Effects of harvest date and growth stage on triticale forages in the southwest USA: kinetics of in vitro disappearance of fiber and dry matter

Abstract

Recently, there has been interest in including triticale (X Triticosecale Wittmack) within forage programs in the southwest USA. Our objectives were to evaluate in vitro disappearance kinetics of neutral detergent fiber (NDF) and dry matter (DM) for cultivars identified during 2019 as positively or negatively deviant from typical cultivars, based specifically on regressions of 48-h in vitro disappearance of NDF on growth stage (GRST). All NDF analyses included the use of heat-stable α-amylase and sodium sulfite, as well as correction for residual ash (asNDFom). Seven triticale cultivars were established on December 18, 2019 at the University of Arizona Maricopa Agricultural Center, located near Maricopa, AZ. Forage plots were arranged in a randomized complete block design with three complete blocks (replications), and then harvested on seven dates the following late-winter and spring (February 26, March 17, April 1, April 14, April 28, May 12, and May 26). Based on a linear model, GRST was highly variable among cultivars on March 17 (44 ± 10.6), April 1 (57 ± 12.1), April 14 (67 ± 8.9), and April 28 (79 ± 7.2) compared with other harvest dates (SD ≤ 1.7). For concentrations of asNDFom, all cultivars exhibited linear (P ≤ 0.042) and quadratic (P < 0.001) polynomial contrasts in response to harvest date, and all cultivars except Merlin Max (P ≥ 0.063) exhibited at least one additional cubic or quartic effect (P ≤ 0.015). A contributing factor to the unique response by Merlin Max was the numerically greater maximum canopy height (145 ± 9.8 cm) compared with the mean of all cultivars (107 ± 17.7 cm), which also was associated with greater percentages of stem, as well as reduced percentages of DM partitioned within the grain head. Regressions of asNDFom disappearance after 30- or 48-h incubations on GRST indicated this was an effective independent variable (R² ≥ 0.927), and responses were most often linear in nature. Generally, relationships for DM disappearance were quadratic, ostensibly due to the complicating effect of grain fill, but GRST was again an effective predictor variable with R² statistics ≥ 0.852 for 12 of 14 combinations of cultivar and incubation time. Predicted percentages of digestible DM attributed to asNDFom disappearance were ≥50.3% through the fully flowered stage of growth, but digestible contributions from nonfiber components following the onset of grain fill profoundly affected overall DM digestibility among cultivars harvested at later GRST.

Lay Summary

Recently, there has been increased interest in including triticale within forage programs throughout the southwest USA. Unless there is an urgency for removing the triticale crop, such as those created by a feed shortage or need to establish a secondary crop, harvest management decisions should be based on plant growth stage, and not calendar date. Assuming a common growth stage, this work suggests that most triticale cultivars will differ only modestly with respect to digestibility before the onset of grain fill. However, producers should be cautious of cultivars with unique or atypical phenotypic traits, such as exceptional canopy height, which may cause exceptions to the previous generalization. If yield is a critical management objective, harvest should most likely be delayed until after the onset of grain fill, but cultivar selection can become more complicated at that time because varying contributions from the filling grain head can radically affect overall digestibility of dry matter (DM). In this respect, producers should carefully evaluate their nutritional and production goals to assess whether their needs prioritize digestible fiber or overall DM digestibility, the latter of which can have limited contributions from digestible fiber.

Assuming a common growth stage, most triticale cultivars will differ only modestly with respect to digestibility before the onset of grain fill. However, appropriate cultivar selection can become more complicated with grain fill as varying contributions from the filling grain head can radically affect overall digestibility of dry matter.

Introduction

Recently, triticale (X Triticosecale Wittmack) frequently has been incorporated into cropping strategies for dairy operations, particularly within the mid-Atlantic region of the USA. This trend is partially associated with mitigation of environmental concerns, such as a desire to capture nutrients originating from land-applied manure, or to provide winter ground cover that may improve soil health and general land stewardship. Very recently, there has been increased interest in including triticale or other cereal-grain forages within dairy management programs throughout the southwest USA. One motivation for this interest is a desire to improve water-use efficiencies (Frisvold, 2015), which differs (in part) from the most common environmental stewardship and double-cropping priorities suggested previously. Frisvold (2015) determined that water application rates for wheat and barley were 26% less than needed for other crops, and specifically 24% less than required for corn silage (Zea mays L.). However, rigorous in-depth evaluations of agronomic and forage nutritive value characteristics have been conducted only sporadically throughout the region when compared against the north-central and mid-to-upper Atlantic portions of the USA, and more detailed evaluations of winter cereals as forages have been solicited by the dairy industry in Arizona.

For triticale, numerous studies have described relationships between various measures of nutritive value and plant part, calendar date, growth stage, or growing degree days (Khorasani et al., 1997; Cherney and Marten, 1982a, 1982b; Coblentz et al., 2018a, 2018b). Generally, dry matter (DM) digestibility and energy density of triticale forages often stabilize or increase as plants partition nonstructural carbohydrate into the filling grain head, but digestibility of neutral detergent fiber (NDF) is largely independent of this process and declines with plant maturity. As a result, extension recommendations for including these forages in the diets of lactating dairy cows often specify harvesting at the boot stage of growth (Kilcer et al., 2010). Unfortunately, applying this management strategy in central Wisconsin carries a (70%) yield reduction compared with a soft-dough harvest timed several weeks later, but an earlier harvest does permit a double-crop of corn (Coblentz et al., 2018a), most likely to be harvested and fermented into silage.

In a companion study (Coblentz et al., XXXX), 14 triticale cultivars grown during 2019 in Maricopa, AZ, were evaluated for agronomic and nutritive value traits. Emphasis was placed on identifying cultivars that deviated significantly from the mean or typical cultivar for concentrations of NDF determined with sodium sulfite and heat-stable α-amylase, and then corrected for residual ash (asNDFom), acid detergent lignin (ADL), energy density, and in vitro determinations of DM and asNDFom disappearance at specific 30- and 48-h incubation intervals. These response variables were all related closely to both harvest date, as well as growth stage at harvest. Deviations from the mean or typical cultivar for energy density (NE_L) were largely related to cultivar maturity rate when regressed against harvest date (days from February 1). Alternatively, the elimination of maturation rate, by regressing on a linear growth stage model, suggested that plant height, as well as percentages (by weight) of stem and grain head were important factors in explaining NE_L. Our objectives for this study were to further evaluate cultivars that demonstrated positive (Goldrush 91 and Legend), negative (Merlin Max and Swift 77), or no (SY115T, SY158T, and 770-001) deviations from typical cultivars (TYPCUL) during the 2019 harvest season with specific respect to in vitro disappearance of asNDFom following a 48-h incubation. These selected cultivars were further evaluated in a condensed experiment conducted during 2020 that evaluated in vitro kinetics of asNDFom and DM disappearance using the ANKOM Daisy^II Incubator System.

Materials and Methods

For in vitro disappearance assays, standards of care for donor animals and procedures for obtaining inoculum were approved by the Research Animal Resources Committee (RARC) of the University of Wisconsin-Madison (Protocol #A01006).

Cultivar selection

Based on the results from the companion screening study conducted with forages harvested during the previous spring of 2019 (Coblentz et al., XXXX), seven cultivars were selected for more extensive analysis. Selections were based specifically on positive (POS), negative (NEG), or no significant deviations from TYPCUL based on the linear coefficient of the cubic regression model of 48-h in vitro disappearance of asNDFom on growth stage (GRST). Positive and negative deviations imply characteristics of fiber digestibility that are better or worse than TYPCUL, respectively. Positive or negative deviations based on other dependent regression variables were not considered as criteria for cultivar selection. As such, a positive deviation (P = 0.004) was observed for Goldrush 91, and a positive tendency for Legend (P = 0.086). Conversely, negative deviations from TYPCUL were observed for Merlin Max (P = 0.002) and Swift 77 (P = 0.043). The TYPCUL group (SY115T, SY158T, and 770-001) did not approach deviation of the linear coefficient based on GRST (P ≥ 0.797). Although cultivar selections were based on the criteria described, the previous associated regressions on days from February 1 (chronological time) potentially establish interesting contrasts within groups. Within the POS group, Goldrush 91 did not deviate (P = 0.372) based on harvest timing, but Legend was greatly affected (P < 0.001). For the NEG group, Merlin Max did not deviate from TYPCUL based on time (P = 0.918), but Swift 77 deviated strongly on this basis (P < 0.001). Among TYPCUL, 770-001 deviated negatively (P = 0.024) based on harvest date, but the other TYPCUL did not (P ≥ 0.708).

Forage establishment and management

Generally, agronomic management of field plots was similar to that described for the cultivar screening study conducted the previous year (Coblentz et al., XXXX). The 7 triticale cultivars selected for this trial were established in 1.8 × 6.1-m plots with a cone planter configured to provide 10 rows per plot spaced at 18-cm intervals. Cultivars were planted into a clean-tilled seedbed on December 18, 2019 at the University of Arizona Maricopa Agricultural Center, located near Maricopa, AZ (33°06ʹN; 112°05ʹW). The soil type at this location was a Casa Grande sandy-loam soil (reclaimed fine-loamy, mixed, superactive, hyperthermic, Typic Natrargid). The plot layout consisted of 21 total plots, subdivided into 3 field blocks, with each cultivar represented once within each field block. The seeding rate for all cultivars was approximately 146 kg/ha. Six of the selected cultivars (SY-115T, SY-158T, Merlin Max, Legend, Swift 77, and Goldrush 91) were commercially available, and obtained from Northern Agri Brands, LLC (Great Falls, MT), while cultivar 770-001 was an experimental line obtained from Arizona Plant Breeders, Inc. (Casa Grande, AZ) that was unavailable to the public at the time of the trial. Plots were irrigated (182 mm) on December 18 to facilitate germination. Subsequent irrigation events occurred on February 5 (101 mm), February 28 (110 mm), March 18 (110 mm), April 2 (66 mm), April 16 (118 mm), and April 30 (80 mm), which totaled 767 mm during the entire trial. Irrigation water was applied by the border flood method, where the quantity of water supplied was calculated by formula that included inputs from the number and diameter of siphon tubes, height of water in the drainage ditch, and the duration of irrigation. Urea (46-0-0) was applied on four dates (February 5, February 28, March 18, and April 16) at respective rates of 113, 57, 40, and 38 kg N/ha. Furthermore, an application of urea-ammonium nitrate solution (32-0-0) was applied at 39 kg N/ha on April 2; therefore, the cumulative N-fertilization rate for the trial was 288 kg N/ha. These fertilization and irrigation management strategies are consistent with current extension recommendations for the area (Husman and Ottman, 2015; Ottman and Thompson, 2015), as well as with those used the previous year at the same location (Coblentz et al. XXXX). No phosphorus fertilizer, herbicides, or pesticides were applied during the trial. During the experiment, mean maximum monthly temperatures were 17.2, 20.0, 22.8, 23.3, 30.6, and 37.2 °C for December, January, February, March, April, and May, respectively. Similarly, respective mean monthly minimum temperatures were 3.9, 2.2, 3.3, 8.9, 12.2, and 17.8 °C, while respective monthly mean temperatures were 10.0, 10.0, 12.2, 16.1, 21.7, and 28.3 °C, where mean temperatures were based on hourly assessments, rather than the simple mean of maximum and minimum temperatures.

Forages were harvested on approximate 2-wk intervals across seven dates throughout the spring of 2020 (February 26, March 17, April 1, April 14, April 28, May 12, and May 26). On each harvest date, 10 plants per plot were assessed for GRST using the linear system described by Stauss (1994), where 30–39 = stem elongation, 40–49 = boot, 50–59 = heading, 60–69 = anthesis, 70–79 = fruit development, 80–89 = ripening, and 90–99 = senescence. Although the maturity ratings for individual cultivars varied on specific harvest dates (Table 1), mean respective GRST on these dates corresponded to the late-tillering (25 ± 0.5), mid-boot (44 ± 10.6), mid-heading (57 ± 12.1), full-flowering or anthesis (67 ± 8.9), late-milk (79 ± 7.2), hard-dough (87 ± 1.7), and fully ripe (89) stages of growth. In addition to assessment for GRST, 10 plants per plot were measured for height (cm), and then separated into leaf, stem, and grain-head tissue types. After hand-separation, tissue samples were dried to constant weight under forced air at 55 °C, and then percentages of leaf, stem, and grain-head tissues were calculated for each experimental plot. Yield of DM was estimated by hand-clipping a 0.46-m segment of two center rows, drying to constant weight as described earlier, and then extrapolating the dry weights to a kg per hectare basis. Dried whole-plant samples were retained for subsequent assessment of nutritive value and in vitro kinetics of asNDFom and DM disappearance.

Table 1.

Growth characteristics of seven selected triticale cultivars grown during 2020 in Maricopa, AZ

Harvest date	Days from February 1	Statistic	Growth stage1	Height	Leaf	Stem	Grain head	Estimated yield
				cm	% of DM			kg DM/ha
All dates (N = 147)	-	Mean	64	87	29.1	42.4	28.5	13,320
		SD	23.1	28.6	21.71	13.16	24.71	7,325.6
		Minimum	25	25	4.6	12.8	0	494
		Maximum	89	151	87.2	76.1	70.2	27,828
February 26 (N = 21)	26	Mean	25	36	70.8	29.2	0	1,649
		SD	0.5	5.8	8.23	8.23	0	552.3
		Minimum	25	25	55.4	12.8	0	494
		Maximum	26	48	87.2	44.6	0	2,685
March 17 (N = 21)	46	Mean	44	64	46.2	49.4	4.4	5,682
		SD	10.6	9.0	9.19	6.35	8.12	964.9
		Minimum	32	50	31.9	37.8	0	3,583
		Maximum	59	78	62.2	60.3	20.2	7,734
April 1 (N = 21)	61	Mean	57	90	29.6	57.4	13.0	10,961
		SD	12.1	13.6	7.63	5.64	11.58	1,556.8
		Minimum	39	67	20.4	47.4	0	7,474
		Maximum	69	113	43.4	67.4	26.3	14,386
April 14 (N = 21)	74	Mean	67	104	21.3	54.8	24.0	15,098
		SD	8.9	11.9	4.44	8.19	11.21	2,384.5
		Minimum	45	84	14.4	48.5	0	11,441
		Maximum	72	122	28.3	76.1	37.2	19,285
April 28 (N = 21)	88	Mean	79	107	17.4	40.6	42.0	18,612
		SD	7.2	17.7	3.95	9.11	12.49	2,195.3
		Minimum	69	87	11.4	30.1	18.2	14,770
		Maximum	85	151	25.0	59.5	58.2	22,804
May 12 (N = 21)	102	Mean	87	105	9.0	34.2	56.8	22,034
		SD	1.7	16.0	2.57	8.76	10.44	2,019.7
		Minimum	85	87	5.9	26.6	35.0	18,885
		Maximum	89	142	14.3	53.4	66.9	26,092
May 26 (N = 21)	116	Mean	89	103	9.1	31.4	59.5	19,207
		SD	–	18.2	2.91	8.16	9.72	3,815.2
		Minimum	89	87	4.6	24.1	34.0	12,640
		Maximum	89	145	18.6	54.2	70.2	27,828

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

Laboratory procedures

Nutritive value

Dried, whole-plant plot samples (N = 147) were ground through a 1-mm screen in a Thomas Model 4 Wiley Mill (Thomas Scientific, Swedesboro, NJ), and stored at room temperature in paper envelopes pending subsequent analysis of nutritive value and in vitro disappearance kinetics. Laboratory DM was determined for each sample by drying each forage (1.0 g) for at least 6 h in a convection oven at 105 °C. Whole-plant ash was then determined by moving these same subsamples to a muffle furnace and combusting them at 500 °C for 6 h. Concentrations of N in each sample were assessed with a rapid combustion procedure (AOAC, 1998, Official Method 990.03; Model TruMac CN; LECO Corp., St. Joseph, MI) followed by conversion to a crude protein (CP) basis by applying a factor of 6.25. Normally, the summative calculation of energy density requires inputs of CP bound within NDF and acid detergent fiber (ADF) residues following nonsequential extraction in neutral and acid detergent (NDICP and ADICP, respectively). Due to the large number of triticale forages to be evaluated, NDICP and ADICP were estimated in triplicate from composites containing equal contributions from all seven cultivars within each harvest date. Concentrations of CP within these NDF and ADF residues were determined with the identical rapid-combustion procedure as described earlier. The preliminary NDF extraction procedure for NDICP did not include sodium sulfite, but did include heat-stable α-amylase. Sodium sulfite cleaves disulfide bonds and dissolves cross-linked proteins, thereby reducing recovery of CP from NDF residues (Van Soest et al., 1991). Concentrations of ADICP were determined following an ADF extraction procedure that did not include a preliminary digestion in neutral detergent (Van Soest et al., 1991). The summative approach (Weiss et al., 1992; NASEM, 2001) was used to calculate energy density (NE_L), where the truly digestible fiber component of total digestible nutrients (TDN) was calculated using the ADL option, and subsequently converted to NE_L by: NE_L (Mcal/kg) = 0.0245 × TDN (%) − 0.12.

Prior to conducting in vitro fermentations, forages were analyzed for asNDFom with sodium sulfite and heat-stable α-amylase included in the analysis, using the batch procedures outlined for an ANKOM200 Fiber Analyzer (ANKOM Technology Corp.; Macedon, NY); these procedures for asNDFom were followed by correcting fibrous residues for residual ash after combusting in a muffle furnace at 500 °C for 6 h. Additionally, ADF and ADL were determined sequentially, again using procedures outlined by ANKOM Technology Corp., but the initial extraction in neutral detergent was omitted.

In vitro incubation procedures

In vitro disappearance kinetics of DM and asNDFom were based on incubations of 6, 12, 24, 36, 48, 72, 144, and 240 h in buffered ruminal fluid using the ANKOM Daisy^II Incubator System (ANKOM Technology Corp.). Prior to incubation, empty F57 filter bags were prerinsed in acetone, and then air-dried prior to loading with 0.25-g subsamples of each forage. After loading, bags were heat sealed, and incubated for the designated time intervals as recommended by the manufacturer. A 0-h incubation time also was included in the kinetic analysis of all forages; disappearance of DM or asNDFom this time point was obtained in an identical manner to other incubation times, but was terminated after the standard 1.0-h equilibration in buffer, but before ruminal inoculum was added to the incubation jars. Incubations were conducted within 6 incubation runs, each evaluating 49 forages (7 cultivars × 7 harvest dates) in singlet at 9 incubation times for a total of 441 samples per run. Ideally, this number of samples requires 18 incubation jars, which exceeds the cumulative capacity of the 4 available incubator boxes (4 jars per box) located in the laboratory. To address this issue, 16 incubation jars containing samples incubated for 6, 12, 24, 36, 48, 72, 144, or 240 h were loaded simultaneously; 2 incubation jars were required to accommodate the 49 forages incubated for each time interval. The 0-h incubation associated with each run did not require ruminal inoculum, and was conducted after jars assigned to early time points had been removed from the incubator boxes. A subsequent, identical incubation run was used to provide a duplicate assessment of disappearance of DM or asNDFom for each forage at each time point, and these were averaged before conducting any statistical analyses. The aforementioned procedures were required to evaluate forages harvested from 1 of the 3 field blocks. As such, incubation runs 1 and 2 were assigned to field block 1, runs 3 and 4 to field block 2, and runs 5 and 6 to field block 3; as a result, a total of 2,646 combinations of field block, incubation run, forage, and incubation time were evaluated. Extra precautions were used to reduce error and limit run-to-run variability. These included: 1) the temperature (39 °C) of each incubator box was standardized against an independent auxiliary thermometer; 2) duplicate incubations of each forage were always conducted in different incubator boxes; 3) incubation times were assigned to whole jars, thus there was no comingling of withdraw times within specific incubation jars; 4) head space in incubator jars was purged with CO₂ after initial insertion of buffers and samples for prewarming, and again after adding the inoculum; and 5) each incubation jar contained both a blank bag plus a bag containing a standard forage.

Ruminal fluid used in the inoculum was obtained from two lactating Holstein cows offered a blended total mixed ration (48.7% DM) comprised of 18.7% (primarily alfalfa) haylage, 32.1% corn silage, 17.1% dry corn, and the balance (24.4%) containing protein supplements, vitamins, minerals, and whey. Ruminal fluid was obtained with a hand-operated siphon pump through a ruminal cannula, and discharged directly into pre-warmed (39 °C) thermos bottles. These containers were immediately placed in insulated chest-type coolers and transported to the laboratory (~15 min). 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 previously to 39 °C. Incubations were terminated by rinsing filter bags containing undigested residues seven times in cold tap water as recommended by the manufacturer of the incubation system. Rinsed bags were then dried to constant weight under forced air at 55 °C before weighing to determine residual DM. Rinsed bags containing residues from standard forages were then dried again overnight under forced air at 105 °C and reweighed; the mean difference between residual weights at the 55 and 105 °C drying temperatures was used as a correction factor to convert all triticale residues to an absolute DM basis (105 °C). Fiber bags containing DM residues from experimental triticale forages were then subjected to a terminal digestion in neutral detergent per procedures for determining asNDFom that were described earlier.

Disappearance kinetics

The in vitro disappearance kinetics of asNDFom and DM were fitted to a modification of the first-order model described by Mertens and Loften (1980), where asNDFom or DM remaining is equal to B × e^{−Kd × (T − L)} + U. In this expression, fraction B is defined as the percentage of asNDFom or DM that disappears from fiber bags at a measurable rate, Kd is the fractional disappearance rate, T is the incubation time, L is the discrete lag time, and U is the percentage of asNDFom or DM that is indigestible. For this analysis, fraction U was fitted by the decay model, and was not equated directly to the actual percentage of asNDFom or DM remaining after a 240-h incubation. A potential extent of asNDFom disappearance also was calculated as 100% − U. For both asNDFom and DM disappearance, some substrate is not accounted for by this decay model (fractions B + U < 100%). While this is expected for DM as soluble components are quickly lost through the filter bags upon insertion into buffers, it also has been noted previously (Hoffman et al., 1993) that some structural plant fiber can be unaccounted for by the decay model (B + U < 100%), particularly when the test forages are immature cool-season grasses. These percentages of fiber are generally small, and usually assumed to be digestible; however, for purposes of this analysis they also have been calculated and defined as fraction A. All disappearance kinetics were calculated by PROC NLIN of SAS (2016), and disappearance at commonly used endpoints (30 or 48 h) were calculated from fitted model parameters as: 100% − [B × e^{−Kd × (T − L)} + U].

Statistics

Descriptive statistics for agronomic characteristics (growth stage, canopy height, estimated DM yield, and percentages of leaf, stem, and grain head) were calculated for the entire study (N = 147) and by harvest date (N = 21), as were concentrations of meaningful nutritive value indices (CP, whole-plant ash, asNDFom, ADF, ADL, TDN, and NE_L) using PROC MEANS of SAS 9.4 (SAS Institute Inc., Cary, NC). Disappearance kinetic parameters (A, B, Kd, lag time, U, and potential extent) as well as the calculated disappearance at typical single-endpoint incubation times (30 or 48 h) were analyzed as a randomized complete block design with cultivars designated as whole-plots within three complete replications (blocks); harvest dates were evaluated further as subplots. Field block was considered a random effect. For disappearance kinetics of asNDFom, parameter estimates generally did not exhibit an interaction of treatment main effects; therefore, only main-effect means are reported and discussed. Orthogonal contrasts were used to evaluate cultivar groupings, and these included: 1) POS vs. TYPCUL; 2) POS vs. NEG; 3) TYPCUL vs. NEG; and 4) linear, quadratic, cubic, and quartic effects of harvest date. Concentrations of asNDFom, as well as kinetic parameters of DM disappearance generally exhibited significant interactions of main effects, and means were evaluated by linear, quadratic, cubic, and quartic effects of harvest date within individual cultivars. Because the triticale cultivars evaluated in this study exhibited varying maturation rates, an additional effort was made to evaluate them at common growth stages most suitable for mechanical harvest (mid-boot, mid-heading, full flower, and soft-dough). To accomplish this, kinetic parameters of both DM and asNDFom were regressed on linear growth stage using quartic, cubic, quadratic, and linear models (PROC REG of SAS, 2016). The best model was selected based on a significant (P < 0.05) F-test for the overall model, as well as significant (P < 0.05) tests for each polynomial coefficient. The intercept also was included in the selected model, irrespective of whether it differed (P < 0.05) from zero. Throughout the study, significance was declared at the P < 0.05 level of confidence, with occasion trends (P < 0.10) discussed when they were especially relevant to the interpretation of results.

Results and Discussion

Agronomic characteristics

For triticale forages, recent extension recommendations have suggested a boot-stage harvest for lactating dairy cows (Kilcer et al., 2010), but harvests of triticale at the late milk or early dough stages of growth have also been suggested in other communications (McDonald et al., 1991; McCartney and Vaage, 1994; Kennelly and Weinberg, 2003). Descriptive statistics for selected triticale cultivars harvested during 2020 (Table 1) indicated a high degree of variability for GRST across March 17, April 1, April 14, and April 28 harvest dates, when the associated SD for these dates ranged from 7.2 to 12.1 compared with February 28 and May harvest dates (≤1.7). This variability was particularly great for March 17 and April 1, when linear GRST assessments ranged from 32 to 59 and 39 to 69, respectively. Thus, during a 2-wk time interval, cultivars ranged from stem elongation (pre-booting) to late-anthesis, indicating that very close attention is required by forage producers wishing to target a specific GRST for harvest, and that maturation rates through the most desirable harvest stages are inconsistent across cultivars. Considerable variability across cultivars also was observed throughout the trial for leaf, stem, and grain-head percentages, as well as estimated DM yield. For example, Swift 77 and 770-001 exhibited measurable head or inflorescence weights as early as March 17, while Merlin Max did not until April 28. Furthermore, there were wide variations in the percentages of DM partitioned within the grain head on the final three harvest dates; ranges observed on April 28, May 12, and May 26 were 18.2% to 58.2%, 35.0% to 66.9%, and 34.0% to 70.2%, respectively. Generally, these proportions of DM partitioned into the grain head exceeded those observed the previous year (Coblentz et al. XXXX), when the maximum observed for any individual cultivar at near-full maturity (GRST = 96) was 61.7%. In addition, a 70% yield drag has been associated with a boot-stage compared with a soft-dough-stage harvest in Wisconsin (Coblentz et al., 2018a), and this appears to be consistent with DM yields observed in Arizona in 2020. Yields of DM observed on March 17 (mean GRST = 44; 5,682 kg/ha) were only 25.8% of those observed on May 12 (mean GRST = 87; 22,034 kg/ha). Therefore, management decisions need to closely consider forage inventory needs against the specific nutritional requirements of various livestock classes to establish proper harvest timing.

Nutritive value

Descriptive statistics for concentrations of CP, whole-plant ash, asNDFom, ADF, and ADL, as well as energy density (TDN or NE_L) are summarized for triticale forages grown in 2020 across all harvest dates, and within individual harvest dates (Table 2). As noted in the companion project conducted the previous year (Coblentz et al., XXXX), a distinguishing characteristic of these triticale forages grown in the southwest USA was the relatively high concentrations of whole-plant ash. Ash declined steadily from a maximum of 16.6 ± 0.96% on February 26, but was not <10% of plant DM until May 12 harvest date (8.8 ± 1.36%). These concentrations exceed substantially those reported for headed triticale silage (9.7%; NASEM, 2001), 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%); furthermore, whole-plant ash has no energy value and acts to directly discount calculated estimates of energy density in forages (NASEM, 2001).

Table 2.

Whole-plant ash, CP, ADL, and energy density (TDN, NE_L) for seven selected triticale cultivars grown during 2020 in Maricopa, AZ

Harvest date	Days from February 1	Statistic	CP	Ash	asNDFom1	ADF	ADL	Ratio2	TDN % DM	NEL Mcal/kg
			% DM	% DM
All dates (N = 147)	–	Mean	13.0	12.4	50.4	34.5	2.85	5.67	63.6	1.44
		SD	6.59	3.24	9.46	7.54	0.984	1.721	5.24	0.129
		Minimum	5.9	6.3	29.3	18.6	0.65	1.57	54.0	1.20
		Maximum	34.8	18.8	67.1	48.0	5.15	8.63	76.8	1.76
February 26 (N = 21)	26	Mean	27.5	16.6	41.3	25.7	1.24	2.99	67.0	1.52
		SD	3.26	0.96	1.61	2.57	0.313	0.715	1.38	0.035
		Minimum	23.7	15.1	35.8	19.5	0.65	1.57	64.6	1.46
		Maximum	34.8	18.8	43.1	28.6	1.90	4.58	70.1	1.60
March 17 (N = 21)	46	Mean	15.8	15.8	53.6	36.6	2.21	4.12	60.6	1.36
		SD	1.79	1.07	2.56	1.94	0.353	0.579	0.98	0.024
		Minimum	13.6	14.2	47.6	31.7	1.32	2.36	58.1	1.30
		Maximum	20.6	17.2	57.5	39.3	2.86	5.02	62.2	1.40
April 1 (N = 21)	61	Mean	11.5	14.0	59.4	41.5	2.82	4.74	59.2	1.33
		SD	1.20	1.28	1.97	1.82	0.487	0.711	1.48	0.036
		Minimum	10.0	11.8	56.1	38.3	2.05	3.55	56.4	1.26
		Maximum	15.5	16.6	62.3	44.4	3.61	5.82	61.3	1.38
April 14 (N = 21)	74	Mean	9.5	12.0	59.2	42.1	3.62	6.14	59.5	1.34
		SD	0.74	1.50	4.28	2.90	0.294	0.686	2.53	0.062
		Minimum	8.5	9.8	52.4	36.2	3.14	4.85	55.0	1.23
		Maximum	11.3	14.7	65.4	46.6	4.15	7.21	64.2	1.45
April 28 (N = 21)	88	Mean	9.2	10.6	50.2	35.5	3.68	7.34	63.2	1.43
		SD	0.75	1.54	9.37	6.69	0.654	0.301	5.03	0.123
		Minimum	7.6	8.1	34.6	25.8	2.68	6.82	54.0	1.20
		Maximum	10.7	13.6	67.1	48.0	5.15	7.87	70.7	1.61
May 12 (N = 21)	102	Mean	8.7	8.8	44.7	30.4	3.24	7.24	68.0	1.55
		SD	0.84	1.36	9.26	6.75	0.749	0.717	5.08	0.124
		Minimum	6.6	6.3	31.9	21.0	1.94	5.53	57.8	1.30
		Maximum	9.9	11.2	64.6	45.4	4.86	8.63	75.4	1.73
May 26 (N = 21)	116	Mean	8.6	8.9	44.2	29.8	3.17	7.13	68.1	1.55
		SD	1.35	1.65	10.49	7.99	0.879	0.509	5.98	0.146
		Minimum	5.9	6.6	29.3	18.6	1.88	6.37	55.8	1.25
		Maximum	11.6	12.3	65.7	45.8	5.03	8.16	76.8	1.76

asNDFom, NDF analysis using heat-stable, α-amylase and sodium sulfite during the extraction in neutral detergent, and correction of the fibrous residue for residual ash.

Quotient of ADL/asNDFom × 100%.

Mean concentrations of asNDFom increased from 41.3 ± 1.61% on February 26 to 59.2 ± 4.28% on April 14 before declining thereafter with the onset of grain fill. Generally, variation among cultivars in this respect was limited through April 1 (SD ≤ 2.56%), but then increased greatly across subsequent harvest dates with strong varietal differences in maturation rate and percentages of plant weight partitioned into the filling grain head (SD = 4.28% to 10.49%). This was particularly evident on the April 28 harvest date (mean GRST = 79), when concentrations of asNDFom ranged from 34.6% to 67.1%, and these varietal differences persisted through the remainder of the trial, with concentrations ranging from 29.3% to 65.7% on the final (May 26) harvest date. Cherney and Marten (1982b), demonstrated that concentrations of cell wall in leaf-blade, leaf-sheath, and stem tissues all continued to increase throughout maturation, but concentrations of cell wall within the inflorescence declined sharply with grain fill, thereby illustrating the competing factors affecting asNDFom balance determined on a whole-plant basis. An in-depth ANOVA for asNDFom indicated strong significance (P < 0.001) for main effects of cultivar and harvest date, as well as their associated interaction (Table 3). Orthogonal trend analysis for each cultivar across harvest dates indicated that all cultivars except Merlin Max exhibited linear, quadratic, and cubic effects (P ≤ 0.042) of harvest date on concentrations of asNDFom. Merlin Max exhibited only a tendency for a cubic response (P = 0.063). Quartic effects also were observed for Legend (P = 0.001), SY115T (P < 0.001), SY158T (P = 0.015), and 770-001 (P = 0.003), but not for other cultivars (P ≥ 0.518). Of these responses, concentrations of asNDFom were most unique for Merlin Max, which reached 63.1% on April 14, but changed only minimally thereafter. As a result, the highest ordered effect for this cultivar was quadratic (P < 0.001), while all other cultivars exhibited more complex responses in response to maturation rate and grain fill.

Table 3.

Concentrations of asNDFom for seven selected triticale cultivars harvested on seven dates during 2020 at Maricopa, AZ

Harvest date	Goldrush 91	Legend	SY115T	SY158T	770-001	Merlin Max	Swift 77
	% of DM
February 26	41.9	42.0	39.9	42.2	41.3	39.9	42.2
March 17	54.9	53.7	49.1	53.8	54.7	52.3	56.9
April 1	61.4	57.5	56.5	60.3	61.3	58.4	60.6
April 14	55.9	64.9	61.7	57.7	58.7	63.1	52.4
April 28	43.9	58.6	50.5	44.8	49.3	66.4	38.1
May 12	42.9	50.1	41.8	39.1	42.3	62.6	34.1
May 26	36.7	50.6	42.2	37.0	46.7	64.0	32.1
SEM	1.28
Contrasts1	P > F
Harvest date: linear	<0.001	0.004	0.042	<0.001	0.003	<0.001	<0.001
Harvest date: quadratic	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
Harvest date: cubic	<0.001	0.001	<0.001	<0.001	<0.001	0.063	<0.001
Harvest date: quartic	0.621	0.001	< 0.001	0.015	0.003	0.518	0.982

Linear, quadratic, cubic, and quartic effect of harvest date.

Grain fill similarly affected calculated energy estimates for these forages (Table 2). For February 26 harvest date, the mean NE_L (1.52 ± 0.035 Mcal/kg) was consistent with other calculated estimates for winter cereals prior to jointing. Coblentz et al. (2020) reported 3-yr means for vegetative rye, triticale, and wheat harvested before winter in Wisconsin of 1.52, 1.55, and 1.56 Mcal/kg, respectively. In the present study, mean energy estimates for triticale cultivars declined with reproductive development, reaching minima of 1.33 ± 0.036 and 1.34 ± 0.062 Mcal/kg on April 1 and 14, respectively, but then increased thereafter in response to grain fill. However, these responses also included a high degree of variability among cultivars. For example, the mean estimate on May 12 (1.55 ± 0.124 Mcal/kg) was similar to that observed pre-jointing (February 26), but the SD was 3.5 times greater. The variability among cultivars on this late harvest date can be associated with varying percentages of leaf, stem, and grain head (Table 1), but only minimally with maturation rate, as cultivars ranged narrowly from GRST 85–89 on May 12.

In vitro disappearance kinetics of asNDFom

Generally, in vitro parameters associated with disappearance of asNDFom (Table 4) were not affected by the interaction of main effects (P ≥ 0.082); therefore, only main-effect means are presented and discussed. Exceptions to this generalization were observed for the indigestible fraction (U) and the potential extent of disappearance (P < 0.001); however, these interaction effects were dwarfed by main-effect responses, represented only minor variations to overall trends, and were ignored to simplify the presentation of results.

Table 4.

Main effect means for in vitro disappearance kinetics of asNDFom for seven selected triticale cultivars assigned as positive-, typical-, or negative-deviant types and fitted to a two-pool disappearance model¹

Treatment	Type	Fraction				Kd6 /h	Lag time h	NDFD307	NDFD488
		A2	B3	U4	Extent5
		% of asNDFom						% of asNDFom
Cultivar
Goldrush 91	Positive	2.6	75.8	21.6	78.4	0.042	2.71	50.9	62.7
Legend	Positive	2.5	76.4	21.1	78.9	0.045	3.05	53.1	65.1
SY115T	Typical	2.3	74.9	22.8	77.2	0.048	2.72	53.4	64.3
SY158T	Typical	2.9	72.5	24.5	75.5	0.039	2.63	48.4	60.0
770-001	Typical	1.7	73.3	25.0	75.0	0.042	2.51	49.0	60.4
Merlin Max	Negative	1.3	75.1	23.6	76.4	0.046	2.82	52.5	63.9
Swift 77	Negative	3.5	73.4	23.1	76.9	0.034	1.37	45.5	56.5
SEM		0.64	0.99	0.53	0.53	0.0029	0.654	2.24	1.76
Harvest date
February 26	–	2.1	88.4	9.5	90.5	0.088	2.34	82.0	88.5
March 17	–	2.1	82.1	15.8	84.2	0.058	3.45	65.5	77.0
April 1	–	1.6	77.3	21.1	78.9	0.042	4.55	52.0	66.0
April 14	–	0.6	71.9	27.5	72.5	0.035	3.13	43.7	56.7
April 28	–	2.2	68.8	29.0	71.0	0.027	1.43	38.8	50.8
May 12	–	4.2	66.2	29.6	70.4	0.024	1.50	35.8	47.1
May 26	–	3.9	66.7	29.3	70.7	0.023	1.40	35.4	46.9
SEM	–	0.53	0.76	0.47	0.47	0.0022	0.439	1.08	0.90
Contrasts		P > F
Type: positive vs. typical		0.741	0.014	<0.001	<0.001	0.750	0.672	0.412	0.172
Type: positive vs. negative		0.864	0.080	0.002	0.002	0.228	0.251	0.198	0.054
Type: negative vs. typical		0.884	0.464	0.132	0.132	0.314	0.400	0.537	0.405
Harvest date: linear		0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
Harvest date: quadratic		0.001	<0.001	<0.001	<0.001	<0.001	0.002	<0.001	<0.001
Harvest date: cubic		0.496	0.130	0.103	0.103	0.002	<0.001	0.046	0.034
Harvest date: quartic		0.005	0.569	0.020	0.020	0.153	0.788	0.699	0.499

Cultivars were harvested during 2020 in Maricopa, AZ.

A, percentage of asNDFom not recovered within B and U pools.

B, percentage of asNDFom disappearing at a measurable rate.

U, percentage of asNDFom considered indigestible, based on the nonlinear regression model.

Extent = 100% − U.

Kd, fractional disappearance rate.

NDFD30, percentage of asNDFom disappearing at 30 h of in vitro incubation in buffered rumen fluid as determined by the nonlinear decay model.

NDFD48, percentage of asNDFom disappearing at 48 h of in vitro incubation in buffered rumen fluid as determined by the nonlinear decay model.

Cultivar effects

Fraction A (P ≥ 0.741), Kd (P ≥ 0.228), lag time (P ≥ 0.251), and in vitro disappearance of asNDFom after 30 h of incubation (NDFD30; P ≥ 0.198) did not exhibit differences between cultivar groupings. Fraction B was greater for POS compared with TYPCUL (76.1% vs. 73.6% of asNDFom; P = 0.014), but only tended to differ from NEG (76.1% vs. 74.3% of asNDFom; P = 0.080), largely due to the inconsistent responses observed between NEG cultivars. Conversely, the indigestible percentage of asNDFom (U) was greater for TYPUL compared with POS (24.1% vs. 21.4% of asNDFom; P < 0.001), and greater for NEG compared with POS (23.4% vs. 21.4% of asNDFom; P = 0.002), but TYPCUL and NEG did not differ (P = 0.132). As a result, the potential extent of asNDFom disappearance was greater (P ≤ 0.002) by 2.8 and 2.0 percentage units for POS compared with TYPCUL or NEG, respectively. Although there were no differences among cultivar groupings for NDFD30 (P ≥ 0.198), a tendency for difference between POS and NEG cultivars was observed for asNDFom disappearance after a 48-h incubation (NDFD48; 63.9% vs. 60.2% of asNDFom; P = 0.054). The noted tendency was affected by the diverse responses within the NEG cultivar grouping, where Merlin Max exhibited numerically greater disappearance across harvest dates than Swift 77 (63.9% vs. 56.5% of asNDFom).

Harvest date effects

All parameters fitted by the two-pool disappearance model exhibited linear (P < 0.001) and quadratic (P ≤ 0.002) effects of harvest date. Significant higher ordered effects (P ≤ 0.046) also were observed for all model parameters except for fraction B (P ≥ 0.130). Quartic (P ≤ 0.020) effects were detected for fractions A, U, and the potential extent of disappearance, while cubic effects (P ≤ 0.046) were detected for Kd, lag time, NDFD30, and NDFD48. Generally, the effects of harvest date were consistent with normal expectations for maturing forage plants; fraction B and the potential extent of disappearance declined with time, as did Kd, NDFD30, and NDFD48. The indigestible fraction (U) increased from 9.5% of asNDFom on February 26 to 29.6% of asNDFom on May 12, but from a practical standpoint, changed only minimally after April 14. For Kd, the disappearance rate was approximately four times faster for pre-jointed plants on February 26 (0.088/h) compared with May 26 (0.023/h). Similarly, NDFD30 declined sharply across harvest dates from 82.0% to 35.4% of asNDFom. Although disappearance increased by a mean of 11.4 percentage units with a longer 48-h incubation interval, NDFD48 also exhibited a sharp decline across harvest dates (88.5% to 46.9% of asNDFom). It is noteworthy that the rate of change for most parameter estimates slowed or stabilized over the final three harvest dates, which likely contributed to the detection of higher ordered, nonlinear contrasts.

Regressions on growth stage

While the ANOVA described previously is useful in describing cultivar differences and responses to harvest date, it remains somewhat confounded by differences in maturation rate that are not considered within the ANOVA. To address this issue, and to better evaluate cultivars at a common maturity, concentrations of asNDFom, parameters obtained from the two-pool disappearance model, as well as NDFD30 and NDFD48 calculated for each cultivar from the fitted decay model parameters were regressed on GRST (Table 5). Concentrations of asNDFom were best fitted to quadratic models (P ≤ 0.026) for all cultivars except Merlin Max, which was cubic in nature (Y = 0.000248 x³ − 0.0570 x² + 4.19 x − 33.5; R² = 0.996). Generally, there were few significant quartic, cubic, quadratic, or linear regressions detected for fraction A or lag time; the only exceptions were a quadratic response by Swift 77 for fraction A (Y = 0.00336 x² − 0.320 x + 8.0; R² = 0.975), and a quartic response by SY158T for lag time (Y = 0.0000129 x⁴ − 0.00294 x³ + 0.235 x² − 7.69 x + 88.61; R² = 0.996). For other disappearance model parameters, the most appropriate regression model was most often a linear relationship with GRST, with a minority proportion of quadratic responses. For fractions B and U, as well as Kd, a linear model met the conditions of selection in 16 of 21 cases (76.2%), while a quadratic model was selected in 4 of 21 cases (19.0%). A lone cubic response was observed between Kd and GRST for Goldrush 91. Relationships between NDFD30 and NDFD48 and GRST also were explained by linear (64.3%) or quadratic models (35.7%). Regardless of the dependent variable, GRST was a very effective predictor; coefficients of determination (R²) were > 0.9 for 34 of 42 individual regressions reported in Table 5, and for all regressions of NDFD30 or NDFD48 on GRST (R² ≥ 0.927). A unique observation throughout this analysis was that Merlin Max exhibited a cubic response for asNDFom, and quadratic responses for all regressions, and was the only cultivar to do so.

Table 5.

Concentrations of asNDFom and kinetic parameters for in vitro disappearance kinetics of asNDFom for seven selected triticale cultivars assigned as positive-, typical-, or negative-deviant types and fitted to a two-pool disappearance model regressed on growth stage^1,2,3

Cultivar	Type	Mean4	Model5	P > F6	RMSE7	R2	Predicted value at key growth stages
							Mid-boot8	Mid-heading9	Full flower10	Soft dough11
asNDFom, % of DM
Goldrush 91	Positive	48.2	Y = −0.0204 x2 + 2.33 x − 6.0	0.008	3.34	0.910	56.5	60.5	59.4	44.9
Legend	Positive	53.9	Y = −0.0165 x2 + 1.96 x − 4.8	0.015	3.13	0.879	58.7	62.8	62.7	52.5
SY115T	Typical	48.8	Y = −0.0183 x2 + 2.09 x − 1.0	0.023	3.90	0.847	55.1	58.7	57.6	44.5
SY158T	Typical	47.8	Y = −0.0193 x2 + 2.15 x − 0.7	0.002	2.42	0.955	56.2	59.4	57.7	42.9
770-001	Typical	50.6	Y = −0.0165 x2 + 1.97 x + 0.6	0.026	3.86	0.838	54.7	58.9	58.8	48.6
Merlin Max	Negative	58.1	Y = 0.000248 x3 − 0.0570 x2 + 4.19 x − 33.5	0.001	0.88	0.996	61.2	66.1	66.5	63.7
Swift 77	Negative	45.2	Y = −0.0230 x2 + 2.51 x − 7.8	0.002	3.10	0.951	57.5	60.6	58.0	39.1
Fraction B, % of asNDFom12
Goldrush 91	Positive	75.8	Y = −0.307 x + 96.7	<0.001	1.97	0.943	83.5	79.8	76.8	70.6
Legend	Positive	76.5	Y = −0.322 x + 95.6	<0.001	2.08	0.950	81.8	77.9	74.7	68.3
SY115T	Typical	74.9	Y = −0.400 x + 99.1	0.001	3.49	0.907	81.9	77.1	73.1	65.1
SY158T	Typical	72.5	Y = −0.405 x + 99.1	<0.001	2.37	0.953	81.7	76.8	72.8	64.7
770-001	Typical	73.3	Y = −0.353 x + 97.7	0.001	2.86	0.902	82.5	78.3	74.7	67.7
Merlin Max	Negative	75.1	Y = 0.00640 x2 − 1.06 x + 110.8	<0.001	1.38	0.983	76.9	71.6	68.7	66.6
Swift 77	Negative	73.4	Y = −0.304 x + 94.7	0.002	2.72	0.883	81.6	77.9	74.9	68.8
Kd, /h13
Goldrush 91	Positive	0.042	Y = −0.000000584 x3 + 0.000113 x2 −0.00773 x + 0.227	0.001	0.0027	0.994	0.057	0.046	0.042	0.028
Legend	Positive	0.045	Y = −0.000887 x + 0.098	0.010	0.0139	0.761	0.060	0.049	0.041	0.023
SY115T	Typical	0.048	Y = −0.00106 x + 0.112	0.001	0.0089	0.914	0.066	0.053	0.043	0.021
SY158T	Typical	0.039	Y = 0.0000104 x2 − 0.00210 x + 0.127	<0.001	0.0018	0.996	0.056	0.043	0.035	0.024
770-001	Typical	0.042	Y = −0.000991 x + 0.110	<0.001	0.0022	0.992	0.067	0.056	0.046	0.026
Merlin Max	Negative	0.046	Y = 0.0000264 x2 − 0.00383 x + 0.162	0.008	0.0082	0.912	0.046	0.032	0.025	0.028
Swift 77	Negative	0.034	Y = −0.00102 x + 0.105	<0.001	0.0035	0.981	0.062	0.049	0.039	0.019
U, % of asNDFom14
Goldrush 91	Positive	21.6	Y = 0.295 x + 1.5	0.001	2.60	0.899	14.2	17.8	20.7	26.6
Legend	Positive	21.1	Y = 0.300 x + 3.2	<0.001	1.54	0.968	16.1	19.7	22.7	28.7
SY115T	Typical	22.9	Y = 0.392 x − 0.9	0.001	3.27	0.914	16.0	20.7	24.6	32.5
SY158T	Typical	24.5	Y = 0.345 x + 1.9	<0.001	1.77	0.964	16.0	20.2	23.6	30.5
770-001	Typical	25.0	Y = 0.349 x + 0.8	0.001	2.97	0.893	15.9	20.0	23.5	30.5
Merlin Max	Negative	23.6	Y = −0.0080 x2 + 1.25 x − 16.4	<0.001	1.42	0.982	22.4	28.0	30.8	31.7
Swift 77	Negative	23.1	Y = 0.231 x + 7.0	0.014	3.44	0.731	16.9	19.7	22.0.	26.6
NDFD30, % of asNDFom15
Goldrush 91	Positive	50.9	Y = −0.720 x + 99.9	<0.001	2.71	0.980	68.9	60.3	53.1	38.7
Legend	Positive	53.1	Y = −0.634 x + 90.9	0.001	4.99	0.927	63.7	56.0	49.7	37.0
SY115T	Typical	53.4	Y = 0.0094 x2 − 1.85 x + 125.8	<0.001	1.93	0.994	63.7	52.6	45.4	36.7
SY158T	Typical	48.4	Y = 0.00681 x2 − 1.48 x + 113.0	<0.001	1.84	0.992	61.8	52.0	45.4	36.1
770-001	Typical	49.0	Y = −0.777 x + 102.8	< 0.001	3.15	0.974	69.4	60.0	52.3	36.7
Merlin Max	Negative	52.5	Y = 0.0173 x2 − 2.63 x + 135.3	0.002	4.30	0.959	54.3	43.1	37.5	36.8
Swift 77	Negative	45.5	Y = −0.775 x + 99.6	< 0.001	3.56	0.966	66.3	57.0	49.3	33.8
NDFD48, % of asNDFom16
Goldrush 91	Positive	62.7	Y = −0.656 x + 107.3	<0.001	2.69	0.976	79.1	71.2	64.7	51.6
Legend	Positive	65.1	Y = −0.575 x + 99.3	< 0.001	2.82	0.970	74.6	67.7	62.0	50.5
SY115T	Typical	64.3	Y = −0.00566 x2 − 1.35 x + 122.5	<0.001	1.92	0.992	74.7	65.1	58.4	48.3
SY158T	Typical	60.1	Y = −0.658 x + 103.2	<0.001	2.07	0.986	74.9	67.0	60.4	47.3
770-001	Typical	60.4	Y = −0.697 x + 108.5	<0.001	3.77	0.954	78.6	70.2	63.2	49.3
Merlin Max	Negative	63.9	Y = 0.0126 x2 − 2.03 x + 130.5	0.001	3.10	0.973	66.5	57.0	51.8	49.0
Swift 77	Negative	56.5	Y = −0.729 x + 107.5	<0.001	4.16	0.949	76.1	67.3	60.0	45.5

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

A small portion of asNDFom was not recovered for some forages (B + U_model pools < 100%), and was designated as Fraction A. This fraction was related to growth stage only for the Swift 77 cultivar (Y = 0.00336 x² − 0.320 x + 8.0; P = 0.001; R² = 0.975), yielding predicted values of 0.5%, 0.6%, 1.4%, and 5.1% of asNDFom at the mid-boot, mid-heading, full-flower, and soft-dough stages of growth, respectively. The overall dependent mean for all other cultivars was ≤2.9% of asNDFom.

Lag time was related to growth stage only for the SY158T cultivar (Y = 0.0000129 x⁴ − 0.00294 x³ + 0.235 x² − 7.69 x + 88.61; P = 0.008; R² = 0.996), yielding predicted values of 3.53, 6.58, 6.04, and 2.93 h at the mid-boot, mid-heading, full-flower, and soft-dough stages of growth, respectively. The overall dependent mean for all other cultivars was ≤3.05 h.

Mean of dependent regression variable.

Most appropriate of quartic, cubic, quadratic, or linear regression models.

P > F for the selected regression model.

RMSE, root mean square error.

Mid-boot stage = 43.

Mid-heading stage = 55.

Full-flower stage = 65.

Soft-dough stage = 85.

Fraction B, percentage of asNDFom disappearing at a measurable rate.

Kd, fractional disappearance rate of Fraction B.

U, percentage of asNDFom considered indigestible, based on the nonlinear regression model.

NDFD30, in vitro disappearance of asNDFom after a 30-h incubation in buffered rumen fluid.

NDFD48, in vitro disappearance of asNDFom after a 48-h incubation in buffered rumen fluid.

In vitro disappearance kinetics of DM

Analysis of variance

For parameters of DM disappearance, an interaction of main effects was detected for all model parameters (P < 0.001) except lag time (P = 0.372); therefore, only interaction means (harvest date within cultivar) are presented (Table 6) and discussed. In vitro DM disappearance after 30- (IVDMD30) or 48-h (IVDMD48) incubations also exhibited interactions (P < 0.001) of main effects, and means are presented (Table 7) and discussed similarly. Parameters derived from the two-pool disappearance model generally exhibited one or more detectable higher ordered effects of harvest date within specific cultivars. In no case was a linear effect the only detectable contrast. However, no detectable polynomial effects of any order were observed for fraction B (P ≥ 0.158) for the Legend and Merlin Max cultivars, or for lag time (P ≥ 0.083) within the Legend cultivar. For IVDMD30 and IVDMD48 (Table 7), all cultivars exhibited a cubic or quartic response (P ≤ 0.046) across harvest dates, except for Merlin Max, for which the highest ordered contrast was quadratic (P < 0.001). For all cultivars except Merlin Max, IVDMD30 and IVDMD48 generally declined through mid-April before increasing thereafter, ostensibly due to the partitioning of nonstructural carbohydrates within developing grain heads. This oscillating pattern of response is at least partially responsible for the previously noted higher ordered orthogonal contrasts; uniquely, DM disappearance for Merlin Max declined steadily through late April before plateauing thereafter, thereby yielding only significant linear (P < 0.001) and quadratic (P < 0.001) effects. Similar response patterns were noted for Kd, where cultivars generally exhibited declining disappearance rates into April, but again increased thereafter. As observed for IVDMD30 and IVDMD48, Merlin Max exhibited no such increase in Kd during late April or May.

Table 6.

Analysis of variance for in vitro disappearance kinetics of DM for seven selected triticale cultivars harvested on seven dates during 2020 at Maricopa, AZ

Harvest date	Goldrush 91	Legend	SY115T	SY158T	770-001	Merlin Max	Swift 77
Fraction A, % of DM
February 26	37.9	37.4	46.0	35.5	37.4	37.8	36.1
March 17	28.4	27.8	35.0	28.2	27.6	31.9	25.6
April 1	24.9	26.8	27.7	22.5	23.6	26.4	23.9
April 14	28.3	20.3	22.4	26.1	23.0	23.5	28.5
April 28	22.3	21.4	25.4	23.9	22.5	21.6	23.3
May 12	19.2	18.8	20.6	22.2	20.5	18.0	23.2
May 26	20.0	19.4	20.7	22.4	20.6	18.0	24.5
SEM	1.32
Contrasts1	P > F
Harvest date: linear	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
Harvest date: quadratic	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
Harvest date: cubic	0.015	0.135	0.003	0.001	<0.001	0.640	0.001
Harvest date: quartic	<0.001	0.238	0.883	0.051	0.086	0.567	<0.001
Fraction B, % of DM
February 26	55.6	55.3	47.8	55.9	54.9	54.5	55.6
March 17	58.9	59.9	52.8	56.5	57.3	53.9	57.1
April 1	58.0	57.8	57.5	59.7	56.4	56.3	56.5
April 14	52.4	59.5	56.8	51.5	53.4	54.5	49.0
April 28	60.2	57.6	50.6	55.0	54.1	50.6	61.5
May 12	62.3	59.5	61.3	60.0	61.3	56.5	60.4
May 26	63.1	57.9	59.4	57.9	59.0	54.6	60.2
SEM	1.68
Contrasts1	P > F
Harvest date: linear	<0.001	0.354	<0.001	0.255	0.016	0.987	0.001
Harvest date: quadratic	0.022	0.158	0.353	0.132	0.057	0.601	0.025
Harvest date: cubic	0.609	0.347	0.013	0.345	0.491	0.345	0.271
Harvest date: quartic	0.001	0.168	0.130	0.004	0.001	0.501	<0.001
Fraction U, % of DM
February 26	6.5	7.3	6.3	8.6	7.7	7.7	8.4
March 17	12.7	12.4	12.2	15.2	15.2	14.2	17.3
April 1	17.2	15.3	14.9	17.8	20.0	17.3	19.6
April 14	19.4	20.2	20.8	22.4	23.7	22.0	22.5
April 28	17.5	21.0	24.0	21.1	23.3	27.8	15.2
May 12	18.5	21.7	18.2	17.8	18.2	25.5	16.4
May 26	16.9	22.7	19.8	19.6	20.5	27.4	15.3
SEM	1.13
Contrasts1
Harvest date: linear	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	0.017
Harvest date: quadratic	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
Harvest date: cubic	0.126	0.871	0.639	0.060	0.021	0.458	<0.001
Harvest date: quartic	0.840	0.512	0.048	0.063	0.011	0.750	0.728
Kd, /h
February 26	0.060	0.061	0.076	0.055	0.048	0.063	0.051
March 17	0.048	0.049	0.061	0.048	0.050	0.062	0.046
April 1	0.038	0.044	0.041	0.037	0.038	0.046	0.034
April 14	0.034	0.035	0.032	0.032	0.035	0.041	0.036
April 28	0.042	0.033	0.041	0.040	0.042	0.038	0.035
May 12	0.043	0.043	0.044	0.043	0.043	0.034	0.043
May 26	0.048	0.041	0.044	0.052	0.041	0.034	0.044
SEM	0.0037
Contrasts1
Harvest date: linear	0.015	<0.001	<0.001	0.266	0.064	<0.001	0.123
Harvest date: quadratic	<0.001	<0.001	<0.001	<0.001	0.028	0.020	<0.001
Harvest date: cubic	0.134	0.730	0.080	0.966	0.519	0.303	0.465
Harvest date: quartic	0.460	0.223	0.012	0.220	0.011	0.171	0.168
Lag time, h
February 26	4.06	3.68	9.01	3.49	3.12	3.11	3.86
March 17	4.89	4.19	5.77	5.65	5.48	5.80	5.04
April 1	6.53	6.06	4.62	6.40	5.67	6.26	6.00
April 14	3.67	4.64	4.65	5.28	3.52	4.91	2.91
April 28	2.81	3.35	3.05	3.38	4.02	6.78	2.02
May 12	2.95	3.87	3.57	3.45	3.11	4.28	2.92
May 26	3.03	4.25	3.98	3.95	3.20	4.27	3.16
SEM	0.987
Contrasts1
Harvest date: linear	0.018	0.714	0.001	0.177	0.170	0.828	0.022
Harvest date: quadratic	0.346	0.354	0.021	0.088	0.138	0.006	0.936
Harvest date: cubic	0.026	0.083	0.606	0.007	0.049	0.299	0.010
Harvest date: quartic	0.829	0.657	0.411	0.991	0.324	0.570	0.381

Linear, quadratic, cubic, and quartic effect of harvest date.

Table 7.

Analysis of variance for predicted in vitro disappearance of DM after 30- and 48-h incubations for seven selected triticale cultivars harvested on seven dates during 2020 at Maricopa, AZ

Harvest date	Goldrush 91	Legend	SY115T	SY158T	770-001	Merlin Max	Swift 77
30-h IVDMD, % of DM
February 26	81.2	81.5	83.7	78.4	77.2	82.2	77.0
March 17	69.4	70.3	75.5	66.9	68.1	73.7	63.8
April 1	58.6	64.4	64.5	57.1	57.3	63.6	55.3
April 14	59.0	55.3	53.9	54.2	54.7	57.4	58.2
April 28	63.3	54.6	59.3	59.7	58.1	51.1	61.4
May 12	61.8	58.7	62.7	62.8	61.9	50.4	64.0
May 26	65.3	56.8	61.0	65.3	59.3	49.8	66.1
SEM	2.11
Contrasts1	P > F
Harvest date: linear	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	0.001
Harvest date: quadratic	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
Harvest date: cubic	<0.001	0.350	0.255	0.001	<0.001	0.315	<0.001
Harvest date: quartic	0.847	0.033	<0.001	0.046	0.002	0.442	0.994
48-h IVDMD, % of DM
February 26	89.0	88.9	90.9	86.6	85.9	89.0	85.8
March 17	79.7	80.2	83.6	77.1	78.0	81.8	74.2
April 1	70.3	75.4	75.0	69.3	68.3	74.1	66.7
April 14	68.7	66.8	64.9	64.4	64.5	67.4	66.8
April 28	73.5	65.2	67.9	69.4	67.7	61.5	72.3
May 12	72.2	69.1	73.2	73.1	72.2	61.0	74.1
May 26	75.2	67.3	71.4	74.4	69.6	60.2	76.3
SEM	1.64
Contrasts1	P > F
Harvest date: linear	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	0.001
Harvest date: quadratic	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
Harvest date: cubic	0.003	0.940	0.604	0.010	0.002	0.138	<0.001
Harvest date: quartic	0.367	0.027	<0.001	0.007	<0.001	0.446	0.471

Linear, quadratic, cubic, and quartic effect of harvest date.

Regressions on growth stage

Consistent with regressions of fiber-related disappearance parameters on GRST, the use of GRST as the independent variable was an effective predictor of DM-related disappearance parameters (Table 8). For the 35 individual regressions evaluated, about two-thirds (62.9%) exhibited R² > 0.9. In only one case, the relationship between Kd and GRST for cultivar 770-001, was the selection criteria not met for linear, quadratic, cubic, or quartic regression models (P ≥ 0.063; overall mean = 0.042/h). For IVDMD30 and IVDMD48, most regression models (10 of 14) were quadratic (P ≤ 0.022) in nature, while (regardless of incubation time) those for cultivar SY158T were cubic (P ≤ 0.001), and those for cultivar 770-001 were linear (P ≤ 0.023). Generally, IVDMD30 and IVDMD48 responses to GRST were of higher order than those observed for disappearance of NDFD30 or NDFD48, which is partially explained by the effects of grain fill. Based on the work by Coblentz et al. (2018b), grain fill is not an important consideration affecting the regression relationship between in vitro asNDFom disappearance from triticale forages at specific time intervals (24, 30, or 48 h) and GRST using the large batch culture Daisy^II Incubator System. However, Mertens and Loften (1980) noted that addition of purified starch to individual incubation flasks decreased the potential extent of NDF disappearance from alfalfa, bermudagrass, tall fescue, and orchardgrass forages, but this effect appeared modest compared with other reports from in vivo assessments of NDF digestibility. In part, this discrepancy may be related to the relative stability of pH within buffered incubation flasks, which were maintained at pH = 6.8 compared with the possible changes that can occur for actual ruminal pH (Mertens and Loften, 1980). In a review, Hoover (1986) has summarized that a rapid fermentation causing pH to decrease to 6.0 or below is likely to negatively affect fiber digestion.

Table 8.

Kinetic parameters for in vitro disappearance kinetics of DM for seven selected triticale cultivars assigned as positive-, typical-, or negative-deviant types and fitted to a two-pool disappearance model regressed on growth stage^1,2,3

Cultivar	Type	Mean4	Model5	P > F6	RMSE7	Predicted value at key growth stages
						R2	Mid-boot8	Mid-heading9	Full flower10	Soft dough11
Fraction A, % of DM12
Goldrush 91	Positive	25.9	Y = −0.255 x + 43.2	0.001	2.29	0.895	32.2	29.2	26.6	21.5
Legend	Positive	24.6	Y = −0.240 x + 38.9	0.003	2.88	0.846	28.6	25.7	23.3	18.5
SY115T	Typical	28.3	Y = 0.00688  x2 − 1.13 x + 68.0	0.002	2.46	0.953	32.0	26.5	23.4	21.4
SY158T	Typical	25.8	Y = −0.175 x + 37.3	0.009	2.51	0.773	29.8	27.7	25.9	22.4
770-001	Typical	25.0	Y = 0.00243 x2 − 0.54 x + 50.0	<0.001	0.77	0.989	31.1	27.5	24.9	21.4
Merlin Max	Negative	25.3	Y = 0.00639 x2 − 1.01 x + 58.0	0.003	2.05	0.948	26.4	21.8	19.4	18.4
Swift 77	Negative	26.4	Y = −0.184 x + 39.3	0.007	2.31	0.794	31.4	29.2	27.3	23.6
Kd, /h13
Goldrush 91	Positive	0.045	Y = 0.0000139 x2 − 0.00187 x + 0.101	0.014	0.0035	0.884	0.046	0.040	0.038	0.042
Legend	Positive	0.044	Y = 0.0000148 x2 − 0.00195 x + 0.099	0.004	0.0029	0.936	0.042	0.036	0.035	0.040
SY115T	Typical	0.048	Y = 0.0000228 x2 − 0.00306 x + 0.138	0.001	0.0035	0.964	0.048	0.039	0.035	0.042
SY158T	Typical	0.044	Y = 0.000000731 x3 − 0.000111  x2 + 0.00472 x − 0.005	0.012	0.0022	0.963	0.050	0.040	0.033	0.041
770-001	Typical	0.042	NS14	≥0.06315
Merlin Max	Negative	0.045	Y = −0.000428 x + 0.069	0.008	0.0064	0.781	0.051	0.045	0.041	0.033
Swift 77	Negative	0.041	Y = 0.00000108 x3 − 0.000187 x2 + 0.00962 x − 0.092	0.019	0.0020	0.950	0.061	0.050	0.039	0.036
U, % of DM16
Goldrush 91	Positive	15.5	Y = −0.00418 x2 + 0.66 x − 8.2	0.002	1.08	0.962	12.6	15.6	17.3	18.0
Legend	Positive	17.2	Y = −0.00315 x2 + 0.58 x − 4.2	< 0.001	0.91	0.983	14.8	18.0	20.0	22.1
SY115T	Typical	16.6	Y = −0.00639 x2 + 0.93 x − 13.1	0.008	2.21	0.909	15.2	18.9	20.6	20.1
SY158T	Typical	17.5	Y = −0.00484 x2 + 0.73 x − 7.0	0.010	1.76	0.902	15.4	18.4	19.9	19.9
770-001	Typical	18.4	Y = 0.211 x + 3.8	0.015	3.16	0.728	12.9	15.4	17.5	21.7
Merlin Max	Negative	20.3	Y = −0.00838 x2 + 1.24 x − 18.2	< 0.001	1.05	0.987	19.8	24.9	27.3	27.0
Swift 77	Negative	16.4	Y = −0.00735 x2 + 0.96 x − 12.0	0.024	2.10	0.846	15.9	18.9	19.7	16.9
IVDMD30, % of DM17
Goldrush 91	Positive	65.5	Y = 0.0106 x2 − 1.52 x + 114.4	0.002	2.12	0.952	68.7	63.0	60.6	62.2
Legend	Positive	63.1	Y = 0.0111 x2 − 1.60 x + 113.0	0.001	2.06	0.971	64.5	58.4	55.7	56.9
SY115T	Typical	65.8	Y = 0.0132 x2 − 1.85 x + 122.4	0.002	2.50	0.960	67.3	60.6	57.9	60.5
SY158T	Typical	63.5	Y = 0.000381 x3 − 0.0523 x2 − 1.64 x + 64.1	< 0.001	0.46	0.998	68.4	59.7	54.7	60.3
770-001	Typical	62.4	Y = −0.290 x + 82.4	0.020	4.72	0.695	70.0	66.5	63.6	57.8
Merlin Max	Negative	61.2	Y = 0.0137 x2 − 2.04 x + 124.6	0.001	2.47	0.975	62.1	53.7	49.6	49.8
Swift 77	Negative	63.7	Y = 0.0109 x2 − 1.47 x + 108.3	0.016	3.03	0.873	65.3	60.6	59.0	62.4
IVDMD48, % of DM18
Goldrush 91	Positive	75.5	Y = 0.0089 x2 − 1.28 x + 117.3	0.005	2.20	0.933	78.6	73.6	71.4	72.4
Legend	Positive	73.3	Y = 0.0089 x2 − 1.32 x + 115.3	0.001	1.77	0.973	75.1	69.7	67.2	67.4
SY115T	Typical	75.3	Y = 0.0119 x2 − 1.66 x + 125.9	0.002	2.30	0.957	76.4	70.5	68.2	70.7
SY158T	Typical	73.5	Y = 0.000389 x3 − 0.0559 x2 − 1.99 x + 65.6	0.001	0.90	0.992	78.9	70.8	65.8	70.1
770-001	Typical	72.3	Y = −0.270 x + 91.0	0.023	4.55	0.679	79.4	76.1	73.4	68.0
Merlin Max	Negative	70.7	Y = 0.0115 x2 − 1.75 x + 125.8	< 0.001	1.95	0.980	71.8	64.4	60.7	60.3
Swift 77	Negative	73.7	Y = 0.0104 x2 − 1.38 x + 115.3	0.022	3.05	0.852	75.0	70.6	69.2	72.7

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

Fraction B, which is the percentage of DM disappearing at a measurable rate, was not related to growth stage for any cultivar. Dependent means across all harvest dates were 58.6, 58.2, 55.2, 56.6, 56.6, 54.4, and 57.2 for Goldrush 91, Legend, SY115T, SY158T, 770-001, Merlin Max, and Swift 77 cultivars, respectively.

Lag time was related to growth stage only for the SY158T (Y = 0.00000529 x⁴ − 0.00118 x³ + 0.091 x² − 2.76 x + 32.2; P = 0.003; R² = 0.999) and SY115T cultivars (Y = −0.0668 x + 8.99; P = 0.020; R² = 0.696). The dependent mean for all other cultivars averaged across seven harvest dates ranged from 3.70 to 5.06 h.

Mean of dependent regression variable.

Most appropriate of quartic, cubic, quadratic, or linear regression models.

P > F for the selected regression model.

RMSE, root mean square error.

Mid-boot stage = 43.

Mid-heading stage = 55.

Full-flower stage = 65.

Soft-dough stage = 85.

Fraction A, percentage of DM disappearing at a rate too fast to measure during an in vitro incubation in buffered rumen fluid.

Kd, fractional disappearance rate of Fraction B.

NS, not significant.

P-values for linear, quadratic, cubic, and quartic regression models.

U, percentage of DM considered indigestible, based on the nonlinear regression model.

IVDMD30, in vitro disappearance of DM after a 30-h incubation in buffered rumen fluid.

IVDMD48, in vitro disappearance of DM after a 48-h incubation in buffered rumen fluid.

Although harvest date clearly affected the results of this experiment, it is equally clear that interpretation strictly on that basis is confounded by differences in maturation rate among triticale cultivars. To compare cultivars on a common GRST, response variables were regressed on GRST, such that concentrations of asNDFom, model parameters for disappearance of asNDFom and DM, NDFD30, NDFD48, IVDMD30, and IVDMD48 could be predicted at potentially distinctive GRST for harvest (mid-boot, mid-heading, full-flower, and soft-dough; Tables 5 and 8). On this basis, concentrations of asNDFom ranged within 8.9 percentage units across cultivars at the mid-boot (54.7% to 61.2%), mid-heading (58.7% to 66.1%), and full-flower (57.6% to 66.5%) stages of growth; however, both Legend and Merlin Max consistently occupied the upper portion of these ranges. Differences among cultivars for asNDFom were accentuated at the soft-dough stage of growth, when Merlin Max (63.7%) and Legend (52.5%) exhibited greater predicted asNDFom than all other cultivars (mean = 44.0%), and particularly Swift 77 (39.1%). For Merlin Max, this response was related to greater plant height, percentages of stem, and reduced percentages of grain head across harvest dates compared with other cultivars. For example, the greatest overall mean canopy height on any individual harvest date was 107 ± 17.7 cm, but the maximum canopy for Merlin Max was 145 ± 9.8 cm, or about 36% greater. This unique growth habit compared with other cultivars can be linked directly to the associated greater percentages of stem and reduced percentages grain head compared with the mean of all cultivars. For Legend, maximum canopy height (107 ± 6.8 cm) was identical to the overall mean maximum canopy height (107 ± 17.7 cm), and the somewhat elevated concentration of asNDFom for this cultivar can be linked directly to reduced percentages of grain head (31.0 ± 3.59%, 49.0 ± 1.30%, and 54.7 ± 5.08% on April 28, May 12, and May 26, respectively) compared with the respective associated means for all cultivars on those dates (42.0 ± 12.49%, 56.8 ± 10.44%, and 59.5 ± 9.72%).

The overall effects of these phenotypic differences among cultivars are shown in Figures 1–4, where the predicted IVDMD48, as well as their associated contributions from both digestible asNDFom and digestible nonfiber portions of the forage are illustrated for the two POS (Goldrush 91, Legend) and two NEG (Merlin Max, Swift 77) cultivars. At the mid-boot stage of growth (Figure 1), the predicted IVDMD48 for these cultivars was ≥71.8% of DM, with similar and majority percentages of the total DM disappearance associated with asNDFom digestibility (56.7% to 58.4%). Although total predicted IVDMD48 declined with maturation, contributions from digestible asNDFom remained ≥50.3% of the total digestible DM through the full-flower stage of growth (Figures 2 and 3). It is important to note that across these three growth stages, Merlin Max consistently exhibited poorer IVDMD48 than the mean of all other cultivars, and this differential was 5.4- (71.8% vs. 77.2%), 7.5- (64.4% vs. 71.9%), and 8.5-percentage units (60.7% vs. 69.2%) at the mid-boot, mid-heading, and full-flower stages of growth, respectively. This can be attributed directly to greater concentrations of asNDFom, as well as poorer NDFD48 (Table 5) at each of those growth stages when compared against all other cultivars.

Figure 1.

Predicted in vitro disappearance of DM for four triticale cultivars harvested at a mid-boot stage of growth. Bars for each cultivar represent digestible DM contributions from fiber (asNDFom) and nonfiber portions of the forage, as well as the total digestibility estimate. Data labels for each forage indicate the percentage of the total digestible DM attributed to digestible fiber.

Figure 4.

Predicted in vitro disappearance of DM for four triticale cultivars harvested at the soft-dough stage of growth. Bars for each cultivar represent digestible DM contributions from fiber (asNDFom) and nonfiber portions of the forage, as well as the total digestibility estimate. Data labels for each forage indicate the percentage of the total digestible DM attributed to digestible fiber.

Figure 2.

Predicted in vitro disappearance of DM for four triticale cultivars harvested at a mid-heading stage of growth. Bars for each cultivar represent digestible DM contributions from fiber (asNDFom) and nonfiber portions of the forage, as well as the total digestibility estimate. Data labels for each forage indicate the percentage of the total digestible DM attributed to digestible fiber.

Figure 3.

Predicted in vitro disappearance of DM for four triticale cultivars harvested at the anthesis (flowering) stage of growth. Bars for each cultivar represent digestible DM contributions from fiber (asNDFom) and nonfiber portions of the forage, as well as the total digestibility estimate. Data labels for each forage indicate the percentage of the total digestible DM attributed to digestible fiber.

From a strict nutritional perspective, these results suggest that digestibility differences among most cultivars before the onset of grain fill are limited, provided they are compared at common growth stages. Only those cultivars with unique phenotypic traits, such as Merlin Max, are likely to demonstrate distinctive differences in digestibility. However, the onset of grain fill changes this dynamic sharply, and is best illustrated by the distinctive differences exhibited by the two NEG cultivars (Merlin Max and Swift 77) at the soft-dough stage of growth (Figure 4). Within this context, Swift 77 exhibited the poorest predicted NDFD48 (45.5% of asNDFom), but the greatest predicted IVDMD48 (72.7%) of any cultivar evaluated. The poor NDFD48 was easily overcome by the low concentration of asNDFom (39.1%), and the large contribution of digestible nonfiber (presumably from a filling grain head) to the total digestible DM pool. For Swift 77 at this advanced GRST, digestible asNDFom comprised only 24.4% of the total digestible DM pool. In contrast, a much greater percentage of plant DM was partitioned into structural fiber (63.7%) for Merlin Max, resulting in a predicted IVDMD48 of only 60.3%; however, a high percentage of this digestible DM (51.7%) was associated specifically with digestible asNDFom. These stark differences between NEG cultivars can be further supported by comparing concentrations of ADL on both a percentage of total DM and asNDFom basis. For example, concentrations of ADL on the (relatively late) May 12 harvest date were 4.62 ± 0.227% and 2.27 ± 0.316% of DM for Merlin Max and Swift 77, respectively, thereby suggesting dilution of this structural fiber component by the greater grain fill observed for Swift 77. However, respective percentages of ADL expressed on a % of asNDFom basis were similar on the same date (7.4 ± 0.16% and 6.7 ± 0.99% of asNDFom). The effects of differences in grain fill also were evident on the same late date for calculated estimates of energy density for these cultivars (NEL = 1.31 ± 0.015 and 1.70 ± 0.030 Mcal/kg, respectively).

A similar, but less pronounced, contrast occurred for the two POS cultivars (Goldrush 91 and Legend) at the soft-dough stage of growth (Figure 4). Both cultivars exhibited NDFD48 ≥ 50.5%, which were numerically greater than all other cultivars; however, Goldrush 91 received a greater contribution to the total digestible DM pool from nonfiber plant components, yielding a better final predicted IVDMD48 (72.4% vs. 67.4%). These findings further confirm work by Cherney and Marten (1982b); for small-grain crops, decreases in forage quality for stem, leaf-blade, and leaf-sheath that occur as plants mature can be offset by the formation of highly digestible inflorescence, and the proportion that inflorescence contributes to the total plant DM.

Conclusions

Unless there is an urgency for removing the triticale crop, such as those created by a feed shortage or need to establish a secondary crop, harvest management decisions should be based on plant growth stage, and not calendar date. Assuming a common growth stage, this work suggests that most triticale cultivars will differ only modestly with respect to digestibility before the onset of grain fill. However, producers should carefully consider cultivars with unique phenotypic traits, such as the tall-growing character of Merlin Max, which is an exception to the previous generalization. If yield is a critical management objective, harvest should most likely be delayed until after the onset of grain fill, but cultivar selection becomes more complicated because varying contributions from the filling grain head can radically affect overall digestibility of DM. In this respect, producers should carefully evaluate their nutritional and production goals to assess whether their needs prioritize digestible fiber or overall DM digestibility, the latter of which may have limited contributions from digestible asNDFom. The poorer NDFD48 and IVDMD48 exhibited by Merlin Max should not necessarily be viewed as a negative; it may be a more appropriate choice for livestock classes with lower energy requirements, such as pregnant dairy heifers, or when a fiber source is needed within the blended diet to maintain proper rumen function.

Glossary

Abbreviations

ADF

acid detergent fiber

ADICP

acid detergent insoluble crude protein

ADL

acid detergent lignin

CP

crude protein

DM

dry matter

GRST

growth stage

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.