Physiology & Behavior 105 (2012) 1117–1123

Contents lists available at SciVerse ScienceDirect

Physiology & Behavior

j ourna l homepage: www.e lsev ie r .com/ locate /phb

Genetic differences in temperament determine whether lavender oil alleviates orexacerbates anxiety in sheep

P.A.R. Hawken a,⁎, C. Fiol b, D. Blache a

a UWA Institute of Agriculture (Animal Production), The University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australiab Departamento de Bovinos, Facultad de Veterinaria, Montevideo, Uruguay

⁎ Corresponding author at: M085, School of AnimWestern Australia, 35 Stirling Highway, Crawley, W8 6488 3588; fax: +61 8 6488 1029.

E-mail addresses: penny.hawken@uwa.edu.au (P.A.R(C. Fiol), dominique.blache@uwa.edu.au (D. Blache).

0031-9384/$ – see front matter © 2011 Elsevier Inc. Alldoi:10.1016/j.physbeh.2011.12.005

a b s t r a c t

a r t i c l e i n f o

Article history:Received 21 March 2011Received in revised form 23 November 2011Accepted 7 December 2011

Keywords:AnxietyDepressionTemperamentIsolationStressAlternative therapies

Growing concerns about the risk of addiction to benzodiazepines have led to increasing interest in alternativetherapies to treat anxiety and depression. Lavender oil (Lavendula augustifolia) is reportedly anxiolytic in anumber of species but little is known about how it affects individuals that are more or less anxious whenfaced with a stressor. In this study, we used changes in locomotor activity and the plasma concentrationsof cortisol to test whether lavender oil would reduce behavioral and endocrine correlates of anxiety in calmand nervous sheep exposed to an isolation stressor. During the non-breeding season, ‘calm’ or ‘nervous’ femalesheep from the UWA temperament flock were exposed to a mask containing either 1 mL of 10% lavender oil(calm: n=8; nervous: n=8) or peanut oil (calm: n=8; nervous: n=8). After 30 min, each sheep was isolatedfor 5 min and then returned to the group. Blood was sampled prior to the mask, prior to isolation, 1 min and30 min after isolation to profile changes in the plasma concentrations of cortisol. Agitation score, locomotoractivity and vocalizations were recorded as correlates of anxiety associated with the isolation stressor. Irrespec-tive ofwhether theywere exposed to lavender oil, calm sheep had a lower agitation score (Pb0.001), crossed thecentral lines of the isolation box less frequently (Pb0.001), expressed fewer vocalizations (Pb0.001) and hadlower plasma concentrations of cortisol immediately after isolation (Pb0.001) than nervous sheep. Exposureof calm sheep to lavender oil decreased the agitation score (Pb0.001), frequency of vocalizations (Pb0.05),decreased the number of crosses of the central lines of the isolation box (Pb0.05), and the plasma concentrationsof cortisol prior to isolation (Pb0.05) (after mask application) compared to calm control sheep. Exposure ofnervous sheep to lavender oil increased the frequency of vocalizations (Pb0.05), the number of sheep attempt-ing to escape (Pb0.05) and the plasma concentrations of cortisol 30 min after isolation (Pb0.05) compared tonervous control sheep. We conclude that genetic differences in temperament determine whether lavender oilalleviates or exacerbates the behavioral and/or endocrine correlates of anxiety in sheep.

© 2011 Elsevier Inc. All rights reserved.

1. Introduction

Anxiety can be interpreted as the emotional capacity of an individualto anticipate an aversive situation that is unpredictable and outside oftheir control [1]. Abnormally high levels of anxiety are associated withdysfunction of the hypothalamic–pituitary–adrenal (HPA) axis, and, inhumans, affective disorders such as pathological anxiety and depression[1]. Anxiety disorders affect 40 million people annually in the USA, andare often co-morbidly expressed with depression [2].

Pathological anxiety and depression are frequently treated withbenzodiazepines which act on the gamma-aminobutyric acid complex(GABA) to suppress neural activity in regions of the brain that mediate

al Biology, The University ofA 6009, Australia. Tel.: +61

. Hawken), cfiol@yahoo.com

rights reserved.

anxiety [3]. In the short term, benzodiazepines are effective at reducinganxiety but long term use can be associated with a number of adverseside effects and a relatively high risk of abuse and addiction (review;[4]). Furthermore, polymorphisms in the enzymes and receptorsresponsible for mediating the effects of anti-depressant or anti-psychotic drugs reduce the reliability and predictability of the drugresponse [5].

In recent years, there has been increasing interest in the use of al-ternative therapies to treat a number of medical conditions, includinganxiety disorders and depression [6]. Lavender oil (Lavendula augusti-folia) is commonly used in aromatherapy and allegedly has anxiolyticeffects in a number of species including rats [7], mice [8], Mongoliangerbils [9], dogs [10] and humans [11]. Both the mechanism of action[12] and anxiolytic effect of lavender oil are reputedly similar to thebenzodiazepines, making lavender oil a potentially viable alternativeto benzodiazepines for the treatment of anxiety [9,13,14].

In humans, lavender oil appears to be calmative for young infants[15] and reduces the anxiety associated with visiting the dentist [16]

1118 P.A.R. Hawken et al. / Physiology & Behavior 105 (2012) 1117–1123

or taking examinations [17]. However, interpretation of human studiescan be difficult because lavender oil is often combinedwith other relax-ation strategies such asmassage [18], foot-baths [19] or baths [15,20]. Arecent study also showed that humans expect lavender oil to be calm-ative and that this expectation, rather than the aroma per se, was re-sponsible for mediating the relaxing effects of lavender oil [21]. Theseissues, combined with the difficulty in sourcing an adequate placebo,make it difficult to differentiate between the real and perceived effectsof lavender oil on anxiety in humans [21].

Animal models provide a viable alternative to test the anxiolyticeffects of lavender oil because we can directly measure behavioraland endocrine correlates of anxiety and expectation is unlikely to bean issue. We can also use natural variation in anxiety within a popu-lation to select animals for extremes in this trait, thus allowing theanxiolytic effects of lavender oil to be tested in individuals that aremore or less anxious when faced with a stressor. HAB/LAB rats havebeen selected for decades for extremes in anxiety related behaviorwhen faced with a mild stressor [1,22,23]. Specifically, HAB ratshave a lower propensity to enter the open arms of the elevated plusmaze than LAB rats and secrete more corticosterone when forced todo so [24]. The repeatability and heritability of anxiety related behav-ior of the two lines within and between generations thus providespredictability over the level of anxiety associated with a stressor.

Sheep at the University of Western Australia have been divergentlyselected for 17 generations on the basis of their behavioral responses tosocial isolation and human presence. Isolation and human presenceelicit a strong fear response in sheep [26], that can be quantified byincreased locomotor activity and vocalization frequency [26,27]. Thescores from the behavioral tests used to select the sheep are repeatableover time (r=0.40–0.76) [28] and moderately heritable (h=0.45)[28]. A recent cross-fostering study found the phenotype of calm andnervous sheep to be evident at 1 week of age and minimally affectedby non-genetic factors [29]. Fear and anxiety are closely related emo-tions that describe the reaction of an individual to an actual or potentialthreat [30]. The stressors used to select the UWA temperament flockpose no actual, physical danger to the sheep, so one could argue thatnervous sheep are predisposed to view an ambiguous situation asmore threatening than calm sheep. This type of negative bias is charac-teristic of high anxiety in rats, which combined with the divergence inthe behavioral [29,31,32] and endocrine [32] responses of calm andnervous sheep to social isolation supports the notion that nervoussheep are more ‘anxious’ than calm sheep when faced with an isolationstressor. The calm and nervous sheep of the UWA temperament flockthus draw many parallels to the HAB/LAB rats (strong phenotypic dif-ferences of genetic origin that are evident early in life) [25] and assuch may provide an opportunity to study anxiety in a non-rodentspecies.

Genetic factors play an integral role in determining the level ofanxiety associated with a stressor [22] and the phenotypic expressionof the drug response [33] so may, in turn, affect the anxiolytic proper-ties of lavender oil. In this study, we tested the effect of lavender oilon the behavioral and endocrine correlates of anxiety in calm andnervous sheep exposed to an isolation stressor. The nervous sheephave been genetically selected for high expression of locomotor activ-ity [29,31] when faced with an isolation stressor that is associatedwith significantly higher plasma concentrations of cortisol thanthose observed in calm sheep [32]. Therefore, we hypothesized thatexposure to lavender oil would reduce locomotor activity and plasmaconcentrations of cortisol in sheep selected for nervous temperamentduring exposure to an isolation stressor. Calm sheep express lowlevels of locomotor activity when faced with an isolation stressor[29,31] that are associated with significantly lower plasma concentra-tions of cortisol than those observed in nervous sheep [32]. Therefore,we hypothesized that exposure to lavender oil would be unlikely tofurther reduce the behavioral and endocrine correlates of anxiety incalm sheep during exposure to an isolation stressor.

2. Materials and methods

2.1. Animals

These experiments were carried out in accordance with theAustralian Code of Practice for the Care and Use of Animals for ScientificPurposes (7th Edition, 2004) and were approved by the Animal EthicsCommittee of The University of Western Australia (RA3/100/947).

2.1.1. The UWA temperament flockThe UWA temperament flock areMerino sheep kept at the Universi-

ty of Western Australia (UWA) research farm in Wundowie (latitude31°46′S, longitude 116°29′E; 330 m in altitude) that have been diver-gently selected for ‘calm’ or ‘nervous’ temperament for 17 generations[29,31,34]. In this context, we define ‘temperament’ as the behavioralreactivity of sheep to the psychosocial stress of social isolation in thepresence or absence of humans [31,34]. For the purpose of geneticselection, behavioral reactivity to stress is measured at 14–16 weeksof age using a series of tests that assess the animal's relative ability tocope with stress. The first test is a 3 min arena test that quantifies theexpression of locomotor activity and vocalizations when faced withthe conflict of approaching a human to gain contact with a smallgroup of con-specifics [31,34]. In the second test, the animal is placedin visual isolation in a solid plywood box (1.5 m3) for 1 min. Physicalactivity inside the box (agitation score) is recorded using a digital agita-tion meter that is calibrated prior to the test for low, medium and highlevels of activity [31,34]. The scores from the two tests are then used in aselection index to formulate a score for the temperament of the sheepas described in detail by Beausoleil et al. [35]. Males with the mostextreme scores within the calm and nervous lines are used for breedingthe next generation of sheep when they reach sexual maturity. Calmand nervous sheep are maintained as one flock at pasture and subjectto conventional Australian farm practice.

We have screened over 10,000 sheep across Australia using the iso-lation box test (test 2) and found that the progeny bred fromunselectedor ‘wild type’ sheep that were ‘more or less reactive’ to the isolationstressor, were also ‘more or less’ reactive to the isolation stressor, asindicated by differences in their agitation score. These data indicatethat the genes underlying calm and nervous temperament in thismodel are likely to be common in the general population, thus support-ing the heritability, relevance and validity of our animal model [28].

2.1.2. Experimental animalsDuring August 2010 (non-breeding season; southern hemi-

sphere), females from the calm (n=16) or nervous (n=16) lines ofthe UWA temperament flock were allocated to one of four treat-ments; calm and nervous sheep exposed to 10% lavender oil (CalmLav; n=8; Nerv Lav; n=8) or exposed to peanut oil as a control(Calm Con; n=8; Nerv Con; n=8). The groups were balanced forage (14–16 months) and live weight within temperament (Table 1).The sheep were initially selected based on their parental tempera-ment (i.e. calm or nervous) and their agitation score in the standardisolation box at weaning (3–4 months). Before they were finally allo-cated to their treatment groups, the sheep were screened during a1 min test in a smaller (0.5 m×1.2 m×1.2 m), portable version ofthe isolation box (screening agitation score) to ensure that thescreening agitation scores were balanced within temperament acrossthe lavender oil and control groups (Table 1).

2.2. Experimental procedures

The isolation box was set up in an enclosed shed under naturallight with access to yards for holding the sheep before and after theisolation stressor. The isolation box was stood on a base of four tiresof equal size and pressure to provide suspension and stability formovement within the box. On the day of the test, a mechanical device

Table 1Mean±SEM live weight and screening agitation score in calm and nervous sheep ex-posed to wool scented with lavender oil (Calm Lav, Nerv Lav) or peanut oil (CalmCon, Nerv Con). Different superscripts (a,b) indicate significant differences betweentreatment groups (Pb0.05).

Live weight (kg) Screening agitation score

Calm Con 29.4±2.0 38.4±5.2a

Calm Lav 31.7±2.5 44.3±8.1a

Nerv Con 35.4±1.1 187.1±14.8b

Nerv Lav 35.4±1.0 193.9±13.4b

Fig. 1. Top panel: Example of a mask fitted to a sheep (1a) and unzipped (1b), showingthe placement of the mesh bag containing the cotton wool infused with either lavenderoil or peanut oil. Bottom panel: Example of a sheep in the isolation box (1c) and a sche-matic diagram of the floor of the isolation box — the bold lines represent those used toquantify the crosses of the central line of the box (1d).

1119P.A.R. Hawken et al. / Physiology & Behavior 105 (2012) 1117–1123

fitted with four piston driven legs was placed inside the box thatmimicked a low, medium and high level of activity inside the box.The sensitivity of the agitation meter was then adjusted so that low,medium and high levels of activity during 1 min resulted in scoresof approximately 60, 90 and 120 arbitrary units respectively (agita-tion score). The calibration was repeated at the end of the day to con-firm the consistency of the agitation meter over the experiment. Calmand nervous sheep were assigned to one of two treatment groups(control and lavender oil) that were held in pens at least 50 m apartand managed by two teams of people to minimize the risk of crosscontamination of odor between the lavender oil and control groups.The same, standard size isolation box was used for both groups tominimize variation in the isolation stimulus and because the risk ofodor transmission from inside the mask to the box was minimal. Ap-proximately 30 min prior to the isolation stressor, each test animalwas moved from the holding pen into the test pen (approximately2 m×2 m) with companion sheep (n=4/5). The test sheep wasthen fitted with a mask containing wool scented with either lavenderoil or peanut oil that remained on the sheep for the duration of theexperiment. The timing of initial mask application was coordinatedto ensure that control and lavender exposed sheep from each temper-ament were subjected alternatively to the isolation stressor, approxi-mately 30 min after initial exposure to the mask. During the isolationstressor, the animal was isolated for 5 min the standard isolation boxthat prevented visual communication with other sheep, although au-ditory communication was possible. During the test, the behavior ofthe animal was recorded using a video camera fixed to the top ofthe isolation box. After 5 min, the test agitation score was recordedand the animal was released and returned to the companion sheep.The mask remained on the sheep for another 30 min after exitingthe isolation box.

2.2.1. Exposure to lavender/peanut oilSheep exposed to lavender oil were fitted with a mask containing

cotton wool infused with 1 mL of 100% lavender oil (L. augustifolia; OilGarden Aromatherapy, Victoria, Australia) diluted in peanut oil at aratio of 1:9. This concentration of lavender oil was selected basedon the work of Diego et al., [36] in humans. Peanut oil was used asthe diluent because it has a neutral aroma and was thus unlikely to af-fect any of the behavioral or endocrine correlates of anxiety. Controlsheep were fitted with an identical mask containing a hand sizedamount of cotton wool (Fig. 1) infused with 1 mL of peanut oil. Themasks were made of fabric with a mesh cover at the end to allow air-flow through the mask. A mesh bag containing the cotton wool in-fused with oil was attached to the mesh-covered end of the mask tooptimize inhalation of the aroma and keep the wool at a fixed dis-tance (1–2 in.) from the nose of the sheep (Fig. 1). Neither themask, wool or peanut oil were expected to affect the behavioral or en-docrine responses of the sheep to the isolation stressor, so an emptymask or no mask control were not included in this study.

2.2.2. Quantification of data from the video footageThe video footage was watched twice by an observer who was un-

aware of the temperament or treatment (i.e. control or lavender) ofthe test animal; firstly to quantify the frequency of all vocalizations

and secondly to quantify locomotor activity within the isolation box.The observer recorded the number of times that the front feet of theanimal crossed the central lines of the floor of the isolation boxshown in Fig. 1. The observer also recorded the frequency of urina-tions, pawing and whether the animal tried to escape from the box.An escape attempt was defined as any jumping movement towardsthe wall with a minimum of two feet off the ground.

2.2.3. Blood collectionTo profile the effects of treatment (i.e. lavender or control) and

temperament on the plasma concentration of cortisol, blood wassampled by jugular venepuncture using 1″ 18 G needles and 4 mLvacutainers coated with lithium heparin (Interpath, WA, Australia)at 4 time points; prior to fitting the animal with the mask; prior to en-tering the isolation box (30 min after fitting the mask): immediately(1 min) post isolation, and 30 min post isolation. All blood sampleswere taken in the test pen with the sheep held in a standing positionby an experienced handler with the head and neck slightly elevatedfor ease of access to the jugular vein. The necks of the sheep wereclipped prior to the experiment and the blood sample taken within1 min of initial restraint of the animal, thus minimizing the impactof the procedure on the plasma concentrations of cortisol in the sam-ple collected. Blood was stored immediately on ice and centrifugedwithin 30 min of collection for 10 min at 3000 rpm. Plasmawas then harvested and frozen in plastic tubes at −20 °C untilimmunoassay.

2.3. Immunoassay

Plasma concentrations of cortisol were quantified using commer-cial radioimmunoassay kits (Diasorin Australia Ltd. NSW) modifiedand validated for sheep as described by Beausoleil et al. [35]. The sen-sitivity of the assay was 1.1 ng/mL. All samples were processed withinone assay. The mean intra assay variation for quality control samplesof plasma cortisol (low:14.0 ng/mL; medium: 29.1 ng/mL) were 9%and 7.9% respectively.

Table 2Mean±SEM test agitation score, vocalization frequency and crosses of the central linesin calm and nervous sheep exposed to wool scented with lavender oil (Calm Lav, NervLav) or peanut oil (Calm Con, Nerv Con). Different superscripts (a,b,c) indicate signifi-cant differences between treatment groups (at least Pb0.05).

Calm Con(n=8)

Calm Lav(n=8)

Nerv Con(n=8)

Nerv Lav(n=8)

Test agitation score 67.8±20.03a 28.1±7.46b 299±43.84c 374±71.98c

Crosses of centralline of isolation box

16.6±4.23a 3.42±0.81b 53.8±4.38c 57.2±6.18c

Vocalizations(bleats per min)

1.95±0.86a 0.34±0.19b 5.60±1.01c 9.2±1.13d

Numbers of animals observed:Urinating(%)

0(0)

0(0)

1(13)

2(25)

Pawing(%)

1(13)

1(13)

2(25)

3(38)

Escape attempts(%)

1ab

(13)0a

(0)0a

(0)4b

(50)

1120 P.A.R. Hawken et al. / Physiology & Behavior 105 (2012) 1117–1123

2.4. Statistical analysis

All the tests used in the data analysis were run in Genstat 5 (Sec-ond Edition, Lawes Agricultural Trust, Rothamsted Experimental Sta-tion, Hertfordshire UK) or Minitab ® 14.1 (Minitab Inc, State College,Pennsylvania, USA). The distributions of all data were initiallyassessed for normality using the Shapiro Wilk test. Where the datawere not from a normal distribution, they were transformed (seebelow) and the tests were repeated to ensure the validity of thedata for parametric analysis. Agitation score, crosses of the centralline and vocalizations were subject to rank transformation prior toanalysis with two-way ANOVA to determine the effect of treatment(i.e. +/− lavender) and temperament and any interaction betweenthe two factors. Live weight directly affects the agitation score sowas included as a covariate in analysis of this parameter. Where a sig-nificant effect of treatment, temperament or interaction wasdetected, data were compared between each combination of treat-ment groups by Students t-test (e.g. Calm Lav versus Calm Con;Calm Lav versus Nerv Con; Calm Lav versus Nerv Lav). Non-transformed data are presented in the tables for ease ofinterpretation.

The numbers of sheep attempting to escape from the box werecompared between treatment groups using a chi-squared test. Thelimited numbers of sheep observed pawing or urinating preventedstatistical analysis of these parameters relative to temperament ortreatment (i.e. +/− lavender).

Cortisol data were subject to square root transformation prior toanalysis with two way repeated measures ANOVA with time, temper-ament and treatment (i.e. +/− lavender) as factors and live weight asa covariate. Where a significant effect of time (pre-mask, pre-isolation, 1 min and 30 min post isolation), treatment (i.e. +/− lav-ender), temperament (calm, nervous) or interaction was detected,data were compared between time points within each treatmentgroup by paired t-test (e.g. pre-mask sample versus pre-isolationsample within Calm Lav sheep) or between each combination oftreatment groups by Students t-test (e.g. Calm Lav versus Calm Conat each time point). Non-transformed data are presented in the tablesfor ease of interpretation.

3. Results

3.1. Behavior

3.1.1. Test agitation scoreTwo-way ANOVA revealed an effect of temperament (F (1, 31)=

1017.95; Pb0.001), no effect of treatment (i.e. +/− lavender oil;F (1, 31)=51.61; P>0.1) and an interaction between on tempera-ment and treatment (F (1, 31)=114.43; Pb0.05) on the test agitationscore. Further analysis revealed that both nervous control sheep andnervous sheep exposed to lavender oil had a higher mean agitationscore than calm control sheep (Nerv Con: t14=5.18; Pb0.001; NervLav: t14=4.45; Pb0.01; Table 2) and calm sheep exposed to lavenderoil (Nerv Con: t14=8.46; Pb0.001; Nerv Lav: t14=7.21; Pb0.001;Table 2). Calm sheep exposed to lavender oil had a lower agitationscore than control sheep of the same temperament (t14=4.81;Pb0.001; Table 2).

3.1.2. Crosses over the central lines of the isolation boxTwo-way ANOVA detected an effect of temperament (F (1, 31)=

10,507.6; Pb0.001) and an interaction between temperament andtreatment (i.e. +/− lavender oil; F (1, 31)=687.7; Pb0.05) but noeffect of treatment (i.e. +/− lavender oil; F (1, 31)=214.4; P>0.1)on the number of crosses of the central lines of the isolation box. Subse-quent analysis revealed that both nervous control sheep and nervoussheep exposed to lavender oil crossed the central lines of the boxmore frequently than calm control sheep (Nerv Con: t14=4.93;

Pb0.001; Nerv Lav: t14=4.60; Pb0.001; Table 2) and calm sheep ex-posed to lavender oil (Nerv Con: t14=6.58; Pb0.001; Nerv Lav:t14=5.86; Pb0.001; Table 2). Calm sheep exposed to lavender oilcrossed the central lines of the box less frequently than calm controlsheep (t14=−2.26; Pb0.05; Table 2). There were no differences inthis variable between nervous sheep exposed to lavender oil and con-trol sheep of the same temperament (t14=−0.68; P>0.1; Table 2).

3.1.3. VocalizationsTwo-way ANOVA detected an effect of temperament (F (1, 31)=

3349.4; Pb0.001) and an interaction between temperament andtreatment (i.e. +/− lavender oil; F (1, 31)=825.4; Pb0.001) but noeffect of treatment (i.e. +/− lavender oil; F (1, 31)=93.4; P>0.1)on the frequency of vocalizations during the isolation stressor. Subse-quent analysis revealed that nervous control sheep and nervoussheep exposed to lavender oil vocalized more frequently than calmcontrol sheep (Nerv Con: t14=2.49; Pb0.05; Nerv Lav: t14=5.08;Pb0.001; Table 2) and calm sheep exposed to lavender oil (NervCon: t14=5.46; Pb0.001; Nerv Lav: t14=8.93; Pb0.001; Table 2).Nervous sheep exposed to lavender oil vocalized more frequentlythan nervous control sheep (t14=2.46; Pb0.05; Table 2). Conversely,calm sheep exposed to lavender oil vocalized less frequently thancalm control sheep (t14=−2.26; Pb0.05; Table 2).

3.1.4. Other behaviorsThe number of sheep attempting to escape from the box was

greater among nervous sheep exposed to lavender oil than nervouscontrol sheep or calm sheep exposed to lavender oil (Chi squarevalue=10.19; Pb0.05; Table 2). The limited numbers of sheep ob-served pawing or urinating prevented any meaningful analysis ofthese parameters relative to temperament or treatment (Table 2).

3.2. Cortisol secretion

Two way, repeated measures ANOVA detected significant effects oftime (F (3, 81)=50.76; Pb0.001) and temperament (F (1, 27)=46.76; Pb0.001) but no effect of treatment (i.e. +/− lavender oil;F (1, 27)=0.35; P>0.1) on the mean plasma concentrations of cor-tisol. There were also interactions between temperament and time(F (3, 81)=2.93; Pb0.01), treatment (i.e. +/− lavender oil) andtime (F (3, 81)=2.68; Pb0.05) and treatment (i.e. +/− lavenderoil), temperament and time (F (3, 81)=7.53; Pb0.001) on themean plasma concentrations of cortisol over the experimentalperiod.

1121P.A.R. Hawken et al. / Physiology & Behavior 105 (2012) 1117–1123

3.2.1. Differences in cortisol secretion between temperaments andtreatments

3.2.1.1. Pre-mask sample.Mean plasma concentrations of cortisol priorto the mask being fitted did not differ among any of the treatmentgroups (P>0.05; Table 3). Nervous control sheep tended to havehigher mean plasma concentrations of cortisol than calm controlsheep but this difference failed to reach significance (t14=−1.86;Pb0.1; Table 3).

3.2.1.2. Pre-isolation sample. Mean plasma concentrations of cortisolwere lower in calm sheep exposed to lavender oil than calm controlsheep (t14=−2.73; Pb0.05; Table 3), nervous control sheep(t14=−5.79; Pb0.05; Table 3) and nervous sheep exposed to lavenderoil (t14=−3.93; Pb0.01; Table 3).Mean plasma concentrations of cor-tisol did not differ between calm control sheep and nervous controlsheep (t14=0.155; P>0.1; Table 3) or nervous sheep exposed tolavender oil (t14=−0.88; P>0.1; Table 3).

3.2.1.3. Post-isolation (1 min) sample. Both calm sheep exposed to lav-ender oil and calm control sheep had lower mean plasma concentra-tions of cortisol than nervous control sheep (Calm Con: t14=−10.61;Pb0.001; Calm Lav: t14=−5.55; Pb0.001; Table 3) and nervoussheep exposed to lavender oil (Calm Con: t14=−5.99; Pb0.001;Calm Lav: t14=−5.03; Pb0.001; Table 3). The plasma concentrationsof cortisol did not differ between calm control sheep and calm sheepexposed to lavender oil (t14=0.90; P>0.1; Table 3) or between ner-vous control sheep and nervous sheep exposed to lavender oil (t14=−1.08; P>0.1; Table 3).

3.2.1.4. Post-isolation (30 min) sample.Mean plasma concentrations ofcortisol were greater in nervous sheep exposed to lavender oil thannervous control sheep (t14=3.53; Pb0.01; Table 3), calm controlsheep (t14=4.82; Pb0.001; Table 3) and calm sheep exposed to lav-ender oil (t14=5.67; Pb0.001; Table 3). Mean plasma concentrationsof cortisol 30 min post isolation in calm sheep exposed to lavender oildid not differ from nervous control sheep (t14=1.35; P>0.1;Table 3) and were only numerically lower than calm control sheep(t14=−2.10; Pb0.1; Table 3).

3.2.2. Differences in cortisol secretion over time

3.2.2.1. Calm control. Mean plasma concentrations of cortisol in-creased between the pre-isolation sample and the pre-mask sample(t7=4.51; Pb0.01; Table 3) and between the pre-mask sample andthe post isolation sample taken 1 min after the isolation stressor(t7=2.96; Pb0.05; Table 3). Mean plasma concentrations of cortisol30 min post isolation were lower than those reported at 1 min postisolation (t7=−11.59; Pb0.001; Table 3) and did not differ signifi-cantly from the pre-mask sample (t7=−1.53; P>0.1; Table 3).

Table 3Mean±SEM plasma concentrations of cortisol (ng/mL) in calm and nervous sheep ex-posed to wool scented with lavender oil (Calm Lav, Nerv Lav) or peanut oil (Calm Con,Nerv Con). Different superscripts (a,b,c,d) indicate differences within rows (i.e. withintreatment group over time as indicated by paired t-test; at least Pb0.05). Different su-perscripts x,y indicate differences within columns (i.e. between treatment groups at aspecific time point as indicated by Student's t-test; at least Pb0.05).

Pre-masksample

Pre-isolationsample

Post-isolation(1 min) sample

Post isolation(30 min) sample

Calm Con 11.38±2.31a,x 24.36±3.41b,x 34.59±1.01c,x 13.48±1.55a,x

Calm Lav 18.10±2.21a,x 14.55±0.90a,c,y 32.04±2.79b,x 8.41±1.94c,x

Nerv Con 19.61±3.72a,x 31.10±2.87a,x 53.29±1.45b,y 14.76±2.99a,x

Nerv Lav 15.73±3.11a,x 28.88±3.97b,x 58.52±4.23c,y 37.11±5.82d,y

3.2.2.2. Calm lavender. There was no change in the mean plasma con-centrations of cortisol between the pre-isolation and pre-mask samplein calm sheep exposed to lavender oil (t7=1.61; P>0.1; Table 3). Ex-posure to the isolation stressor increased the mean plasma concentra-tions of cortisol in the post isolation sample (1 min) above thosereported in the pre-mask sample (t7=6.23; Pb0.001; Table 3). Meanplasma concentrations of cortisol 30 min post isolation were lowerthan those reported at 1 min post isolation (t7=−5.75; Pb0.001;Table 3) and in the pre-mask sample (t7=−3.27; Pb0.05; Table 3).

3.2.2.3. Nervous control. Mean plasma concentrations of cortisol didnot increase significantly between the pre-isolation sample and thepre-mask sample (t7=2.22; Pb0.1; Table 3). However, the meanplasma concentrations of cortisol did increase between the pre-mask sample and the post isolation (1 min) (t7=6.65; Pb0.001;Table 3). Mean plasma concentrations of cortisol 30 min post isolationwere lower than those reported at 1 min post isolation (t7=−8.30;Pb0.001; Table 3) and did not differ from the pre-mask sample(t7=0.90; P>0.1; Table 3).

3.2.2.4. Nervous lavender. Mean plasma concentrations of cortisolincreased between the pre-isolation sample and the pre-mask sample(t7=2.35; Pb0.05; Table 3) and between the pre-mask sample andthe post isolation (1 min) sample (t7=6.98; Pb0.05; Table 3).Mean plasma concentrations of cortisol 30 min post isolation werelower than those reported at 1 min post isolation (t7=−5.87;Pb0.001; Table 3) but were higher than those reported during thepre-mask sample (t7=4.02; Pb0.01; Table 3).

4. Discussion

Lavender oil (L. augustifolia) is reportedly anxiolytic in a number ofspecies [7,16,37,38] but this is the first study to investigate the effectsof lavender oil on animals that differ genetically in their behavioralreactivity to a specific stressor. We originally hypothesized that expo-sure to lavender oil would reduce the behavioral and endocrine corre-lates of anxiety in nervous sheep exposed to an isolation stressor.However, exposure of nervous sheep to lavender oil appeared toincrease rather than decrease the anxiety associated with the stressor,reflected by a higher frequency of vocalizations, more animals attempt-ing to escape and higher plasma concentrations of cortisol 30 min afterisolation compared to control sheep of the same temperament. In con-trast, exposure of genetically ‘calm’ sheep to lavender oil appeared toreduce the anxiety associated with the stressor, reflected by reducedlocomotor activity and frequency of vocalizations compared to controlsheep of the same temperament. Polymorphisms in genes involved inthe processing of anti-depressant and anti-psychotic drugs directly af-fect the proportion of individuals showing positive or adverse reactionsto these drugs [5]. It is therefore possible that the divergent effects oflavender oil on the level of anxiety associatedwith the isolation stressorare an artifact of co-selection for polymorphisms in the enzymes and/orreceptors responsible for mediating the effects of lavender oil on anxi-ety in sheep. Further studies are required to accept or reject this hypoth-esis, including direct comparisons with sheep of the same breed notselected for calm or nervous temperament.

Exposure of calm sheep to lavender oil appeared to reduce theplasma cortisol response of calm sheep to mask application and wasassociated with significantly lower plasma concentrations of cortisol30 min after isolation than pre-treatment values. These observationsindicate that lavender oil had an anxiolytic effect on calm sheep, atleast in terms of cortisol secretion. However, the absence of an effectof lavender oil on the plasma concentrations of cortisol immediatelyafter the isolation stressor indicates that, at least in calm sheep,there may be a threshold concentration of cortisol above which theanxiolytic effects of lavender oil are ineffective. This concept is sup-ported by Bradley et al. [39] who found that an oral dose of lavender

1122 P.A.R. Hawken et al. / Physiology & Behavior 105 (2012) 1117–1123

oil reduced anxiety in humans exposed to neutral film clips but failedto reduce anxiety when the same individuals were exposed to filmclips expected to induce anxiety in humans. However, lavender oilreduced locomotor activity and vocalization frequency in calmsheep during exposure to the isolation stressor, indicating a discon-nection between the behavioral and endocrine correlates of anxiety.A similar disconnection between the behavioral and endocrineresponse of animals to a stressor was reported in pigs, where expo-sure to lavender straw reduced the behavioral expression of travelsickness but not the effects of transport on cortisol secretion [40].We therefore propose that 10% lavender oil was sufficient to reducethe behavioral correlates of anxiety in calm sheep exposed to anisolation stressor but that its impact on cortisol secretion was depen-dent on the plasma concentrations of cortisol.

The ‘nervous’ sheep were more active than ‘calm’ sheep, both interms of their behavior and HPA activity when faced with an isolationstressor, indicating that they have been selected for a pro-active strate-gy to cope with social stressors [41]. This type of coping strategy inrodents is associated with higher levels of aggression and activation ofthe HPA axis when faced with a social stressor than those with a morereactive coping strategy [42]. Reports of positive, neutral or negativeeffects of drugs are common in the literature, especially with respectto the expression of aggressive behavior [43,44]. For example, benzodi-azepine treatment of an aggressive species of mice (Omychomys leuco-gaster) increased the expression of aggressive behavior in response toelectric shock but had no effect on less aggressive, albino white mice[43].We propose that the divergent effects of lavender oil on the behav-ioral and endocrine correlates of anxiety in calm and nervous sheepmay be at least in part due to differences in the strategies that theyuse to cope with stress.

An alternative explanation for the greater expression of the behav-ioral and endocrine correlates of anxiety in nervous sheep exposed tolavender oil is the novelty of the aroma. Sheep are fearful of novel envi-ronments and objects, and these stimuli are frequently used to induce afear response in sheep [26,45]. Little work to date has studied theimpact of novel odors on fear and anxiety in sheep, but their responsesto unfamiliar feeds may be mediated, at least in part, by the novelty ofthe odor [46]. However, the concentrations of plasma cortisol duringthe pre-isolation sample (i.e. after 30 min of exposure to lavender oil)did not differ between nervous control sheep or nervous sheep exposedto lavender oil, indicating that the novelty of the aroma was unlikely tobe responsible for the elevated secretion of cortisol in nervous sheep.

Parallels are frequently drawn between lavender oil and benzodiaz-epines because they have similar effects on anxiety inMongolian gerbilsand humans [9,13] that are mediated, at least in part, via the gamma-aminobutyric acid complex [12]. There is currently no information onthe effects of benzodiazepines on anxiety in sheep selected for calm ornervous temperament. We propose that further studies are warrantedto directly compare the effects of anxiolytic drugs, such as benzodiaze-pines, on the behavioral and endocrine correlates of anxiety in sheep.

The elevation in plasma concentrations of cortisol 30 min after ini-tial exposure to themask in all treatment groups, except calm sheep ex-posed to lavender oil, indicates a degree of stress associated with themask and/or experimental procedure. We did not include a non-maskor empty mask control because masks have been successfully used insheep with no apparent effect on hormone secretion [47,48]. However,changes in plasma concentrations of LH were the end point of thesestudies [47,48] rather than plasma concentrations of cortisol, as wasthe case in this study. In spite of this issue, calm and nervous sheepclearly differed in their cortisol response to the isolation stressor butwe acknowledge that the mask itself may have influenced the magni-tude of the changes in cortisol secretion.

Lavender oil appears to have bi-directional effects on anxiety in calmand nervous sheep exposed to an isolation stressor. This outcome raisesthe question of whether this pattern is likely to be mirrored during ex-posure to other stressors? The two lines of sheep have been selected for

their behavioral reactivity to social isolation and human presence butthere are limited studies into their reactivity to other stressors andthese have yielded variable results. For example, Beausoleil et al. [35]found that nervous sheep were more active when faced with theconflict of approaching a human to access a pen of companion sheepbut that the cortisol responses of the two lines to this stressor did notdiffer. Bickell [32] found that calm and nervous sheep initially had asimilar increase in plasma concentrations of cortisol when faced withisolation combinedwith a novel stressor (white,flapping plastic). How-ever, they found that the magnitude of this cortisol response decreasedduring subsequent exposures to the same stressor in calm sheep but notin nervous sheep, even when the novel stressor was removed [32]. Thelatter study demonstrated an important divergence between calm andnervous sheep in their propensity to interpret an ambiguous situationas negative or threatening, similar to that purported for rodent modelsof anxiety [25]. Social isolation clearly appears to be the dominantstressor in the index used to select calm and nervous sheep but furtherstudies are required to determine whether the behavioral and endo-crine correlates of anxiety differ between the two lines during initialand repeated exposure to different stressors. Only once this baselineinformation has been established can we conclusively determinewhether lavender oil has a bi-directional effect on anxiety in calm andnervous sheep.

5. Conclusions

Exposure to lavender oil had divergent effects on the behavioral andendocrine correlates of anxiety in calm and nervous sheep. Specifically,lavender oil reduced locomotor activity and vocalization frequency incalm sheep during exposure to an isolation stressor but only appearedto affect cortisol secretion when plasma concentrations were relativelylow. In nervous sheep, lavender oil increased locomotor activity, vocal-ization frequency and plasma concentrations of cortisol 30 min after theisolation stressor relative to control sheep. We conclude that geneticdifferences in temperament determine whether lavender oil alleviatesor exacerbates anxiety in sheep. It may therefore be important to con-sider the emotional state of an individual prior to using lavender oil totreat symptoms of anxiety.

Acknowledgments

This research was supported by operating funds from the UWALefroy fellowship. We thank S Gray for his assistance in the care andmanagement of the animals, M Blackberry for her assistance withthe cortisol assays, and S Plug and M Shikh Maidin for their assistancein data collection.

References

[1] Landgraf R. HAB/LAB rats: an animal model of extremes in trait anxiety and de-pression. Clin Neurosci Res 2003;3:239–44.

[2] Kessler RC, Chiu WT, Demler O, Walters EE. Prevalence, severity, and comorbidityof twelve-month DSM-IV disorders in the National Comorbidity Survey Replica-tion (NCS-R). Arch Gen Psychiatry 2005;62:617–27.

[3] Roy-Byrne PP. The GABA-benzodiazepine receptor complex: structure, function,and role in anxiety. J Clin Psychiatry 2005;66(Suppl 2):14–20.

[4] Lader MH. Limitations on the use of benzodiazepines in anxiety and insomnia: arethey justified? Eur Neuropsychopharmacol 1999;9(Suppl 6):S399–405.

[5] Kirchheiner J, Nickchen K, Bauer M, Wong M-L, Licino J, Roots I, et al. Pharmaco-genetics of antidepressants and antipsychotics: the contribution of allelic varia-tions to the phenotype of drug response. Mol Psychiatry 2004:442–73.

[6] Cline M, Taylor JE, Flores J, Bracken S, McCall S, Ceremuga TE. Investigation of theanxiolytic effects of linalool, a lavender extract, in the male Sprague–Dawley rat.AANA J 2008;76:47–52.

[7] Shaw D, Annett JM, Doherty B, Leslle JC. Anxiolytic effects of lavender oil inhalationon open-field behaviour in rats. Phytomedicine 2007;14:613–20.

[8] Kim Y, Kim M, Kim H, Kim K. Effect of lavender oil on motor function and dopaminereceptor expression in the olfactory bulb of mice. J Ethnopharmacol 2009;125:31–5.

[9] Bradley BF, Starkey NJ, Brown SL, Lea RW. Anxiolytic effects of Lavendula augusti-folia odour on the Mongolian gerbil elevated plus maze. J Ethnopharmacol2007;111:517–25.

1123P.A.R. Hawken et al. / Physiology & Behavior 105 (2012) 1117–1123

[10] Graham L, Wells DL, Hepper PG. The influence of olfactory stimulation on the be-haviour of dogs housed in a rescue shelter. Appl Anim Behav Sci 2005;91:143–53.

[11] Field T, Diego M, Hernandez-Reif M, Cisneros W, Feijo L, Vera Y, et al. Lavenderfragrance cleansing gel effects on relaxation. Int J Neurosci 2005;115:207–22.

[12] Aoshima H, Hamamoto K. Potentiation of GABAA receptors expressed in Xenopusooctyes by perfume and phytoncid. Biosci Biotechnol Biochem 1999;63:743–8.

[13] Woelk H, Schlafke S. A multi-center, double blind randomised study of the Laven-der oil preparation Silexan in comparison to Lorazepam for generalized anxietydisorder. Phytomedicine 2010;17:94–9.

[14] Kasper S, Gastpar M, Muller WE, Volz HP, Moller HJ, Dienel A, et al. Efficacy andsafety of silexan, a new, orally administered lavender oil preparation, in sub-threshold anxiety disorder— evidence from clinical trials. Wien MedWochenschr2010;160:547–56.

[15] Field T, Field T, Cullen C, Largie S, Diego M, Schanberg S, et al. Lavender bath oilreduces stress and crying and enhances sleep in very young infants. Early HumDev 2008;84:399–401.

[16] Lehrner J, Marwinski G, Lehr S, Johren P, Deecke L. Ambient odors of orange andlavender reduce anxiety and improve mood in a dental office. Physiol Behav2005;86:92–5.

[17] McCaffrey R, Thomas DJ, Kinzelman AO. The effects of lavender and rosemary es-sential oils on test-taking anxiety among graduate nursing students. Holist NursPract 2009;23:88–93.

[18] van der Ploeg ES, Eppingstall B, O'Connor DW. The study protocol of a blindedrandomised-controlled cross-over trial of lavender oil as a treatment of beha-vioural symptoms in dementia. BMC Geriatr 2010;10:49.

[19] Saeki Y. The effect of foot-bath with or without the essential oil of lavender on theautonomic nervous system: a randomized trial. Complement Ther Med 2000;8:2–7.

[20] Morris N. The effects of lavender (Lavendula angustifolium) baths on psychologicalwell-being: two exploratory randomised control trials. Complement Ther Med2002;10:223–8.

[21] Howard S, Hughes BM. Expectancies, not aroma, explain impact of lavender aro-matherapy on psychophysiological indices of relaxation in young healthywomen. Br J Health Psychol 2008;13:603–17.

[22] Landgraf R, Wigger A. Born to be anxious: neuroendocrine and genetic correlatesof trait anxiety in HAB rats. Stress 2003;6:111–9.

[23] Landgraf R, Wigger A. High vs low anxiety-related behavior rats: an animal modelof extremes in trait anxiety. Behav Genet 2002;32:301–14.

[24] Landgraf R, Wigger A, Holsboer F, Neumann ID. Hyper-reactive hypothalamo–pituitary–adrenocortical axis in rats bred for high anxiety-related behaviour.J Neuroendocrinol 1999;11:405–7.

[25] Landgraf R. Animal models of anxiety. Stress 2003;6:73–5.[26] Boissy A. Fear and fearfulness in animals. Q Rev Biol 1995;70:165–91.[27] Romeyer A, Bouissou MF. Assessment of fear reactions in domestic sheep, and influ-

ence of breed and rearing conditions. Appl Anim Behav Sci 1992;43:93–119.[28] Blache, D., Ferguson, D. M. Increasing sheep meat production efficiency and ani-

mal welfare by selection for temperament In: SHGEN.025 Fr, editor. Meat andLivestock Australia. Sydney 1995.

[29] Bickell SL, Poindron P, Nowak R, Chadwick A, Ferguson D, Blache D. Genotyperather than non-genetic behavioural transmission determines the temperamentof Merino lambs. Anim Welf 2009;18:459–66.

[30] Lang PJ, Davis M, Ohman A. Fear and anxiety: animal models and human cognitivepsychophysiology. J Affect Disord 2000;61:137–59.

[31] Murphy PM, Purvis IW, Lindsay DR, Le Neindre P, Orgeur P, Poindron P. Measuresof temperament are highly repeatable in Merino sheep and some are related tomaternal behaviour. Proc Aust Soc Anim Prod 1994;20:247–50.

[32] Bickell, S. L. Calm Sheep will be Less Stressed When Faced with a Novel, UnknownObject. Honours Thesis, University of Western Australia, Perth, Australia; 2005.

[33] Serretti A, Benedetti F, Zanardi R, Smeraldi E. The influence of Serotonin Trans-porter Promoter Polymorphism (SERTPR) and other polymorphisms of the sero-tonin pathway on the efficacy of antidepressant treatments. ProgNeuropsychopharmacol Biol Psychiatry 2005;29:1074–84.

[34] Murphy, P. M. Maternal Behaviour and Rearing Ability of Merino Ewes can be Im-proved by Strategic Feed Supplementation During Late Pregnancy and Selectionfor Calm Temperament. PhD Thesis, University of Western Australia, Perth, Aus-tralia; 1999.

[35] Beausoleil NJ, Blache D, Stafford KJ, Mellor DJ, Noble ADL. Exploring the basis ofdivergent selection for ‘temperament’ in domestic sheep. Appl Anim Behav Sci2008;109:261–74.

[36] Diego M, Jones NA, Field T, Hernandez-Reif M, Schanberg S, Kuhn C, et al. Aroma-therapy positively affects mood, EEG patterns of alertness and math computa-tions. Int J Neurosci 1998;96:217–24.

[37] Toda M, Morimoto K. Effect of lavender aroma on salivary endocrinological stressmarkers. Arch Oral Biol 2008;53:964–8.

[38] Lee IS, Lee GJ. Effects of lavender aromatherapy on insomnia and depression inwomen college students. Taehan Kanho Hakhoe Chi 2006;36:136–43.

[39] Bradley BF, Brown SL, Chu S, Lea RW. Effects of orally administered lavender es-sential oil on responses to anxiety-provoking film clips. Hum Psychopharmacol2009;24:319–30.

[40] Bradshaw RH, Marchant JN, Meredith M, Broom DM. Effects of lavender straw onstress and travel sickness in pigs. J Altern Complement 1998;4:271–5.

[41] Koolhaas JM, de Boer SF, Coppens CM, Buwalda B. Neuroendocrinology of copingstyles: towards understanding the biology of individual variation. Front Neuroen-docrinol 2010;31:307–21.

[42] Veenema AH, Neumann ID. Neurobiological mechanisms of aggression and stresscoping: a comparative study in mouse and rat selection lines. Brain Behav Evol2007;70:274–85.

[43] DiMascio A. The effects of benzodiazepines on aggression: reduced or increased?Psychopharmacologia 1973;30:95–102.

[44] Matte AC. Effects of hashish on isolation induced aggression in wild mice. Psycho-pharmacologia 1975;45:125–8.

[45] Forkman B, Boissy A, Meunier-Salaun MC, Canali E, Jones RB. A critical review offear tests used on cattle, pigs, sheep, poultry and horses. Physiol Behav 2007;92:340–74.

[46] Burritt EA, Provenza FD. Effect of an unfamiliar location on the consumption ofnovel and familiar foods by sheep. Appl Anim Behav Sci 1997;54:317–25.

[47] Cohen-Tannoudji J, Locatelli A, Signoret JP. Non-pheromonal stimulation by themale of LH release in the anoestrous ewe. Physiol Behav 1986;36:921–4.

[48] Cohen-Tannoudji J, Lavenet C, Locatelli A, Tillet Y, Signoret JP. Non-involvement ofthe accessory olfactory system in the LH response of anoestrous ewes to maleodour. J Reprod Fertil 1989;86:135–44.