Skip to main content
Dryad logo

Continuous bite monitoring method (GPEP)

Citation

Soares Bolzan, Anderson Michel et al. (2022), Continuous bite monitoring method (GPEP), Dryad, Dataset, https://doi.org/10.5061/dryad.573n5tb73

Abstract

Determining herbage intake is pivotal for studies on grazing ecology. Direct observation of animals allows describing the interactions of animals with the pastoral environment along the complex grazing process. The objectives of the study were to evaluate the reliability of the continuous bite monitoring (CBM) method in determining herbage intake in grazing sheep compared to the standard double-weighing technique (DW) method during 45-min feeding bouts; evaluate the degree of agreement between the two techniques; and to test the effect of different potential sources of variation on the reliability of the CBM. The CBM method has been used to describe the intake behaviour of grazing herbivores. In this study, we evaluated a new approach to this method, i.e., whether it is a good proxy for determining the intake of grazing animals. Three experiments with grazing sheep were carried out in which we tested for different sources of variations, such as the number of observers, level of detail of bite coding grid, forage species, forage allowance, sward surface height heterogeneity, experiment site, and animal weight, to determine the short-term intake rate (45 min). Observer (Pexp1 = 0.018, Pexp2 = 0.078 and Pexp3 = 0.006), sward surface height (Pexp2 < 0.001), total number of bites observed per grazing session (Pexp2 < 0.001 and Pexp3 < 0.001) and sward depletion (Pexp3 < 0.001) were found to affect the absolute error of intake estimation. The results showed a high correlation and agreement between the two methods in the three experiments, although intake was overestimation by CBM on experiment 2 and 3 (181.38 and 214.24 units, respectively). This outcome indicates the potential of CBM to determining forage intake with the benefit of a greater level of detail on foraging patterns and components of the diet. Furthermore, direct observation is not invasive nor disrupts natural animal behavior.

Methods

MATERIALS AND METHODS

Site, treatments and experimental design

Three independent studies were conducted in which we tested for different sources of variation: number of observers, level of detail of BC grid (Figure 1), forage species, forage allowance, sward surface height (SSH) heterogeneity, experiment site, and animal breed. Experiment 1 was carried out between 15 September and 15 October 2014, on an area of approximately 0.50 ha of self-seeding Italian ryegrass (Lolium multiflorum Lam.) at the experimental farm of the Federal University of Rio Grande do Sul, Brazil (30°05´27” S, 51°40’18” W). The area was divided into two paddocks of 0.25 ha (experimental areas) with salt and water freely available. In this protocol, there were no defined criteria for managing the sward pre-grazing structure. Only average sward surface height measurements were collected at the time of observation.

Experiments 2 and 3 were carried out at the Canguiri experimental station of the Federal University of Paraná, Brazil (25°26'30''S and 49°7'30''W). The experiments were established in a 0.3 ha experimental area of tall fescue cv. INIA Aurora (Schedonorus arundinaceus [Schreb.] Dumort) sown in June 2015 on a prepared seedbed, at 55 kg ha-1. Beginning in September 2015, the experimental area was managed under continuous stocking with sward surface height maintained between 10 and 15 cm, except just prior to and during the grazing events when the different pre-grazing SSH treatments were imposed. Experiment 2 was carried out between 24 June and 12 July 2016 and Experiment 3 was carried out between 15 and 24 November 2016. In Experiment 2, five homogeneous pre-grazing SSH (14, 17, 20, 23, and 26 cm) were evaluated in a randomized complete block design with four replicates. In Experiment 3, five levels of depletion (0, 20, 40, 60, and 70%) of average SSH (20 cm) were evaluated in a randomized complete block design with four replicates, through grazing with non-experimental animals prior to grazing sessions to measure behaviour. A more detailed description management protocol can be found in Szymczak et al. (2020).

Sward measurements

Five hundred SSH measurements were taken in the two experimental paddocks of Experiment 1 to characterize the vegetation structure. Mean SSH was 38.0 cm (±13.12 cm) and 38.7 cm (±12.98 cm) for paddocks 1 and 2. For Experiments 2 and 3, 150 points pre-grazing SSH within each sampling unit were measured. Pre-grazing SSH were 14.2 (±0.19), 17.3 (±0.20), 19.7 (±0.27), 22.8 (±0.28) and 25.9 (±0.26) cm, for treatment of 14, 17, 20, 23 and 26 cm in experiment 2, respectively. The measured pre-grazing SSH were 20.2 (±0.18), 16.5 (±0.52), 12.2 (±0.52), 8.3 (±0.48) and 5.9 (±0.37) cm, for treatments 0, 20, 40, 60 and 70% of depletion in experiment 3, respectively.

Intake and grazing behavior evaluations

Animals and experimental procedures

Procedures involving the experimental animals were conducted under the Guidelines for the Use of Animals (2012) and complied with ethical guidelines published by the International Society for Applied Ethology. All procedures involving animals were approved by the Commission for Ethics in the Use of Animals of the Sector of Agricultural Sciences of the Federal University of Paraná (024/2016).

Two methodologies were used simultaneously during grazing tests to measure short-term intake rate (STIR): the double weighing technique (DW) as the reference practice (Penning & Hooper, 1985) and the continuous bite monitoring (CBM) method (Agreil & Meuret, 2004; Bonnet et al. 2015). In Experiment 1, eight Texel ewes (42.07 ± 3.15 kg LW) were used. Sixty days before the data collection, ewes were allocated on an adjacent Italian ryegrass pasture for acclimation to forage and adaptation to observers and equipment. During the experimental procedure, animals were distributed in two groups of four testers per paddock, where two testers per paddock were used for evaluation of the continuous bite monitoring and double-weighing. After the evaluations, animals were placed back on the adjacent pasture for the remaining of the day. In Experiments 2 and 3, six White Dorper x Suffolk ewes were used with an average weight of 61.9 ± 5.5 kg. Two animals were chosen as testers, all previously adapted to the experimental procedure and maintained in an area similar and adjacent to the experimental paddocks.

Continuous bite monitoring

Experiment 1 involved three observers: one with previous experience on the methodology with wild herbivores and cattle (EO; Bonnet et al., 2015), and two new observers (TO1 and TO2) were trained by EO. Experiments 2 and 3 involved four different observers, all inexperienced. Prior to the beginning of the experiments, a mutual familiarization phase was adopted for three weeks. During this phase, animals were handled daily to acclimate to observers and protocols, and for observers to familiarize with pasture and grazing behavior. Once the tester animals were identified and familiarized with the evaluators, all bites observed were described and classified into categories based on the observation of the animals' intake behavior before the experiments under the following aspects: i) structural and nutritional distribution of the components in the sward; ii) the nature, size, density, and position of selected plant parts by animals, as a set of leaves, isolated leaves or inflorescences; and iii) handling (gathering herbage into the mouth, severing the herbage, ingestive mastication and swallowing, Laca et al., 1994). Simple codes were established for each bite category agreed upon by all observers, composing a grid of codes for the identification of bites in real-time (Fig. 1A and 1B). The level of detail for each BC grid differed based on SSH, phenological stages, plant density heterogeneity, and species diversity. For Experiments 2 and 3 the same grid was used, but the dimensions and masses varied (Annex 1 and 2).

After three weeks of training, both experienced and naïve observers were able to codify with confidence every bite observed (Bonnet et al., 2011). The three observers in Experiment 1 (one observer per tester animal per paddock in each session evaluated) and four observers on Experiments 2 and 3 (two observers per experiment, one observer per tester animal, and one tester per paddock) collected data by standing close to the animals (within 1 m), during 45-min grazing sessions. Thirty-two (Experiment 1) and twenty (Experiments 2 and 3) sessions were conducted. These sessions were blocked into morning and afternoon periods, arranged in a completely randomized design. After each grazing session had finished and while the animals remained in a common pen for determining insensible weight losses, each bite category was simulated (minimum 22 hand-plucks for each type of bite, for bite codes more frequent we replicated the samples). Samples were collected in paper bags and placed in a thermal box and weighed immediately after collection to estimate fresh matter (FM) intake. Intake was calculated as a sum of FM for all recorded bites. Data were registered using a Sony ICD-PX312 (Sony Corp., Japan) digital voice recorder and, subsequently, transcribed using the J Watcher software (www.jwatcher.ucla.edu).

Short-term intake rate

During the experimental period each day around 5:40 am (Experiment 1) or 6:30 am (Experiments 2 and 3), animals were moved to the handling area, fitted with harnesses for a total collection of urine and feces, and weighed at t1 (W1 = initial weight for estimating the rate of insensitive weight losses (H2O evaporation, CO2 and CH4 losses; RIWL pre-grazing). After being weighed, the animals remained in a common pen for 45 min without access to feed or water and then weighed again at t2 (W2 = final weight for pre-grazing RIWL and pre-grazing weight). Immediately after, all the animals were conducted and allotted to their paddocks for the 45-min grazing session (ET). Once the grazing session was finished, the animals were led to the handling area and the tester animals were weighed at t3 (W3 = post-grazing weight and initial weight for the post-grazing RIWL). The tester animals then remained in a common area without access to feed, water or shade for 45 min until being weighed at t4 (W4= final weight for post-grazing RIWL). The harnesses were then immediately removed, and the animals returned to the adjacent area. This same procedure was repeated in the afternoon (between 2:15 pm and 6:30 pm for Experiment 1 and between 2:30 and 6:30 pm, for Experiments 2 and 3). The animals were weighed using an electronic scale (MGR-3000 Junior, Toledo, Canoas, Brazil) with a capacity of 200 kg (5-g increments). Short-term intake rate (g FM min-1; Eq. 1) was calculated by measuring the weight change, corrected for insensible weight loss, and the time spent grazing, according to Penning & Hooper (1985).

STIR=W2-W1 (t2-t1)+W3-W4(t4-t3) X t2-t1ET        Eq.1

Statistical Analysis

The data were analyzed using the R software (R Development Core Team, 2016). Animal test group was the experimental unit. We systematically verified normality and homogeneity of the residuals. Pearson correlation was used as mean accuracy between the methods and was considered poor (<0.4), reasonable (0.4 to 0.6), good (0.6 to 0.8), or excellent (0.8 to 1.0). Bland–Altman plots were created to indicate the degree of agreement between the two techniques (Bland and Altman, 1999). The limits of agreement were determined by calculating the bias and standard deviation of the paired differences. The standard deviation was multiplied by the 1.96 quantiles of a normal distribution and then, the amount of the calculated average was added or subtracted to provide the upper or lower limits, respectively. Thus, the agreement limits were calculated as bias ± standard deviation. One Sample T-Test, at a significance level of 95%, was performed to check if there was a significant difference from zero, for the comparison between the methods.

Funding

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior

Ministry of Science, Technology and Innovation | Conselho Nacional de Desenvolvimento Científico e Tecnológico