Skip to main content
NCBI home page
Journal of Animal Science logoLink to Journal of Animal Science
. 2017 Nov;95(11):4703–4707. doi: 10.2527/jas2017.1835

Technical note: Instantaneous sampling intervals validated from continuous video observation for behavioral recording of feedlot lambs1

A N Pullin *, M D Pairis-Garcia *,2, B J Campbell *, M R Campler *, K L Proudfoot †
PMCID: PMC6292281  PMID: 29293740

Abstract

When considering methodologies for collecting behavioral data, continuous sampling provides the most complete and accurate data set whereas instantaneous sampling can provide similar results and also increase the efficiency of data collection. However, instantaneous time intervals require validation to ensure accurate estimation of the data. Therefore, the objective of this study was to validate scan sampling intervals for lambs housed in a feedlot environment. Feeding, lying, standing, drinking, locomotion, and oral manipulation were measured on 18 crossbred lambs housed in an indoor feedlot facility for 14 h (0600–2000 h). Data from continuous sampling were compared with data from instantaneous scan sampling intervals of 5, 10, 15, and 20 min using a linear regression analysis. Three criteria determined if a time interval accurately estimated behaviors: 1) R2 ≥ 0.90, 2) slope not statistically different from 1 (P > 0.05), and 3) intercept not statistically different from 0 (P > 0.05). Estimations for lying behavior were accurate up to 20-min intervals, whereas feeding and standing behaviors were accurate only at 5-min intervals (i.e., met all 3 regression criteria). Drinking, locomotion, and oral manipulation demonstrated poor associations (R2) for all tested intervals. The results from this study suggest that a 5-min instantaneous sampling interval will accurately estimate lying, feeding, and standing behaviors for lambs housed in a feedlot, whereas continuous sampling is recommended for the remaining behaviors. This methodology will contribute toward the efficiency, accuracy, and transparency of future behavioral data collection in lamb behavior research.

Keywords: behavior, feedlot, lambs, validation

INTRODUCTION

Behavioral observations of sheep in confined environments, such as feedlots and indoor barns, have been increasingly used as a tool to assess sheep production and welfare parameters. From a production standpoint, behavioral data contributed to discussions in assessing feed supplements (Ítavo et al., 2011; Nicory et al., 2015), pen design (Aguayo-Ulloa et al., 2014; Jongman et al., 2017), and stocking density (da Cunha Leme et al., 2013). Additionally, the prevalence of shy-feeder lambs (Rice et al., 2016a) and development of abnormal behaviors (Lauber et al., 2012; Teixeira et al., 2014) present animal welfare considerations. These studies have used a variety of behavioral methodologies, such as continuous sampling and instantaneous sampling at varying time intervals (i.e., 5, 10, 15, and 30 min). Continuous sampling is considered the gold standard method because observation for the entire duration of the sample period generates a true record of behavior, and it has been used for live behavioral observations for decades (Altmann, 1974). However, instantaneous sampling offers an alternative method to improve efficiency, where data is collected at predetermined time intervals (e.g., 5- or 10-min intervals) within the sample period, potentially for more than one animal at a time (Martin and Bateson, 2007). The researcher must consider the selected time interval's approximation of the behavior when compared with continuous sampling, though, to avoid inaccuracies in data interpretation. The behavior of interest as well as who performs it and the temporal and spatial conditions are considerable variables that impact the appropriate sampling methodology (Lehner, 1992). Therefore, it is imperative to validate the methodology to ensure that accurate conclusions are drawn from the data (Mitlöhner et al., 2001).

Previous literature has validated behavioral methodologies for other ruminant animals in confined environments, such as feedlot cattle (Mitlöhner et al., 2001), dairy calves (Miller-Cushon and DeVries, 2011), and dairy cows (Chen et al., 2016). To the authors' knowledge, only 1 prior study has validated behavioral observation methodologies for lambs housed on pasture (i.e., Pullin et al., 2017), and no studies have previously focused on lambs housed in a feedlot. Therefore, the objective of this study was to validate the accuracy of 4 different instantaneous scan sampling intervals (i.e., 5, 10, 15, and 20 min) compared with continuous sampling for lambs housed in a feedlot.

MATERIALS AND METHODS

The Ohio State University Institutional Animal Care and Use Committee approved the protocol for this study (protocol 2015A00000061). The animals were cared for in accordance with the Guide for the Care and Use of Agricultural Animals in Agricultural Research and Teaching (FASS, 2010).

Animals and Housing

This study was conducted at the Ohio Agricultural Research and Development Center Sheep Unit in Wooster, OH, in July 2015. Eighteen crossbred twin lambs (Hampshire × Dorset and Suffolk × Dorset; 17.8 kg [SD 2.7] and 62.7 d of age [SD 5.4]) were blocked by sex and BW (small, 15.6 ± 1.2 kg; medium, 17.0 ± 1.0 kg; and large, 20.7 ± 1.0 kg) and randomly assigned to 1 of 3 pens in an indoor covered research feedlot facility (n = 6 lamb per pen; 18 lambs total). The feedlot pens (4.1 m long by 1.5 m wide) consisted of expanded metal flooring with 3 metal gates and a wooden fence line feed bunk (3.7 m long by 0.3 m wide by 0.3 m deep) on the fourth side. Lambs were provided ad libitum access to water with an automatic waterer (0.3 m long by 0.2 m wide; Ritchie Industries, Inc., Conrad, IA) and fed a custom diet of 70% whole shelled corn, 15% supplement pellet, 10% alfalfa pellets, and 5% soybean hulls. The supplement was a separate pellet that was custom formulated to provide additional protein, minerals, and vitamins to meet, or exceed, recommended nutrient requirements (NRC, 2007).

Behavioral Measurements

Behavior was continuously recorded for 14 h (0600–2000 h) using 2 color wireless Internet protocol (IP) video cameras (model F19805P; Amcrest Technologies, Houston, TX) recording at 30 frames/s. Behaviors may have occurred outside of this time window, but video recordings had low visibility in the feedlot after 2000 h, which did not permit reliable observations between 2000 and 0600 h. One camera was centrally positioned in front of 2 pens and a second camera was centrally positioned in front of the remaining pen (1.8 m from pen front) using an elbow bracket at a height of 3.0 m from the pen floor. Digital video recordings were collected and stored using a portable laptop with external USB hard drives. Video output was viewed with Amcrest Technologies software (version 4.1) to ensure picture clarity and camera positioning prior to the behavioral recording session.

Behavioral data (Table 1) were collected through continuous focal animal sampling using The Observer software (version 5.0.25; Noldus Information Technology B.V., Wageningen, the Netherlands) for each of the 18 feedlot lambs. Lambs were identified by unique letters marked onto the fleece on both sides of the ribs (Marksman Livestock Spray Marker; Nettex, Rochester, UK).

Table 1.

Ethogram to identify behaviors performed by 18 crossbred, feedlot-housed lambs in a 14-h day (0600–2000 h) for validation of instantaneous sampling intervals from continuous video observation

Behavior Description
Feeding Head within the feed bunk.
Standing Body supported by 4 extended legs not in motion.
Lying Full contact of body on ground and body not supported by all 4 legs.
Locomotion Body supported by 4 legs while in motion; excludes standing. Motion defined as 1 full leg extended. Excludes intermediate movements (scratching and/or slight shifts in body weight).
Drinking Standing with muzzle within the watering cup.
Oral manipulation Physical contact of lamb's muzzle to an inanimate object or pen mate while head is in motion.
Other Unable to identify or observe lamb due to technical or environmental interference.
Open in a new tab

Statistical Analysis

Data were analyzed using SAS software (version 9.4; SAS Inst. Inc., Cary, NC) considering the individual lamb as the experimental unit (n = 18). The continuous data were converted to samples at 1-s intervals using PROC EXPAND, extrapolated into instantaneous samples for 4 intervals (5, 10, 15, and 20 min), and converted to the total duration estimated by each interval for each behavior in the 14-h period (PROC MEANS; Chen et al., 2016). Data were not normally distributed for drinking, locomotion, or oral manipulation behaviors. Transformations did not improve normality of these behaviors, so original values were used for analysis.

To evaluate the accuracy and bias of each sampling interval, a linear regression analysis (PROC REG) was conducted. For each behavior, pairwise comparisons were made between the behavioral estimates from each sampling interval (5, 10, 15, and 20 min) and the continuous sampling data. A tested sampling interval was considered to accurately estimate the behavior if the following criteria were met: R2 ≥ 0.90, slope not statistically different from 1 (P > 0.05), and intercept not statistically different from 0 (P > 0.05; Ledgerwood et al., 2010). The combination of these values reflects the strength of association (R2), linear relationship (slope), and over- or underestimation of the duration values of each behavior (intercept; Ledgerwood et al., 2010; Daigle and Siegford, 2014).

RESULTS AND DISCUSSION

The evaluation of the tested instantaneous sampling intervals against the regression analysis criteria is shown in Table 2. The pairwise relationship (R2) between each interval and continuous sampling is illustrated in Fig. 1. Intervals ≤20 min accurately estimated the amount of time feedlot-housed lambs performed lying behavior. Lying is performed in longer bouts (10.1 min in this study; unreported in other studies) for a majority of the day (60.8% in this study, 63–70% in the study of Bøe et al. [2006], and 78% in the study of Aguayo-Ulloa et al. [2014]). However, it should be noted that lying patterns can be affected by additional factors in confined environments, such as stocking density (Bøe et al., 2006) and type of bedding (Teixeira et al., 2013). Lying behavior plays a critical, diurnal role for rumination and rest, where most animals perform the behavior in the early morning and early evening (Shreffler and Hohenboken, 1980; da Cunha Leme et al., 2013). As such, longer, infrequent sampling intervals will accurately estimate lying behavior in confined ruminant animals (Chen et al., 2016; Pullin et al., 2017).

Table 2.

Means (SD) for the total minutes that behavior was performed by 18 crossbred, feedlot-housed lambs in a 14-h day (0600–2000 h) among different instantaneous sampling intervals that were extrapolated from continuous observations. Values for median and first quantile (Q1) through third quantile (Q3) are presented; data for drinking, locomotion, and oral manipulation behaviors were not normal

Sampling interval
Behavior Continuous 5 min 10 min 15 min 20 min Q1 Median Q3
Drinking 5.5 (3.9) 4.2 (4.6) 2.2 (4.3) 4.2 (8.6) 2.2 (6.5) 0.0 0.0 5.0
Feeding 79.4 (45.4) 81.7 (47.8)1 85.0 (52.7) 93.3 (54.3) 88.9 (64.4) 42.5 90.0 130.0
Locomotion 16.0 (6.4) 15.3 (10.5) 17.8 (14.0) 14.2 (12.0) 23.3 (23.0) 2.5 15.0 28.3
Lying 492.1 (85.9) 522.2 (92.0)1 515.0 (90.6)1 525.8 (101.9)1 506.7 (88.7)1 450.5 521.5 581.8
Oral manipulation 13.3 (13.1) 13.6 (13.2) 16.1 (16.5) 14.2 (24.4) 11.1 (14.1) 0.0 10.0 20.0
Other 32.4 (19.1) 37.5 (23.7) 36.7 (21.1) 37.5 (33.8) 40.0 (27.4) 20.0 30.0 55.6
Standing 170.5 (65.3) 177.8 (70.8)1 181.7 (78.6) 173.3 (90.4) 178.9 (90.4) 114.8 180.0 220.0
Open in a new tab
1

Sample intervals that met all 3 criteria from linear regression analysis: R2 ≥ 0.90, intercept not significantly different from 0 (P > 0.05), and slope not significantly different from 1 (P > 0.05).

Figure 1.

Figure 1.

Open in a new tab

Strength of association (R2) between the continuous sampling and tested instantaneous scan sampling intervals when estimating the amount of time (total minutes) that feedlot-housed lambs performed a behavior in a 14-h d (0600–2000 h). The dotted line at R2 = 0.90 represents the cutoff criteria for an instantaneous time interval to accurately estimate behavior compared with continuous sampling (R2 ≤ 0.90).

Feeding and standing behaviors were accurately estimated using a 5-min interval. In a feedlot environment, sheep reportedly feed for short bouts (0.5 min in this study and 1.8–3.8 min in the study of Rice et al. [2016b]) for a small proportion of the day (9.8% in this study and 6.7% in the study of Aguayo-Ulloa et al. [2014]). These findings are similar to those of previous studies with dairy calves and cows that concluded a 5-min interval was the longest interval to accurately estimate feeding behavior (Miller-Cushon and DeVries, 2011; Chen et al., 2016). However, our findings are more conservative compared with a study evaluating instantaneous intervals for feeding behavior in beef cattle housed in a feedlot (at 15-min intervals; Mitlöhner et al., 2001). These differences may be due to animal species as well as the length of behavioral recordings (three 2-h sessions in the study of Mitlöhner et al. [2001] compared with one 14-h session in this study). Standing is a posture generally associated with feeding and rumination and is not commonly reported in the literature as an independent behavior for confined lambs (Das et al., 1999; da Cunha Leme et al., 2013). Similar to feeding, standing was also performed for short bouts (0.3 min in this study and 9 min ruminating while standing in the study of Das et al. [1999]) for a moderate proportion of the day (21.1% in this study, 20–37% in the study of Teixeira et al. [2013], and 7.2% in the study of Aguayo-Ulloa et al. [2014]). The SD for feeding and standing behaviors increased as the sampling interval became longer, indicating greater variation and potential inaccuracies when approximating these behaviors with intervals longer than 5 min.

Drinking, locomotion, and oral manipulation were not accurately estimated using any of the time intervals tested in the present study. Instantaneous sampling is not recommended for discrete, rare behaviors of short durations (Martin and Bateson, 2007). Collectively these behaviors encompassed 4.3% of the daily time budget for the feedlot-housed lambs in this study (0.7, 2.0, and 1.6% for drinking, locomotion, and oral manipulation, respectively) and were performed in bouts of ≤10 s. These findings are similar to those of Das et al. (1999), who reported that feedlot lambs perform drinking in bouts of 1 min for 0.3% of a 24-h day. Additionally, Pullin et al. (2017) found that locomotion and drinking behavior of lambs housed on pasture were not accurately estimated using 5-, 10-, 15-, or 20-min intervals. Results from the present study support methodologies in previous literature, where researchers used continuous sampling to collect data on stereotypies or abnormal behavior in confined lambs (Vasseur et al., 2006; Aguayo-Ulloa et al., 2014; Teixeira et al., 2014).

Although the results from this study are similar to those from previous studies, we recognize that limitations from our methodology did arise. This study observed behaviors only during daylight hours and therefore did not account for nocturnal behavioral activities that may have occurred between 2000 and 0600 h. An observation period of a single 14-h day is a smaller time period than previous validation studies, which we strove to account for by using a similar or greater number of animals (18 lambs in our study, 18 cows in the study by Chen et al. [2016], 10 calves in the study by Miller-Cushon and DeVries [2011], and 8 heifers in the study by Madruga et al. [2017]). Lastly, the lambs used in this study were of varying weights, so they were grouped by size to reduce the likelihood that social dynamics (e.g., dominance) would influence behavior.

Despite these limitations, collecting behavioral data with video recording allows for a greater collection of quantity of data and greater reliability due to fewer observers that may not have been possible with the logistics of live observations. Validated instantaneous sampling intervals improve efficiency of behavioral data collection and maintain accuracy. This tool can be used by researchers interested in feedlot lamb behavior, such as investigating feedlot housing design, dietary characteristics, social groupings, welfare considerations (e.g., enrichment or human–animal interactions), etc.

Conclusion

The present study concludes that a 5-min instantaneous time interval accurately estimates lying, feeding, and standing behaviors for lambs housed in a feedlot environment. Continuous sampling is more appropriate for drinking, locomotion, and oral manipulation behaviors until future research evaluates instantaneous intervals shorter than 5-min for these behaviors. These results contribute toward the accuracy, efficiency, and transparency of methodologies for collection of behavioral data of lambs housed in a feedlot environment.

LITERATURE CITED

  1. Aguayo-Ulloa L. A., Villarroel M., Pascual-Alonso M., Miranda-de la Lama G. C., María G. A. 2014. Finishing feedlot lambs in enriched pens using feeder ramps and straw and its influence on behavior and physiological welfare indicators. J. Vet. Behav. 9:347–356. doi: 10.1016/j.jveb.2014.07.005 [DOI] [Google Scholar]
  2. Altmann J. 1974. Observational study of behavior: Sampling methods. Behaviour 49:227–266. doi: 10.1163/156853974X00534 [DOI] [PubMed] [Google Scholar]
  3. Bøe K. E., Berg S., Andersen I. L. 2006. Resting behaviour and displacements in ewes—Effects of reduced lying space and pen shape. Appl. Anim. Behav. Sci. 98:249–259. doi: 10.1016/j.applanim.2005.10.001 [DOI] [Google Scholar]
  4. Chen J. M., Schütz K. E., Tucker C. B. 2016. Technical note: Comparison of instantaneous sampling and continuous observation of dairy cattle behavior in freestall housing. J. Dairy Sci. 99:8341–8346. doi: 10.3168/jds.2016-11351 [DOI] [PubMed] [Google Scholar]
  5. da Cunha Leme T. M., Titto E. A. L., Titto C. G., Pereira A. M. F., Neto M. C. 2013. Influence of stocking density on weight gain and behavior of feedlot lambs. Small Rumin. Res. 115:1–6. doi: 10.1016/j.smallrumres.2013.07.010 [DOI] [Google Scholar]
  6. Daigle C. L., Siegford J. M. 2014. When continuous observations just won't do: Developing accurate and efficient sampling strategies for the laying hen. Behav. Processes 103:58–66. doi: 10.1016/j.beproc.2013.11.002 [DOI] [PubMed] [Google Scholar]
  7. Das N., Maitra D. N., Bisht G. S. 1999. Genetic and non-genetic factors influencing ingestive behavior of sheep under stall-feeding conditions. Small Rumin. Res. 32:129–136. doi: 10.1016/S0921-4488(98)00175-8 [DOI] [Google Scholar]
  8. Federation of Animal Science Societies (FASS) 2010. Guide for the care and use of agricultural animals in agricultural research and teaching. Consortium for developing a guide for the care and use of agricultural animals in agricultural research and teaching (3rd ed.). FASS, Champaign, IL. [Google Scholar]
  9. Ítavo C. C. B. F., Morais M. G., Costa C, Ítavo L. C. V., Franco G. L., da Silva J. A., Reis F. A. 2011. Addition of propolis or monensin in the diet: Behavior and productivity of lambs in feedlot. Anim. Feed Sci. Technol. 165:161–166. doi: 10.1016/j.anifeedsci.2011.02.020 [DOI] [Google Scholar]
  10. Jongman E. C., Rice M., Campbell A. J. D., Butler K. L., Hemsworth P. H. 2017. The effect of trough space and floor space on feeding and welfare of lambs in an intensive finishing system. Appl. Anim. Behav. Sci. 186:16–21. doi: 10.1016/j.applanim.2016.10.015 [DOI] [Google Scholar]
  11. Lauber M., Nash J. A., Gatt A., Hemsworth P. H. 2012. Prevalence and incidence of abnormal behaviours in individually housed sheep. Animals (Basel) 2:27–37. doi: 10.3390/ani2010027 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Ledgerwood D. N., Winckler C., Tucker C. B. 2010. Evaluation of data loggers, sampling intervals, and editing techniques for measuring the lying behavior of dairy cattle. J. Dairy Sci. 93:5129–5139. doi: 10.3168/jds.2009-2945 [DOI] [PubMed] [Google Scholar]
  13. Lehner P. N. 1992. Sampling methods in behavior research. Poult. Sci. 71:643–649. doi: 10.3382/ps.0710643 [DOI] [PubMed] [Google Scholar]
  14. Madruga A., Mainau E., González L. A., Rodríguez-Prado M., Ruíz de la Torre J. L., Manteca X., Ferret A. 2017. Technical note: Recording rules for behavioral studies in growing heifers fed high-concentrate diets. J. Anim. Sci. 95:2339–2344. doi: 10.2527/jas.2016.1037 [DOI] [PubMed] [Google Scholar]
  15. Martin P., Bateson P. 2007. Measuring behaviour: An introductory guide. Cambridge Univ. Press, New York: p. 48–61. doi: 10.1017/CBO9780511810893.006 [DOI] [Google Scholar]
  16. Miller-Cushon E. K., DeVries T. J. 2011. Technical note: Validation of methodology for characterization of feeding behavior in dairy calves. J. Dairy Sci. 94:6103–6110. doi: 10.3168/jds.2011-4589 [DOI] [PubMed] [Google Scholar]
  17. Mitlöhner F. M., Morrow-Tesch J. L., Wilson S. C., Dailey J. W., McGlone J. J. 2001. Behavioral sampling techniques for feedlot cattle. J. Anim. Sci. 79:1189–1193. doi: 10.2527/2001.7951189x [DOI] [PubMed] [Google Scholar]
  18. Nicory I. M. C., de Carvalho G. G. P., Ribeiro O. L., Silva R. R., Tosto M. S. L., Costa-Lopes L. S., Souza F. N. C., de Oliveira Nascimento C. 2015. Ingestive behavior of lambs fed diets containing castor seed meal. Trop. Anim. Health Prod. 47:939–944. doi: 10.1007/s11250-015-0812-9 [DOI] [PubMed] [Google Scholar]
  19. NRC 2007. Nutrient requirements of small ruminants: Sheep, goats, cervids, and New World camelids. Natl. Acad. Press, Washington, DC. [Google Scholar]
  20. Pullin A. P., Pairis-Garcia M. D., Campler M. R., Proudfoot K. L. 2017. Validation of scan sampling techniques for behavioral observations of pastured lambs. Anim. Welf. 26:185–190. doi: 10.7120/09627286.26.2.185 [DOI] [Google Scholar]
  21. Rice M., Jongman E. C., Borg S., Butler K. L., Hemsworth P. H. 2016a. Characterisation of shy-feeding and feeding lambs in the first week in a feedlot. Appl. Anim. Behav. Sci. 179:39–45. doi: 10.1016/j.applanim.2016.03.011 [DOI] [Google Scholar]
  22. Rice M., Jongman E. C., Butler K. L., Hemsworth P. H. 2016b. Relationships between temperament, feeding behaviour, social interactions, and stress in lambs adapting to a feedlot environment. Appl. Anim. Behav. Sci. 183:42–50. doi: 10.1016/j.applanim.2016.07.006 [DOI] [Google Scholar]
  23. Shreffler C., Hohenboken W. 1980. Circadian behaviour, including thermoregulatory activities, in feedlot lambs. Appl. Anim. Ethol. 6:241–246. doi: 10.1016/0304-3762(80)90025-5 [DOI] [Google Scholar]
  24. Teixeira D. L., Miranda-de la Lama G. C., Pascual-Alonso M., Aguayo-Ulloa L., Villarroel M., María G. A. 2013. A note on lamb's choice for different types of bedding materials. J. Vet. Behav. 8:175–179. doi: 10.1016/j.jveb.2012.06.007 [DOI] [Google Scholar]
  25. Teixeira D. L., Miranda-de la Lama G. C., Villarroel M., Escós J., María G. A. 2014. Lack of straw during finishing affects individual and social lamb behavior. J. Vet. Behav. 9:177–183. doi: 10.1016/j.jveb.2014.02.008 [DOI] [Google Scholar]
  26. Vasseur S., Paull D. R., Atkinson S. J., Colditz I. G., Fisher A. D. 2006. Effects of dietary fibre and feeding frequency on wool biting and aggressive behaviors in housed Merino sheep. Aust. J. Exp. Agric. 46:777–782. doi: 10.1071/EA05320 [DOI] [Google Scholar]

Articles from Journal of Animal Science are provided here courtesy of Oxford University Press

RESOURCES