Grazing behavior and production characteristics among cows differing in residual feed intake while grazing late season Idaho rangeland

Abstract

The objectives were to determine whether cows previously classified during a postweaning test as either low or high residual feed intake (LRFI or HRFI) differed in BW, BCS, and winter grazing activity while consuming poor-quality forage. Thirty Hereford × Angus (LRFI = 16; HRFI = 14) 2-yr-old mid- to late-gestation cows (pregnant with second calf) grazed sagebrush steppe for 78 d beginning 29 September 2015. BW and BCS were collected before and after grazing. Five cows of each RFI classification were fitted with global positioning system (GPS) collars on 16 November 2015 with data collection commencing 3 d later and continuing for 25 d in a 323-ha pasture. The GPS units collected location coordinates every 2 min from which total daily travel distance (DTD) was calculated. Visual counts for bite rate were obtained from collared cows over 8 d. Coordinate data, daily bite rate, BW, and BCS were analyzed as repeated measures using a mixed model, which included RFI group, day, and RFI group × day as fixed effects and cow within RFI group as the random effect. Change in BW and BCS was analyzed by ANOVA with RFI group as the main effect. Cow BCS and BW differed for both day (P < 0.0001) and day × RFI (P < 0.05). Body condition was less (P < 0.05) in LRFI cows at the beginning (5.8 ± 0.13 vs. 6.2 ± 0.14 BCS), but similar (P = 0.67) to HRFI at the end of the study (4.6 ± 0.13 vs. 4.6 ± 0.14). BW among the RFI groups did not differ (P = 0.20) prior to going to range. However, BW-change and BCS-change differed (P < 0.05) between RFI groups. Not only did the LRFI cows lose less BW (−50.0 ± 5.41 kg vs. −66.6 ± 5.78 kg) over the trial, they also were less variable with respect to BW loss. Cows did not differ (P > 0.21) by RFI for DTD or bite rate, but day was significant (P < 0.0001) with cows increasing bite rate as the season of year progressed (55.2 ± 5.63 bites/min for day 4 vs. 84.8 ± 5.32 bites/min for day 21) and increasing DTD as snow storms occurred. Although LRFI cows were leaner than HRFI cows at the commencement of the project, they lost less BW in a late season rangeland environment.

Introduction

Residual feed intake (RFI) is expressed as the difference between expected feed intake (based upon BW and growth) and actual feed intake (Koch et al., 1963). Although intensive research and industry adoption (Wulfhorst et al., 2010) of selection for RFI has occurred over the last two decades, comprehensive research has not been done with RFI cattle in a grazing environment (Herd et al., 1998, 2002, 2004; Meyer et al., 2008; Lawrence et al., 2012, 2013; Knight et al., 2015; Manafiazar et al., 2015; Moore, 2018). Biological, mechanistic drivers of cow performance may differ by age, stage of production, and availability of feed resources. For example, Randel and Welsh (2013) stated “Selection for low residual feed intake results in selection of leaner heifers that reach puberty at older ages. These leaner heifers calve later in their first and subsequent calving seasons.” Yet, when evaluating cows over 10 production cycles, which produced low-RFI (LRFI) and high-RFI (HRFI) progeny, Basarab et al. (2007) found that cows which produced LRFI (more efficient) progeny had 2 to 3 mm more back fat than dams that produced HRFI (less efficient) progeny while maintaining similar BW. In an additional research trial conducted in Alberta, Canada (Basarab et al., 2011), these researchers reported a decline in fertility with LRFI heifers in a grazing environment. It appears that there may be a threshold for decreased body energy stores that exists among younger LRFI heifers, which can be overcome as they mature and are able to express an advantage with maintenance requirements in a grazing environment. It is important to determine “fitness” of young 2-yr-old cows differing in feed efficiency when presented with a challenging rangeland environment.

In an Idaho sagebrush-steppe environment, the objectives were to determine whether cows classified as either LRFI or HRFI differed in BW, BCS, and grazing behavior [bite rate and daily travel distance (DTD)] over time.

Materials and Methods

All procedures were approved by the University of Idaho Animal Care and Use Committee (IACUC # 2015-44). Animal husbandry, management, and handling procedures in the research environment were in accordance with the Ag Guide (2010). Figure 1 shows the timeline for this research trial.

Figure 1.

Experimental timeline. HDQ = headquarters; USSES = U. S. Sheep Experiment Station; BR = bite rate; NMCREEC = Nancy M. Cummings Research, Extension & Education Center. Cattle entered experimental pasture for grazing behavior on 18 November and grazing behavior data commenced the following day.

Range site

The study site for this experiment was in a 323-ha pasture on the USDA, ARS, U. S. Sheep Experiment Station (USSES), located about 16 km northeast of Dubois, Idaho (44° 18′ N, 112° 7′ W). Water troughs were located centrally in the pasture and refilled daily, usually around noon. The range site was in the sagebrush-steppe with elevations within the pasture ranging from 1,762 to 1,806 m on slopes <20% but mostly between 0 and 12%. The 20-yr mean annual precipitation (1981 to 2010) near the research site (44° 15′ N, 112° 12′ W, elevation, 1661 m) is 328 mm, with 58% falling during April through September. The pasture is dominated by mountain big sagebrush (Artemisia tridentata Nutt. ssp. vaseyana [Rydb.] Beetle) and threetip sagebrush (A. tripartita Rydb.) with subdominant shrub species including antelope bitterbrush (Purshia tridentata [Pursh] DC.), yellow rabbitbrush (Chrysothamnus viscidiflorus [Hook.] Nutt.), and spineless horsebrush (Tetradymia canescens DC.). Dominant perennial grasses include Great Basin wildrye (Leymus cinereus [Scribn. & Merr.] A. Löve), Idaho fescue (Festuca idahoensis Elmer), sandberg bluegrass (Poa secunda Presl), thickspike wheatgrass (Elymus lanceolatus [Scribn. & J.G. Sm.]), bluebunch wheatgrass (Pseudoroegneria spicata [Pursh] A. Löve), and needle and thread (Hesperostipa comata (Trin. & Rupr.) Barkworth), with only trace amounts of cheatgrass (Bromus tectorum L.). Common forbs on this site include lupines (Lupinus spp. L.), milkvetches (Astragalus spp. L), fleabanes (Erigeron spp. L.), and pussytoes (Antennaria spp. Gaertn.). The soils within this site are predominantly loamy with numerous shallow soil areas, rocky soils, and exposed rock outcrops due to underlying volcanic deposits of basalt.

Forage production and quality

Forage production was estimated at the beginning of the grazing period by hand clipping. Ten randomized 0.16 m² quadrats were sampled from both recently burned (2005) and unburned areas. All perennial and annual graminoids rooted within the quadrat frame were clipped to ground level and dried for 24 h at 65 °C. Shrubs were not sampled for production.

CP (Padmore, 1990a,b; Gavlak et al., 1996; Miller et al., 1997) was determined on replicate samples (n = 5 clipped plots/replicate) of clipped forage by a commercial lab (Ward Laboratories, Inc., Kearney, NE). Forage digestibility of the clipped forage samples at the same lab was estimated in vitro from ADF using the Ankom 200/220 Fiber Analyzer (Ankom Co., Macedon, NY) and following the procedures of Mertens (1992). Forage mineral concentrations for Ca, P, Mg, K, Na, Fe, Zn, Cu, Mn, Mo, S, and Co were analyzed at the same lab using inductively coupled atomic plasma analysis (Campbell and Plank, 1991; Kovar, 2003). Poor replication of Co analysis samples among forage replicates resulted in these samples being excluded from the study. Samples were analyzed for Se at the South Dakota Agricultural Laboratories (Brookings, SD) using fluorometric procedures (Olson et al., 1975; Koh and Benson, 1983; Palmer and Thiex, 1997; AOAC, 2016).

Animals and grazing behavior

The 30 Hereford × Angus cows (2-yr old) used in this study were selected based upon their RFI rank as growing heifers during a 49-d trial period (Hall et al., 2015). Briefly, 104 heifers (11 to 13 mo old) were classified for RFI using the GrowSafe Systems (Calgary, Alberta, Canada) at the Nancy M. Cummings Research, Extension and Education Center at Carmen, Idaho (45° 17.322′ N, 113° 52.697′ W). The diet during the RFI trial consisted of 80% alfalfa hay, 10% wheat middlings, and 10% liquid supplement as a total mixed ration. Heifers were allowed a 10-d warm-up period followed by a 49-d RFI trial from 10 February to 31 March. At the end of the classification period, 40 heifers (combination of 16 LRFI, 14 HRFI, and 10 average RFI) were retained as herd replacements. For the current study, all available 14 HRFI and 16 LRFI cows were used to compare young cows for rangeland adaptability which varied greatly in feed efficiency.

Previous to placement of these divergently ranked RFI cattle on rangeland, cattle grazed irrigated, predominantly cool-season grass pasture at the Nancy M. Cummings Research, Extension and Education Center. Major forage species consisted of orchardgrass (Dactylis glomerata L.), red clover (Trifolium pretense L.), smooth brome (Bromus inermis Leyss.), and Kentucky bluegrass (Poa pratensis L.).

Preceding and following data collection on rangeland, all cattle were weighed and scored for BCS (1 to 9, 9 = fattest; 1 trained observer, same each time) on 21 September and 16 December 2015. Calf weaning weights for the HRFI vs. LRFI cattle did not differ (264 ± 6.1 kg for LRFI; 273 ± 5.2 kg for HRFI; P > 0.10). All 2-yr-old cows used in this study were pregnant at weaning with the expected calving season beginning in February of the subsequent year. On 29 September 2015, 20 d after weaning of calves, cattle were shipped with the cowherd (n = 242) from the Nancy M. Cummings Research, Extension and Education Center to USSES rangeland pastures (44° 25.325′ N, 111° 47.583′ W) near Kilgore, Idaho for a 46-d pretrial grazing period. Pastures were similar in shrub and herbaceous understory to the 323-ha experimental pasture with patches of Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) and quaking aspen (Populus tremuloides Michx.) along the north and west fringes. After 46 d, cattle were trailed ~31 km southwest over a 2-d period to the 323-ha experimental pasture located on the USSES Headquarters range. Cattle were grazed in this pasture for 27 d. Prior to trailing, five of the HRFI and five of the LRFI cows were randomly chosen and instrumented with custom global positioning system (GPS) collars (Clark et al., 2006; Clark, 2018) programmed to obtain GPS positions at 2-min intervals from which DTDs were calculated over 25 d.

Cattle were not provided any protein supplement throughout this trial but were provided a loose free-choice mineral–vitamin supplement (Table 1). The mineral–vitamin supplement was placed near the water source in five tubs and was accompanied by white iodized salt in one additional tub. The mineral–vitamin supplement was offered ~3 d/wk to achieve an average intake per cow of 85 g/d plus an additional 17 g/d of white salt. On the day cattle arrived in the USSES experimental pasture on 18 November, hay was provided to allow cattle to fill up following the 2-d cattle drive. The collared cattle accompanied the rest of the experimental cattle throughout the remainder of the grazing period.

Table 1.

Mineral supplement analysis

Ingredient	Minimum analysis	Maximum analysis
Ca, %	6	7
P, %	6
Salt, %		5
Na, %	2	3
Mg, %	5
Cu, ppm	2,000
Zn, ppm	2,000
Iodine, ppm	125
Se, ppm	38
Vitamin A, IU/kg	92,517
Vitamin D, IU/kg	4,989
Vitamin E, IU/kg	514

List of ingredients: corn distillers dried grains, monocalcium phosphate, magnesium oxide, calcium carbonate, mineral oil, hydrated sodium calcium aluminosilicate, copper sulfate, vitamin E supplement, zinc sulfate, zinc hydroxychloride, copper hydroxychloride, vitamin A supplement, vitamin D3 supplement, salt, ethylenediamine dihydriodide, sodium selenite, sodium chloride, calcium iodate, calcite. Manufactured by Van Beek Nutrition, Twin Falls, ID.

Eight days of grazing observations on collared cattle were conducted on horseback by one observer for 3 to 4 h in morning (about 0730 to 1130 h local) and afternoon (1300 to 1700 h). The same observer and horse were used on all days of the study. Observational data were collected on days 4, 6, 7, 14, 15, 16, 20, and 21 of the trial corresponding to 21, 23, and 24 November and 1, 2, 3, 7, and 8 December 2015. Cattle were able to be approached closely by the horseback observer (to within 6 m) and did not exhibit flight response or agitation during the observation process. On some days (n = 4), one or two cows were unable to be found. On other days (n = 2), 75% of the cattle were observed twice (i.e., AM and PM).

Focal sampling for bite rate (bites/min) was conducted on single animals (Sprinkle et al., 2000) during the AM and PM observation time periods for ~10 to 15 min. At least four replicate samples per observation period were acquired whenever possible. Beginning and ending times for each replicate were recorded in the field on a tablet computer using a spreadsheet with an integrated timestamp. Sometimes cattle commenced resting, walking to water, or ruminating in the midst of an observed grazing bout, so it was not always possible to obtain multiple sample replicates of four or greater within each grazing observation period. Bite rate frequency data were averaged over each AM or PM time period and then daily bite rate was averaged over both AM and PM grazing.

Grazing collar data management

The GPS data from the 2-d trailing event and the first and last days of grazing were omitted due to the confounding effects of herding on cattle behavior. Data files retrieved from GPS receivers were imported into Excel software and processed further for importation into ArcMap software (v. 10.2.2, ESRI Inc., Redlands, CA). Briefly, the latitude/longitude coordinates were converted to Universal Transverse Mercator format so that travel distances can be calculated from successive waypoints using the Pythagorean theorem (Ganskopp and Johnson, 2007; Knight et al., 2018). The GPS positions appearing >5 m outside of the mapped fenceline within ArcMap were treated as outliers and deleted. Positions ≤5 m outside of the fence were kept since the location outside the fence was most likely due to a combination of minor GPS and map errors. To help identify the paths traveled, lines were generated from point to point and split by date. After visually inspecting all points with positional dilution of precision (PDOP) values >6.0 for the first collar processed, it was determined that all points with a PDOP ≥ 10.0 could be automatically deleted from the remaining data sets before conducting further review (D’Eon and Delparte, 2005). The PDOP is an index of the quality of the GPS fix in terms of geometric precision and is based on the three-dimensional distribution or topology of satellites visible to the GPS receiver at the time of the fix. While lower PDOP values indicate a well-dispersed satellite constellation resulting in better triangulation and higher quality fixes, higher PDOP values indicate poorer satellite geometry and lower quality fixes. As such, PDOP can be a good indicator of the positional accuracy of GPS positions. All remaining points with a PDOP > 6.0 were flagged and individually inspected for spatial correspondence with the surrounding positions. Flagged positions were visualized along the travel trajectory line formed by the surrounding positions. Flagged positions which sharply diverged (i.e., substantial displacement occurring at acute angles which probably do not describe the true shape of the actual animal movement path) from the general directionality of the trajectory were deleted. In some instances where several consecutive points had similar high PDOP values, the number of satellites used for each point was also used to determine which positions to keep or discard. Ideally, each GPS location should have had at least five satellites present, but many were based on only three or four satellites. Several points with PDOP values between 6.0 and 7.0 were kept because of the higher number of satellites used and the good fit with the surrounding points. Additional points with PDOP values 6.1 to 7.0 and a few around 8.0 and 9.0 were kept to maintain the general shape of the path traveled in instances where several consecutive points had PDOP values > 6.0.

Obvious point outliers due to GPS error were deleted. For example, points sharply diverging from the trajectory path by ≥20 m, while an animal appeared to be stationary, were deleted. Furthermore, points in line with surrounding points, but having a high PDOP value, were either deleted due to its poorly rated quality and low influence on the overall path, or kept to maintain the overall shape of the path if it was part of a group of points with high PDOP values. Also, some points with PDOP of <6.0 were deleted if points were over 100 m from a cluster of other points collected from an apparent resting cow. Approximately 7.2% to 10.9% of the 17,000 to 17,413 data points collected for each collar were deleted. After deletion of erroneous GPS positions, a new set of directional path lines were generated for each cow and examined for any additional outliers to be deleted. From the final corrected path line, DTD for each cow was calculated.

Statistical Analyses

Daily bite rate, DTD, and production (cow BW and BCS) data were analyzed using a restricted maximum likelihood-based mixed effects model for repeated measures (v. 9.4, SAS Inst., Inc., Cary, NC) with the categorical, fixed effects of RFI group and day of study (or day of year for repeated production data) and the interaction between RFI group × day of study (or day of year). Cow within RFI group was included as a random effect and was the repeated subject. The denominator degrees of freedom for treatment F-statistics were approximated using the Kenward–Roger’s method. For all these models except DTD, a heterogeneous autoregressive structure was used as a covariance structure to model the relationships between repeated observations. In order to get DTD to properly converge, a simplified compound symmetry covariance structure was used for this model. The change in BW and change in BCS had no repeated data and were analyzed using the GLM procedure of SAS (v. 9.4, SAS Inst., Inc.) with RFI group as a main effect. The change in BW was tested for heteroskedasticity using both the Bartlett and Levenes tests in SAS. Least squares treatment means for all statistical models were separated using the pairwise contrasts (PDIFF, v. 9.4, SAS Inst., Inc.). Letter assignments for differences in least square means (P < 0.05) were produced using the pdmix800.sas macro as originally described by Saxton (1998).

Results and Discussion

Climatic Data

The USSES weather station, located at the Headquarters 7.8 km South Southwest of the experimental pasture, recorded a mean maximum and minimum temperature of 0.6 and −9.7 °C from 19 to 30 of November, 2015 and 2.5 and −5.4 °C from 1 to 14 of December, 2015. The highest maximum and minimum temperatures in November were 7.2 °C (24 November) and −14.4 °C (30 November) and for December were 11.1 °C (9 December) and −13.9 °C (1 December). Actual temperatures were probably lower at the experimental pasture, which was ~122 m higher in elevation. For example, the temperature at 0730 h at the experimental pasture on 1 December was −16.1 °C. Recorded snowfall at the weather station, which accounted for all precipitation, was 13 mm in November and 180 mm in December.

Forage Production

Forage production with the accompanying 90% CI for perennial and annual grasses combined was estimated at 331 ± 95 and 491 ± 242 kg/ha for unburned (25%) and burned (75%; 2005 burn date) areas of the experimental pasture, respectively. For calculating stocking rates, a weighted (based on burned and unburned areas) average of 451 ± 205 kg/ha was used. Assuming maximal average forage intake was 2.1% of BW for the entire 242 head of cattle in this pasture, then DMI was estimated to be 12.14 kg/cow for cows averaging 578 kg BW at a BCS = 5. Over 26 d (cows fed hay on the first day they arrived in the pasture), maximal forage intake was projected to be ~76,385 kg which was ~52% of the available forage (145,673 kg). Due to the forage quality present in the experimental pasture (Figure 2), it was more likely that forage intake was below maximal projections, and thus, overall forage harvested was <50%. At this level of forage utilization, cattle would not be forced to do extensive searching to obtain sufficient daily forage.

Figure 2.

Forage quality for burned and unburned sections of experimental pasture. Forage digestibility was based upon in vitro acid detergent fiber digestibility. Both CP and digestibility are expressed on a DM basis. Laboratory values were obtained on 10 clipped samples for both burned and unburned pasture locations which were composited into two replicate samples for each classification for analysis. The SDs for burned replicate samples were 0.56 and 0.64 for digestibility and CP and for unburned replicate samples, 0.35 for digestibility and 0.49 for CP, respectively.

The average cow weight at a BCS of 5 for 2-yr-old cattle on this trial was 526 kg. Ferrell and Jenkins (1984) reported that 9-yr-old nonlactating, nonpregnant Angus × Hereford cattle had maintenance requirements of 130 kcal ME·kg^−0.75·d⁻¹. Assuming slightly higher maintenance requirements of 140 kcal ME·kg^−0.75·d⁻¹ for younger cattle or Angus × Hereford cattle with greater milk production potential and/or late in gestation, and multiplying by 1.3 for increased maintenance requirements for free-ranging cattle (Osuji, 1974), would set estimated maintenance requirements (at thermoneutral zone) for average 2-yr-old cattle in this trial at 20 Mcal ME/d. It can be assumed that the 2-yr-old cattle in this trial were not meeting maintenance requirements since they all lost weight (Table 2). If energy mobilized from stored body tissue is estimated to be 3.51 Mcal ME/kg at 70% efficiency of turnover (Moe et al., 1971; Moe and Tyrell, 1974; Alderman et al., 1982; Thompson et al., 1983; NRC, 2001), then with an average weight loss for all 2-yr-old cows in this trial of 0.75 kg/d (Table 2), cattle would have supplied 2.62 Mcal ME/d from mobilized tissue. Accounting for this tissue mobilization, 17.4 Mcal of ME would have needed to come from rangeland forage. With a weighted forage digestibility of 44% (or 1.59 Mcal ME/kg forage) for this pasture (Figure 1), then forage intake for the younger cows would have needed to ~2.1% of BW. At this level of forage quality, the aforementioned forage intake was unlikely. If cattle were able to select a better quality diet than what was predicted from the clipped forage laboratory analyses (Sprinkle et al., 2000), then forage intake would have been less (e.g., 50% forage digestibility would have forage intake equal 1.83% of BW).

Table 2.

Effects of postweaning residual feed intake classification on performance data of grazing cows¹

Item	LRFI cows2	HRFI cows2	P-value
BW, kg3
Beginning	556.8 ± 10.1	576.8 ± 10.8	0.197
Ending	506.8 ± 9.8	509.8 ± 10.5	0.836
BCS3
Beginning	5.8 ± 0.13	6.2 ± 0.14	0.020
Ending	4.6 ± 0.13	4.6 ± 0.14	0.667
BW change, kg	-50.0 ± 5.41	-66.6 ± 5.78	0.046
BCS change	-1.2 ± 0.11	-1.6 ± 0.12	0.030

¹Cows were on rangeland for 78 d, RFI = residual feed intake.

²LRFI = 16 low RFI (efficient cows); HRFI = 14 high RFI (inefficient cows).

³There was a significant day × RFI group interaction present (P < 0.05).

Forage Quality

Figures 2 and 3 show the forage quality for representative clipped forage samples from previously burned and unburned sections of the pasture. CP of the clipped samples was below (7% of DM) what is considered necessary to maintain adequate rumen function (Leng, 1990; Cochran, 1995) and digestibility was below that necessary (52%) to prevent weight loss for nonlactating, pregnant cattle (Sprinkle, 2015). For forage mineral concentrations (Figure 3), only Ca and Mn were considered adequate (NASEM, 2016). Antagonistic minerals for Cu absorption [Mo if > 2 ppm (NASEM, 2016) and Fe if > 400 ppm (Corah and Dargatz, 1996)] were also present, with the unburned clipped forage samples having 2.8 ppm Mo and the burned clipped forage samples having 639 ppm Fe.

Figure 3.

Forage mineral concentrations expressed as a percentage of beef cattle daily requirements for burned and unburned sections of the experimental pasture. Circled minerals were deficient. Based upon Nutritional Requirements for Beef Cattle (NASEM, 2016). Calcium and P requirements are dependent upon cow size, physiological state, and milk production. Estimate for Ca and P is for a 567 kg cow with a 38.5 kg calf birth weight.

Animal Performance

Day of year (P < 0.0001) and the RFI × day of year interaction (P < 0.031) were significant for BCS. At the beginning of the grazing period, the BCS was less (P < 0.020) for LRFI cows compared with HRFI cows. However, BCS were not affected by RFI classification at the end of the grazing period (Table 2). The LRFI and HRFI cows both lost (P < 0.0001) BCS during the grazing trial, but the loss in BCS was greater (P < 0.030) for HRFI than LRFI cows. It is presumed that the greater loss of BCS for HRFI cattle was due to greater maintenance requirements. Simple means (Hall et al., 2015) for the yearling heifer RFI scores in units of SD were 1.17 ± 0.2 for HRFI and −0.88 ± 0.1 for the LRFI heifers.

As with BCS, BW was significant for the day of year effect (P < 0.0001) and the interaction (P = 0.048) of RFI group with day of year (Table 2). In contrast to the results for BCS, BW for the different RFI cows did not differ (P = 0.197) prior to going to range (Table 2). However, BW change and change in BCS differed significantly (P = 0.046 for BW and P = 0.030 for BCS change) among the LRFI and HRFI cows in this study following 78 d of grazing late season Idaho rangeland (Table 2). Not only did the LRFI cows lose less BW over the trial, they also were less variable with respect to BW loss (Figure 4), indicating more opportunity for making positive changes in overall feed efficiency in the cow herd by careful selection among selected tested sires.

Figure 4.

Box plot of distribution of weight loss data for low residual feed intake (RFI) vs. high RFI 2-yr-old cows grazing late season Idaho rangeland for 78 d. Center line in each box is median value, highest and lowest point of whiskers indicate maximum and minimum values. Upper section of box is the first quartile of data and lower section is the third quartile of data. Least square means for low RFI and high RFI were −50.0 ± 5.41 kg and −66.6 ± 5.78 kg, respectively. Heteroskedasticity was tested and was not significant (P > 0.12).

In agreement with much of the published literature (Richardson et al., 1998; Herd and Bishop, 2000; Herd et al., 2003; Kerley, 2010), our 2-yr-old LRFI cows in this study were leaner (P = 0.020) prior to going to rangeland. However, the LRFI cattle demonstrated an ability to more favorably adapt to the diminished forage quality on late-season rangeland.

Few studies have examined the effects of RFI classification on performance of cows while grazing native rangeland, particularly late-season rangelands characterized by poor-quality forage. To our knowledge, this is one of the first, if not the first research considering this question. Manafiazar et al. (2015) evaluated the performance of divergently ranked RFI pregnant yearling heifers on improved, fertilized summer pasture in Alberta, Canada. In their study, LRFI cattle had greater measured back fat (P = 0.036) than HRFI cattle at the end of the grazing period. Meyer et al. (2008) compared mid- to late gestation, LRFI vs. HRFI 2-yr-old Hereford cows on summer fescue pasture and found no differences in either initial (P ≥ 0.55) or post grazing trial BCS (P ≥ 0.19). Lawrence et al. (2012) evaluated LRFI, medium RFI, and HRFI heifers on improved pasture in Ireland during the autumn of their first lactation. At the end of the grazing period, medium RFI cattle had greater (P = 0.02) BCS than did LRFI and HRFI. In New South Wales, Australia, Herd et al. (1998) compared LRFI vs. HRFI 3-yr-old lactating (third month of lactation) cows on irrigated oat pastures and found that LRFI cows weighed 7% more (P < 0.05) and had similar subcutaneous fat (P > 0.05) when compared with HRFI cows. Knight et al. (2015) reported that there were no differences in BCS between LRFI and HRFI cows which grazed Arizona rangelands during the time period spanning the advent of summer monsoon rainstorms. All of the above research trails can be characterized as having forage quality sufficient to meet maintenance requirements. Consideration of the lower maintenance requirements for LRFI cattle becomes much more important as cattle move into a negative energy balance. In the current study, pregnant cows were grazed on poor-quality rangeland without protein supplementation to examine the effects of postweaning RFI classification on performance in a nutritionally stressful environment. Our intent was to discover if the presumably lower maintenance requirements of LRFI cattle could be advantageous when facing nutritional stress. If LRFI cattle can better meet the limitations imposed by a challenging rangeland environment, then there is an advantage for producers applying some selection pressure for this trait. Basarab et al. (2007) compared cattle over 10 production cycles and reported that cows that produced LRFI progeny had 2 to 3 mm more back fat (P = 0.017) than cows that produced HRFI progeny. All of this points to a possibility to better fit cattle to a limiting rangeland environment by considering cow efficiency when selecting replacement bulls and heifers.

Grazing Behavior

A concern for young efficient cattle is whether these leaner animals will function as well as inefficient animals in a nutritionally limited rangeland environment. During RFI testing in a feedlot, inefficient cattle often exhibit more aggressive feed intake characteristics such as feeding duration, frequency of feeding, and head down eating time (Golden et al., 2008; Durunna et al., 2013). The greater body condition common among inefficient cattle at younger ages combined with their more aggressive feeding behavior could potentially offer an advantage in a rangeland environment. However, there were no differences (Table 3) between the LRFI and HRFI groups for either DTD (P = 0.532) or bite rate (P = 0.216). Similarly, the interaction of RFI group with the day of the study was not significant for DTD (P = 0.826) or bite rate (P = 0.592). However, the influence of the day of the study was highly significant (P < 0.0001) for both behavioral responses.

Table 3.

Grazing behavior for 2-yr-old cows differing in feed efficiency¹

Date	Day of study	n	Daily traveled distance, km/d	SEM	P-value
Distance traveled by day
19-Nov-2015	2	10	4.21l	0.422	<0.0001
20-Nov-2015	3	10	4.89kl
21-Nov-2015	4	10	6.57fghi
22-Nov-2015	5	10	6.04hij
23-Nov-2015	6	10	6.99efgh
24-Nov-2015	7	10	6.00hij
25-Nov-2015	8	10	11.32b
26-Nov-2015	9	10	11.47b
27-Nov-2015	10	10	11.27b
28-Nov-2015	11	10	7.81de
29-Nov-2015	12	10	6.40fghij
30-Nov-2015	13	10	9.01c
1-Dec-2015	14	10	6.02hij
2-Dec-2015	15	10	8.32cd
3-Dec-2015	16	10	6.40fghij
4-Dec-2015	17	10	5.53jk
5-Dec-2015	18	10	5.70ijk
6-Dec-2015	19	10	7.09efg
7-Dec-2015	20	10	6.28ghij
8-Dec-2015	21	10	4.88kl
9-Dec-2015	22	10	7.02efgh
10-Dec-2015	23	10	8.13cd
11-Dec-2015	24	10	7.33def
12-Dec-2015	25	10	8.24cd
13-Dec-2015	26	10	14.35a
Distance traveled by treatment group
LRFI	All day	5	7.34a	0.305	0.5320
HRFI	All day	5	7.63a
Date	Day of study	n	Bites/min		P-value
Bite rate by day of study
21-Nov-2015	4	9	55.2 ± 5.63e		<0.0001
23-Nov-2015	6	9	66.3 ± 6.21ab
24-Nov-2015	7	7	74.3 ± 5.49ac
1-Dec-2015	14	10	64.2 ± 5.63b
2-Dec-2015	15	9	65.6 ± 5.35b
3-Dec-2015	16	10	81.3 ± 5.24cd
7-Dec-2015	20	10	82.9 ± 5.14d
8-Dec-2015	21	10	84.8 ± 5.32d
Bite rate by treatment group
LRFI	All day	5	78.4 ± 3.72a		0.2158
HRFI	All day	5	65.3 ± 9.31a

¹LRFI = low residual feed intake; HRFI = high residual feed intake.

^a–lMeans within a column by dependent variable without a common superscript differ (P < 0.05). Winter storms occurred on 25 November, 30 November, and 13 December, 2015.

As the season of year progressed, bite rate increased for all cows (P < 0.05; Table 3). Since forage supply was not limiting, we interpret the increased bite rate to be a behavioral response promoted by the need to increase harvesting efficiency as temperatures decreased and daylight hours declined.

Significant increases (P < 0.05) in DTD were associated with snow storms which occurred on 25 November, 30 November, and 13 December 2015 (Table 3). It is apparent by reviewing these data that the cows in this study changed behavior as a result of the winter storms, likely due to seeking shelter and being driven by the wind (Valentine, 2000; Rubio et al., 2008). Substantial changes in travel distance were evident when data for the day prior to the storm event were contrasted with that from the day of the storm. For example, when a big snow storm moved in on December 13, 2015, travel increased (P < 0.0001) from 8.2 ± 0.42 km/d the day before the storm to 14.3 ± 0.42 km/d on the day of the storm.

Very little research has been conducted comparing cows differing in RFI on rangeland. Knight et al. (2015) evaluated grazing behavior for LRFI (efficient) vs. HRFI (inefficient) cows on Arizona rangeland. Efficient cows in that study traveled further (P = 0.005) than inefficient cows (6.23 km/d vs. 5.84 km/d) on pinyon-juniper rangelands but there was no difference (P ≥ 0.06) for distance traveled, distance from water, and elevation for Ponderosa pine rangeland. Cattle in the Arizona study grazing pinyon-juniper rangelands had less abundant forage of lesser forage quality than when grazing Ponderosa pine rangeland and the researchers hypothesized that efficient cows may have been engaged in more search grazing on the pinyon-juniper rangelands. Additional research data are needed to better describe conditions that drive and determine foraging efficiency on rangeland in different environments. In our study, we did not see any differences in distance traveled on late season rangeland in Idaho with non-lactating 2-yr-old cows differing in RFI. Since our late-season rangeland was of poor quality that did not meet maintenance requirements, it is likely that all cattle in our study were engaged in search grazing.

The young 2-yr-old LRFI cows in our study were leaner than HRFI cows prior to being moved to rangeland. However, the LRFI cows lost less weight while on rangeland and ended up at the same BCS as the HRFI cows at the end of the grazing period. Furthermore, LRFI cows appear to have less variability for weight loss than do HRFI cows, leading one to conclude there is opportunity for selection of efficient cows that eat less and also fit a rangeland environment. Limited research has been conducted comparing these divergently ranked cattle in a rangeland environment. Our research suggests that efforts to select more efficient beef cattle can yield benefits in rangeland livestock production. However, further studies need to be done with lactating 2-yr-old cows on rangeland to see if these leaner cows can thrive in a rangeland environment and remain in the herd as productive, mature cows.

Funding

This work was supported by the USDA National Institute of Food and Agriculture, Hatch project # 1010550. Mention of a proprietary product does not constitute a guarantee or warranty of the product by the Idaho Experiment Station, University of Idaho, USDA ARS US Sheep Experiment Station, or the authors and does not imply its approval to the exclusion of other products that may also be suitable. We wish to acknowledge Rob Laird, Boyd Leonard, and Jack Hensley for animal care and management and Mark Williams for project coordination.

Conflict of interest statement. None declared.