vasconcellos et al. 2009;Quadros et al. 2014).Such differing reactions might be due to diverse coping styles (e.g.Koolhaas et al. 1999;Ferreira et al. 2016).
In contrast to the apparently contradictory results mentioned previously, avoidance/hiding behavioursas reported for the female in this studyhave been consistently linked with fear, stress or apathy (Young 2003;Forkman et al. 2007;Sherwen et al. 2015c).Little penguins (Eudyptula minor) increased their distance from the visitor area and spent more time hiding in the presence of visitors (Sherwen et al. 2015c).Our data also corroborates studies with jaguarundis (Buhr 2018) and quokkas (Learmonth et al. 2018) in which hiding behaviours were correlated with visitor presence, and the noise they created.

Animal Welfare
Foraging behaviours are an important component of the behavioural repertoire of any species, since they are essential for survivorship (Young 2003).A reduction in the exhibition of foraging in response to visitor noise has been also reported in pumas (Ricci et al. 2018) and tigers (Panthera tigris: Kerley et al. 2002).On visiting days, a giant otter (Pteronura brasiliensis) showed lower frequency of eating, aquatic activity, playing and rolling, behaviours that tend to be associated with good welfare (Oliveira & Carpi 2016).
Although some behavioural alterations reported in this study could lead to conflicting interpretations if taken in isolation (increase in locomotion and maintenance) when seen as a whole, and considering species' characteristics, our data suggest that our study animals were in a state of restlessness due to noisea condition with the potential to impact negatively on welfare.Such results have already been reported in walk-through enclosures for squirrels (Woolway & Goodenough 2017).The possible restlessness, in conjunction with an increase in hidden (out of view) time for the female and the tendency for decreased foraging as a medium-term noise response for the male, suggests a stress response, which can be related to reduced welfare.Discomfort due to interactions with visitors has already been reported for sloths (brown three-toed sloths [Bradypus variegatus]), with individuals performing vigilance and limb stretching, behaviours either not reported for the species, or performed at lower rates in the wild (Carder et al. 2018).Such results have also been interpreted by those authors as possible evidence of fear and stress.

Animal welfare implications and conclusion
Although our results cannot be generalised due to our small sample size, they suggest the maintenance of two-toed sloths in walk-through enclosureswithout any acoustic control (i.e.sonic barrier or management of visitor behaviour)might be detrimental to their welfare.Yuri Garcia de Abreu Rezende et al.

Figure 1 .
Figure 1.Mean acoustic parameters measured along three weeks, in three periods (morning, noon, and afternoon), in 15-minute sessions inside a walk-through enclosure of Choloepus didactylus, in a zoo in the United Kingdom.A -LAeq (equivalent continuous noise level); B -LAmax (maximum sound pressure level); C -LApeak (peak value).

Figure 2 .
Figure 2. Duration (s) of locomotion of the male Choloepus didactylus housed in a walk-through enclosure, in a zoo in the UK, as a function of the noise level (LApeak) during the 15min sessions of noise recording.

Figure 3 .
Figure 3. Duration (s) of maintenance behaviours of the male Choloepus didactylus housed in a walk-through enclosure, in a zoo in the UK, as a function of the noise level (LApeak) during the 15-min sessions of noise recording.

Figure 4 .
Figure 4. Duration (s) of the time the female Choloepus didactylus spent out of view, in a walk-through enclosure, in a zoo in the UK, as a function of the noise level (LApeak) during the 15-min sessions of noise recording.

Figure 5 .
Figure 5. Duration (s) of foraging behaviour of the male Choloepus didactylus housed in a walk-through enclosure, in a zoo in the UK, as a function of the noise level (LAmax) within 24 h of noise exposure. 6

Table 1 .
Ethogram of Choloepus didactylus observed in this study* One arm is extended aggressively toward another sloth, followed by a scratch or bite Non visible Non visible The animal is not visible to the observer, inside one of the sleeping boxes *Adapted from Hayssen (2011), Silva et al. (2013) and Clark and Melfi (2012).2Yuri Garcia de Abreu Rezende et al.