DO

Oa

b

a

ARRA

KABCGHO

1

ccorc

f

0h

Veterinary Parasitology 193 (2013) 214– 222

Contents lists available at SciVerse ScienceDirect

Veterinary Parasitology

jou rn al h om epa ge: www.elsev ier .com/ locate /vetpar

o the changes in the behaviours of cattle during parasitism withstertagia ostertagi have a potential diagnostic value?

llie Szyszkaa,∗, Bert J. Tolkampb, Sandra A. Edwardsa, Ilias Kyriazakisa

School of Agriculture Food and Rural Development, Newcastle University, Newcastle upon Tyne, NE1 7RU, UKAnimal Health Group, Scottish Agricultural College, Edinburgh, EH9 3JG, UK

r t i c l e i n f o

rticle history:eceived 21 August 2012eceived in revised form 24 October 2012ccepted 26 October 2012

eywords:ctivityehaviourattleastrointestinal parasitismealth detectionstertagia ostertagi

a b s t r a c t

We investigated the magnitude of temporal changes in activity, posture and feedingbehaviour of cattle infected with Ostertagia ostertagi, and their reversal after treatmentwith an anthelmintic. Twenty-six, 3-month-old, Holstein-Friesian bulls were allocated toone of three treatment groups. Bulls in two of those (groups P and PA) received 100,000larvae on three occasions (Days 0, 7 and 14) and the remaining animals served as controls(C). The PA group also received an anthelmintic on Day 31. Parasite eggs appeared in thefaeces of P and PA bulls from Day 17; from approximately the same time blood pepsinogenlevels increased and body weight (BW) gain decreased (P < 0.001). The reduction in BW gainpersisted until Day 45 for P animals only. There was a decrease in the number of steps takenfor P and PA animals, as well as lying and standing episode frequency, by 41 and 44% respec-tively (P < 0.001) from Day 21 onwards. The average lying and standing episode durationincreased by 52 and 55% respectively (P < 0.001) from the same time in P and PA comparedto C bulls. In addition, meal frequency showed a tendency to decrease for P animals only(P = 0.039) from Day 39, and this was the only aspect of feeding behaviour affected by par-asitism. All behaviours, returned to control levels within a week of anthelmintic drenchingof PA bulls, apart from the number of steps taken. Although BW gain and pepsinogen also

started to recover after drenching, these had not returned to control levels by Day 45. Themagnitude of the changes in activity, and standing and lying episode frequency and durationsuggest that these might have a diagnostic value, especially as all can now be monitored byautomated means. However, these behaviours did not show the rapid changes we expectedbefore parasitism manifested clinically and following recovery.

. Introduction

Animal behaviour may be one of the first things thathange when an animal is affected by a health or welfarehallenge, and can precede any clinical signs of stress

r disease (Kyriazakis and Tolkamp, 2010). A number ofecent papers have demonstrated that several aspects ofattle behaviour may be modified by (mainly) bacterial

∗ Corresponding author. Tel.: +44 0191 2227402;ax: +44 0191 2226720.

E-mail address: ollie.szyszka2@newcastle.ac.uk (O. Szyszka).

304-4017/$ – see front matter © 2012 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.vetpar.2012.10.023

© 2012 Elsevier B.V. All rights reserved.

challenges (Quimby et al., 2001; Borderas et al., 2008;González et al., 2008; Fogsgaard et al., 2012) and that thesechanges may have diagnostic value. González et al. (2008),for example, have shown that measurable changes in thefeeding behaviour of dairy cattle occur in cases of lamenessand ketosis several days before anything can be detected byfarm personnel, which allows for early action to be taken.The rate of change in such behaviours is dependent on thenature of the health challenge (González et al., 2008).

The preliminary and short term study of Szyszka et al.(in press) is the only one that has investigated severalbehavioural changes that may occur in cattle parasitizedby helminths. It was found that several aspects of cattle

y Parasit

O. Szyszka et al. / Veterinar

posture and activity, such as lying and standing behaviour,and the number of steps taken, were affected to a variableextent by subclinical parasitism with Ostertagia ostertagi.In the present study we were interested in the temporalaspects and the magnitude of similar changes in behaviourthat might take place during such infections. In additionto measurements of activity and posture, we also mea-sured feeding behaviour because (i) a reduction in foodintake is a consistent feature of such infections (Fox et al.,1989; Kyriazakis et al., 1998) and (ii) changes in the graz-ing behaviour of dairy cattle as a consequence of parasitismhave been reported by Forbes et al. (2007). We wereinterested in whether such changes in behaviours arisebefore conventional signs of parasitism or clinical symp-toms occur, and also in how quickly these changes arereversed when animals are drenched with an anthelmintic.Some abomasal damage by the parasites can occur beforeany eggs appear in the faeces, concurring with the devel-opment of larvae (Murray et al., 1970); this could manifestas a change in animal behaviours, before conventionaldetection of parasitism. In addition, Kyriazakis et al. (1996)have shown that the food intake of sheep parasitized withTrichrostonglylus coubriformis recovered within 5 days postdrenching with an anthelmintic. Thus knowledge of suchtemporal changes may have diagnostic value for the assess-ment of the effectiveness of a treatment. However, suchbehavioural changes must be of sufficient magnitude ifthey are to have diagnostic validity (González et al., 2008).The hypotheses tested here, therefore, were: (1) changes ofsignificant magnitude in aspects of behaviour due to para-sitism will appear before any clinical signs, or the presenceof eggs in the faeces and (2) such changes will be reversedvery rapidly after dosing with an anthelmintic.

2. Materials and methods

The experiment took place at the facilities of Newcas-tle University after approval of the experimental protocolsby the Animal Experiments Committee and under licenseaccording to the UK Animals (Scientific Procedures) Act forexperimental challenge and regulated procedures.

2.1. Animals and housing

The animals used were 26 Holstein-Friesian bulls,approximately 3 months of age, with an average weightof 117 ± 25 kg, derived from a commercial farm. All ani-mals were housed together in a single straw-bedded penmeasuring 10 m × 4 m. Both food and water were availableon an ad libitum basis with the feeder providing 23 feedingspaces, in accordance to the space requirements set by theUK Animals (Scientific Procedures) Act (Wolfensohn andLloyd, 2003). The food offered was a total mixed ration,consistent throughout the experiment, containing 31.25%barley, 18.45% sugar beet pellets, 15.98% soya bean meal,13.42% barley cereal, 9.76% distillers maize, 7.41% molassesand 3.75% chopped barley straw. The chemical composi-

tion of the food was 13.01 MJ Metabolizable Energy and197 g Crude Protein per kg dry matter as estimated fromAFRC (1993) feed tables. The animals had not receivedany prior challenge with parasites and were treated with

ology 193 (2013) 214– 222 215

2 ml of the anti-inflammatory dexamethasone (Rapidexon,Eurovet, Cambridge, UK) and 8 ml of the antibiotic flor-fenicol (Nuflor, Shering-Plough, Milton Keynes, UK) priorto being parasitized, to reduce the risk of potentially con-founding bacterial infections; such a treatment was notexpected to affect the development of parasitism (Greeret al., 2005).

2.2. Experimental design

The animals were randomly assigned, taking intoaccount their initial body weight, to one of four treatmentgroups. The experiment was considered to start when theanimals received the first challenge (designated as Day 0).Prior to the application of the health challenges (Day −8)animals were fitted with a pedometer (IceRobotics, SouthQueensferry, UK) secured with Velcro on their front leftleg, and video recordings of feeding behaviour started inorder to provide background data before the parasite chal-lenge. Also on the same day (Day −8) the animals received7.5 mg/kg body weight of the anthelmintic albendazole(Albenil, Virbac, Woolpit, UK).

The first treatment group (P) consisted of seven animals,which received a trickle dose of 300,000 L3 O. ostertagi lar-vae in total, administered by gavage in doses of 100,000 L3on Days 0, 7 and 14 of the experiment. The dose and itsexpected consequences on feeding behaviour and activityhad been established in a previous experiment (Szyszkaet al., in press). Trickle dosages of O. ostertagi have beenpreviously reported in literature to mimic a repeated chal-lenge (Fox et al., 1989; Forbes et al., 2009), which is morelikely to be encountered in the field. The animals in thisgroup remained infected for the duration of the exper-iment (Day 45). The second treatment group (PA) alsoconsisted of seven animals and received the same trickleinfection as group P. However, on Day 31 of the experiment,these animals received 7.5 mg/kg of the broad-spectrumanthelmintic ablendazole (Albenil, Virbac, Woolpit, UK).

The third (C) and fourth (CA) treatment groups, whichacted as unchallenged controls, each consisted of six ani-mals. They were given by gavage 20 ml water on Days 0, 7and 14. Group CA was also treated with the anthelminticon Day 31 (Albenil, Virbac, Woolpit, UK) to coincide withthe drenching of the second treatment group, thereby con-trolling for any potential behavioural side effects of theanthelmintic dosing on the animals. The L3 larvae usedas the challenge were obtained from Ridgeway Research(Gloucestershire, UK) and were of an Ivermectin suscepti-ble strain that was isolated in South Gloucestershire, UK, 3months before use (reference label OOSG10). Upon arrivalthe 3.6 M larvae were split in six glass beakers and dilutedin 500 ml of water each, which was changed every otherday until use. Just prior to dosing, 410 ml of surplus waterwas removed from the top leaving six doses of 15 ml with100,000 L3 per dose. Each dose was topped up with 5 ml ofwater and administered to each animal by gavage.

The experiment lasted for 45 days, throughout which

faecal samples and body weight measurements were takentwice a week, and blood samples were taken once a week;the same measurements were taken on the last day of theexperiment. All animal handling took place in the morning,

2 y Parasit

btswpisetLlffmV

2

itaGa(smoeies7dwradmf

2

pdap4

gtctS

2

oe

16 O. Szyszka et al. / Veterinar

etween 9.30 and noon and their frequency ensured thathe animals were accustomed to the procedures. The faecalamples were taken from the rectum and their consistencyas recorded on a subjective three point scale prior to beinglaced in labelled pots. The counting of the number of eggs

n the faeces (FEC) took place on the same or the next day ofampling, in which case the samples were kept in the refrig-rator overnight. The 10 ml blood samples were taken fromhe jugular vein with a plain tube (Vacutainer, BD, Franklinakes, NJ, USA) for serum collection, and the samples wereeft overnight in the fridge, after which they were spunor 15 min at 1500 rpm at 18 ◦C; the serum samples wererozen for subsequent analysis. At the end of the experi-

ent all animals were treated with anthelmintic (Albenil,irbac, Woolpit, UK).

.3. Behavioural observations

The behaviours observed were posture, levels of activ-ty and feeding. The focus was on these behaviours becausehey can be automatically monitored and are potentiallyffected by health challenges in cattle (Quimby et al., 2001;onzález et al., 2008; Szyszka et al., in press). Posturend activity levels were measured with the use of IceTagIceRobotics, South Queensferry, UK) pedometers that tookecond-to-second readings throughout the experiment,easuring the number of steps taken and the posture

f the animal. These data were read at the end of thexperiment when the pedometers were removed. Feed-ng behaviour was monitored with 24-h video recordingquipment, which was watched using continuous focalampling. Observations of feeding behaviour during Days–9 (period 1, prepatent parasitism), 31–33 (period 2, pre-icted peak of parasitism) and 39–41 (period 3, post dosingith anthelmintic) were considered in the analysis of the

esults. Feeding episodes were considered to start whenn animal put its head into the feeder and end upon with-rawal. The animals were distinguished by their individualarkings. All behaviours were analysed for duration and

requency.

.4. Blood and faecal measurements

The collected serum samples were used to assessepsinogen levels. Plasma pepsinogen concentration wasetermined using the modified method of Paynter (1992)nd expressed in international units (iu), per litre. The sam-les analysed were those taken on Days −5, 21, 28, 35 and2.

The FEC were expressed as the number of eggs perram (epg) of collected fresh faeces. They were assessed byhe flotation method, as described in the Ministry of Agri-ulture “Manual of veterinary parasitological laboratoryechniques” (Ministry of Agriculture, 1977) and detailed inzyszka et al. (in press).

.5. Statistical analysis

The CA group was used to control for any potential effectf the anthelmintic treatment per se on the PA group. How-ver, it was found that there was no significant (P > 0.05)

ology 193 (2013) 214– 222

effect of the anthelmintic treatment per se on the controlanimals, as all measurements taken were similar betweenthe two control groups. For this reason, C and CA groupswere combined for subsequent analysis.

The activity and posture data acquired from the IceTagswere downloaded with the IceRobotics software in a formatof one summary record per minute. Each record provideda date, time and percentage of time spent lying and stand-ing and the number of steps taken. The lying and standingdata were consolidated into episodes with the use of pur-pose written FORTRAN programs (Tolkamp et al., 2010).These episodes were calculated by assuming that a con-tinuous series of records that showed 100% either lying orstanding behaviour, were part of the same episode. Whenboth lying and standing occurred in the same minute, itwas assumed that this was a transition minute in which thebehaviour during the first part of the minute was the sameas that during the last part of the previous minute. Shortlying episodes (those under 4 min) were deleted becausethese were previously verified with video footage not tocorrespond to real lying behaviour (Tolkamp et al., 2010;Szyszka et al., in press). This resulted in a sharp reduction inthe number of episodes without any considerable impacton the total lying and standing times, since many deletedepisodes lasted only a few seconds. The analysis appliedwas on the total number of steps taken, the total lying orstanding time (which are reciprocal), and the frequencyof standing episodes (which by definition is identical tothe frequency of lying episodes) and duration of lying andstanding episodes per day.

Feeding behaviour episodes can be grouped into mealsafter estimating a meal criterion, i.e. the maximum lengthof non-feeding intervals that are acceptable as within-meal intervals (Tolkamp et al., 2011). Meal criteria werecalculated after fitting mixed models to the frequency dis-tribution of log-transformed lengths of intervals betweenfeeding episodes measured in seconds. Both a two-population model, consisting of one Gaussian distribution(for within-meal intervals) and one Weibull distribution(for between-meal intervals) and a three-population modelwere fitted to the data. The latter contained an additionalGaussian to describe the population of within-meal inter-vals during which animals drink (Tolkamp and Kyriazakis,1999). The models are described in detail by Yeates et al.(2001) and were fitted using the SAS 9.1 (SAS InstituteInc., Cary, USA) program of González et al. (2008). Modelfit was compared by statistical analysis based on the like-lihood values for the models. Since likelihood values arevery small, they are expressed in SAS as minimum func-tion values (MFV, i.e., twice the negative log-likelihood).Because the two-population model (with 5 parameters)was completely nested in the three-population model(with 8 parameters), a likelihood ratio test, assuming a �2

distribution of the test statistic (Kleinbaum et al., 1988)was possible. This test showed that the addition of thethird population resulted in a significantly (P < 0.001) bet-ter fit. From the parameters of the three-population model,

the meal criterion was estimated at 19.41 min using themethod of Yeates et al. (2001) and all intervals shorterthan this were considered as within meal intervals. Bothepisodes and meals were analysed for their duration and

y Parasitology 193 (2013) 214– 222 217

Fig. 1. Eggs per gram of fresh faeces against experimental day withconfidence intervals for bulls challenged with 100,000 L3 larvae of Osterta-gia ostertagi on Days 0, 7 and 14 of the experiment. The PA (parasite

trols remained low at 0.49 (SED = 0.34) iu/l. For the Pgroup these remained elevated throughout the experi-ment, whereas for the PA group levels started to decreasepost dosing (Day 42), as shown in Fig. 2. Despite this

Fig. 2. Pepsinogen levels (iu/l) against experimental day for unchallengedcontrols (control; n = 12) and bulls challenged with 100,000 L3 larvae ofOstertagia ostertagi (P and PA) on Days 0, 7 and 14 of the experiment. The

O. Szyszka et al. / Veterinar

frequency. Feeding behaviour data were averaged per ani-mal across the 3 days within each period considered.

Body weight (BW), pepsinogen levels, posture, activityand feeding behaviour were all analysed with the use of arepeated measures ANOVA (SPSS 15, IBM, USA) with treat-ment as a fixed and day (or period) as a random effect. ForBW and BW gain analysis, the Day 0 BW value was used as acovariate. Activity as well as lying and standing behaviourused the mean of the 8 days prior to the experiment (Days−8 until −1) as a covariate to account for variation betweenindividuals. Furthermore the data on the activity, as num-ber of steps taken, FEC, average meal duration, and lyingand standing episode duration were log-transformed priorto analysis in order to normalize their distribution. Theseresults are reported as back-transformed means with 95%confidence intervals (CI).

In addition, the area under the curve (AUC) (Matthewset al., 1990) was calculated for the FEC for each individualin the P and PA groups over specified time periods. Thetime periods used were Days 21–31 for both the P and PAgroups, and Days 31–45 and 21–45 for the P group only.The AUC values were correlated to the corresponding meanfor individual activity and posture measurements indicatedabove through a Pearson correlation (SPSS 15, IBM, USA).

To allow a correction to be made for the fact that a largenumber of variables were analysed, increasing the proba-bility of false-positive results (Type 1 error), a Bonferronicorrection was applied. This resulted in an adjusted P-valueof 0.002 for the 5% significance threshold.

3. Results

One animal from group P was removed from the experi-ment due to poor growth even before parasitism developedand its data were treated as a missing value.

3.1. Faecal egg counts

The consistency of the faeces varied throughout the trialand between treatments: up to Day 17 the faeces were sim-ilarly solid for all groups. However, after Day 17 a numberof animals in both P and PA groups showed signs of diar-rhoea (maximum 8 out of 14 on Day 28); faecal consistencyreturned to normal in the PA animals post dosing withthe anthelmintic. The egg counts for the control animalsremained at zero throughout the experiment. The FEC ofthe parasitized animals were positive from Day 17, almost3 weeks after infection. Between Days 17 and 31 the aver-age FEC was 190 eggs/g (CI: 70.2–516) for the P and 183eggs/g (CI: 76.7–439) for the PA group (P > 0.05). After thePA group was drenched on Day 31, their FEC returned to,and remained at 0 throughout (Fig. 1). The FEC for the Pgroup continued to be high until the end of the experi-ment, having an average of 527 eggs/g (CI: 286–968), forDays 31–45.

3.2. Pepsinogen

The pepsinogen levels in the serum showed a signifi-cant (P < 0.001) time effect, treatment effect and treatmentby time interaction. Animals on the P and PA groups

interrupted; n = 7) group was drenched with an anthelmintic on Day 31,whereas P (parasitized; n = 7) group remained undrenched throughout.Drenching is indicated by an arrow.

had elevated pepsinogen levels by Day 21, 2.17 and2.19 iu/l respectively, while pepsinogen levels of the con-

PA (parasite interrupted; n = 7) group was drenched with an anthelminticon Day 31, whereas P (parasitized; n = 7) group remained undrenchedthroughout. The bar is the standard error of the difference (SED) andshown on the control treatment. Drenching is indicated by an arrow.*P < 0.05 P and PA compared to control. #P < 0.05 P compared to PA.

218 O. Szyszka et al. / Veterinary Parasitology 193 (2013) 214– 222

Fig. 3. Body weight (kg) against experimental day for unchallenged con-trols (control; n = 12) and bulls challenged with 100,000 L3 larvae ofOstertagia ostertagi (P and PA) on Days 0, 7 and 14 of the experiment. ThePA (parasite interrupted; n = 7) group was drenched with an anthelminticon Day 31, whereas P (parasitized; n = 7) group remained undrenchedts*

de

3

omcm9sso(tgtr

3

t(bi(cpB1du

Fig. 4. The 3-day rolling mean for the total number of steps taken perday for unchallenged controls (control; n = 12) and bulls challenged with100,000 L3 larvae of Ostertagia ostertagi (P and PA) on Days 0, 7 and 14 ofthe experiment. The PA (parasite interrupted; n = 7) group was drenchedwith an anthelmintic on Day 31, whereas P (parasitized; n = 7) groupremained undrenched throughout. The bar is the confidence interval (CI)

hroughout. The bar is the standard error of the difference (SED) andhown on the control treatment. Drenching is indicated by an arrow.P < 0.05 P and PA compared to control.

ecrease, they had not returned to control levels by thend of the trial.

.3. Body weight gain

The change in BW throughout the experiment is shownn Fig. 3. There was a significant interaction between treat-ent and time (P < 0.001) on BW and the effect of the

ovariate was also significant (P < 0.001). Up to Day 21 ani-als in all groups gained weight at a similar rate, which was

35 (SEM = 81.5) g/day. From Day 21 the BW of the animalstarted to diverge with infected bulls gaining weight at alower rate (P < 0.001) than the controls. The growth ratef the animals from Day 21 to 31 was −33, −357 and 1775SED = 364) g/day for the P, PA and control respectively. Forhe same groups the growth rates from Day 31, when the PAroup was treated with the anthelmintic, until the end ofhe experiment was −200, 895 and 1172 (SED = 231) g/day,espectively.

.4. Activity and posture

Overall activity, as measured by the number of stepsaken per day, was not affected by time or treatmentP > 0.05); however it did show a significant decrease foroth the P and PA treatments from Day 21, which resulted

n a significant time by treatment interaction (P < 0.001)Fig. 4); in addition the effect of the covariate was signifi-ant (P < 0.001). Before Day 21 the average number of stepser day was similar across groups at 2150 (CI: 1830–2525).

etween Days 21 and 31 this was 1391 (CI: 1044–1853),206 (CI: 918–1585) and 1866 (CI: 1555–2239) steps peray for the P, PA and controls respectively, and contin-ed to be lower for the parasitized treatments after Day

associated with the back-transformed means and shown on the controltreatment. Drenching is indicated by an arrow. *P < 0.05 P and PA com-pared to control.

31 at 1252 (CI: 892–1756), 1156 (CI: 861–1551) and 1934(CI: 1610–2324) steps per day, respectively. There was nosignificant difference between the P and PA groups afterdosing (P > 0.05).

There was no effect of time and no time by treat-ment interaction on total standing time (P > 0.04). Theeffect of the covariate was significant for total standingtime (P < 0.002). Because an increase in total standing timeimplies a decrease in total lying time, the latter is not dealtwith to any further extent. The total daily standing timewas 506 (CI: 441–581) min across treatments.

The frequency of the lying and standing episodes,which by definition is identical, was significantly (P < 0.001)affected by time, treatment and the interaction betweentreatment and time; the effect of the covariate was also sig-nificant (P < 0.001). The interaction was due to a decreasein frequency by the P and PA animals after Day 21 (Fig. 5).The average frequency of episodes before Day 21 was 16.8episodes per day (SEM = 0.95) across treatments. BetweenDays 21 and 31 the average frequency of episodes was 12.5,10.7 and 19.7 (SED = 1.58) for the P, PA and control groupsrespectively. After Day 31, the average frequencies were9.02, 12.1 and 18.6 (SED = 1.99), respectively. The PA grouphad returned to the same values as the control by Day 39.

There was a significant time effect (P < 0.001) on theaverage lying, however not for the average standingepisode duration. Average lying (P = 0.002) duration, butnot average standing duration (P = 0.013), was significantly

affected by treatment, with a significant time by treatmentinteraction for both average lying and standing duration(P < 0.001); the effect of the covariate was also significant(P < 0.001). The time by treatment interaction was due to

O. Szyszka et al. / Veterinary Parasitology 193 (2013) 214– 222 219

Fig. 5. Average lying episode frequency (number per day) againstexperimental day for unchallenged controls (control; n = 12) and bullschallenged with 100,000 L3 larvae of Ostertagia ostertagi (P and PA) onDays 0, 7 and 14 of the experiment. The PA (parasite interrupted; n = 7)group was drenched with an anthelmintic on Day 31, whereas P (par-asitized; n = 7) group remained undrenched throughout. The bar is the

Fig. 6. Duration of average lying episodes in minutes against experimen-tal day for unchallenged controls (control; n = 12) and bulls challengedwith 100,000 L3 larvae of Ostertagia ostertagi (P and PA) on Days 0, 7and 14 of the experiment. The PA (parasite interrupted; n = 7) group wasdrenched with an anthelmintic on Day 31, whereas P (parasitized; n = 7)group remained undrenched throughout. The bar is the confidence inter-

standard error of the difference (SED) and shown on the control treat-ment. Drenching is indicated by an arrow. *P < 0.05 P and PA compared tocontrol. #P < 0.05 P compared to PA.

an increase in average standing and lying episode dura-tion for the P and PA groups, which started on Day 22 andpersisted whilst infestation continued (Fig. 6). The averagelying time per episode up to Day 22 was 43.1 (CI: 37.7–49.2)min and the average standing time per episode was 15.0(CI: 12.3–18.2) min across groups. Between Days 22 and31, the average lying time was 60.1 (CI: 47.8–76.3), 72.6(CI: 50.4–105) and 37.0 (CI: 31.0–42.9) min, and the averagestanding time was 21.5 (CI: 14.2–32.6), 25.2 (CI: 16.6–38.3)and 11.0 (CI: 8.8–13.8) min for the P, PA and controlrespectively. For the same groups post Day 31, the aver-age lying episode durations were 69.4 (CI: 38.3–126), 55.0(CI: 33–91.9) and 35.9 (CI: 28.1–45.8) min, and the aver-age standing episode durations were 34.0 (CI: 20.2–57.2)27.0 (CI: 15.7–46.3) and 12.4 (CI: 9.7–15.9) min, respec-tively. The PA group had returned to the same averagelying and standing episode values as the control by Day39. The analysis of activity therefore showed that the Pand PA groups resulted, on average, in less frequent andlonger lying and standing episodes, showing less alterationbetween the postures.

3.5. Feeding behaviour

Feeding behaviour was analysed per period for bothfeeding episodes and meals. There was a significant(P < 0.001) time effect for both the average and total feeding

episode duration; this was not the case for the feed-ing episode frequency (P > 0.05). There was no significanteffect of treatment and no interaction between treatmentand time (P > 0.05) either on average and total feeding

val (CI) associated with the back-transformed means and shown on thecontrol treatment. Drenching is indicated by an arrow. *P < 0.05 P or PAcompared to control.

episode duration, or on feeding episode frequency. Theaverage feeding episode duration was 14.9, 16.5 and 17.9(SEM = 0.8) min; total feeding episode duration was 73.4,99.6 and 115 (SEM = 8.19) min; the frequency of feedingepisodes was 24.8, 22.6 and 22.6 (SEM = 2.20) episodes perday for periods 1, 2 and 3 respectively.

There was only a time effect on total meal duration(P < 0.001) and a tendency for a time effect on average mealduration (P = 0.003). Treatment (P > 0.05) and treatment bytime interaction (P > 0.05) did not significantly affect thesemeasurements. The average meal duration was 17.7 (CI:12–26), 18.5 (CI: 12–29) and 21.5 (CI: 15–30) min and thetotal meal duration was 120, 138 and 165 (SEM = 11.1) minper day for periods 1, 2 and 3 respectively. However, therewas a non-significant trend for time (P = 0.027) and time bytreatment interaction (P = 0.039) effects on meal frequency.This was caused by a decrease in the meal frequency inthe third period for the P group (Fig. 7), whereas meal fre-quency for the PA and control animals continued to rise,leading to 6.6, 7.8 and 8.1 (SED = 0.67) meals per day for P,PA and control respectively during the third period.

3.6. Correlations between FEC and activity and posture

The AUC of the FEC showed a non-significant (P = 0.007),but high positive correlation with the average lying

episode duration between Days 21 and 31 (r = 0.71). Non-significant, but high negative correlations were also seenbetween the AUC of the FEC over the time period betweenDays 31 and 45 and activity (number of steps taken)

220 O. Szyszka et al. / Veterinary Parasit

Fig. 7. Meal frequency against experimental day for unchallenged con-trols (control; n = 12) and bulls challenged with 100,000 L3 larvae ofOstertagia ostertagi (P and PA) on Days 0, 7 and 14 of the experiment. ThePA (parasite interrupted; n = 7) group was drenched with an anthelminticots

(i

4

rdoaGtbnuipett

iueiltp2sgt

n Day 31, whereas P (parasitized; n = 7) group remained undrenchedhroughout. The bar is the standard error of the difference (SED) andhown on the control treatment. #P < 0.05 P compared to PA and control.

r = −0.882; P = 0.02) on the one hand, and lying and stand-ng frequency on the other (r = −0.891; P = 0.017).

. Discussion

The objectives of the study were to identify any tempo-al changes that occur in several types of behaviour of cattleuring abomasal parasitism and to quantify the magnitudef such changes. The latter was undertaken with the aim tossess the diagnostic value of these behavioural changes.onzález et al. (2008) have suggested that changes of 2.5

imes the standard deviation of the mean value of theehaviour of unchallenged animals should have a diag-ostic value for several health challenges in cattle. Wesed the device of interruption of parasitism through dos-

ng with an anthelmintic to test how quickly behavioursotentially affected by parasitism returned to ‘normal’ lev-ls (Kyriazakis et al., 1996). Fast recovery in the behavioursested could also have diagnostic use for the assessment ofhe effectiveness of treatments (Forbes et al., 2004).

The effects of parasitism on the standard diagnosticndicators of FEC and pepsinogen levels were not apparentntil Day 21 post infection, when both were significantlylevated. Because abomasal damage, of which pepsinogens an indicator, concurs with the establishment of infectivearvae (Jennings et al., 1966), a rise in pepsinogen is likelyo occur before Day 21 (Forbes et al., 2009; Szyszka et al., inress) and before any eggs appear in the faeces (Fox et al.,

002). Differences between treatment groups in BW wereeen from around Days 21–24 as a result of the absence ofain or even BW loss in infected bulls and persisted untilhe end of the experiment. The effect of infection on BW

ology 193 (2013) 214– 222

gain was more severe than expected, both on the basisof the literature (Fox et al., 2002) and the authors’ priorexperience (Szyszka et al., in press). This, in combinationwith the detection of low dry matter faeces in some ani-mals, shows that the established infection developed intoa clinical one for some animals. The removal of infectionthrough the administration of the anthelmintic resulted inan immediate reduction in FEC and an increased BW gain.The issue is how the changes in these measurements relateto any changes in the behaviours measured.

The focus was on measurements of overall activity andposture, captured through automated means, and feedingbehaviour, captured through the use of video cameras. Thelatter could also be captured by automated means, becausethe technology to achieve this is available in confined ani-mals (Sowell et al., 1999; Quimby et al., 2001; Weary et al.,2009). We expected that changes in activity and/or posturewould be observed before more conventional detectionof parasitism, through FEC and decreased BW gain. Thiswas based on the expectation that behaviour would beaffected by, for example, damage caused by the establish-ment of infective larvae, which generally occurs within afew days (Murray et al., 1970). This expectation however,did not materialize. The changes caused by parasitism inall measurements of activity and posture were observedto begin at approximately Day 22 post infection. In somecases the differences between infected and control animalsattained formal significance only a few days later, presum-ably due to variations in individual responses. These effects,manifested as a time by treatment interaction, were insummary: (i) a decrease in the number of steps taken, (ii) adecrease in lying and standing episode frequency and (iii)an increase in average lying and standing episode dura-tion in infected compared to control animals. The decreasein steps taken was expected, because activity levels fre-quently decrease when an animal is faced with a parasitichealth challenge (Hutchings et al., 2002; Reiner et al.,2009; Szyszka et al., in press), although the occurrence ofthe decrease may depend on the nature of the pathogenchallenge (Hart, 1990). It is still unclear what causes thisdecrease in activity during parasite or other infectiouschallenges. It has been suggested that it may arise fromlethargy that accompanies the physiological changes asso-ciated with “sickness behaviour” (Hart, 1988). Animalsmay try to conserve energy during exposure to pathogens,especially because sickness behaviour is frequently accom-panied by a reduction in voluntary food intake (see below;Sandberg et al., 2006), probably as a parasite-avoidancestrategy (Kyriazakis et al., 1998). The effect of parasitism onposture resulted in a reduction in transition between thelying and standing postures, thereby decreasing the energyrequirements for posture changes (Hart, 1988) and thepattern distribution of behaviour sequences, when com-pared to the controls. This pattern distribution if viewedover time, also known as the complexity of behaviour, haspreviously been observed in environmentally challengedanimals, such as parasitized wild goats (ibex, Alados et al.,

1996) and food deprived chickens (María et al., 2004).A decrease in behavioural complexity is considered anindicator of stress that results from health and welfare chal-lenges (Alados et al., 1996).

y Parasit

O. Szyszka et al. / Veterinar

The above changes in activity and posture due to par-asitism persisted throughout the infection period for thenumber of steps taken and increased over time for the lyingand standing episode frequency and average episode dura-tion. Furthermore there was a high, but non-significantcorrelation between the level of the FEC and the degree towhich activity and posture were affected, indicating thatanimals with higher FEC took fewer steps, laid for longerand changed posture less frequently. The infection estab-lished in this experiment, was for a number of animals atleast, clinical. A previous study (Szyszka et al., in press)showed similar effects, but of lesser magnitude on thebehaviour of cattle with a sub-clinical O. ostertagi infec-tion. All the above raise the possibility that the magnitudeof behavioural changes observed may be dose depend-ent. There is a dose dependent response in behaviourduring micro-parasitic health challenges on social explo-ration (Larson and Dunn, 2001) and activity (Skinner et al.,2009). The effects of increased doses of macro-parasiteson the behaviour of animals have yet to be systemati-cally addressed. It is possible that the relationship betweenparasite dose and effect on behaviour is of the form sug-gested by Sandberg et al. (2006), where they assumechanges in feed intake to occur at a gradient dependenton dose.

Drenching parasitized animals with an anthelminticfailed to result in the expected and hypothesizedrapid recovery in activity and posture. Although afteranthelmintic drenching the behavioural levels for theaverage standing and lying episode duration returned tosimilar values to those of the uninfected animals, thisoccurred approximately 1 week post-drenching. For thetotal frequency of lying and standing, the recovery startedimmediately, though taking the same amount of time toreturn to control levels. Our hypothesis was based on thefindings that certain behaviours such as feeding behaviour,seem to recover within a couple of days post administra-tion of an anthelmintic in parasitized sheep (Kyriazakiset al., 1996). It is possible, however, that the time course ofrecovery of the behaviour post-treatment depends on theseverity of the pathology caused by the parasite challengeand hence the physiological recovery, and is thus differentfor different health challenges.

Feeding behaviour was measured through the use ofvideo cameras and the measurements were taken overthree periods, each 3 days long. Period 1 covered Days 7–9,to provide baseline measurements because the behaviourshould not be different between the treatment groups atthis stage as was shown in a previous study (Szyszka et al.,in press). Period 2 (Days 31–33), was chosen with theexpectation that the effects of parasitism on activity andposture would then peak. Finally, period 3 (Days 39–41),was selected because by this time the bulls in the PAgroup were expected to have recovered from any possiblechanges in behaviour due to parasite infection. The hypoth-esis was that duration and possibly frequency of the feedingevents would decrease during infection, as suggested by

González et al. (2008) in cows suffering from subclinicalketosis or lameness, and by Forbes et al. (2000, 2004, 2007)in parasitized grazing calves and dairy cattle. Such changeswould be consistent with an expected decrease in food

ology 193 (2013) 214– 222 221

intake of cattle parasitized with O. ostertagi (Fox et al., 1989,2002).

However, the only weak effect we saw on feedingbehaviour was a decrease in meal frequency in period 3 forthe animals parasitized throughout the trial, 7 weeks afterthe initial challenge. The reduction was around 17% in theP group compared with the other two (parasite-free) treat-ment groups. A reduction in meal frequency could lead toa reduced intake as assumed by the weight gain; howeverno measures of food intake or feeding rate were taken inthis experiment.

We also expected a rapid effect on feeding behaviourfollowing the administration of the anthelmintic, basedon the study of Kyriazakis et al. (1996) who found veryrapid recovery in food intake of parasitized sheep post-dosing. Forbes et al. (2004) likewise recorded an increaseof feeding episode duration in animals that had receivedtreatment with an anthelmintic compared to infected ani-mals. Although there was an increase in meal frequency7 days after the anthelmintic treatment in the PA groupcompared to the untreated parasitized animals, the mealfrequency of the control animals also continued to increaseover time, albeit at a slower rate. In view of the rapidadvances in the monitoring of cattle feeding behaviour(Weary et al., 2009), the absence of effects of infection onfeeding behaviour in our study were disappointing.

The magnitude of behavioural changes in activity andposture was large enough to be statistically significantwhich suggests that they might be suitable candidates fordisease detection. Activity, as measured by the numberof steps taken, showed a 41% decrease in the parasitizedanimals once established. There was an increase by 52%in average lying and by 55% in standing episode dura-tion, while the frequency of lying and standing episodesdecreased by 44%. González et al. (2008) suggested thatchanges of at least 2.5 standard deviations from the previ-ous 7-day rolling average of total individual daily feedingtime have strong diagnostic value. They were able to iden-tify more than 80% of cows with acute disorders at least1 day before diagnosis by farm staff. The changes in thebehaviours observed in this experiment are of greater mag-nitude than such deviations. Although, in this experiment,the changes in the behaviour coincided with the onset ofclinical measurements, such as FEC, they may have a diag-nostic value where clinical signs are not apparent (Forbeset al., 2004, 2007). When the infection is acquired naturally,it would be expected that the animal takes in a lower num-ber of larvae on a greater number of occasions (Mansouret al., 1992) as well as a mixture of parasite species (Forbeset al., 2004). Therefore, in order to eventually progress tothe early detection and the detection of subclinical para-site infection through behaviour, the magnitude and rateof change in behaviours that accompany a natural infectionwill also need to be quantified.

This experiment constitutes the first step towardsaddressing the potential diagnostic value of changes inbehaviour to parasitism in cattle. There are several steps

that need to be taken in order to assess the value of suchchanges as a diagnostic indicator, including their sensitiv-ity and specificity. In addition, the magnitude and directionof such changes will have to be assessed in less controlled

2 y Parasit

eat

C

A

L

R

A

A

B

F

F

F

F

F

F

F

G

G

H

H

H

22 O. Szyszka et al. / Veterinar

nvironments, such as those on pasture. In our view, this isn area of research where effort could be usefully directedo.

onflict of interest statement

The authors declare no conflicts of interest.

cknowledgement

This Project was sponsored by EBLEX (English Beef andamb Executive).

eferences

FRC, 1993. Energy and Protein Requirements of Ruminants: An AdvisoryManual Prepared by the AFRC Technical Committee on Responses toNutrients. CAB International, Wallingford, UK.

lados, C.L., Escos, J.M., Emlen, J.M., 1996. Fractal structure of sequentialbehaviour patterns: an indicator of stress. Anim. Behav. 51, 437–443.

orderas, T.F., Passilé, A.M., Rushen, J., 2008. Behavior of dairy calves aftera low dose of bacterial endotoxin. J. Anim. Sci. 86, 2920–2927.

ogsgaard, K.K., Røntved, C.M., Sørensen, P., Herskin, M.S., 2012. Sicknessbehaviour in dairy cows during Escherichia coli mastitis. J. Dairy Sci.95, 630–638.

orbes, A.B., Huckle, C.A., Gibb, M.J., 2004. Impact of eprinomectin ongrazing behavior and performance in dairy cattle with sub-clinicalgastrointestinal nematode infections under continuous stocking man-agement. Vet. Parasitol. 125, 353–364.

orbes, A.B., Huckle, C.A., Gibb, M.J., 2007. Evaluation of the effect ofeprinomectin in young dairy heifers sub-clinically infected with gas-trointestinal nematodes on grazing behaviour and diet selection. Vet.Parasitol. 150, 321–332.

orbes, A.B., Huckle, C.A., Gibb, M.J., Rook, A.J., Nuthall, R., 2000. Eval-uation of the effects of nematode parasitism on grazing behaviour,herbage intake and growth in young grazing cattle. Vet. Parasitol. 90,111–118.

orbes, A.B., Warren, M., Upjohn, M., Jackson, B., Jones, J., Charlier, J.,Fox, M.T., 2009. Associations between blood gastrin, ghrelin, leptin,pepsinogen and Ostertagia ostertagi antibody concentrations and vol-untary feed intake in calves exposed to a trickle infection with O.ostertagi. Vet. Parasitol. 162, 295–305.

ox, M.T., Gerrelli, D., Pitt, S.R., Jacobs, D.E., Gil, M., Gale, D.L., 1989. Osterta-gia ostertagi infection in the calf: effects of a trickle challenge onappetite, digestibility, rate of passage of digesta and liveweight gain.Res. Vet. Sci. 47, 294–298.

ox, M.T., Uche, U.E., Vaillant, C., Ganabadi, S., Calam, J., 2002. Effectsof Ostertagia ostertagi and omeprazole treatment on feed intake andgastrin-related responses in the calf. Vet. Parasitol. 105, 285–301.

onzález, L.A., Tolkamp, B.J., Coffey, M.P., Ferret, A., Kyriazakis, I., 2008.Changes in feeding behavior as possible indicators for the auto-matic monitoring of health disorders in dairy cows. J. Dairy Sci. 91,1017–1028.

reer, A.W., Stankiewicz, M., Jay, N.P., McAnulty, R.W., Sykes, A.R., 2005.The effect of concurrent corticosteroid induced immune-suppressionand infection with the intestinal parasite Trichostrongylus colubri-formis on food intake and utilization in both immunologically naïveand competent sheep. Anim. Sci. 80, 89–99.

art, B.L., 1988. Biological basis of the behavior of sick animals. Neurosci.Biobehav. Rev. 12, 123–137.

art, B.L., 1990. Behavioral adaptations to pathogens and parasites: five

strategies. Neurosci. Biobehav. Rev. 14, 273–294.

utchings, M.R., Gordon, I.J., Kyriazakis, I., Robertson, E., Jackson, F., 2002.Grazing in heterogeneous environments: infra- and supra- parasitedistributions determine herbivore grazing decisions. Oecolagia 132,453–460.

ology 193 (2013) 214– 222

Jennings, F.W., Armour, J., Lanson, D.D., Roberts, R., 1966. ExperimentalOstertagia ostertagi infections in calves: studies with abomasal can-nules. Am. J. Vet. Res. 27, 1249–1257.

Kleinbaum, D.G., Kupper, L.L., Muller, K.E., 1988. Applied Regression Anal-ysis and Other Multivariate Methods. PWS-Kent Publishing Company,Boston, USA.

Kyriazakis, I., Tolkamp, B.J., 2010. Disease. In: Mills, D.S. (Ed.), The Ency-clopedia of Applied Animal Behaviour and Welfare. CAB International,Wallingford, UK, pp. 176–177.

Kyriazakis, I., Tolkamp, B.J., Hutchings, M.R., 1998. Towards a functionalexplanation for the occurrence of anorexia during parasitic infections.Anim. Behav. 56, 265–274.

Kyriazakis, I., Anderson, D.H., Oldam, J.D., Coop, R.L., Jackson, F., 1996.Long-term subclinical infection with Trichostrongylus colubriformis:effects on food intake, diet selection and performance of growinglambs. Vet. Parasitol. 61, 297–313.

Larson, S.J., Dunn, A.J., 2001. Behavioral effects of cytokines. Brain Behav.Immun. 15, 371–387.

Mansour, M.M., Dixon, J.B., Rowan, T.G., Carter, S.D., 1992. Modulation ofcalf immune responses by Ostertagia ostertagi: the effect of diet duringtrickle infection. Vet. Immunol. Immunopathol. 33, 261–269.

María, G.A., Escos, J., Alados, C.L., 2004. Complexity of behaviouralsequences and their relation to stress conditions in chickens (Gal-lus gallus domesticus): a non-invasive technique to evaluate animalwelfare. Appl. Anim. Behav. Sci. 86, 93–104.

Matthews, J.N.S., Altman, D.G., Campbell, M.J., Royston, P., 1990. Anal-ysis of serial measurements in medical research. Br. Med. J. 300,230–235.

Ministry of Agriculture, Fisheries and Food, 1977. Manual of VeterinaryParasitological Laboratory Techniques. London, UK.

Murray, M., Jennings, F.W., Armour, J., 1970. Bovine ostertagiasis: struc-ture, function and mode of differentiation of the bovine gastric mucosaand kinetics of the worm loss. Res. Vet. Sci. 11, 417–427.

Paynter, D.I., 1992. Pepsinogen Activity: Determination in Serum andPlasma. Australian Standard Diagnostic Techniques, Australia, Benalla.

Quimby, W.F., Sowell, B.F., Bowman, J.P.G., Branine, M.E., Hubbert, M.E.,Sherwood, H.W., 2001. Application of feeding behaviour to predictmorbidity of newly received calves in a commercial feedlot. Can. J.Anim. Sci. 81, 315–320.

Reiner, G., Hübner, K., Hepp, S., 2009. Suffering in diseased pigs asexpressed by behavioural, clinical and clinical-chemical traits, in awell defined parasite model. Appl. Anim. Behav. Sci. 118, 222–231.

Sandberg, F.B., Emmans, G.C., Kyriazakis, I., 2006. A model for predictingfeed intake of growing animals during exposure to pathogens. J. Anim.Sci. 84, 1552–1566.

Skinner, G.W., Mitchell, D., Harden, L.M., 2009. Avoidance of physical activ-ity is a sensitive indicator of illness. Physiol. Behav. 96, 421–427.

Sowell, B.F., Branine, M.E., Bowman, J.P.G., Hubbert, M.E., Sherwood, H.E.,Quimby, W.F., 1999. Feeding and watering behavior of healthy andmorbid steers in a commercial feedlot. J. Anim. Sci. 77, 1105–1112.

Szyszka, O., Tolkamp, B.J., Edwards, S.A., Kyriazakis, I. The effects of acuteversus chronic health challenges on the behavior of beef cattle. J. Anim.Sci., in press. http://dx.doi.org/10.2527/jas.2011-4765

Tolkamp, B.J., Kyriazakis, I., 1999. To split behaviour into bouts, log-transform the intervals. Anim. Behav. 57, 807–817.

Tolkamp, B.J., Haskell, M.J., Langford, F.M., Roberts, D.J., Morgan, C.A., 2010.Are cows more likely to lie down the longer they stand? Appl. Anim.Behav. Sci. 124, 1–10.

Tolkamp, B.J., Allcroft, D.J., Barrio, J.P., Bley, T.A.G., Howie, J.A., Jacobsen,T.B., Morgan, C.C., Schweitzer, D.P.N., Wilkinson, S., Yeates, M.P., Kyr-iazakis, I., 2011. The temporal structure of feeding behavior. Am. J.Physiol. Regul. Integr. Comp. Physiol. 301, R378–R393.

Weary, D.M., Huzzey, J.M., von Keyserlingk, M.A.G., 2009. Using behaviourto predict and identify ill health in animals. J. Anim. Sci. 87,770–777.

Wolfensohn, S., Lloyd, M., 2003. Handbook of Laboratory Animal Manage-ment and Welfare. Blackwell Publishing Limited, Oxford, UK.

Yeates, M.P., Tolkamp, B.J., Allcroft, D.J., Kyriazakis, I., 2001. The use ofmixed distribution models to determine bout criteria for analysis ofanimal behaviour. J. Theor. Biol. 213, 413–425.