Subclinical nematode parasitism affects activity and rumination patterns in ﬁrst-season grazing cattle

Sickness behaviour has been suggested as an applicable indicator for monitoring disease. Deviating feeding behaviour and activity can provide information about animals’ health and welfare status. Recent advances in sensor technology enable monitoring of such behaviours and could potentially be utilized as an indicator of gastrointestinal nematode ( GIN ) infections. This study investigated activity and rumination patterns in ﬁrst-season grazing steers exposed to subclinical infection levels of the GIN Ostertagia ostertagi and Cooperia oncophora. At turnout, animals were allocated to one of four experimental groups and were faced with ‘‘high” ( H1 , n = 15; H2 , n = 17) or ‘‘low” ( L1 , n = 17; L2 , n = 11) levels of parasite exposure by grazing in similar enclosures contaminated with overwintering third stage ( L3 ) GIN larvae. Animals in H1 and H2 ( HP ) received a 1:1 mix of approximately 10,000 O. ostertagi and C. oncophora L3 at turnout; whereas the animals in L1 and L2 ( LP ) were treated monthly with ivermectin. Activity and rumination patterns were monitored by ﬁtting animals with leg-(IceQube) and neck-mounted (Heatime) sensors. BW was recorded every fortnight, whereas faecal and blood samples were examined every four weeks for nematode faecal egg count and serum pepsinogen concentrations ( SPCs ). There was an interaction effect of exposure level and period ( P < 0.0001) on average lying daily time across the entire grazing time. A higher mean daily lying time ( P = 0.0037) was found in HP compared with LP during the ﬁrst 40 days on pasture. There was also interaction effects of treatment and day since turnout on rumination time ( P < 0.0001) and rumination change ( P = 0.0008). Also mean daily steps ( P < 0.0001) and mean daily motion index ( P < 0.0001) were markedly higher in HP during days 62–69, coinciding with peaking SPC in HP. Strongyle eggs were observed both in HP and LP from 31 days after turnout. Eggs per gram ( EPG ) differed between parasite exposure levels ( P < 0.0001), with mean EPG remaining low in LP throughout the experiment. Similarly, an increase in SPC was observed ( P < 0.0001), but only in HP where it peaked at day 56. In contrast, no difference in BW gain ( BWG ) ( P = 0.78) between HP and LP was observed. In conclusion, this study shows that behavioural measurements monitored with sensors were affected even at low infection levels not affecting BWG. These combined results demonstrate the potential of automated behavioural recordings as a tool for detection of subclinical parasitism. (cid:1) 2021 The Authors. Published by Elsevier B.V. on behalf of The Animal Consortium. Thisisanopenaccess article


Introduction
Gastrointestinal nematode (GIN) parasites have severe negative effects on the overall animal health and welfare in ruminants and thereby pose a significant economic burden to the global livestock industry (Charlier et al., 2020b).The current control of GIN https://doi.org/10.1016/j.animal.2021.1002371751-7311/Ó 2021 The Authors.Published by Elsevier B.V. on behalf of The Animal Consortium.This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

Animal
The international journal of animal biosciences infections is primarily based on the use of anthelmintic drugs (Sutherland and Scott, 2010).Although anthelmintic resistance was previously mainly a problem in the sheep sector, it is today emerging also in cattle nematodes such as Ostertagia ostertagi and Cooperia oncophora (Areskog et al., 2014;Demeler et al., 2009;Peña-Espinoza et al., 2016;Ramos et al., 2016;Waghorn et al., 2006).Targeted selective treatment (TST), where only individual animals within a group are treated based on indicators such as faecal egg count (FEC) or weight gain, has been proposed as a sustainable strategy to yield individual benefits to animal health and furthermore decrease the risk for development of anthelmintic resistance.The implementation of TST approaches in practice is today limited by the lack of user-friendly, reliable and affordable animal-side indicators (Charlier et al., 2014).
Advancements of Precision Livestock Farming (PLF) enable realtime monitoring of animal activity and behaviour (Berckmans, 2017).Monitoring of sickness behaviours has been suggested as an applicable indicator of disease in cattle.For example, physical activity (resting and movements) and feeding behaviour (rumination and grazing time) can provide specific information about animals' health and welfare status (von Keyserlingk et al., 2009).The knowledge of responses in host activity and feeding behaviour in relation to parasite infections is today limited and PLF-systems will have to be developed and explored for parasite management to be integrated (Vercruysse et al., 2018).
Today, there is only a handful of studies investigating host response with animal sensors to monitor altered activity patterns and feeding behaviour in relation to GIN infections.Studies on housed Holstein-Friesian calves experimentally trickle infected with 300 000 O. ostertagi third stage larvae (L3) for three weeks showed a decrease in number of steps taken as well as a decrease in number of standing and lying bouts 21 days postinfection (Szyszka et al., 2013).Recently, studies on first-season grazing cattle (FSG) exposed to different levels of O. ostertagi and C. oncophora on pasture for 21 weeks showed a decrease in total activity during three two-week periods, from day 46 until day 114 after turnout, and an increase in number of recorded lying bouts 74-86 days after turnout (Högberg et al., 2019).Heifer calves naturally challenged with GIN for 12 weeks showed a decrease in grazing time 70 days after turnout (Forbes et al., 2000), whereas dairy cows naturally infected with GIN showed a decrease in eating time, meal duration and total bites (Forbes et al., 2004).The development of PLF-systems now allows continuous monitoring of animals on pasture using affordable commercially available solutions, underlining the need to further explore the possible integration into parasite management.
The aim of this study was to investigate how activity patterns (i.e.lying time, number of steps, number of lying bouts and total activity), and rumination patterns change along with standard diagnostic indicators such as BW gain (BWG), nematode faecal egg count (FEC) and serum pepsinogen concentration (SPC), in naïve grazing cattle, when exposed to two different levels of O. ostertagi and C. oncophora under grazing conditions.We predicted that total activity would decrease, whereas lying time and number of lying bouts would increase in FSG exposed to a higher level of GIN.Furthermore, it was hypothesized that FSG exposed to a higher dose of GIN would show a decrease in rumination time.

Material and methods
The study took place at Götala Beef and Lamb Research Centre, Sweden (58°42 0 N, 13°21 0 E; elevation 150 m.a.s.l.) from May 2nd until September 18th 2018.The study was approved by the Committee on Animal Experiments in Gothenburg (registration number 824-2017).

Experimental design
The experiment involved replicated grazing groups exposed to two different levels of GIN.Before turnout, animals were randomized according to age and BW.Each group was released into one of four separate pasture enclosures of 7-10 ha naturally contaminated with nematodes the previous year.Group size was adjusted to result in a similar stocking rate in every group.The high level exposure groups (HP), H1 (n = 15) and H2 (n = 17), were each primed at turnout with approximately 10 000 infective L3 of O. ostertagi and C. oncophora (1:1).The animals in low parasite exposure groups (LP), L1 (n = 17) and L2 (n = 11), were instead treated with an ivermectin pour-on solution (Ivomec Ò Pour-on, Boehringer Ingelheim) at the recommended dose rate of 0.5 mg per kg BW, at four-week intervals from turnout until the end of the 20 week grazing trial.

Pasture
The pasture consisted of permanent semi-natural pastures, which in the previous year had been grazed for 22 weeks by untreated cattle.For the present experiment the pasture was split up into four similar enclosures, consisting of approximately 20% dry, 60% mesic and 20% wet areas.The pasture was mainly open, but included small areas of deciduous trees.The dominant plant species was Deschampsia cespitosa (tufted hairgrass), but Festuca rubra (red fescue) was also prominently present.The sward height and chemical composition of the herbage were measured at turnout, housing and at four-week intervals in between.In each enclosure, sward height measurements were made according to Frame (1993), with 120-150 recordings performed with a rising plate meter.To estimate chemical composition, 25-30 herbage mass samples were cut in 3-m diameter circles along the route.The samples were analysed for concentration of crude protein, neutral detergent fibre and in vitro organic matter digestibility.The crude protein was determined according to Dumas (1831) and neutral detergent fibre was determined according to Chai and Udén (1998).Metabolizable energy concentration was calculated in vitro disappearance of rumen organic matter according to Lindgren (1979).

Activity and rumination data
One week prior to turnout, the steers were fitted with IceQube Ò 3D-accelerometers (IceRobotics Ltd, Edinburgh, UK; Validated by: Borchers et al., 2016;Kok et al., 2015;Ungar et al., 2018) on the left hind leg above the fetlock.Sensor dimensions were 55 Â 55 Â 27 mm and 75 g.The tri-axial accelerometer operates using a sample rate of 4 Hz with a time resolution of 15 min, with a 9-day internal memory (i.e. one period).It continuously recorded Lying (indicates whether the animal is lying down or not), Steps (number of steps/leg movements), Lying bouts (indicates the start of a lying bout) and Motion Index (the measured net acceleration, indicates total activity).Recordings from IceQubes, expressed as minutes per 24 h and numbers of steps and lying bouts per 24 h, were downloaded using an IceReader when the animals were handled.This was done at weekly intervals during the first six weeks (periods 1-6), and thereafter every fortnight (periods 7-13).Recordings from the first day of period 1 and the last day of period 6, as well as recordings from the first and last day for each of periods 7-13, were not included in the analysis so that each analysed day contained 24 h of data and that each period contained seven days.
Animals were also fitted with an Heatime Ò HR-LD activity and rumination collar (SCR Engineers Ltd., Netanya, Israel) (validated by: Burfeind et al., 2011;Schirmann et al., 2009).It measured neck activity (the measured net acceleration, indicates total activity of the neck) and rumination time using a 3D-accelerometer and a microphone containing a microprocessor, respectively.Data were calculated and summarized at 2-h intervals and automatically downloaded via a long-distance antenna to the Heatime HR system (SCR Engineers Ltd., Netanya, Israel).The system gives the raw rumination time (min) and activity level, scaled from 0 to 255 at 2-h intervals.In addition, the system calculates activity change and rumination change, determining the individual behavioural change by comparing raw data of the given 2-h interval to data from the corresponding 2-h interval of previous days, presented as the change in standard deviations adjusted to a À100 to 100 scale.It should be noted that this process is not fully transparent due to intellectual property rights.Activity and rumination data, raw and change respectively, were summarized on a 24 h basis and downloaded at the end of the study.

Infection levels and weight gain
BW was recorded to the nearest kg using a calibrated scale, at turnout (start of experiment) and after 142 days at housing (end of experiment), as well as at every fortnight in between.BW gain between every pair of consecutive weighings was calculated.In connection with every second weighing, rectal faecal samples and 2 Â 5 ml blood were collected from the coccygeal vein (BD Vacutainer Ò Plus Plastic Serum Tubes, Becton Dickinson).FEC was determined according to a modified McMaster technique based on 5 g of faeces with a minimum detection level of 20 nematode eggs per gram (EPG).In addition, pooled group samples containing approximately 5 g of faeces per animal were mixed with Vermiculite Ò and kept under humid conditions at 20 °C for 10 days to culture the eggs to the L3 stage.The larvae were harvested using the Petridish method (Elmahalawy et al., 2018), concentrated and stored in separate 1.5 ml Eppendorf tubes.Total DNA was extracted using the NucleoSpin XS Tissue kit, following the guidelines issued by the manufacturer (Macherey Nagel, Germany).The fractional abundances or proportion of DNA copies of O. ostertagi and C. oncophora in the internal transcriped spacer region 2 were then determined in duplex reactions using a droplet digital PCR assay (BioRad), as described by Baltrušis et al. (2019).The limit of detection for the method, defined as the lowest threshold for the detection of each parasite genus DNA in a mixed sample, is 2.2% of fractional DNA copy number for Coperia and 0.67% for Ostertagia, respectively.This means that 1 molecule of Ostertagia DNA can be idientified within a mix of 45 DNA molecules, whereas a single Cooperia DNA molecule can be identified within a mix of roughly 149 DNA molecules.Sera was analysed at Ghent University (Faculty of Veterinary Medicine, Department of Virology, Parasitology & Immunology, Laboratory for Parasitology) to determine SPC expressed as IU tyrosin according to a micro-method (Charlier et al., 2011).The method is based on the hydrolysing effect of serum on buffered bovine albumin substrate.The result is expressed as units of tyrosine (U Tyr) with a threshold of >3.5 U Tyr representing clinical ostertagiosis.

Statistical analysis
The statistical analyses were performed using R studio (v.1.2.5033).Assumptions of variance homogeneity and normal distribution of residuals of the collected data were checked by inspection of residual plots.Differences in activity and rumination measurement were analysed using mixed models with repeated measures using the LME function in the NLME package (Pinheiro et al., 2021).For IceQube data (Lying time, Lying bouts, Steps and Motion Index), treatment (HP, LP) reflecting different levels of GIN exposure and Periods (1-13) were treated as fixed factors with the experimental group and individual as nested random effects.Start weight was treated as a covariate in the model.Data from one IceQube sensor during periods 7-13 were missing due to a malfunction, as well as from one animal treated for interdigital phlegmons in L1 that was excluded for the prior and following week to treatment.These data points were treated as missing data in the analysis.
For Heatime data, treatment and day (from turnout) were treated as fixed factors.In addition, IceQube and Heatime data from turnout to day 40 were analysed separately with treatment and day as fixed factors.To account for time autocorrelation, a continuous autoregressive structure for a continuous time covariate (Cor-Car1) was fitted for both models, respectively.Final model selection was based on the Akaike Information Criterion.Pairwise differences were compared with ANOVA in the NLME package.Differences between HP and LP during the different periods were compared using Tukey's pairwise comparisons with the emmeans package (Lenth, 2021).Activity recordings of time spent standing were not analysed, as they are a direct mirroring of time spent lying.BWG, EPG counts and SPC were compared as dependent factors in a repeated measure mixed model with exposure level (HP, LP) and day as fixed effects and the individual animal as a random effect using the LME function in the NLME package (Pinheiro et al., 2021).A continuous time covariate (corCar1) was also fitted to account for autocorrelation.All graphical illustrations were made using the ggplot2 package (Wickham, 2016).

IceQube data
There was an interaction effect of exposure level and period (P < 0.0001) on average lying daily time across the entire grazing time (Fig. 1a).Animals in HP showed an increased daily lying time by 28.5 ± 7.7 min (P = 0.037) during the first 40 days compared with those in LP.No effect on the number of daily lying bouts (P = 0.29) or interaction (P = 0.29) was recorded across the grazing period (Fig. 1b).Although there was only a tendency (P = 0.057) for a difference in the average daily number of steps with animals in HP taking 138 ± 121 more steps per day across the grazing period (Fig. 1c), there was a highly significant difference from day 62 to day 69 (Period 8) (P < 0.0001).During this period animals in HP took 913 ± 148 more steps per day than those in LP (H1: 4 529 ± 2 005, range: 2 608-10 261; H2: 4 983 ± 2 000, range: 2 483-11 134; L1: 3 851 ± 552, range: 2 429-5 407; L2: 3 532 ± 933, range: 2 258-10 047).Overall there was also a tendency (P = 0.051) of parasite exposure effect recorded on daily Motion Index (Fig. 1d), with a highly significant (P < 0.0001) difference from day 62 to day 69 (Period 8).Then HP animals were showing a daily increase in Motion Index with 3 148 ± 539 compared with those in LP.

Heatime data
The average rumination time and rumination change are shown in Fig. 2. No effect of treatment on average rumination time (P = 0.68) and average rumination change (P = 0.45) was shown across the entire grazing time.However, there was an interaction effect on both rumination (P < 0.0001) and rumination change (P = 0.0008) during the first 40 days with animals in HP showing a larger variation (0.25 ± 21.9) in rumination change compared with those in LP (À1.31 ± 20.4).On the other hand, no effect of parasite exposure was observed on activity (P = 0.86), or activity change (P = 0.49) throughout the grazing period (Fig. 2).However, an interaction effect (P = 0.0009) on activity and a tendency (P = 0.065) of a difference in activity change between the treat- ments were recorded during the first 40 days.During this period, HP animals (0.068 ± 11.1) showed a larger individual decrease in activity than those in LP (0.86 ± 12.6).

Infection levels and weight gain
The BW for all animals are shown in Fig. 3a.At turnout the mean weight in HP was 341 ± 47 kg and in LP 342 ± 52 kg.There was an increase in BW in both HP and LP during the course of the study (P < 0.0001), but no significant difference (P = 0.97) or interaction (P = 0.97)) in BWG between HP and LP was found.The average growth rate from turnout until housing was 283 ± 13 g/day for HP and 281 ± 13 g/day for LP.The FEC is shown in Fig. 3b.Strongyle eggs appeared in both treatments 23 days after turnout.The FEC was higher (P < 0.0001) in HP than in LP throughout the study, peaking at day 31 and then declining.The proportions of O. ostertagi and C. oncophora (Ost/Coop) in both groups are shown in Fig. 4.Both parasites were found in HP and LP throughout the study, with proportions shifting from approximately 1:2 at day 31 to 4:1 at day 141 in HP.SPC levels (Fig. 3b) One treatment (HP) was primed at turnout with %10 000 infective third stage larvae of Ostertagia ostertagi and Cooperia oncophora (1:1) and thereby exposed to a high parasite challenge (n = 32), whereas the other treatment (LP) was dewormed with Ivomec Ò (0.5 mg kg À1 ) monthly, thus being exposed to a lower parasite challenge (n = 28).
were higher in HP (P < 0.0001) than in LP.The mean levels in HP ranged from 0.81 ± 0.40 to 1.78 ± 0.74 IU tyrosin, whereas in LP they ranged from 0.59 ± 0.59 to 0.74 ± 0.39 IU tyrosin.

Discussion
In this study, we investigated activity and rumination patterns in experimental groups of first year grazing cattle exposed to different levels of GIN whilst grazing in nearby pasture allocations for 21 weeks.Although both FEC and SPC were elevated in HP from day 31 until day 113, no differences in BWG were observed.In HP FEC peaked at 287 EPG at day 31, and clinical levels (>3.5 U Tyr) of ostertagiosis were only observed at one occasion at day 31.Still major behavioural differences were identified.For example, by using sensor data an increase in lying time was observed in HP during the first 40 days on pasture, as well as an increase in step count and motion index during day 62-69.Furthermore, interaction effects on rumination and activity were observed during the first 40 days.This suggests that activity in FSG is affected by subclinical levels of GIN infections even though BWG is unaffected.
As subclinical GIN infections in Sweden usually are associated with decreased BW gain (Dimander et al., 2003;Höglund et al., 2013;Larsson et al., 2007), our present finding indicates either a low level of parasite exposure also in HP animals and/or that the full growth potential in LP was not exploited.The average daily weight gain in both groups was extremely low, i.e. 283 ± 13 g for HP and 281 ± 13 g for LP, respectively.For example, in a comparable study at the same pasture, corresponding daily weight gain was 542 ± 29 g for animals exposed to high levels of GIN and 636 ± 29 g for animals exposed to low levels of GIN, respectively (Högberg et al., 2019).Nevertheless, the differences in both FEC and SPC clearly indicate a difference in parasite exposure levels in HP and LP.Performance in grazing cattle is a complex interaction between, among other, genetics and pasture availability and quality (Lawrence et al., 2012).The low weight gains in the present study could be explained by severe drought during the summer of 2018 (SLU Fältforsk, 2020) and lower nutrient concentration in the herbage, compared with previous studies at the same pasture (Höglund et al., 2013;Högberg et al., 2019), outweighing a possible detrimental effect from parasite infection.The dry condition could also have affected the development of L3 larvae on pasture resulting in a low larval challenge (Rossanigo and Gruner, 1995).Nevertheless, EPG levels in the dewormed group, which was not primed at turnout, indicates the presence of GIN larvae on pasture.This may indicate that the behavioural differences observed were primarily due to the larval challenge emerging from priming at turnout, rather than from massive pasture contamination.Nevertheless, the proportion of O. ostertagi and C. oncophora in HP were in line with previous studies in Sweden (Dimander et al., 2003).
It has been suggested that a behavioural response to GIN infections might be connected to the mucosal phase of the infection (Högberg et al., 2019).L3 of O. ostertagi enter the gastric glands six hours after uptake with the grass, resulting in swelling and  oedema after 12-24 hours.In FSG they then leave the mucosa following moulting to L4, after approximately three weeks and with the majority of glands in the fundic region of the mucosa normalizing after seven weeks (Osborne et al., 1960).This could explain why differences in lying time and tendencies of differences in activity change and rumination change were only observed the first 40 days.These observations are in line with earlier studies suggesting that challenge with GIN following experimental trickle infections will result in a deviation in activity and posture around three weeks post infection in housed animals (Szyszka et al., 2013).
Still, HP animals also showed a higher number of recorded steps and Motion Index between day 62 and day 69, soon after the highest SPC levels were observed.These findings are in line with the previous study of Högberg et al. (2019), where a similar discrepancy from classical sickness behaviour was detected in conjunction with high SPC levels.This may indicate that the increase in number of steps and Motion Index probably reflects an increase in leg movement, possibly abdominal kicking, rather than an increase in locomotor movement (i.e. a leg movement used to move from one place to another) (Ungar et al., 2018).This is further highlighted by the results that no difference in activity was observed during the same period.
It has previously been shown that GIN infections reduce the voluntary feed intake and efficiency of feed utilization (Coop and Holmes, 1996;Fox et al., 1989) and decrease the grazing time in FSG cattle (Forbes et al., 2000(Forbes et al., , 2007)).At the same time it has been suggested that cattle compensate a decrease in eating time with an increase in rumination time (Dado and Allen, 1994).Therefore, rumination time could be a potential indicator for GIN infection.However, the compensation is affected by the fibre concentration in the feed, where a higher negative correlation is expected at semi-natural pastures compared with leys.In an earlier study FSG cattle challenged at grass with O. ostertagi and C. oncophora did not show any difference in rumination time compared with dewormed animals 14 weeks after turnout (Forbes et al., 2007).In contrast, our results revealed an interaction effect on rumination and rumination change, as well as a tendency of increase in rumination change in HP during the first 40 days on pasture.However, it should be noted that there is a normal animal variability of daily rumination time, where the coefficient of variation has been reported to be between 16% (Dado and Allen, 1994) and 48% (Byskov et al., 2015).This underlines that individual variations need to be taken into account when developing and evaluating PLF-systems for parasite control programmes.The determination of thresholds should therefore be based on individual variation, especially as it is well known that GIN infections are overdispersed (Charlier et al., 2020a).
In conclusion, this study provides a first attempt to evaluate short-and long-term effects of GIN infection on activity and rumination patterns in FSG cattle using different on-animal sensors in a replicated grazing trial.The results show that several behavioural measurements were affected at least in a short-term perspective.However, because infection levels were low and not resulting in subclinical disease also in HP animals, it is necessary to further investigate how reoccurring larval challenges affect long-term activity and rumination patterns.Although additional research is required with more focus on changes at the individual level, we have demonstrated the potential of automated behaviour recordings as a sensitive diagnostic tool for detection of GIN and the possibility to integrate it into future parasite control programmes.

Fig. 3 .
Fig.3.Mean ± SD (a) BW (kg), and (b) serum pepsinogen (international units of tyrosine) in two exposure treatments of first-season grazing steers.One treatment (HP) was primed at turnout with %10 000 infective third stage larvae of Ostertagia ostertagi and Cooperia oncophora (1:1) and thereby exposed to a high parasite challenge (n = 32), whereas the other treatment (LP) was dewormed with Ivomec Ò (0.5 mg kg À1 ) monthly, thus being exposed to a lower parasite challenge (n = 28).

Fig. 4 .
Fig.4.Mean gastrointestinal nematode faecal egg counts (EPG) ± SD and corresponding species abundance (%) in two exposure treatments of first-season grazing steers.One treatment (HP) was primed at turnout with %10 000 infective third stage larvae of Ostertagia ostertagi and Cooperia oncophora (1:1) and thereby exposed to a high parasite challenge (n = 32), whereas the other treatment (LP) was dewormed with Ivomec Ò (0.5 mg kg À1 ) monthly, thus being exposed to a lower parasite challenge (n = 28).