Effects of harvest date and growth stage on triticale forages in the southwest United States: agronomic characteristics, nutritive value, energy density, and in vitro disappearance of dry matter and fiber

Abstract

Recently, there has been increased interest in including triticale (X Triticosecale Wittmack) or other winter cereals within forage programs throughout the southwest United States. Our objectives were to screen 14 diverse triticale cultivars for agronomic and nutritive characteristics with specific emphasis on identifying normal, as well as deviant, responses to the calendar date and plant maturity for forages seeded in December and harvested from late February throughout May at Maricopa, AZ. Fourteen cultivars were established in a randomized complete block design with each cultivar represented within each of three field blocks. Plots were clean tilled and established on December 18, 2018, and then harvested at 2-wk intervals beginning on February 27 and ending May 23, 2019. Across all harvest dates, forage (N = 315) energy density (NE_L) exhibited strong negative correlations with growth stage (r =  −0.879), plant height (r =  −0.913), head weight (r =  −0.814), and estimated dry matter (DM) yield (r =  −0.886) but was positively associated with percentages of leaf (r = 0.949), and weakly associated with percentages of the stem (r = 0.138). Through April 10, similar correlations were observed within individual harvest dates (N = 45) for growth stage, leaf percentage, and plant height but not for stem or head-weight percentages. Within later harvest dates, only sporadic correlations with NE_L were observed. Primarily cubic regression relationships for neutral detergent fiber, acid detergent lignin, 30- and 48-h in vitro disappearance of DM and fiber, and NE_L were fit for the mean or typical cultivar using both days from February 1 and growth stage as independent variables. Coefficients of determination (R² ≥ 0.860) in all cases indicated a good fit for the polynomial models. For NE_L, deviation from the typical cultivar when days from February 1 was used as the independent regression variable was largely affected by cultivar maturation rate. When the growth stage was substituted as the independent variable, plant height, stem percentage beginning at anthesis, and low grain-head percentage were associated with the maximum negative deviant cultivar (Merlin Max). The 0.23 Mcal/kg difference between maximum positive and negative deviant cultivars at a common late-boot/early-heading stage of growth suggests that some attention should be placed on cultivar selection as well as forage inventory needs and overall cropping goals.

Lay Summary

Recently, there has been increased interest in using triticale within forage programs in the southwest United States. Our objectives were to screen 14 triticale cultivars for agronomic and nutritive value characteristics with specific emphasis on identifying typical, as well as deviant, responses to the calendar date and plant maturity. Regression relationships for neutral detergent fiber, acid detergent lignin, 30- and 48-h in vitro disappearance of dry matter and fiber, and net energy of lactation (NE_L) were fit for the mean or typical cultivar using both days from February 1 or growth stage at harvest as independent regression variables. Deviant cultivars usually demonstrated rapid or slow maturation rates, which were often accompanied by physical characteristics reflective of advanced or slow maturation, respectively. Overall, there were a limited number of cultivars that deviated from typical with respect to NE_L, but the total range in energy density at a common late-boot/early-heading stage of growth (0.23 Mcal/kg) suggests that some attention should be placed on cultivar selection, especially when specific cultivars display atypical growth characteristics, such as greater canopy height. However, either positive or negative deviation with respect to energy density may be desirable depending on the energy needs of the targeted livestock class.

Recently, there has been increased interest in using triticale within forage programs in the southwest United States, in part because of reduced requirements for water. An evaluation of 14 triticale cultivars indicated that the net energy of lactation may vary by 0.23 Mcal/kg at a common late-boot/early-heading stage of growth, indicating that some attention should be devoted to appropriate cultivar selection.

Introduction

Triticale (X Triticosecale Wittmack) has been incorporated into dairy cropping strategies more routinely in recent years, partly in response to environmental pressures, such as a desire to capture nutrients originating from land-applied manure or to provide winter ground cover that improves land stewardship. In addition, triticale is recognized as an excellent forage crop with the potential for use across a wide range of livestock classes; however, this is also dependent on proper synchronization of forage nutrient composition with the nutrient demands of specific livestock classes (Khorasani et al., 1997). Studies conducted with triticale over the last 50 yr have described relationships between various measures of nutritive value and calendar date, growth stage, or growing degree days (Cherney and Marten, 1982a, 1982b; Khorasani et al., 1997; Coblentz et al., 2018a, 2018b). Generally, these studies suggest that the energy density of triticale forages either stabilizes or increases in association with grain fill, but neutral detergent fiber (NDF) digestibility is relatively independent of this process and continues to decline with plant maturity. For lactating dairy cows, extension recommendations include harvesting at the boot stage when the flag leaf is fully emerged, but no heads are visible (Kilcer et al., 2010). This management approach may carry a significant (70%) yield reduction compared with a later harvest date at the soft-dough stage of growth (Coblentz et al., 2018a); however, it would also allow for a double crop of corn (Zea mays L.) or soybeans [Glycine max (L.) Merr.] in northern states, such as Wisconsin or New York. In California, Drake and Orloff (2005) found that a single simulated spring grazing event at 0.15-m of vegetative growth did not reduce dry matter (DM) yields for a subsequent harvest of hay at anthesis, but multiple simulated grazing events and/or later initial defoliation dates (0.30-m height, jointing, or boot stage) negatively affected production from a final hay harvest.

From a management perspective, it is unlikely that triticale would be offered to lactating cows as a sole forage in confinement facilities. Recently, Harper et al. (2017) demonstrated that triticale harvested at the boot stage of growth could support daily milk yields greater than 41 kg/d when included at 10% of the total dietary DM (replacing corn silage). In that study, DMI was not affected, but the inclusion of triticale silage reduced milk yield and milk yield efficiency by relatively small, but statistically significant, amounts (1.5 kg milk/d and 0.06 milk yield/DMI [kg/kg]). Ostensibly, this occurred because dietary starch was replaced with digestible fiber. Inclusion of triticale within the diet also increased excretion of urinary urea and milk urea nitrogen (MUN), while reducing milk N efficiency; however, it was concluded that the double-cropping concept of triticale or wheat followed by corn silage was a suitable management strategy for dairy farms needing additional forage.

Recently, there has been increased interest in including triticale or other cereal-grain forages within dairy management programs throughout the southwest United States; however, the impetus for this interest is a more sustainable approach to water use (Frisvold, 2015) rather than the environmental stewardship and double-cropping priorities mentioned previously. In a comprehensive analysis (Frisvold, 2015), water application rates for wheat (Triticum aestivum L.) and barley (Hordeum vulgare L.) were 26% less than needed for all crops, and 24% less than needed for corn harvested as silage. However, in-depth evaluations of agronomic and nutritive value characteristics have been only sporadic throughout the region when compared with the north-central and mid-to-upper Atlantic portions of the United States. As such, these intensive evaluations of winter cereals as forages have been requested recently by the dairy industry in Arizona. Our objectives were to screen 14 triticale cultivars for agronomic and nutritive value characteristics with specific emphasis on identifying normal, as well as deviant, responses to the calendar date and plant maturity for forages seeded in December and harvested from late February through May at Maricopa, AZ.

Materials and Methods

All procedures associated with in vitro disappearance assays that were related specifically to obtaining inoculum and care of donor animals were approved by the Research Animal Resources Committee (RARC) of the University of Wisconsin-Madison (Protocol #A01006).

Forage establishment and management

Forty-five 1.8 × 6.1-m plots were established in three field blocks (15 plots/block) on a Casa Grande sandy-loam soil (reclaimed fine-loamy, mixed, superactive, hyperthermic, and Typic Natrargid) at the University of Arizona Maricopa Agricultural Center, located near Maricopa, AZ (33°06ʹN; 112°05ʹW; altitude = 362 m). Fourteen triticale cultivars were obtained from Northern Agri Brands, LLC (Great Falls, MT) and Arizona Plant Breeders, Inc. (Casa Grande, AZ). Cultivars obtained from Northern Agri Brands included “105,” “SY-115T,” “SY-158T,” “Merlin Max,” “Legend,” “Swift,” and “Goldrush 91,” all of which are available commercially. Cultivars obtained from Arizona Plant Breeders were “470-113,” “470-133,” “470-249,” “470-269,” “470-285,” “770-001,” and “770-113.” These were all experimental lines unavailable to the public at the time the trial was conducted. Each cultivar was assigned randomly to 1 of 15 plots within each block, except for SY-158T, which was established twice within each block.

The experimental site was fallowed prior to establishing the trial, and then clean-tilled before seeding. Soils were sampled in early November of 2018, and tests (Brookside Laboratories, Inc.; New Bremen, OH) indicated soil pH = 8.7, organic matter (OM) = 0.6%, Olsen P = 2 ppm, and K = 310 ppm. Plots were planted with a cone-type planter on December 18, 2018, where each plot was 10 drill rows wide with rows spaced at 18-cm intervals. This planting date generally is consistent with an approximate 45-d window for maximization of grain yields from small-grain crops near Maricopa, AZ (Ottman, 2015), where grain yields can be depressed by frost damage when planting occurs too early (before November 15), or by elevated temperatures at critical points in plant development when planted too late (after January 1).

Plots were seeded at an actual rate of 146 kg/ha and irrigated (181 mm) on December 18 to facilitate germination. Subsequent irrigation events occurred on February 1 (130 mm), March 1 (63 mm), March 15 (87 mm), March 29 (90 mm), April 12 (116 mm), April 26 (154 mm), and May 10 (71 mm), totaling 892 mm across the entire trial. Irrigation water was applied using a border flood method, where the quantity of water supplied was estimated by a formula that included inputs from the number and diameter of siphon tubes, the height of water in the drainage ditch, and the duration of irrigation. Triple superphosphate (0-45-0) was applied at a rate of 56 kg P₂O₅/ha on December 18, while urea (46-0-0) was applied on five dates (December 18, February 1, March 1, March 15, and March 29) at a rate of 56 kg N/ha for each application event, resulting in a cumulative trial rate of 280 kg N/ha. These fertilization (Ottman and Thompson, 2015) and irrigation (Husman and Ottman, 2015) management strategies were consistent with current extension recommendations for the area. No herbicides or pesticides were applied during the trial.

Plants were harvested on February 27, March 14, March 27, April 10, April 24, May 8, and May 23, 2020, which corresponded generally to early stem elongation, late stem elongation, boot, late heading, watery ripe grain fill, early dough, and full maturity based on the linear staging system of Stauss (1994), which is suitable to serve as an independent regression variable. In this system, a range is established for each major stage of development, such that 30 to 39 = stem elongation, 40 to 49 = boot, 50 to 59 = heading, 60 to 69 = anthesis, 70 to 79 = fruit development, 80 to 89 = ripening, and 90 to 99 = senescence. Throughout the sampling period, a premium was placed on avoiding border effects and reducing uneven light penetration within the plots. As such, on each harvest date, two adjacent rows from a 0.46-m section of each plot were hand-clipped. Either the third and fourth or seventh and eighth rows were harvested, depending on the designated harvest date. The remaining unclipped portion of each 0.46-m section was not removed to limit the effects of uneven light penetration throughout the remainder of the plot on subsequent harvest dates. During the experiment, the mean maximum monthly temperatures were 18.3, 18.9, 17.2, 24.4, 30.6, and 30.6 °C for December, January, February, March, April, and May, respectively. Similarly, respective mean monthly minimum temperatures were 2.2, 2.8, 3.3, 7.2, 12.8, and 14.4 °C, while respective monthly mean temperatures were 9.4, 10.0, 10.0, 16.1, 22.2, and 22.8 °C, where mean temperatures were based on hourly assessments, rather than the simple mean of maximum and minimum temperatures.

Clipped forages were dried to a constant weight under forced air at 55 °C. Dry weights were then extrapolated to a kg/ha basis as an estimate of forage DM yield, and samples were retained for nutritive analysis. Similarly, 10 plants of each plot were selected randomly, measured for height (cm), assessed for growth stage (GRST; Stauss, 1994), and then separated into leaf, stem, and grain-head tissue types. Separated tissue samples were dried as described previously, and then percentages of leaf, stem, and grain head were calculated for each experimental plot.

Laboratory procedures

Dried whole-plant plot samples (N = 315) were ground through a 1-mm screen in a Thomas Model 4 Wiley Mill (Thomas Scientific, Swedesboro, NJ) and then placed in storage envelopes pending subsequent analysis for nutritive value. Concentrations of laboratory DM were determined by drying 1.0-g samples of each forage overnight in a convection oven at 105 °C; subsequently, whole-plant ash was determined from these same subsamples by combustion in a muffle furnace at 500 °C for 6 h. Forages were analyzed for NDF (asNDFom) with sodium sulfite and heat-stable α-amylase included in the analysis procedures using the batch procedures outlined for an ANKOM200 Fiber Analyzer (ANKOM Technology Corp.; Macedon, NY); digestion procedures were followed by a correction for residual ash based on combustion of residual fiber in a muffle furnace at 500 °C for 6 h. Concentrations of acid-detergent fiber (ADF) and acid-detergent lignin (ADL) were determined through a sequential analysis, using procedures outlined by ANKOM Technology Corp. but omitting the initial extraction in neutral detergent. A rapid combustion procedure (AOAC, 1998, Official Method 990.03; Model TruMac CN; LECO Corp., St. Joseph, MI) was used to quantify concentrations of N within each forage sample, and then a conversion factor of 6.25 was applied to calculate crude protein (CP).

In vitro disappearance of DM and asNDFom were conducted based on 30- (IVDMD30, NDFD30) and 48-h (IVDMD48, NDFD48) incubations in buffered ruminal fluid using the ANKOM Daisy II Incubator System (ANKOM Technology Corp.). Subsamples (0.25 g) of each forage were heat-sealed in F57 fiber filter bags and incubated for 30 or 48 h as recommended by the manufacturer. Empty F57 filter bags were pre-rinsed in acetone and then air-dried prior to loading with the sample. The 30- and 48-h incubations were conducted in three identical incubator boxes, such that: 1) each forage was represented (in singlet) within each of 6 independent runs (3 runs per incubation time); 2) each forage was incubated in a different incubator box during each of the 3 runs per incubation time; and 3) the temperature (39 °C) of each incubator box was standardized against an independent auxiliary thermometer. These procedures were extra precautions against any biases potentially created by minor differences in incubator boxes, and to limit run-to-run variability. The ruminal fluid used in the inoculum was obtained from two lactating Holstein cows offered a total mixed ration (53.7% DM) with a nutrient composition of 17.7% CP, 28.8% NDF, 42.8% non-fiber carbohydrate, 4.6% fat, 74.3% total digestible nutrients (TDN), and 1.72 Mcal/kg net energy of lactation (NE_L) (University of Wisconsin Soil and Forage Laboratory; Marshfield, WI). The ruminal fluid was obtained from each cow through a ruminal cannula with a hand-operated siphon pump that discharged directly into pre-warmed (39 °C) thermos bottles. Insulated thermos bottles were then transported to the laboratory (15 min) in chest-type insulated coolers. After filtering the ruminal fluid through four layers of cheesecloth, equal contributions from each donor cow totaling about 400 mL were added to each incubation jar containing buffers and samples that had been pre-warmed to 39 °C. The head-space within each incubation jar was twice purged with CO₂; one purge occurred immediately after buffers, and sample bags were added to the incubation jars for pre-warming, and the second purge occurred immediately after the ruminal fluid was added to each incubation jar, which was immediately before final lid placement. For each of the 315 experimental units (45 plots × 7 harvest dates), DM and asNDFom disappearance values were averaged across 3 runs prior to conducting any statistical analyses. Subsequently, indigestible asNDFom following a 240-h in vitro incubation was determined on a subset of cultivars using similar procedures; however, samples were analyzed only in duplicate for this response variable.

The summative approach (Weiss et al., 1992; NASEM, 2001) was used to calculate energy density (TDN), where truly digestible fiber was calculated as asNDFom (% of DM) × NDFD48 (% of asNDFom), and NE_L was calculated as described by NASEM (2001). Normally, the summative computation of forage energy density requires inputs of CP bound within NDF and ADF residual fibers following non-sequential extraction in neutral and acid detergent (NDICP and ADICP, respectively). Because of the large number of experimental units within this project, NDICP and ADICP were estimated in triplicate from composites that included similar contributions from all cultivars for each harvest date. Concentrations of CP within each fibrous residue were quantified with the identical rapid-combustion procedure described previously for whole-plant samples. Furthermore, the NDF extraction procedures used to quantify NDICP included heat-stable α-amylase, but not sodium sulfite, which is known to cleave disulfide bonds and dissolve cross-linked proteins, thereby reducing recovery of CP from NDF residues (Van Soest et al., 1991).

Statistics

Descriptive statistics (mean, SD, minimum, and maximum) were calculated by harvest date for agronomic characteristics, as well as concentrations of ash, CP, asNDFom, ADL, TDN, and NE_L using the PROC MEANS procedure of SAS 9.4 (SAS Institute Inc., Cary, NC). Correlations between NE_L and plant growth characteristics (GRST, canopy height, leaf, stem, and grain head percentages) were conducted across all harvest dates (N = 315) and within harvest date (N = 45) with the PROC CORR procedure.

Random coefficient models were fit in PROC GLIMMIX to evaluate the relationship between asNDFom, ADL, NE_L, NDFD30, NDFD48, IVDMD30, and IVDMD48 with two independent variables that included days from February 1 (DYS) and GRST. Separate models were fit for each combination of response and independent variables. At the outset, the basic modeling framework included a cubic-fixed-effect model (i.e., intercept, linear, quadratic, and cubic terms) and cubic-random-effect terms, where the latter were specific for each cultivar. Generally, the random effect parameters for each cultivar represent deviations from the corresponding fixed-effect parameter. Estimates derived from only the fixed effects represented a mean or “typical” cultivar (TYPCUL), whereas all of the random-effect components assume that their mean value is zero. The examination of the random effects for each cultivar can, therefore, be considered a general assessment of whether a specific cultivar is different from a TYPCUL with respect to the relationship (shape of the curve) between the response and independent variables. After assessing the various model components, the cubic fixed effects were retained in all models in addition to the linear component of the random effects. Deviations from higher-ordered coefficients were not meaningful and were not considered further. One exception was the model relating asNDFom with GRST, where a fourth-degree polynomial fixed effect term was added. In addition, the independent variables were natural-log transformed for models with NE_L as the dependent variable to attain proper model convergence. It should be further noted that the graphical presentation (Figures 1–7) of the relationships between dependent response variables (asNDFom, ADL, NE_L, NDFD30, NDFD48, IVDMD30, or IVDMD48) and DYS or GRST includes a response curve for the maximum deviant (positive and/or negative) cultivars. These are provided for visual perspective and are specific to the deviant cultivar only; therefore, those deviant response curves are not generated with data inputs from all cultivars that would be expected within the general random coefficient modeling concept.

Figure 1.

Relationships between asNDFom and days from February 1 (top) or growth stage (bottom), where the regression line (▬▬) and equation represent the response for a typical or mean cultivar. Gray circles and hashed line (- - - -) indicate the response for the only cultivar (Merlin Max) that deviated from typical (P = 0.001). Abbreviation: asNDFom, neutral detergent fiber conducted with heat-stable α-amylase and sodium sulfite, and with correction for residual ash.

Figure 7.

Relationships between NE_L and the natural log (ln) of days from February 1 (top) or growth stage (bottom), where the regression line (▬▬) and equation represent the response for a typical or mean cultivar. Gray or open circles with hashed lines (- - - -) are associated with cultivars exhibiting the greatest positive and negative deviations, respectively, from typical (P ≤ 0.015). For days from February 1, respective positive and negative deviant cultivars were “Legend” and “770-001,” while for growth stage, deviant cultivars were “Goldrush 91” and “Merlin Max.” Abbreviation: NE_L, net energy of lactation.

After conducting the aforementioned analyses, several cultivars were designated as representatives of three categories: 1) positive deviants from TYPUL based on responses for DM and asNDFom disappearance, as well as NE_L; 2) negative deviants from TYPCUL; or 3) TYPCUL. This subset of cultivars was analyzed for indigestible asNDFom following a 240-h in vitro incubation in buffered ruminal fluid. Results for this response variable were analyzed in a randomized complete block design with harvest dates treated as sub-plots. Interaction means were compared with logical contrast statements. For all analyses, significance was declared at the P < 0.05 level of confidence, and any tendencies important to the explanation of results were noted at P < 0.10.

Results and Discussion

Agronomic characteristics

Although recent extension recommendations have targeted a boot-stage harvest of triticale for lactating dairy cows (Kilcer et al., 2010), harvests of triticale at the late-milk or early-dough stages of growth have also been reported (McDonald et al., 1991; McCartney and Vaage, 1994; Kennelly and Weinberg, 2003). Considered across the 14 cultivars evaluated in this study, the boot stage of growth generally occurred in late March (Table 1), with the mean responses on March 14 (GRST = 37 ± 1.7) and March 27 (GRST = 45 ± 5.3) corresponding to the flag leaf just visible, but still rolled, and the flag leaf sheath swollen, respectively (Stauss, 1994). However, differences in developmental rate produced a range of maturities from GRST 39 (flag leaf fully unrolled) to 55 (half of the inflorescence emerged) on the March 27 harvest date. Concomitantly, there were wide ranges in canopy height (109 vs. 75 cm), leaf percentage (55.5% vs. 25.2%), stem percentage (72.8% vs. 44.5%), and grain-head percentage (11.0% vs. nil) on the March 27 harvest date that were more extreme than observed 2 wk earlier. On the May 8 harvest date, which generally corresponded to the soft-dough stage of growth (GRST = 84 ± 4.9), there also were wide ranges with respect to descriptive growth characteristics. Perhaps the most notable of these was the grain-head percentage, which ranged widely from 19.9% to 56.9% of total plant weight (overall mean = 47.4 ± 8.43%). Generally, the yield drag associated with a boot-stage compared with a later soft-dough-stage harvest observed in this study was similar to a 70% reduction described previously in Wisconsin (Coblentz et al., 2018a). Mean DM yields on March 14 (5,596 ± 1,402.1 kg DM/ha) and March 27 (8,046 ± 1,432.5 kg DM/ha) were only 32.3% and 46.4%, respectively, of those observed on May 8 (17,334 ± 1,923.2 kg DM/ha). As a result, forage needs/inventories must be weighed against livestock nutritional requirements, as well as the need to accommodate the maturation requirements of a subsequent crop, to develop appropriate harvest management decisions that are specific to each production situation.

Table 1.

Growth characteristics of 15 triticale cultivars grown during 2019 in Maricopa, AZ

Harvest date	Days from February 1	Statistic	Growth stage1	Height, cm	Leaf	Stem	Grain head	Estimated yield, kg DM/ha
					--------- % of DM3 ----------
February 27	27	Mean2	31	40	60.7	39.3	0	2,534
		SD	0.3	5.7	4.17	4.17	0	557.4
		Minimum	31	25	50.4	32.6	0	1,583
		Maximum	32	52	67.4	49.7	0	4,066
March 14	42	Mean	37	66	52.6	47.4	0	5,596
		SD	1.7	8.6	3.88	3.88	0	1,402.1
		Minimum	35	50	44.4	31.0	0	2,956
		Maximum	40	85	69.1	55.6	0	9,231
March 27	55	Mean	45	91	40.2	59.5	0.3	8,046
		SD	5.3	10.0	7.85	7.51	1.68	1,432.5
		Minimum	39	75	25.2	44.5	0	4,358
		Maximum	55	109	55.5	72.8	11.0	11,567
April 10	69	Mean	58	109	27.9	53.5	18.6	10,674
		SD	5.9	13.3	5.60	4.46	7.65	1,759.2
		Minimum	46	85	19.0	47.6	0	6,847
		Maximum	65	135	42.5	67.8	27.7	14,843
April 24	83	Mean	71	110	20.7	46.7	32.6	12,638
		SD	5.8	11.0	4.04	4.35	6.98	1,894.7
		Minimum	59	92	11.9	40.7	13.7	9,029
		Maximum	77	132	29.8	61.5	45.8	17,860
May 8	97	Mean	84	111	17.3	35.3	47.4	17,334
		SD	4.9	12.8	3.34	6.18	8.43	1,923.2
		Minimum	72	91	12.6	28.6	19.9	12,089
		Maximum	88	135	26.1	57.3	56.9	21,683
May 23	112	Mean	96	112	14.7	32.3	53.1	16,277
		SD	3.5	12.2	2.90	6.91	8.12	2,305.4
		Minimum	87	96	9.6	19.2	20.9	11,346
		Maximum	99	137	23.1	56.2	61.7	21,474

Stage of growth identified as: 20 to 29, tillering; 30 to 39, stem elongation; 40 to 49, booting; 50 to 59, heading; 60 to 69, anthesis; 70 to 79, seed development; 80 to 89, seed ripening; 90 to 99, senescence (Stauss, 1994).

N = 45.

DM, dry matter.

Nutritive value

The descriptive statistics for concentrations of whole-plant ash, CP, asNDFom, and ADL, as well as energy density (TDN or NE_L), are summarized by harvest date in Table 2. A somewhat unique observation of these triticale forages grown in Arizona was the magnitude of ash concentrations. Harvest-date means ranged from 12.8% to 17.0% of DM, which is greater than those reported by NASEM (2001) for headed triticale silage (9.7%), and by Coblentz et al. (2018a) for triticale forages harvested over 2 yr, and a similar range of maturities in Wisconsin (4.6% to 11.5%). A comparison of headed triticale silage (NASEM, 2001) with plants harvested at a comparable GRST (58) harvested on April 10 from the present study would suggest a discounted energy density (TDN) of about 4.5 percentage units based on elevated ash concentrations alone because ash makes no contribution to energy (NASEM, 2001). Although not analyzed statistically, concentrations of ash appeared to decline with plant maturation, which is consistent with other reports for this species (Coblentz et al., 2018a).

Table 2.

Whole-plant ash, CP, ADL, and energy density (TDN, NE_L) for 15 triticale cultivars grown during 2019 in Maricopa, AZ¹

Harvest date	Days from February 1	Statistic	Ash	CP	asNDFom1	ADL	Ratio2	TDN, % of DM	NEL, Mcal/kg
			---------------- % of DM ----------------
February 27	27	Mean3	15.8	22.0	41.1	1.64	3.99	76.5	1.75
		SD	1.13	1.54	1.98	0.187	0.420	1.73	0.042
		Minimum	13.4	19.0	35.2	1.28	3.30	72.1	1.65
		Maximum	17.9	24.9	45.0	2.03	4.85	79.4	1.82
March 14	42	Mean	17.0	21.1	47.3	2.79	5.89	71.0	1.62
		SD	0.93	1.83	1.94	0.606	1.260	1.57	0.039
		Minimum	15.1	17.4	43.5	1.74	3.74	66.7	1.51
		Maximum	19.3	24.0	51.9	4.89	10.58	73.4	1.68
March 27	55	Mean	16.0	16.0	54.5	3.64	6.67	64.4	1.46
		SD	1.17	1.13	2.48	0.649	1.067	1.95	0.048
		Minimum	13.6	13.4	51.1	2.68	5.04	60.0	1.35
		Maximum	18.1	18.0	60.2	5.52	9.96	68.3	1.55
April 10	69	Mean	14.2	13.1	58.5	4.41	7.54	59.3	1.33
		SD	1.26	0.95	1.76	0.487	0.773	2.59	0.063
		Minimum	11.1	11.3	55.0	3.31	5.81	52.9	1.18
		Maximum	17.1	15.8	63.9	5.43	9.13	66.4	1.51
April 24	83	Mean	12.8	9.8	55.3	5.10	9.23	57.7	1.29
		SD	0.66	0.79	2.70	0.472	0.750	1.85	0.045
		Minimum	10.5	8.1	50.7	4.05	7.77	53.9	1.20
		Maximum	14.1	11.2	63.6	6.40	11.26	60.9	1.37
May 8	97	Mean	13.2	7.5	56.2	5.11	9.10	55.5	1.24
		SD	0.97	0.92	1.72	0.447	0.794	2.07	0.051
		Minimum	10.5	5.9	50.8	4.29	7.68	50.7	1.12
		Maximum	14.8	10.0	59.9	6.35	11.51	61.3	1.38
May 23	112	Mean	13.1	6.9	55.7	4.97	8.92	55.3	1.24
		SD	0.96	0.58	2.28	0.701	1.176	1.80	0.044
		Minimum	10.8	5.9	50.4	3.93	7.12	51.4	1.14
		Maximum	14.9	8.6	60.3	6.67	12.14	58.7	1.32

ADL, acid-detergent lignin; asNDFom, NDF analysis using heat-stable, α-amylase and sodium sulfite during the extraction in neutral detergent, and subsequent correction for residual ash; CP, crude protein; DM, dry matter; NE_L, net energy of lactation; TDN, total digestible nutrients.

Quotient of ADL/asNDFom × 100%.

N = 45.

Mean concentrations of asNDFom ranged tightly from 54.5% to 58.5% across the last five harvest dates, which was a time interval of nearly 2 mo (57 d). Similar responses were observed for concentrations of ADL, which also ranged tightly (4.4% to 5.1%) across a 43-d period from April 10 to May 23. Although concentrations of these fiber components were relatively stable across later harvest dates, these responses obscure the changing proportions of leaf, stem, and grain-head tissues during this time period (Table 1), and especially the effects of dilution facilitated by a filling grain head that stabilize or reduce concentrations of fiber components during that time interval (Cherney and Marten, 1982a, 1982b). The mean percentage of asNDFom comprised of ADL ranged from 4.0% on February 27 up to 9.2% on April 24; according to NASEM (2001), triticale silage harvested at the heading stage of growth should contain 60% NDF and about 6% ADL, suggesting that about 10% of the structural cell wall is comprised of ADL. Within the present study, this percentage was approached at the early stages of grain fill but declined slightly thereafter. Similarly, energy density estimates suggested by NASEM (2001) for headed triticale silage (57.2% TDN) corresponded closely to the mean observed on April 24 (57.7 ± 1.85%), but there was little change observed thereafter as grain fill worked to offset the typical depressions in energy density associated with advancing plant maturity.

Identification of deviant cultivars

Concentrations of asNDFom

The TYPCUL exhibited cubic and quartic relationships with DYS and GRST, respectively (Figure 1). Coefficients of determination were ≥ 0.860, indicating that both independent variables were effective in explaining changes in concentrations of asNDFom, but the quartic relationship with GRST provided a superior fit (R² = 0.921). The higher-ordered nature of both response curves was largely created across later harvest dates by the competing processes of normal plant maturation and deposition of nonstructural carbohydrate in the developing grain head. Cherney and Marten (1982a) have described a collective, but similar, response to maturation/grain fill for four cereal-grain species, including triticale, in which cell wall concentrations increased until approximately 7 d after the first appearance of inflorescence, and then stabilized at < 55% of DM thereafter. Furthermore, in a companion report (Cherney and Marten, 1982b), concentrations of cell wall in leaf-blade, leaf-sheath, and stem tissues all continued to increase throughout plant maturation, but concentrations within the inflorescence declined sharply with grain fill, thereby explaining the relative balance between the competing factors observed on a whole-plant basis. Recently, Coblentz et al. (2018a) described a similar, and close (R² = 0.975), quartic relationship between concentrations of NDF and GRST for triticale forages grown in Wisconsin in which NDF declined by nearly 10 percentage units between the anthesis and soft-dough stages of growth. Similarly, Khorasani et al. (1997) described a curvilinear NDF response to maturity, where concentrations of NDF were maximized approximately 3 wk after boot stage at about 60% of DM before declining thereafter.

When evaluated with DYS as the independent variable, there were no linear coefficients for cultivars that deviated from a TYPCUL (P ≥ 0.101; Table 3); however, Merlin Max exhibited a significant (P = 0.001) positive deviation from TYPCUL when GRST was substituted as the independent variable (Table 4). This response is also illustrated in Figure 1 and indicates that concentrations of asNDFom generally ranged from 2 to 4 percentage units greater for Merlin Max compared with the TYPCUL after reaching a mid-boot stage of growth, although asNDFom rarely exceeded 60% of DM for any specific cultivar, and then only at the anthesis or flowering, and not at full maturity.

Table 3.

Deviations from the typical linear fixed effect coefficient based on an overall cubic model and their associated P-values (P > |t|) for 15 triticale cultivars grown during 2019 in Maricopa, AZ, where the fixed effect was days from February 1¹

Cultivar	---- asNDFom2 ------		------- ADL2 --------		---- NDFD302 ----		--- NDFD482 ----		--- IVDMD302 ---		--- IVDMD482 ---
	Estimate	P > |t|	Estimate	P > |t|	Estimate	P > |t|	Estimate	P > |t|	Estimate	P > |t|	Estimate	P > |t|
	----------------- % of DM -----------------				--------------- % of asNDFom --------------				-------------------- % of DM -----------------
Typical cultivar
Coefficient	1.393	—	0.0663	—	−0.401	—	0.250	—	−0.354	—	−0.107	—
Deviations from typical
105	0.002	0.833	−0.0014	0.371	−0.006	0.613	0.003	0.797	0.001	0.899	0.003	0.673
SY 115T	−0.010	0.231	0.0010	0.488	−0.002	0.896	−0.001	0.910	0.005	0.568	0.006	0.434
SY 158T	−0.005	0.498	0.0002	0.902	−0.001	0.913	−0.004	0.708	−0.002	0.819	−0.010	0.213
Merlin Max	0.013	0.101	0.0031	0.041	0.010	0.392	0.001	0.918	−0.002	0.801	0.000	0.987
Legend	−0.004	0.591	−0.0016	0.290	0.053	<0.001	0.054	<0.001	0.022	0.009	0.028	0.001
Swift 77	−0.002	0.806	0.0011	0.457	−0.043	0.001	−0.048	<0.001	−0.018	0.033	−0.027	0.001
Goldrush 91	0.011	0.180	−0.0010	0.497	0.000	0.994	0.009	0.372	−0.001	0.922	0.002	0.837
SY 158T	−0.006	0.437	0.0012	0.447	0.001	0.951	−0.005	0.625	0.001	0.940	−0.002	0.794
470-113	−0.004	0.603	−0.0031	0.042	0.035	0.005	0.032	0.003	0.007	0.368	0.011	0.150
470-133	−0.010	0.201	−0.0020	0.182	0.012	0.315	0.012	0.235	0.006	0.441	0.011	0.177
470-249	−0.002	0.821	0.0001	0.942	−0.008	0.511	−0.009	0.408	−0.007	0.428	−0.009	0.263
470-269	−0.006	0.438	−0.0036	0.019	0.005	0.695	0.003	0.755	0.008	0.310	0.006	0.432
470-285	0.009	0.256	−0.0008	0.612	−0.014	0.242	−0.015	0.158	−0.001	0.880	0.000	0.955
770-001	0.008	0.324	0.0032	0.035	−0.027	0.026	−0.024	0.024	−0.010	0.230	−0.018	0.027
770-113	0.007	0.363	0.0035	0.022	−0.014	0.246	−0.009	0.376	−0.010	0.212	−0.002	0.758

Deviations from the typical quadratic or cubic coefficients were not meaningful and are not shown.

Abbreviations: ADL, acid-detergent lignin; asNDFom, NDF assay included heat-stable, α-amylase and sodium sulfite, and subsequent correction for residual ash; DM, dry matter; NDFD30 and NDFD48, in vitro asNDFom disappearance following 30- and 48-h incubations, respectively; and IVDMD30 and IVDMD48, in vitro DM disappearance following 30- and 48-h incubations, respectively.

Table 4.

Deviations from the typical linear fixed effect coefficient based on an overall cubic model and their associated P-values (P > |t|) for 15 triticale cultivars grown during 2019 in Maricopa, AZ, where the fixed effect was growth stage at harvest¹

Cultivar	--- asNDFom2,3 ---		----- ADL2 -----		---- NDFD302 ----		--- NDFD482 ----		--- IVDMD302 ---		--- IVDMD482 ---
	Estimate	P > |t|	Estimate	P > |t|	Estimate	P > |t|	Estimate	P > |t|	Estimate	P > |t|	Estimate	P > |t|
	----------------- % of DM ---------------				-------------- % of asNDFom ------------				----------------- % of DM --------------
Typical cultivar
Coefficient	7.362	—	0.4326	—	−4.405	—	−2.743	—	−3.280	—	−2.488	—
Deviations from typical
105	−0.002	0.825	−0.0027	0.141	0.004	0.617	0.015	0.149	0.009	0.247	0.012	0.072
SY 115T	−0.013	0.099	0.0016	0.385	−0.002	0.840	−0.003	0.797	0.005	0.562	0.006	0.410
SY 158T	−0.008	0.311	0.0000	0.996	0.000	0.977	−0.001	0.946	−0.002	0.833	−0.010	0.153
Merlin Max	0.029	0.001	0.0064	0.002	−0.014	0.124	−0.035	0.002	−0.023	0.009	−0.024	0.002
Legend	−0.002	0.861	0.0006	0.740	0.009	0.298	0.019	0.086	0.008	0.345	0.011	0.149
Swift 77	−0.009	0.236	−0.0007	0.680	−0.009	0.283	−0.020	0.043	−0.007	0.365	−0.014	0.040
Goldrush 91	0.013	0.102	−0.0028	0.110	0.013	0.139	0.030	0.004	0.011	0.146	0.016	0.021
SY 158T	−0.008	0.313	0.0013	0.490	0.000	0.985	−0.004	0.730	0.000	0.970	−0.002	0.759
470-113	−0.002	0.817	−0.0015	0.438	0.003	0.742	0.003	0.778	−0.007	0.405	−0.005	0.477
470-133	−0.012	0.137	−0.0016	0.384	0.002	0.797	0.006	0.566	0.003	0.710	0.008	0.269
470-249	−0.002	0.820	0.0001	0.954	−0.004	0.655	−0.007	0.506	−0.007	0.376	−0.009	0.187
470-269	−0.012	0.128	−0.0044	0.017	0.005	0.585	0.006	0.564	0.012	0.127	0.010	0.172
470-285	0.012	0.119	−0.0014	0.434	−0.003	0.708	−0.007	0.487	0.003	0.690	0.006	0.410
770-001	0.006	0.420	0.0016	0.376	−0.001	0.924	0.001	0.927	0.002	0.812	−0.004	0.514
770-113	0.009	0.242	0.0036	0.050	−0.004	0.658	−0.003	0.793	−0.008	0.300	0.002	0.810

Deviations from the typical quadratic or cubic coefficients were not meaningful and are not shown.

Abbreviations: ADL, acid-detergent lignin; asNDFom, NDF assay included heat-stable, α-amylase and sodium sulfite, and subsequent correction for residual ash; NDFD30 and NDFD48, in vitro asNDFom disappearance following 30- and 48-h incubations, respectively; and IVDMD30 and IVDMD48, in vitro DM disappearance following 30- and 48-h incubations, respectively.

Based on a quartic relationship between asNDFom and growth stage, rather than the cubic model.

Acid detergent lignin

Concentrations of ADL increased as cubic functions of DYS or GRST with similar coefficients of determination (R² = 0.917 and 0.912, respectively; Figure 2). Although increases in ADL within triticale forages are expected to occur with plant maturity (Cherney and Marten, 1982a; Khorasani et al., 1997; Coblentz et al., 2018a), and are influenced positively by ambient temperature (Ford et al., 1979; Van Soest, 1982), the competing effects of grain fill on whole-plant concentrations of ADL are often inconsistent. Cherney and Marten (1982b) described sharp declines in ADL within the inflorescence during grain fill, but the collective effect of this process on whole-plant ADL is dependent on the competing increases in lignification within leaf, leaf sheath, and stem tissues. Concentrations of ADL declined between the anthesis and soft-dough stages of growth within triticale forages grown in Wisconsin (Coblentz et al., 2018a), but this response was less clear in the present study. In contrast, Khorasani et al. (1997) described a curvilinear relationship between ADL and time, where concentrations increased continuously through the soft-dough stage of growth. Using DYS as the independent regression variable, five cultivars exhibited significant deviations from the TYPCUL, with the maximum positive and negative deviations observed for 770-113 (P = 0.022) and 470-269 (P = 0.019), respectively (Table 3). Trendlines (Figure 2) indicated that both TYPCUL and 470-269 declined between May harvest dates, but 770-113, which deviated positively from TYPCUL, did not. This resulted in a 1.31-percentage unit differential between the maximum deviant cultivars on the final May 23 harvest date. An even greater differential (1.64 percentage units) in ADL was observed at the most advanced GRST (Figure 2); within this context, the maximum positive and negative deviation from TYPCUL were observed for Merlin Max (P = 0.002) and 470-269 (P = 0.017), respectively (Table 4).

Figure 2.

Relationships between ADL and days from February 1 (top) or growth stage (bottom), where the regression line (▬▬) and equation represent the response for a typical or mean cultivar. Gray or open circles with hashed lines (- - - -) are associated with cultivars exhibiting maximum positive or negative deviations from typical, respectively (P ≤ 0.022). Positive and negative deviant cultivars are identified in Tables 3 and 4. Abbreviation: ADL, acid detergent lignin.

In vitro asNDFom disappearance

The regression of NDFD30 (% of asNDFom) on DYS was best fit to a cubic model with a high coefficient of determination (R² = 0.964; Figure 3). Cultivars for which the linear coefficient deviated positively from TYPCUL were Legend (P < 0.001) and 470-133 (P = 0.005), while Swift 77 (P = 0.001) and 770-001 (P = 0.026) deviated negatively (Table 3). The differential in NDFD30 between the most extreme deviant cultivars (Legend and Swift 77) was about 13 percentage units of asNDFom throughout dates in March and April; during this interval, both cultivars maintained approximately equivalent deviations from the TYPCUL. From a practical perspective, Legend, Swift 77, and the TYPCUL exhibited parallel responses to DYS throughout those 2 mo. Unlike the relationship between NDFD and DYS, the cubic model for regressions on GRST yielded no significantly deviant cultivars (P ≥ 0.124; Table 4), and the overall fit (R² = 0.975) indicated that GRST is an excellent predictor variable for in vitro asNDFom disappearance (Figure 3), which has been noted in other studies with triticale (Coblentz et al., 2018b).

Figure 3.

Relationships between in vitro disappearance of asNDFom following a 30-h incubation (NDFD30) and days from February 1 (top) or growth stage (bottom), where the regression line (▬▬) and equation represent the response for a typical or mean cultivar. Gray or open circles with hashed lines (- - - -) are associated with cultivars exhibiting maximum positive (Legend) or negative (Swift 77) deviations from typical, respectively (P ≤ 0.001). Abbreviations: asNDFom, neutral detergent fiber conducted with heat-stable α-amylase and sodium sulfite, and with correction for residual ash; NDFD30, in vitro disappearance of asNDFom after 30 h.

As expected, the longer incubation time (48 h) resulted in generally increased asNDFom disappearance; the overall mean NDFD48 increased to 65.1% of asNDFom (6.9 percentage units) by extending the incubation time from 30 to 48 h. Regressions of NDFD48 on DYS generally yielded similar responses to those discussed for the 30-h incubation (Table 3; Figure 4). Significant positive deviations for the linear regression coefficient were again observed for Legend (P < 0.001) and 470-113 (P = 0.003), whereas negative deviations from the TYPCUL were again observed for Swift 77 (P < 0.001) and 770-001 (P = 0.024). The overall cubic regression model for the TYPCUL exhibited a similar fit compared with the 30-h incubation (R² = 0.966). The difference in NDFD48 between the most deviant cultivars (Legend and Swift 77) reached a maximum of 12.2 percentage units of asNDFom throughout mid-April. Generally, Swift 77 deviated from a TYPCUL in a relatively consistent manner throughout the entire 3-mo sampling period, ranging narrowly from −3.2 to −5.9 percentage units of asNDFom compared with the TYPCUL. In contrast, Legend exhibited less consistent differences from the TYPCUL, ranging from + 0.4 percentage units on February 27 to a maximum of + 7.2 percentage units in mid-April when most triticale plants were actively flowering.

Figure 4.

Relationships between in vitro disappearance of asNDFom following a 48-h incubation (NDFD48) and days from February 1 (top) or growth stage (bottom), where the regression line (▬▬) and equation represent the response for a typical or mean cultivar. Gray or open circles with hashed lines (- - - -) are associated with cultivars exhibiting the maximum positive or negative deviations from typical, respectively (P ≤ 0.004). Positive and negative deviant cultivars are identified in Tables 3 and 4. Abbreviations: asNDFom, neutral detergent fiber conducted with heat-stable α-amylase and sodium sulfite, and with correction for residual ash; NDFD48, in vitro disappearance of asNDFom after 48 h.

Exchanging GRST for DYS as the independent regression variable resulted in a cubic model exhibiting a similar fit to that described for DYS (R² = 0.975; Figure 4). Three cultivars deviated significantly from a TYPCUL, where Goldrush 91 deviated positively (P = 0.004), and both Swift 77 (P = 0.043) and Merlin Max (P = 0.002) deviated negatively (Table 4). For NDFD48, the maximum differential between the most extreme deviants (Goldrush 91 and Merlin Max) was 13.3 percentage units of asNDFom, which occurred at the first appearance of inflorescence (GRST = 50 to 51); however, this differential was much smaller earlier during stem elongation, or later during grain fill.

In vitro DM disappearance

Much like responses discussed for NDFD30 and NDFD48, the cubic relationship between (apparent) IVDMD30 (% of DM) and DYS exhibited an R² statistic indicative of a good overall fit (R² = 0.949; Figure 5). Only two cultivars exhibited statistically significant deviations from the TYPCUL (Table 3); these included a positive deviation by Legend (P = 0.009) and a negative deviation by Swift 77 (P = 0.033). Generally, Legend, Swift 77, and the TYPCUL exhibited parallel responses to DYS through the April 10 harvest date; during this time interval, the maximum range between positive and negative deviants was 8.1-percentage units of DM, which occurred in late March when triticale plants corresponded generally to the booting stage of growth. After April 10, Legend maintained a consistent advantage over the TYPCUL that ranged narrowly from 2.5 to 3.5 percentage units of DM through the end of the sampling period. In contrast, the differential between Swift 77 and the TYPCUL narrowed after April 10, ranging from 1.1 to 1.8 percentage units of DM during May. The substitution of GRST for DYS as the independent variable yielded a similar excellent fit for the cubic regression relationship (R² = 0.966); however, only one cultivar (Merlin Max) deviated from the typical response (P = 0.009; Table 4). Merlin Max did not exhibit a consistently negative deviation across all GRST, deviating by a maximum of 4.9 percentage units of DM at the late-flowering to early-fruit-development stages of growth, but only minimally at both early and advanced plant maturities (Figure 5).

Figure 5.

Relationships between in vitro disappearance of DM following a 30-h incubation (IVDMD30) and days from February 1 (top) or growth stage (bottom), where the regression line (▬▬) and equation represent the response for a typical or mean cultivar. Gray or open circles with hashed lines (- - - -) are associated with cultivars exhibiting the maximum positive or negative deviations from typical, respectively (P ≤ 0.033). Positive and negative deviant cultivars are identified in Tables 3 and 4. Abbreviations: DM, dry matter; IVDMD30, in vitro DM disappearance after 30 h.

Averaged over all cultivars and harvest dates, the extension of incubation time to 48 h resulted in increased IVDMD48 compared with the shorter (30-h) incubation time (71.7% vs. 66.0% of DM). Generally, the cubic response curves using DYS as the independent variable exhibited similar trends to those described for the 30-h incubation, with Legend (P = 0.001) and Swift 77 (P = 0.001) again exhibiting positive and negative deviations from the TYPCUL (Figure 6; Table 3). Following the 48-h incubation, 770-001 also exhibited a negative deviation for the linear coefficient (P = 0.027). With GRST designated as the independent regression variable (Table 4), Merlin Max deviated negatively (P = 0.002) as did Swift 77 (P = 0.040), but to a lesser extent. Unlike the 30-h incubation, Goldrush 91 exhibited a positive deviation (P = 0.021). The maximum differential between Goldrush 91 and Merlin Max was 7.5 percentage units of DM with approximately equivalent deviations from the TYPCUL, and this occurred at the early-heading stage of growth. However, deviations were essentially nil early and late in plant development.

Figure 6.

Relationships between in vitro disappearance of DM following a 48-h incubation (IVDMD48) and days from February 1 (top) or growth stage (bottom), where the regression line (▬▬) and equation represent the response for a typical or mean cultivar. Gray or open circles with hashed lines (- - - -) are associated with cultivars exhibiting the maximum positive or negative deviations from typical, respectively (P ≤ 0.004). Positive and negative deviant cultivars are identified in Tables 3 and 4. Abbreviations: DM, dry matter; IVDMD48, in vitro DM disappearance after 48 h.

Correlations between plant growth characteristics and NE_L

Considered across all harvest dates, the percentage of total plant DM partitioned within the grain head, canopy height, estimated yield, and GRST all exhibited strong negative correlations (r ≤  −0.814; P < 0.001) with energy density, expressed as NE_L (Table 5). Among these correlations, the most counterintuitive is the negative association of the filling grain head with energy; however, within the context of all harvest dates, grain-head percentage is likely more reflective of advancing plant maturity than the positive energy benefits of stored nonstructural carbohydrates at later harvest dates. Percentage of leaf tissue exhibited the strongest positive association with NE_L (r = 0.949; P < 0.001), as it decreased steadily across harvest dates, irrespective of whether this occurred indirectly through accelerated deposition of stem tissue or grain fill, or directly via leaf senescence. The weak positive association between NE_L and percentage of the stem (r = 0.138; P = 0.014) also is counterintuitive, but likely explained by declining percentages of the stem from March 27 (59.5 ± 7.51%) through May 23 (32.3 ± 6.91%), which coincided with reductions in energy density, despite the deposition of nonstructural carbohydrate in the grain head during this time interval.

Table 5.

Pearson correlation coefficients for the relationships between net energy of lactation (NE_L) and plant growth characteristics for 15 triticale cultivars grown during 2019 in Maricopa, AZ

Harvest date	Statistic	Days from February 1	Growth stage	Leaf	Stem	Grain head	Height, cm	Estimated yield, kg DM/ha
				-------- % of dry matter (DM) --------
All dates	N	315	315	315	315	315	315	315
	Mean	69	60	33.4	44.8	21.7	91	10,443
	SD	—	23.1	17.42	10.62	22.1	28.2	5,323.8
	Minimum	27	31	9.6	19.2	0	25	1,583
	Maximum	112	99	69.1	72.8	61.7	137	21,683
	r	−0.918	−0.879	0.949	0.138	−0.814	−0.913	−0.886
	P > │r│	<0.001	<0.001	<0.001	0.014	<0.001	<0.001	<0.001
Individual harvest dates (N = 45)
February 27	r	—	−0.654	0.569	−0.569	—	−0.506	−0.604
	P > │r│	—	<0.001	<0.001	<0.001	—	<0.001	<0.001
March 14	r	—	−0.612	0.394	−0.394	—	−0.578	−0.510
	P > │r│	—	<0.001	0.007	0.007	—	<0.001	<0.001
March 27	r	—	−0.439	0.274	−0.231	−0.246	−0.575	−0.606
	P > │r│	—	0.003	0.068	0.126	0.103	<0.001	<0.001
April 10	r	—	−0.436	0.558	−0.088	−0.357	−0.372	0.162
	P > │r│	—	0.003	<0.001	0.566	0.016	0.012	0.287
April 24	r	—	−0.052	0.231	−0.301	0.054	−0.377	−0.066
	P > │r│	—	0.736	0.126	0.044	0.726	0.011	0.666
May 8	r	—	−0.195	0.312	−0.090	−0.058	−0.161	0.128
	P > │r│	—	0.199	0.037	0.558	0.706	0.292	0.401
May 23	r	—	−0.063	0.274	0.047	−0.138	0.014	0.354
	P > │r│	—	0.680	0.068	0.759	0.365	0.925	0.017

Within harvest date, correlations through March 27 generally paralleled those observed across all harvest dates, with negative associations of GRST (r ≤  −0.439; P ≤ 0.003), canopy height (r ≤  −0.506; P < 0.001), and estimated DM yield (r ≤  −0.510; P < 0.001) with NE_L. The percentage of leaf was positively correlated with NE_L (r ≥ 0.394; P ≤ 0.007) for the February 27 and March 14 harvest dates but only tended to be correlated by March 27 (r = 0.274; P = 0.068). In contrast to trends observed across all harvest dates, percentages of stem were negatively associated with energy on the first two harvest dates (r =  −0.349; P ≤ 0.007), but there was no correlation detected on March 27 (P = 0.126). On April 10, which roughly corresponded to a mean late-heading stage of growth, negative correlations with NE_L were detected for GRST, percentage of grain head, and canopy height (r ≤  −0.357; P ≤ 0.016), whereas a strong positive correlation was detected for the percentage of leaf (r = 0.558; P < 0.001). After that harvest date, only sporadic correlations were detected.

Identification of deviant cultivars for NE_L

Unlike most dependent variables evaluated in this project, a natural-log (ln) transformation of both independent regression variables (DYS and GRST) was required to achieve model convergence in evaluations of NE_L. Within this context, energy density declined over ln-transformed DYS in a cubic pattern with a very high coefficient of determination (R² = 0.960); from a practical perspective, the response was primarily linear in nature, but with NE_L then stabilizing over the final 2 harvest dates. Across cultivars, 5 of the 14 entries deviated from the TYPCUL (Table 6) with Merlin Max, Swift 77, and 770-001 deviating negatively (P ≤ 0.030), whereas Legend and 470-133 deviated positively (P ≤ 0.035). Among these, 770-001 and Legend were the most extreme respective cases (Figure 7), exhibiting a maximum differential of about 0.15 Mcal/kg that was maintained generally from late-March throughout most of April. However, maximum positive and negative deviations from the TYPCUL did not overlap chronologically. Legend displayed the maximum deviation from the TYPCUL (0.09 Mcal/kg) late in the plant development cycle, while 770-001 deviated most negatively (0.10 Mcal/kg) during the month of March. Figure 8 depicts the growth characteristics for Legend, 770-001, and the TYPCUL as a function of harvest date. Within that context, Legend triticale was less mature compared with the TYPCUL or 770-001 across all harvest dates, but especially from March 27 through May 8. The slower maturation rate was also accompanied by reduced canopy height as well as a greater percentage of leaf on all harvest dates. Legend also exhibited a lower percentage of stem tissue across the initial three harvest dates, reaching a maximum differential of about 19 percentage units of DM compared with negatively deviant 770-001 on March 27; however, this relationship with other cultivars was not maintained thereafter, nor did Legend ever exhibit a greater percentage of head weight than 771-001 or TYPCUL. In contrast, the maximum negative deviation in NE_L that was exhibited early in the maturation cycle by 770-001 can be directly associated with a more rapid maturation rate, greater canopy height, and a reduced leaf/stem ratio compared with either TYPCUL or Goldrush 91.

Table 6.

Deviations for net energy of lactation (NE_L) from the typical linear fixed effect coefficient based on an overall cubic model and their associated P-values (P > |t|) for 15 triticale cultivars grown during 2019 in Maricopa, AZ, where the fixed effects were days from 1 February (DYS) or growth stage at harvest (GRST)¹

Cultivar	------- DYS --------		------- GRST --------
	Estimate	P > |t|	Estimate	P > |t|
Typical cultivar
Coefficient	16.372	—	−18.280	—
Deviations from typical
105	0.003	0.306	0.006	0.041
SY 115T	0.004	0.166	0.002	0.473
SY 158T	0.001	0.836	0.001	0.702
Merlin Max	−0.006	0.030	−0.015	<0.001
Legend	0.013	<0.001	0.004	0.259
Swift 77	−0.010	0.001	−0.002	0.551
Goldrush 91	0.002	0.495	0.008	0.015
SY 158T	0.000	0.991	0.000	0.912
470-113	0.003	0.309	−0.004	0.157
470-133	0.006	0.035	0.003	0.350
470-249	−0.004	0.182	−0.003	0.273
470-269	0.003	0.245	0.004	0.194
470-285	0.002	0.388	0.004	0.172
770-001	−0.013	<0.001	−0.005	0.116
770-113	−0.005	0.094	−0.003	0.407

Both independent variables were transformed with a natural-log transformation to ensure model convergence. Deviations from the typical quadratic or cubic coefficients were not meaningful and are not shown.

Figure 8.

Growth stage and physical characteristics of the mean or typical cultivar as well as cultivars exhibiting the maximum positive (Legend) and negative (770-001) deviations from the typical cultivar for the relationship between NE_L and days from February 1 (see Figure 7, top). The cultivar “Legend,” which exhibits the maximum positive deviation from a typical cultivar with respect to NE_L, matures less rapidly, is shorter in height, and has greater percentages of leaf across harvest dates than other cultivars. However, percentages of head weight are also less across harvest dates for Legend. Abbreviation: NE_L, net energy of lactation.

The use of ln-transformed GRST as the regression independent variable theoretically removes the effects of differing maturation rates among cultivars compared with ln-transformed DYS. The overall cubic relationship with NE_L again exhibited a high coefficient of determination (R² = 0.956; Figure 7), but the linear coefficient for only three individual cultivars deviated significantly (P ≤ 0.041) from the TYPCUL response (Table 6). Goldrush 91 exhibited the largest positive deviation (P = 0.015), indicating that it was significantly more energy dense than a TYPCUL; in contrast, Merlin Max exhibited a negative deviation (P < 0.001) from the TYPCUL that was larger in magnitude and reached a maximum of about −0.14 Mcal/kg at the boot/early-heading stage of growth. At that stage, the differential in energy density (NE_L) between Goldrush 91 and Merlin Max cultivars was equivalent to about 0.23 Mcal/kg. Graphical representations of plant physical characteristics for Merlin Max, Goldrush 91, and the TYPCUL are shown in Figure 9; these illustrations indicate that the negative deviations for Merlin Max compared with the TYPCUL or Goldrush 91 were likely associated with a greater plant height across all GRST, greater percentage of stem once plants reached anthesis, and reduced head weights beginning at the anthesis stage of growth.

Figure 9.

Physical characteristics of the mean or typical cultivar (▬▬) as well as cultivars exhibiting the maximum positive (Goldrush 91, ▬▬) and negative (Merlin Max, ▬ ▬ ▬) deviations from the typical cultivar for the cubic relationships between NE_L and growth stage (see Figure 7, bottom). The cultivar “Merlin Max” is taller in height, has greater percentages of stem at advanced growth stages, and exhibits less head weight than other cultivars at comparable growth stages. Abbreviation: NE_L, net energy of lactation.

Indigestible asNDFom

Grant (2015) has described indigestible asNDFom as a baseline of ruminal fill that constrains possible NDF flux, where a minimum pool is necessary to maintain proper rumen function, but exceeding an upper threshold may limit voluntary intake. Within that context, the daily intake of indigestible asNDFom by high-producing cows typically ranges from 0.30% to 0.48% of BW. This concept requires an estimate of asNDFom indigestibility at extended incubation times (240 h) and is also critical to the 3-pool model of fiber digestibility that includes rapidly and slowly digestible pools, coupled with an indigestible fraction (Raffrenato and Van Ambaugh, 2010; Grant, 2015; Mertens, 2016). Indigestible asNDFom for selected cultivars each exhibited a strong linear (P < 0.001) response to harvest date, but higher-ordered effects also were observed in each case, including either significant cubic (P ≤ 0.047), and (less commonly) quartic (P ≤ 0.021) contrasts (Table 7). Perhaps more importantly were comparisons among categories on specific harvest dates. Consistently positively deviant cultivars from TYPCUL with respect NDFD48 exhibited less (P ≤ 0.005) indigestible NDF than TYPCUL on all harvest dates after March 14. Similarly, positive deviants differed (P ≤ 0.023) from negative deviants on all harvest dates except February 27 (P = 0.499). However, there were no differences (P ≥ 0.069) between negative deviants and TYPCUL across all dates except for April 10 (P = 0.006). Overall, indigestible asNDFom increased across harvest dates as plants matured from 6.4% to 29.2% of asNDFom, where the maximum difference between positive-deviant cultivars and TYPCUL was 5.7 percentage units of asNDFom. Previous estimates of indigestible asNDFom for triticale forages have ranged from 11.6% to 44.3% and 9.9% to 37.3% of asNDFom over 2 yr and similar maturity ranges, although those evaluations were terminated at the soft-dough stage of growth (Coblentz et al., 2018b).

Table 7.

Abbreviated analysis of indigestible asNDFom¹ (% of asNDFom) of selected triticale cultivars harvested in Maricopa, AZ, and determined following a 240-h incubation in buffered ruminal fluid

Cultivar	------------------------- Harvest date -------------------------							-------------- Contrasts,2 P > F --------------
	February 27	March 14	March 27	April 10	April 24	May 8	May 23	Linear	Quadratic	Cubic	Quartic
Goldrush 91	6.7	8.2	12.4	20.8	23.8	25.2	25.7	<0.001	<0.001	<0.001	0.021
Legend	5.9	6.1	8.8	14.4	20.1	23.4	23.8	<0.001	0.571	<0.001	0.812
SY 115T	5.1	8.5	14.4	21.5	28.7	31.4	32.5	<0.001	0.001	<0.001	0.609
SY 158T (a)	5.8	9.8	15.4	20.7	26.8	27.1	27.9	<0.001	<0.001	0.002	0.421
SY 158T (b)	6.6	7.3	13.9	20.2	26.4	27.7	30.5	<0.001	0.033	<0.001	0.010
770-001	9.7	11.0	15.6	23.8	29.0	28.9	30.4	<0.001	0.001	<0.001	0.008
Merlin Max	5.3	7.5	13.5	20.5	27.0	30.0	34.4	<0.001	0.584	0.001	0.067
Swift 77	6.1	10.5	16.2	18.6	25.9	27.0	28.1	<0.001	<0.001	0.047	0.570
SEM	------------------------- 0.85 -------------------------
Category means3
Positive	6.3	7.2	10.6	17.6	22.0	24.3	24.8
Typical	6.8	9.2	14.8	21.6	27.7	28.8	30.3
Negative	5.7	9.0	14.9	19.6	26.5	28.5	31.3
Contrasts4	-------------------------P > F -------------------------
Positive vs. Typical	0.461	0.005	<0.001	<0.001	<0.001	<0.001	<0.001
Typical vs. Negative	0.131	0.846	0.999	0.006	0.069	0.742	0.171
Positive vs. Negative	0.499	0.023	<0.001	0.017	<0.001	<0.001	<0.001

asNDFom, neutral detergent fiber conducted with heat-stable α-amylase and sodium sulfite, and with correction for residual ash.

Linear, quadratic, cubic, and quartic effects of harvest date. Harvest dates were assumed to be equally spaced.

Category means: i) Positive, (Legend, Goldrush 91); ii) Negative (Merlin Max, Swift 77); and iii) Typical (SY 115T, SY 158T a,b, 770-001), and category groupings were based on positive, negative, or generally nonsignificant deviations from a typical cultivar for in vitro disappearance of asNDFom following 30- or 48-h incubations in buffered ruminal fluid.

Contrast comparisons of category groupings.

Conclusions

Energy estimates (NE_L) for triticale cultivars based on calendar or harvest date were primarily affected by GRST at the time of harvest; deviant cultivars from TYPCUL usually demonstrated rapid or slow maturation rates, which were often accompanied by physical characteristics reflective of advanced or slow maturation, respectively. Within that context, the maximum positive energy deviant from TYPCUL (Legend) exhibited physical characteristics largely consistent with its slower maturation rate, including reduced plant height, greater leaf percentage, and reduced stem percentage prior to grain fill. The use of GRST as the independent regression variable essentially removes the maturation rate from the analysis. Within that context, traits associated with negative deviation from TYPCUL included greater plant height, greater stem percentage beginning at anthesis and continuing until physiological maturity, and a reduced head-weight percentage. Overall, there were a limited number of cultivars that deviated from TYPCUL with respect to NE_L as well as other response variables; however, the 0.23 Mcal/kg differential between maximum positive and negative deviants for NE_L at a common late-boot/early-heading stage of growth suggests that some attention should be placed on cultivar selection, especially when cultivar growth characteristics are atypical.

Glossary

Abbreviations

ADF

acid detergent fiber

ADICP

acid-detergent insoluble crude protein

ADL

acid-detergent lignin

CP

crude protein

DM

dry matter

NDICP

neutral detergent insoluble crude protein

NE_L

net energy of lactation

TDN

total digestible nutrients

TYPCUL

typical or mean cultivar

Conflict of Interest Statement

The authors declare no conflicts of interest associated with this research.