Relationship between Temperate Grass Sward Characteristics and the Grazing Behavior of Dairy Heifers

Abstract

Sward architecture mediates ruminant grazing behavior in temperate grazing lands. Temperate grasses differ in their sward structure, which may influence the grazing behavior of cattle. We determined relationships between the grazing behavior of dairy heifers and the sward structure of the following temperate grasses: meadow fescue (Schedonorus pratensis (Huds.) P. Beauv.), orchardgrass (Dactylis glomerata L.), quackgrass (Elymus repens (L.) Gould), and reed canarygrass (Phalaris arundinacea L.). Vegetative-stage grasses were rotationally grazed by Holstein heifers (average initial body weight of 460 kg) during 5 day periods in the spring, summer, and fall of 2007 and 2008. The herbage dry matter (DM) allowance was twice the expected daily intake (11 kg DM animal⁻¹ d⁻¹). The sward characteristics were measured before grazing (e.g., the herbage height and mass, vertical distribution of leaf and stem fraction, and nutritive value). The grazing behavior of the heifers was quantified using automatic jaw movement recorders. In this study, the grass species had little effect on the grazing behavior. However, the bite rate was negatively correlated with the herbage mass, while the number of bites was positively correlated with the sward height and herbage mass. These results suggest that when herbage availability is not limited, grazing dairy heifers exhibit similar ingestive and rumination behavior across grass species and seasons, yet jaw movement dynamics may respond to the different characteristics of the swards. The results of this study provide the following benefits: (1) they inform managers about the jaw movement mechanics that can be expected of dairy heifers in temperate forage systems, showing that they are not limited by herbage allowance, and (2) they provide insight for future studies that employ on-animal sensors to evaluate foraging dynamics and animal performance outcomes in temperate forage pasture systems.

1. Introduction

Herbage dry matter (DM) intake (DMI) is a key determinant of animal performance in pasture-based systems [1,2]. Herbage DMI is dependent not only on the herbage nutritive value, but also on the herbage availability [3] and accessibility [4], with the latter being directly related to the sward structure [5]. The sward structure depends on the sward surface height and the organization and proportion of the leaves and stems of the individual plants, which influences the sward bulk density and the tiller density [6]. For example, a greater sward surface height of perennial ryegrass (Lolium perenne L.) has been shown to increase the intake of dairy cows; however, decreasing the sward surface height (from 19 to 10 cm) over four days of grazing resulted in increased grazing time and decreased rumination [7].

Sward height has been shown to significantly affect bite mass [8,9,10,11]. Laca et al. [12] reported that bite mass varied less than bite dimensions due to compensatory effects, among other factors, including the bite area and bite depth. They noted that bite mass was more sensitive to surface height than to bulk density in steers that were offered hand-constructed swards. Conversely, Flores-Lesama et al. [13] found that dairy cows had greater milk production when grazing a perennial ryegrass cultivar with greater leaf mass in the upper strata. These conflicting studies suggest that other sward characteristics may be involved; therefore, research evaluating the sward structure mediation of grazing behavior is needed.

The concepts of grazing behavior that are used to make management decisions [14,15] are often derived from small-scale [8,9,16] studies. However, such studies may not be fully representative of field conditions and must be validated at that level. The handful of studies that do exist at the field level were conducted with monocultures of grass or simple grass–legume mixtures that are not typical of temperate pastures [17,18,19]. For example, Soder et al. [20] evaluated the sward structure and grazing behavior of dairy cows in complex mixtures of temperate grasses and legumes at the pasture scale but were unable to separate out the effects of the varying levels of plant species’ diversity from the effects of the generic sward structure. In a short-term study using hand-constructed sward boxes, Soder et al. [8] evaluated the bite mass of monocultures of the same grass species that were evaluated in the current study in short-term grazing sessions, using sward boxes. They found, under the conditions of the sward boxes, that individual grass species had little influence on bite mass. Rather, generic sward structure features such as bulk density, surface height, leaf length, and herbage mass had a greater influence on fine-scale grazing behavior, as evidenced by the differences in bite mass irrespective of the species over the two years of the sward box study.

Knowledge about how grass sward structure characteristics, as affected by forage species and differences in growth structures by species, influence grazing behavior would permit producers to utilize pasture grasses, tailor animal management and supplementation more efficiently, and inform pasture renovation strategies. The bulk of the literature has been conducted on perennial ryegrass, which is not a common forage species in temperate regions of the United States, such as the Northeast or the Midwest, due to the climate. Little is known about how the temperate grass species commonly used in temperate pasture systems affect grazing behavior from a species, sward structure, or seasonal perspective. Knowledge about the individual grass species across the grazing season at the field scale is needed to capture the potentially greater temporal and spatial variation in grazing behavior patterns. This has not been possible in the previous, small-scale studies. Furthermore, such information can then be used to develop and evaluate forage mixtures to improve the primary and secondary productivity of grazing systems.

We hypothesize that the sward structure of four common temperate grass species, which are found in temperate pasture systems, mediates the grazing behavioral patterns at the field (pasture) scale. Our objective was to determine how the sward characteristics of four diverse, erect-growing temperate grasses influenced the grazing behavior of dairy heifers at the pasture scale. This study is a companion to Brink and Soder’s study [21] as it assesses grazing behavior in response to structural variations and the allowance of herbage. To provide additional background and context to our behavioral study outcomes, we have plotted the results (Table 3; [21]) and have added the post-grazing sward height, the post grazing herbage mass, and the herbage allowance in each season and for each grass-species treatment (Figure 1).

2. Materials and Methods

2.1. Pasture Treatments and Management

The grazing experiment was conducted in the spring, summer, and fall of 2007 and 2008 at the US Dairy Forage Research Center farm near Prairie du Sac, WI (43°20′24″ N–89°43′12″ W), on an 8 ha pasture comprised of the following three soils: McHenry silt loam (fine-loamy, mixed, superactive, and mesic Typic Hapludalf); Richwood silt loam (fine-silty, mixed, superactive, and mesic Typic Argiudoll); and St. Charles silt loam (fine-silty, mixed, superactive, and mesic Typic Hapludalf). A composite sample of the soils at a measurement depth of 10 cm had a mean pH of 6.9, 50 ppm P (Bray P1), and 200 ppm K. Four replicates of 0.4 ha paddocks that comprised monocultures of ‘Bartura’ meadow fescue (MDF; Schedonorus pratensis (Huds.) P. Beauv.), ‘Bronc’ orchardgrass (ORG; Dactylis glomerata L.), common quackgrass (QGR; Elymus repens (L.) Gould), and ‘Palaton’ reed canarygrass (RCG; Phalaris arundinacea L.) were arranged in a randomized complete block design. Orchardgrass is one of the most common temperate grass species found in temperate-region pastures. Meadow fescue is gaining popularity in these regions due to palatability and nutritive quality. Quackgrass and RCG are rhizomatous grasses that possess pseudostems, thereby, they have a very different sward structure (i.e., low stature) from the first two grass species. The latter three grass species all have sparse research available on their relationships with herbivore ingestive behavior. All grass species were seeded a year before this study, in May 2006. Environmental conditions during seeding (2006) were optimum and resulted in weed-free stands of each grass species. They remained in that monoculture condition throughout the next 2 yrs. Precipitation (mm) for spring sampling periods (May) were 29 and 59 mm in 2007 and 2008, respectively. Summer (July) precipitation was 29 and 159 mm for 2007 and 2008, respectively. Fall sampling period (September) precipitation was 76 and 49 mm in 2007 and 2008, respectively.

2.2. Animal Management and Sampling Schedule

All standards for animal handling and care for this experiment were authorized by the Research Animal Resources Committee (RARC) of the University of Wisconsin-Madison (Protocol #A003221).

Prior to the three sampling periods (spring, summer, and fall) over two years (2007, 2008:

Table 1. Sampling schedule at the US Dairy Forage Research Center farm near Prairie du Sac, WI, USA.

Sampling Period	Acclimation Period		Experimental Period
	2007	2008	2007	2008
Spring	30 April	4 May	7 May	11 May
Summer	1 July	6 July	8 July	13 July
Fall	16 September	14 September	23 September	21 September

), 32 bred Holstein heifers (14 to 16 months of age, with average initial body weight of 465 and 453 kg for 2007 and 2008, respectively) were divided into four groups of eight animals with similar total body weight. Each heifer group was randomly assigned to graze two replicates of the assigned grass monoculture during a 7 d acclimation period. Following acclimation, each heifer group was further divided into two sub-groups of four animals of similar total body weight and assigned to graze the remaining two replicates of the same grass over the next 5 d experimental period. Initiation of experimental periods occurred when grasses were at maturity stage of V3 to V4 [22], although the stage of some tillers had reached E0, or the onset of stem elongation during the experimental period. To reduce the irregular effect of animal urine- and feces-derived N on herbage growth, each paddock was fertilized with 40 kg N ha⁻¹ as NH₄NO₃ 21 d before summer and fall study periods in 2007 and before the spring and fall periods in 2008.

At the conclusion of the spring and summer experimental periods in each year, heifers were rejoined into one group and continuously grazed a large (>50 ha) pasture of Kentucky bluegrass (Poa pratensis L.), ORG, and QGR until 7 d before the next experimental period, at which time they were once again randomly divided into four groups of eight animals and acclimated to a randomly assigned treatment grass, as described above.

2.3. Sward Measures and Laboratory Analyses

One day before each experimental period began (during the last day of acclimation), grass sward structure was measured by hand-harvesting herbage at 10 cm intervals to a 10 cm residual sward height within a 25 by 100 cm aluminum quadrat, at three random locations within each paddock. Herbage (n = 3 clippings/paddock/experimental period) was composited by sward layer, placed in a paper bag, dried at 65 °C for 48 h and weighed. The sample was separated into the leaf and stem fraction, and the partial DM herbage mass of each fraction was used to calculate sward leaf and stem bulk density (mg DM cm⁻³). Stem fraction consisted of the leaf sheath in our measurements. Herbage mass within the first area to be grazed was measured with a rising plate meter (0.21 m⁻²; [23]), at 12 random locations within each paddock. Dry matter herbage collected from all pastures was regressed against respective plate measurements to develop an equation for prediction of herbage mass. As additional measurements were made during each study period, the regression equation was updated and the final equation describing the relationship between plate meter height (cm) and herbage mass (g DM below the plate) was y = 2.56x − 6.5; r² = 0.73. The paddock was then divided into sub-paddocks of sufficient area to provide the four heifers with twice their expected daily herbage DMI of 11 kg DM animal⁻¹ d⁻¹ (2.4% of body weight; [24]), or approximately 0.43 kg herbage DM kg⁻¹ animal live weight [25]. Expected daily herbage DMI corresponds to measurement at ground level. Sward height of the sub-paddock was estimated by measuring mean extended leaf height along a 40 m diagonal transect at 2 m intervals (n = 20 leaf height measurements per sub-paddock) [26]. Extended leaf height provides a proxy for the portion of the sward accessible to a grazing animal.

During each 5 d experimental period, each group of four heifers grazed one sub-paddock (total of 5 sub-paddocks per replicate and treatment) for 24 h, beginning at 800 h. Approximately 20 h before the next sub-paddock was grazed, herbage mass and mean sward height were determined as described above. Immediately after each sub-paddock was grazed, residual herbage mass and mean sward height were measured in the same manner as the pregraze measurements, in locations where animals had not defecated or bedded.

At the mid-point of the spring, summer, and fall experimental periods of 2008, 15 cm diameter circular soil cores, with intact grass tillers and roots, were removed to 25 cm depth from three random locations within each paddock. The cores were kept on a laboratory bench at 16 °C during analysis (less than 4 h duration). Within each core, the second collared leaf was removed from two tillers, a 5 cm segment was cut from the middle of each leaf blade, and a 0.5 mm section was excised from each end using a microtome. Using a stereo microscope, leaf thickness was measured at three points on each of the following excised sections: at the vascular bundle farthest from the midrib of the leaf, at the midrib of the leaf, and at the mid-point between the mid-rib and the vascular bundle. The mean of these three measurements were considered the mean leaf cross-sectional thickness. The remainder of each leaf blade was scanned using a CID Bio-Science (Camas, WA, USA) 203 leaf area meter^® to determine its length, width, area, and perimeter and was placed lengthwise between two pneumatic clamps, set 2 cm apart, with a closure pressure of 210 kPa. An MTS Insight 1^® electromechanical testing system (Eden Prairie, MN, USA), equipped with a 100 N load cell, measured the force as the leaf segment was pulled apart at 5 mm min⁻¹ until it severed. Using leaf cross-sectional thickness, length, width, area, and perimeter measurements, the energy (N·mm) required to sever the leaf blade and the leaf tensile strength (MPa) of the leaf blade were calculated using MTS Testworks 4^® software.

A 5 g subsample of dried herbage of each grass, with the original proportion of sward leaf and stem fraction, was subject to a technique to determine resistance to particle size reduction, according to Casler et al. [27]. Briefly, samples were tumbled in a hobby-size rock polishing machine with 200 stainless-steel ball bearings (100 each with a 9.5 or 12.7 mm diameter) for 30 s (s). After tumbling, contents were carefully poured and brushed onto a 6.7 mm screen, which was placed on a 1 mm screen, which was then placed on a solid sieve plate. The stack of sieves was placed on a mechanical shaker for one minute, after which the dried herbage on each sieve was collected and weighed. The particle size reduction index was computed as the percentage of the initial sample weight that passed through the 1 mm screen.

A 10 g subsample of the composited dried herbage of each grass that had the original proportion of sward leaf and stem fraction was ground using rotating blades to pass a 1 mm Wiley mill screen, and then stored in a plastic bottle. Ground samples were analyzed for DM, N [28], neutral detergent fiber (NDF) [29], and in vitro NDF digestibility (NDFd by calibrated near-infrared reflectance spectroscopy) [30].

2.4. Grazing Behavior

Grazing behavior data were collected from 2 heifers per treatment (one from each of the two replicates), over 4 days during the experimental period for each of the three periods in both years. The grazing behavior equipment was fitted to the animals while they were restrained in a holding chute before they began grazing the second sub-paddock. The grazing recorder contained a computer processor that received signals of jaw movements via a graphite-filled, elastic jaw band [31]. Grazing and ruminating behaviors were recorded during four, 24 h periods. Before animals were moved to a new paddock each day, heifers wearing recorders were walked to the handling area adjacent to the grazing paddocks for equipment adjustments and memory card and battery replacement. This process took approximately 1 h each day. Behavior data were analyzed with GRAZE software (version 0.8; [32]). Grazing, ruminating, and idling times (min d⁻¹), and numbers of bites and chews (number d⁻¹) were quantified based on the amplitude, frequency, and shape of the jaw movement waveforms, as determined by the GRAZE software [32]. Calculated variables included total grazing jaw movements (sum of bites plus chews) and bite rate (bites min⁻¹ during active grazing sessions).

2.5. Statistical Analyses

Data were analyzed as a randomized complete block design, by the mixed model procedure of SAS^®, with block considered as a random intercept, and with grass species, season, and year considered as fixed effects [33]. Each 24 h period was considered a separate data point (not repeated measures), therefore, n = 8 for each treatment per period (2 heifers × 4 days), 3 seasons, and 2 years of data. This level of data collection (2–3 test animals per treatment) was validated and used by the developer of the behavior recorder (Institute of Grassland and Environmental Research, North Wyke, UK; [31]). Differences in grazing behavior among grass species were reported with specific p-values, some of which were slightly greater than the 5% type I error rate [34]. LSD values were computed using the exact p-values reported in the tables. Pearson correlation coefficients were calculated for grazing behavior and selected sward variables using replicate means (n = 24).

3. Results

3.1. Physical Sward Characteristics and Herbage Allowance of Grass Species

There were no significant (p > 0.10) grass–season–year interactions for the postgraze sward height, the postgraze herbage mass, or the herbage allowance (Table S1). There were no significant differences (p > 0.10) in the postgraze sward height among the grasses during the spring of 2007, nor during the spring of 2008 (Table S2). The sward height of RCG was greater than that of the other grasses during the summer (p = 0.005), with the exception of ORG, which was greater in the summer of 2008 (p = 0.012). The postgraze sward height of RCG exceeded the heights of the other grasses in the fall period of 2007 (p = 0.012) but not in the fall period of 2008 (p > 0.10). The sward height of QGR was frequently less than that of the other grasses across the year (Figure 1)

The herbage residual (postgraze herbage mass) was not different among the four grasses in spring 2007 (p > 0.10), while the postgrazing mass of RCG was greater than the other grasses in spring 2008, with the exception of MDF (p = 0.061) (Table S2; Figure 1). In the remaining summer (2007 p = 0.003, 2008 p = 0.004) and fall (2007 p = 0.005, 2008 p = 0.087) periods, RCG residual was frequently greater than that of the other three grasses.

The herbage allowance of RCG was not greater than that of the other grasses in spring 2007 (p > 0.10) but was greater than that of the other grasses in summer 2007 (p = 0.032), spring 2008 (p = 0.026), summer 2008 (p = 0.008), and fall 2008 (p = 0.083). The herbage allowance of QGR was less than that of the other grasses in fall 2007 (p = 0.070) (Table S2; Figure 1).

3.2. Grass Species’ Effects on Behavior

RCG ranked the lowest in ruminating time when expressed either as min d⁻¹ (p = 0.056) or as a percentage (%) of the total time (p = 0.055;

Table 2. Effect of temperate grass species (meadow fescue—MDF; orchardgrass—ORG; quackgrass—QGR; reed canarygrass—RCG) on ruminating time, idling time, number of chews, and total grazing jaw movements (TGJM) of grazing dairy heifers. Means are average of three seasons and two years.

	Grass Species				LSD	p-Value
	MDF	ORG	QGR	RCG
Ruminating time
min d−1	391	395	424	358	47	0.056
% of total time	27	27	29	25	3	0.055
Idling time
min d−1	659	659	621	689	74	0.283
% of total time	46	46	43	48	5	0.282
Chews d−1	14,814	⋯ *	11,935	11,875	4416	0.602
TGJM d−1	28,533	30,188	29,426	26,431	8903	0.817

* Missing data point.

), and it was significantly lower than QGR. The grass species had no effect on the idling time, total chews, total bites, or total grazing jaw movements recorded for the dairy heifers.

3.3. Seasonal Effects on Behavior

The ruminating time (min d⁻¹ or % of total time) was lower (p = 0.080) in the spring, while the idling time (min d⁻¹ or % of total time) was greater (p < 0.001) in the spring than it was during the other two seasons (

Table 3. Effect of season on ruminating time, idling time, number of chews, and total grazing jaw movements (TGJM) of grazing dairy heifers. Means are average of two years.

	Season			LSD	p-Value
	Spring	Summer	Fall
Ruminating time
min d−1	353	405	418	41	0.080
% of total time	25	28	29	3	0.080
Idling time
min d−1	782	638	551	64	≤0.001
% of total time	54	44	38	4	≤0.001
Chews, # d−1	11,847	11,680	15,166	2317	0.034
TGJM, # d−1	22,316	29,786	33,832	7710	0.021

). In addition, the number of chews d⁻¹ was highest (p = 0.034) during the fall, while the total grazing jaw movements d⁻¹ ranked lowest (p = 0.021) during the spring.

3.4. Interaction of Grass Species–Season on Behavior

A significant grass species–season interaction occurred for the grazing time (p = 0.070; min d⁻¹ and % of total time), the number of bites d⁻¹ (p = 0.054), and the bite rate (p = 0.066;

Table 4. Grass species–season interaction for grazing time, bites per day, and bite rate of four grass species (meadow fescue—MDF; orchardgrass—ORG; quackgrass—QGR; reed canarygrass—RCG). Means are average of two years.

	Grass Species				LSD 1	p-Value
	MDF	ORG	QGR	RCG
Grazing time (min d−1)
Spring	327	298	261	329	79	0.070
Summer	417	350	437	373
Fall	422	498	489	479
Grazing time (% of total time)
Spring	23	21	18	23	5	0.070
Summer	29	24	30	26
Fall	29	35	34	33
Bites per day
Spring	12,322	10,387	10,798	11,231	59	0.054
Summer	17,165	16,969	19,957	15,049
Fall	13,436	16,750	27,060	23,237
Bite rate (bites min−1)
Spring	33	32	36	34	13	0.066
Summer	42	40	46	34
Fall	32	34	57	49

¹ LSD—least significant difference for comparison of grass species within a season.

). The grazing time ranked lowest for ORG and QGR in the spring, highest for QGR and MDF in the summer, and lowest for MDF in the fall. The number of bites d⁻¹ was highest for MDF and RCG in the spring, highest for QGR in the summer, and highest for QGR and RCG in the fall. The bite rate was highest for QGR in the spring and fall, and lowest for RCG in the summer.

3.5. Sward Physical Characteristics of Grasses

A grass species—season interaction occurred for the leaf tensile strength (p = 0.062), the energy required to sever the leaf blade (p = 0.078), and its resistance to particle size reduction (p < 0.001,

Table 5. Grass species–season interaction for leaf tensile strength, energy required to sever the leaf blade, and resistance to particle size reduction of four grass species (meadow fescue—MDF; orchardgrass—ORG; quackgrass—QGR; reed canarygrass—RCG). Means are average of two years.

	Grass Species				LSD 1	p-Value
	MDF	ORG	QGR	RCG
Leaf tensile strength (MPa)
Spring	224	334	211	253	46	0.062
Summer	308	522	443	391
Fall	280	391	341	303
Energy required to sever leaf blade (N·mm)
Spring	5.02	6.47	5.41	9.26	1.51	0.078
Summer	5.60	7.83	3.94	6.53
Fall	6.23	5.34	3.67	5.31
Resistance (particle size reduction index, %) †
Spring	17.7	11.5	6.4	6.1	2.4	≤0.001
Summer	14.4	5.4	14.5	9.5
Fall	12.8	6.3	17.0	14.0

^† [26]. ¹ LSD—least significant difference for comparison of grass species within a season.

). The leaf tensile strength was highest for ORG, but lowest for MEF (or among the lowest) in all three seasons. The energy required to sever the leaf blade was lowest for QGR in all three seasons. The seasonal peaks of energy required to sever RCG occurred in the spring, for ORG they occurred in the summer, and for MDF they occurred in the fall. The resistance was highest for MDF in the spring and was lowest for both QGR and RCG in the spring. During the summer, the resistance was highest for both QGR and MDF, and lowest for ORG; however, during the fall season the resistance was ranked highest for QGR and lowest for ORG.

3.6. Relationships between Grazing Behavior and Sward Characteristics

The ruminating time was negatively correlated (p ≤ 0.10) with the following five sward characteristics: pregraze sward height, pregraze herbage mass, stem bulk density, leaf tensile strength, and energy required to sever the leaf blade (

Table 6. Correlation coefficients and associated probability for grazing behavior variables (ruminating time—RT, bite rate—BR, and number of bites) and pregraze sward physical characteristics, and nutritive value of four grass species.

	Pregraze		Sward Bulk Density		Leaf Tensile Strength
	Sward Height	Herbage Mass	Leaf	Stem		Energy ŧ	Resistance §	N	NDF †	NDFd
RT (min d−1)	r = −0.46 0.024	r = −0.46 0.071	r = −0.27 0.204	r = −0.57 0.004	r = −0.50 0.013	r = −0.43 0.037	r = −0.25 0.246	r = −0.12 0.568	r = 0.22 0.309	r = 0.28 0.187
BR (bites min−1)	r = −0.03 0.877	r = −0.41 0.050	r = −0.32 0.132	r = −0.02 0.942	r = −0.22 0.291	r = 0.14 0.500	r = −0.26 0.214	r = −0.15 0.480	r = 0.16 0.456	r = 0.18 0.400
No. Bites	r = 0.52 0.009	r = 0.60 0.002	r = 0.06 0.792	r = 0.10 0.627	r = −0.12 0.585	r = −0.10 0.629	r = 0.18 0.408	r = −0.07 0.762	r = −0.08 0.693	r = 0.18 0.403

^† NDF—neutral detergent fiber content; NDFd—neutral detergent fiber content digestibility. ^ŧ Energy required to sever leaf blade (N·mm); ^§ (particle size reduction index, %).

). The bite rate was negatively correlated (p = 0.050) with the pregraze herbage mass. The number of bites was positively correlated with the pregraze sward height (p = 0.009) and the pregraze herbage mass (p = 0.002).

4. Discussion

When these same four grass species were grown in micro-sward boxes, Soder et al. [8] found that the bite mass was impacted by the sward structure but not by the grass species, as evidenced by the differences in the bite mass between the different years, but not among grass species within a year. However, while Soder et al. [8] provided critical preliminary data, their study was conducted under artificial micro-sward box conditions, which may not fully mimic the pasture situations in competitive grazing and the changing environmental conditions, as the structure of these artificial swards that were grown in a greenhouse were not identical to those found in pasture settings. While the bite mass was not quantified in the current study, other behavior parameters may provide an indicator of bite mass. Although there has been evidence of a ‘chew–bite’ where grazing ruminants are simultaneously harvesting one bite while chewing the former bite [35,36], grazing ruminants generally do not regularly bite and chew an accumulated bite at the same time. If the bite mass is high, an animal’s buccal cavity will fill after a small number of bites, and the animal must pause to chew and swallow the accumulated herbage [37]. Due to this increased handling time, in such cases the bite rate will be decreased. This results in a negative relationship between the bite rate and the bite mass, and between the bite rate and the herbage density [37], which agrees with the results of the current study, where the increased bite rate in the QGR, as well as the negative correlation between the bite rate and the herbage mass, corresponded with a lower pregraze sward height and herbage mass.

In a companion study to our current study, Brink and Soder [21] showed that the herbage DMI of these same heifers on the same pastures was similar among the grasses in four of the six seasons of that study. The data for Brink and Soder’s study agree with the grazing behavior data of the current study, where few differences were noted. One possible explanation for the lack of differences between the herbage DMI and the grazing behavior is that the pregraze herbage allowance was set at approximately twice the expected herbage DMI (11 kg animal⁻¹ d⁻¹), which would not restrict intake. The herbage allowance was set high so that it did not confound any behavior–sward structure relationships. However, had the herbage allowance been more limiting, there may have been greater contrasts in the behavior related to the sward structure or grass species. Alternatively, using the sward height alone as a predictor of the DMI is not without risk, for if the sward structure is such that the top canopy is sparse [19,38,39], the DMI may still be limited.

Brink and Soder [21] found that stem bulk density was negatively associated with herbage DMI. In the current study, stem bulk density was negatively correlated with ruminating time, which may have resulted from the decreased herbage DMI, as there was less total herbage to process via rumination. Additionally, the increased energy required to sever the leaf blade and the leaf tensile strength may also have negatively affected rumination (through decreased herbage DMI), as noted by the negative correlations of these sward characteristics with rumination time. The effect of the stem bulk density was most pronounced in the spring and summer periods, resulting in a decreased herbage DMI [21], which also coincided with a decreased grazing time in the current study. These data agree with Benvenutti et al. [40], who reported that one of the major determinants of herbage DMI was stem density and its physical properties in cattle grazing on artificial swards of guineagrass (Megathyrsus maximus).

Grazing cows may have the capacity to adjust their short-term bite dimensions within and across meals to maintain consistent long-term DMI [18]. The results of the current study are a mean of 24 h periods, not of individual grazing bouts. This means that the heifers may have varied their grazing behaviors according to the canopy stratum within each sward, and on a small temporal scale, such that the heifers may have grazed intensively for brief bursts and ingested greater bite masses during particular grazing bouts (i.e., when first turned out on the pasture). However, as the canopy was depleted, the heifers may have decreased their grazing activity because they had increased their time searching for more rewarding bites [6]. However, these temporal variations in grazing behavior would not have been detected in the 24 h means [41].

Barrett et al. [42] established field plots of ryegrass (Lolium perenne, multiflorum, and boucheanum Kunth cultivars) that varied in bulk density and that were then grazed by dairy cows. While the taller swards of the vegetative perennial ryegrass typically had lower bulk densities than the shorter swards, there were no differences among the swards regarding bite mass. This indicates that the cows compensated by grazing to a greater bite depth on the taller swards. This mechanism could partially explain the lack of differences in grazing behaviors among the grass species in our current study as the heifers may have adjusted their bite dimensions to result in a constant amount of DM harvested per bite, or a constant daily herbage DMI, as noted by Brink and Soder [21]. In the current study, the number of bites was negatively correlated with the sward height and herbage mass. Based on the results from previous sward box studies e.g., [40], the heifers may have adjusted their bite depths to selectively graze only the leafy upper strata of the taller swards to consume the higher-quality leaf tips [18,43,44]. This, in turn, may have resulted in a lower bite mass and the corresponding increase in the number of bites needed to meet their intake requirements, despite the plant’s architectural barriers. Bite mass can be affected by bulk density, particularly as paddocks are depleted during grazing, as has been shown by Casey et al. [6]. However, McGilloway et al. [45] reported no relationship between bite mass and bulk density in tall, ungrazed perennial ryegrass swards. They suggested that the bulk density of the grazed sward canopy differed from the bulk density of the entire sward, as this is influenced mainly by the amount of DM in the lower strata of the paddock.

While the grass species–season interaction was significant for the resistance to particle size reduction in the current study, this effect was always the smallest of the fixed- effect sizes and it is likely that it has little biological significance. Contrasting environmental conditions during the various seasons of the year may have contributed to the grass species–season interaction in our study. In 2007, the cumulative precipitation before and during the spring and summer experimental periods was near or below normal, but was nearly 200% above normal before and during the fall period [21]. In contrast, the cumulative precipitation before and during the spring and summer periods of 2008 was 50 to 200% above normal, but was 50% below normal before and during the fall period. The mean minimum and maximum temperatures were usually above normal in 2007, but near normal in 2008.

The sward height and the herbage mass have been shown to impact the behavior of grazing cattle [46]. The productivity of each grass species relative to the others varied with the season in the current study. Reed canarygrass is regarded as one of the highest-yielding temperate grasses, particularly during the summer months [47], and is adapted to a wide range of soil conditions. In the current study, RCG had the greatest pregraze sward height and herbage mass [21]. The result was a lower number of bites per day⁻¹ and a lower bite rate for RCG, particularly during the summer, when compared to the other grass species. Conversely, while QGR was relatively similar to other grass species in its pregraze sward height and mass during the spring, QGR had a lower sward height and mass during the summer and fall, when compared with the other grass species. The greater number of bites d⁻¹ and bite rate observed for QGR during the fall agrees with the data, since heifers would need to take a greater number of bites of a lower mass bite⁻¹ to maintain a similar herbage DMI, as was observed by Brink and Soder [21]. The leaf and stem bulk densities also depended on the grass species and season in our study. The RCG had a lower stem bulk density in the spring, but a greater stem bulk density during the summer and fall, which resulted in the stem bulk density constituting a significant proportion of the pregraze herbage mass during the summer and fall. Orchardgrass had the highest leaf bulk density in the summer of 2008, but not of 2007 [21], which may have corresponded to the decreased grazing time during summer 2007 and resulted in the significant grass–season interaction. Yet despite these differences in the sward characteristics and growth patterns of the four grass species, the heifers were able to maintain a similar herbage DMI [21] and grazing behavior in the current study.

5. Conclusions

The grazing behavior of dairy heifers grazing the monocultures of four temperate grass species (meadow fescue, orchardgrass, quackgrass, and reed canarygrass) in 0.4 ha paddocks differed due to the morphological differences among the grass species. Negative relationships between the bite rate and the herbage mass, as well as a negative relationship between the energy required to sever leaf tissue and ruminating time indicated that the sward characteristics of temperate grass species in this study influenced ingestive and rumination behavior. Despite these influences on ingestion via a reduced bite rate and rumination time, our data suggests that dairy heifers are able to maintain similar and adequate dry matter intake (also reported in a companion study, [21]) under conditions where the herbage allowance is twice the expected dry matter intake on these high-quality vegetative pastures. These findings contribute to our understanding of the sward structure–grazing behavior relationship among common pasture forage species in the temperate climate of the United States’ dairy systems. Because the sward height was partially confounded by the grass species in this study, the conclusions about species differences must be tempered by the fact that the species differed in structure. Due to the large morphological differences among these species, it must be recognized that it may not be possible to completely separate the effects of the species from the effects of the sward structure. Our findings have extended comprehension of the grazing efforts and jaw movement mechanics employed by Holstein dairy animals, beyond the former artificial sward box studies of confined animals, which grazed on hand-constructed temperate grass swards. The results of this study provide the following benefits: (1) they inform managers about the jaw movement mechanics that can be expected of dairy heifers in temperate forage systems that are not limited by herbage allowance, and (2) they provide insight for future studies that employ on-animal sensors to evaluate the foraging dynamics and animal performance outcomes in temperate forage pasture systems.