Grazed rain-fed small-grain cereals as a forage option for small-scale dairy systems in central Mexico

Small-scale dairy systems face reduced availability of water for irrigation of pastures and disruption in the amount or pattern of rains due to climate change, so research on alternative short-cycle rain-fed forages is needed. Grazing reduces feeding costs and small-grain cereals may be an option. The objective was to assess on-farm the performance of dairy cows grazing 6 h/day of three small-grain cereals: rye (RYE), wheat (WHT), and triticale (TRT), and supplemented 4.5 kg dry matter (DM)/cow/day of concentrate. Twelve Holstein cows were used in repeated 3 × 3 Latin squares with 14-day experimental periods. Pasture variables were analysed with a split-plot design, and economic analysis was performed with partial budgets. Sampling of forage (sward height, net herbage accumulation, botanical and chemical composition of herbage) and animal variables (milk yield and composition, live weight, and body condition score) were at the end of each period. The RYE showed a trend (P > 0.05) for higher net herbage accumulation (NHA) with highly significant differences (P < 0.001) among periods. The RYE had higher DM, a lower crude protein (CP) content (P < 0.05), and no differences for other chemical components between treatments (P > 0.05). The RYE proportion in pasture was consistently higher, whilst the lowest cereal proportion was in WHT. There were no differences (P > 0.05) for any animal variable. The RYE and TRT treatments proved useful as rain-fed forage alternatives under grazing. WHT showed lower crop and economic performance. However, given the higher cereal component in pastures, regrowth potential, and post-grazing herbage mass, as well as better economic performance, the RYE treatment was better ranked by the participating farmer.


Introduction
Small-scale cattle systems improve livelihoods worldwide by increasing simultaneously the assets of farming families and resilience to climatic, health, and economic risks (Blümmel et al. 2015).
In Mexico, over 88% of all farms with cattle are small (INEGI 2018), whilst small-scale dairy systems contribute more than 30% of the national milk production, and they enable families to overcome rural poverty (Espinoza-Ortega et al. 2007).
The sustainable intensification of small-scale livestock systems is promoted to improve their contribution to livelihoods and food security and to reduce the environmental footprint (Makkar 2016;FAO 2017); such intensification enables small-scale farmers to better commercialize their produce and increase their income (Rasmussen et al. 2018). Small-scale diversified livestock-crop systemsmay provide both high productivity and ecosystem services to mitigate the environmental footprint (Lemaire et al. 2014).
Grazing irrigated pastures has shown reduced feeding costs and increased profitability in small-scale dairy farms in central Mexico (Prospero-Bernal et al. 2017). However, water for irrigation is becoming scarce limiting the feasibility of long-term pastures, a situation that will worsen due to possible effects of climate change with disruptions in the amount and patterns of rainfall in the highlands of Mexico (Ortiz-Espejel et al. 2015).
Therefore, there is a need to assess alternative forage species better adapted to rain-fed conditions with low water requirements, which may be implemented in small-scale systems with a view to improving their productivity without increasing their environmental footprint (Lemaire et al. 2014).
Small-grain cereals (Payne et al. 2008) meet these requirements, since their short agronomic cycle requires less water than permanent pastures or long cycle annual crops such as maize, plus small-grain cereals are hardy and frost resistant.
Their successful use as silage to complement grazing during the dry season in small-scale dairy systems has been documented (Celis-Alvarez et al. 2016;Burbano-Muñoz et al. 2018;Gómez-Miranda et al. 2020;González-Alcántara et al. 2020), and they may offer a rain-fed alternative for these systems in the rainy season. Aasen et al. (2004) reported that grazing small-grain cereals in Canada is an option that may be resorted to as an emergency feed or to extend the grazing season in dry years. Another benefit is that grazing small-grain cereals is a means to improve soil fertility through nutrient cycling (Henz et al. 2016). McCartney et al. (2008) mentioned that grazing of smallgrain cereals for beef cattle in Canada was implemented more than a hundred years ago to extend grazing seasons, reduce feeding costs, and mitigate climatic effects (mainly dry years).
Over 45 years ago, in an assessment of small-grain cereals as a forage source, Fisher and Fowler (1975) indicated that when assessing small-grain cereals for forage it was not only necessary to take dry matter (DM) yields into account, but also the nutritional value of the forage for the livestock to feed.
Wheat (Triticum spp.) is the most widely cultivated cereal grain in the world. It has an ample adaptation range, growing in diverse environments, and may be sown as a fall or spring crop (Soto et al. 2009), and it has been used as a dualpurpose (grain and grazing) crop to overcome feed shortages in Brazil.
The rye (Secale cereale) has been used as a doublepurpose crop in Canada, as a grazed forage source in fall, spring, or both seasons, and then left to grow for a grain harvest (Kilcher 1982). Fisher and Lessard (1987) stated that forage yield and digestibility make it a useful option for ruminant feeding.
Triticale (X Triticosecale Wittmack) is an intergeneric hybrid bred to conserve the quality and grain yield of wheat with the hardiness of rye (McCartney et al. 2008). In Mexico, triticale has been promoted as a possible forage for conservation given its resistance to low temperatures, drought, pests, and disease, as well as reduced loss of nutritive value as it matures (Ballesteros-Rodríguez et al. 2015;González-Alcántara et al. 2020).
Small-grain cereals are reported for good yields of quality forage on short agricultural cycles that represent an alternative to grazing pastures (GRDC 2018), so the characteristics of these crops (Payne et al. 2008) meet the criteria to be included in cattle feeding strategies better adapted both to current situations of restriction in the availability of irrigation and in the future possible lower rainfall and disrupted rain patterns due to climate change (Thornton et al. 2009).
Although small-grain cereals have been evaluated as conserved forages, wheat, rye, and triticale have not been estimated under grazing in Mexico. Therefore, the objective was to assess the yield of these cereals and dairy cow performance under grazing for small-scale dairy systems.

Materials and methods
Work was an on-farm experiment with a collaborating smallscale dairy farmer in accordance with guidelines for participatory livestock technology development (Conroy 2005), in the municipality of Aculco, State of Mexico, Mexico (20° 10′ N, 99° 48′ W; 2470 m altitude) in the central highlands.
The adoption of innovations increases when farmers participate in research, not only providing their knowledge on the farming system, but also the social, economic, and cultural contexts that eventually determine adoption (Garcia et al. 2018).
The collaborating farmer actively participated in the experiment, had knowledge of the objectives, was duly informed and consulted, and his decisions were respected at all times. The experiment was undertaken on the farmers' farmland, with cows from his herd and under his management conditions.
Fertilization at sowing was 23 N-60 P-40 K/ha; a second fertilization took place 19 days after sowing with 57 kg N/ha as urea; and a third one, 44 days after sowing with 23 kg N/ ha. This way, total fertilization was 103 N 60 P-40 K kg/ha.
The grazing experiment started on day 65 after sowing, i.e. 24 July 2020, when the cereal crops were at least 25 cm in height.

Forage variables
Net herbage accumulation (NHA) was estimated with six exclusion cages per treatment (cereal crop) cutting to ground level with shears all herbage within 0.4 × 0.4 m metal squares outside the cages on day 1, and inside the cages at the end of each experimental period. Sward height was recorded with a rising plate grass metre and with a measuring tape (Plata-Reyes et al. 2018;Vega-García et al. 2020).
For chemical analyses, hand-plucked samples of herbage simulating grazing were taken at the end of each experimental period . Determinations of dry matter (DM), organic matter (OM), crude protein (CP), neutral detergent fibre (NDF), acid detergent fibre (ADF), and in vitro dry matter digestibility (IVDMD) were by standard methods following Celis-Alvarez et al. (2016). The estimated metabolisable energy content (eME) was from AFRC (1993).
Botanical composition was from five random samples per crop, cutting to ground level within a 0.4 × 0.4 m metal square and manually separating herbage into cereal crop, grasses, other plants, and dead tissue, dried in a draught oven at 55 °C for 48 h; results were reported as g/100 g DM (Sanderson et al. 2005).

Treatments and animals
The experimental design was a 3 × 3 Latin square repeated four times with the following treatments: continuous day grazing of rye crop (RYE), wheat (WHT), or triticale (TRT) for 6 h/day (9:00 to 15:00 h). The treatment sequence was randomly assigned; also, each cow within each group (Latin Square) was randomly assigned to the treatment sequence.
Additional to grazing, each cow received 4.5 kg DM/ day of a commercial dairy concentrate, split into two meals of half the allocation after morning (7:30 h) and evening (18:00) milkings, which were carried out manually.
Twelve Holstein cows were used in the experiment in groups of four cows each (for each Latin square), similar in terms of parity, days in milk, daily milk yield (MY), and live weight (LW). Mean pre-experimental variables were 3.6 ± 0.49 calving/cow, daily milk yield of 14.5 ± 1.19 kg/ cow/day, 120.5 ± 23.70 days in milk, 459.5 ± 13.00 kg LW, and 1.7 ± 0.05 body condition score (BCS) on a 1 to 5 scale.

Animal productive performance
Animal variables were recorded following Gómez-Miranda et al. (2020). Milk yield was recorded daily over the last 4 days using a clock spring balance with a 20-kg capacity, and an aliquot sample refrigerated until analyses for milk fat, protein, and lactose content with an ultrasound milk analyser (Ekomilk Bond, Ekomilk, Stara Zagora, Bulgaria). Milk yield was corrected to 3.5% fat content following Dairy Records Management Systems (2014) and Plata-Reyes et al. (2021). Analyses for milk urea nitrogen (MUN) were by the enzymatic colorimetric method (Chaney and Marbach 1962), as reported in previous works (Celis-Alvarez et al. 2016;Plata-Reyes et al. 2018. LW was recorded during the last two consecutive days at the end of each experimental period with an electronic weighbridge, whilst BCS (on a 1 to 5 scale), on the last day of each period, always by the same observer.

Economic analyses
Partial budget analyses were run considering only feeding (including cropping) costs and margins over feeding costs as in previous works (Celis-Alvarez et al. 2016;Prospero-Bernal et al. 2017;Gómez-Miranda et al. 2020).

Statistical analyses
The experiment, for animal and economic variables, followed a 3 × 3 Latin square design repeated four times (Kaps and Lamberson 2004), with 14-day experimental periods (10 days for adaptation and 4 days for sampling).
Latin square designs are useful for research on feeding strategies for dairy cows as they enable maximizing a small number of available experimental units ( where μ = general mean; S = effect due to squares; i = 1, 2, 3, 4; C = effect due to cows in squares j = 1, 2, 3; P = effect due to experimental periods; k = 1, 2, 3; t = effect due to treatment l = 1, 2, 3; and e = residual error term. The experimental design for crop variables was a randomized split-plot design with crop subdivisions as experimental units, crops as main plots, experimental periods as split-plots, and samplings (inside and outside the exclusion where m = general mean; csd = effect of crop subdivision i = 1, 2, 3; T = effect of crop (small-grain cereal) treatment (main plot) j = 1, 2, 3; E = error term for main plots [csd(T)ij]; p = effect of experimental periods (split-plot) k = 1, ..., 3; Tp = interaction term between treatments and measurement periods; Tr = interaction term between treatments and sampling replicates (r m = 1,2); e = experimental error term. Figure 1 shows meteorological conditions in the study area during the experiment. Although there was adequate rainfall (mean of 127 mm per experimental period), it tended to decrease by the end of the experiment. Mean maximal and minimal temperatures were 24.5 °C and 10.6 °C, respectively. Table 1 reports that RYE had a trend (P > 0.05) for the highest net herbage accumulation per period over the experiment with 2.8 ton DM/ha/period, whilst WHT accumulated 1000 kg DM/ha/period less than RYE and TRT exhibited an intermediate yield.

Crop variables
There were highly significant differences (P < 0.001) among experimental periods for the three crops, with the Y ijkl = m + csd i + T j + E k + p l + Tp jl + Tr jm + e ijk lowest NHA in period 1 and the highest in period 2. The RYE also had the higher sward height of the three cereals (P < 0.05).
The RYE showed the best yield in spite of the haphazard rain with dry spells during the experiment, and a greater NHA was from the cereal not from the spontaneous grass or other plants as was the case in WHT (Table 2).
In botanical composition, over the experiment, there was a greater presence of cereal in RYE (mean 78.9%) in comparison with the other two treatments, WHT (mean 19.4%) and TRT (mean 40.8%). In RYE, the presence of cereal decreased as the experiment progressed, whilst the presence of other plants remained constant (mean 15.2%).
The presence of other plants in WHT was higher over the entire experiment (mean 61.2%), with the highest presence of cereal in period 1 with only 21.5%. It is worth mentioning that the presence of grass increased as periods passed, surpassing the proportion of wheat in period 3 when the cereal accounted only for 18.4% of DM.
In TRT, the presence of other plants was also high, even though in this case the presence of cereal was twice as much in comparison with WHT. For this treatment, grass in the crop reached a peak in value in period 2 (7.6%), and the presence of dead tissue was minimal (mean 3.4%) ( Table 2).
The various growth stages of the cereals influence their chemical composition. The RYE was at a more advanced growth stage given its early vigorous growth, so that its DM content was higher than those of WHT and TRT. This stage of growth also caused that CP content was 10% lower in RYE (P < 0.05) in comparison with the other treatments; albeit, there were no differences (P > 0.05) in relation to NDF and ADF and IVDMD. There were differences in all the variables of chemical composition among periods (P < 0.05) except for NDF (Table 3).  Table 4 reports results for animal variables. There were no significant differences (P > 0.05) for any evaluated animal performance variable between treatments as MY, LW, or BCS, although results for LW and BCS are only indicative as the short experimental periods preclude any other interpretation for these variables.

Animal variables
There were no differences (P > 0.05) in milk composition for fat, protein, and lactose. There were no differences (P > 0.05) either for MUN or for milk pH, which was reported as slightly acid.
DM intake during the experiment was similar (P > 0.05) among treatments. The mean intake in pasture was 10.2 kg DM cow/day and the mean total intake was 14.6 kg DM/ cow/day (Table 5). Table 1 Net herbage accumulation (NHA) and sward height of small-grain cereal crops a,b P < 0.05, NS P > 0.05, NHA net herbage accumulation, GM grass metre, MT measuring tape, RYE rye crop, WHT wheat crop, TRT triticale crop, SEMt standard error of the mean for treatments, SEMp standard error of the mean for periods, SEMt*p standard error of the mean of interaction, P-t probability value for treatments, P-p probability value for experimental periods    Table 6 reports results of the economic analyses for feeding costs (in USD$). In the present experiment, treatments accounted for 25.6, 32.0, and 26.0% of the total feeding costs for RYE, WHT, and TRT, respectively. The RYE had the best income margins over feeding costs, although the income/feeding cost ratio in all treatments was above USD$ 2.45.

Discussion
According to Soto et al. (2009), sowing date has a positive effect on the development of forages, as it defines the environmental conditions that plants will have along their life cycle. Additional to climate factors, others that intervene in the final yield of the crops are related to genetics and management.
In this experiment, the early sowing may have affected the poor forage performance of the cereal in WHT and TRT treatments (Table 2), where initial low rainfall from sowing to grazing may have slowed development in WHT (López-Bellido 1991).
The RYE treatment had a trend towards significant higher net forage accumulation per period that over the whole experiment added up to 8.4 ton DM/ha for RYE, 19% over the 5.2 ton DM/ha for WHT, and TRT with an intermediate 7.6 ton DM/ha. The yield of each cereal in the treatments (Table 2) results in an NHA for the cereal of 2.2, 0.3, and 0.9 ton DM/ha/period for RYE, WHT, and TRT, respectively.
These forage yields were achieved in 107 days in total considering 65 days from sowing to the initiation of grazing, plus 42 days grazing (in three experimental periods), which at a stocking rate of 4 cows/ha meant 168 cow grazing days/ha. Total forage accumulation during the experiment was higher than 5.6 ton DM/ha reported by Celis-Alvarez et al. (2017) for triticale and rye in the same study area, though in their work cereals were not used for grazing and were cut at 72 days after sowing. McCartney et al. (2008) mentioned that rye has been recommended as an annual crop in dry zones and produces more DM than triticale, which concurs with this work.
Wheat did not have an acceptable yield to be considered as an annual forage option for grazing in the study area when compared to RYE and TRT, even though it was recommended as a forage for annual pasture in Canada for droughtprone areas given its water-deficit resistance (McCartney et al. 2008).
The WHT treatment in this work had a 1.0 ton DM/ha/ period lower yield than RYE and was severely invaded by grass and weeds. In this regard, Soto et al. (2009) mentioned that wheat for grazing must receive higher dosages  of nitrogen fertilization with a view to stimulating regrowth and not decreasing the final yield, for in a dual-purpose management, cultivations are subject to defoliation of varying intensities (in which forage offer, weight, and number of animals take part) and frequencies, though regardless of the degree, it means stress for the plants. For its part, net triticale accumulation per period was 2.7 ton DM/ha. The lowest NHA of the plant was for period 1 (Table 1). This is because in the pastures the presence of other plants and grass increased as periods passed (Table 2), though RYE, the treatment with the lowest NHA in period 1, had a forage availability of 20.9 kg DM cow/day. For period 3, NHA decreased for the three treatments probably due to lower rainfall (Fig. 1). Fisher and Fowler (1975) mentioned that the parameters of nutritional quality are influenced by the progressive stage of maturity up to the harvest day; in this case, the later maturing characteristics of WHT affected DM content in comparison with RYE and TRT. Conversely, the more advanced growth stage of RYE resulted in a lower content of CP (P < 0.05), though CP for this cereal (136 g/kg DM) is 42% higher than the value reported by Celis-Alvarez et al. (2017). Although there were more days after sowing in the work herein reported compared to Celis-Alvarez et al. (2017), there was active regrowth in RYE after grazing. Juskiw et al. (1999) assessed rye and triticale as monocultures associated with barley and observed a CP content over 190 g/kg DM for both cereals sown in the same season. WHT and TRT treatments had CP values similar to reports by Khorasani et al. (1997) and Henz et al. (2016).
In terms of IVDMD, there were no differences (P > 0.05) among treatments, though differences between periods were detected, with lower digestibility as time passed. Fisher and Fowler (1975) mentioned that chronological age as a maturity measurement does not accurately reflect physiological maturity, neither does a high DM yield mean it is necessarily digestible. The nutritional value of a forage is influenced by its nutrient composition, and for its part, this is better ascertained by a digestibility measure.
The IVDMD values obtained in this work rank cereals as of medium quality, since Celis-Alvarez et al. (2017) reported digestibility for rye and triticale 5.5% higher. Henz et al. (2016) found lower values for wheat (615 g/kg DM) in relation to the IVDMD observed in the work herein reported.
Milk yields from grazing the cereals were similar to those obtained when grazing on grass-clover pastures in these small-scale production systems (Plata-Reyes et al. 2018;Carrillo-Hernández et al. 2020;Muciño-Álvarez et al. 2021). As there were no differences in DM intake or metabolisable energy content of forages, there were no differences in milk production (Ribeiro-Filho et al. 2021).
Grazing time and supplementation enabled a mean pasture DM intake of 10.2 kg DM/cow/day and a total mean DM intake of 14.7 kg DM/cow/day, representing 3.1% of mean live weight, indicating sufficient pasture in the smallgrain cereal plots and that the 6-h access time for grazing did not restrict forage intake by cows, and that when access to pastures is reduced, cows increase time devoted to grazing to maximize intake in expense to rumination or idling time, as reported by Ribeiro-Filho et al. (2021).
The chemical composition of milk was within the parameters established by Mexican Standards for raw milk as regards to milk fat and protein contents. Lactose content was at the lower limit. According to Costa et al. (2019), the factors involved in lactose content are the number of lactations (with decreased lactose content with increased number of lactations; in this work, the average was 3.6 calvings/ cow), days in lactation, as the lactose-content curve behaves as that of milk yield (the experiment began with 120 days in milk on average), somatic cell count, and energy contained in feed.
According to Kohn et al. (2002), MUN contents were higher than recommended, since these authors stated that the adequate concentration level for this variable in dairy herds should be within the 8.5 to 11.5 mg/dL range, although values were within the range for normal MUN between 11 and 18 mg/dL contents reported by Powell et al. (2011) for cows in the USA. In the experiment herein reported, it is possible that the diet might have had a higher protein content than requirements (Powell et al. 2011), albeit Aguilar et al. (2012 mentioned that there are a number of factors besides protein intake that may bring about variations in MUN content (sampling, season, weight, and breed of the cows).
As for the economic analysis, McCartney et al. (2008) mentioned that under annual crop grazing, there are expenses at the beginning to afford machinery, fertilizers, and seeds, plus tillage work in the fields. However, the advantage of grazing is that livestock collect their feed instead of gathering it mechanically, lowering costs. The comparison of the costs of establishing the crops (on the basis of 2020 prices) demonstrated that RYE is the most expensive due to the higher seed price of rye, compared with wheat and triticale seed prices. Therefore, costs for WHT and TRT were 19% and 18% less expensive, respectively. Separately, Aasen et al. (2004) assessed the potential of spring cereals under grazing and found a low cost for crop establishment under Canadian conditions.
The best performance of RYE in forage DM yield turned it into the lowest cost forage per each kg DM, decreasing expenditures in comparison with WHT and TRT. Conversely, WHT had the higher feeding cost per kg milk (4.5% and 5.6% v. TRT and RYE, respectively) and a lower income margin in relation to feeding costs (Table 6), and as has been mentioned, the ratio of income over feeding costs was very favourable since the three treatments were above USD$ 2.45 income/USD$ 1.00 feeding costs.
Another advantage of the three cereal grains assessed was their dual-purpose possibilities for a silage crop after grazing, as the standing herbage mass at the end of the grazing experiment was 9.9 ton DM/ha for RYE, 5.7 ton DM/ha for TRT, and 4.3 ton/ha for WHT, with a cereal presence of 74.7, 41.6, and 17.7%, for RYE, TRT, and WHT, respectively, and a presence of grass and weeds of 19.5, 53.9, and 76.7% in RYE, TRT, and WHT, respectively; the rest was dead forage.
Since there were no differences in the animal variables, the three cereals are put forward as viable options for grazing in these small-scale dairy production systems owing to their high DM yields, with adequate protein content and medium digestibility to sustain milk yields and profitability in mid-lactation low yielding dairy cows.