Forage yield, nutritive value, and ensilability of sweet pearl millet and sweet sorghum in five Canadian ecozones

Sweet pearl millet [Pennisetum glaucum (L.) R. Br.] and sweet sorghum [Sorghum bicolor (L.) Moench], previously tested for ethanol production, were evaluated as high sugar crops for animal feeds to possibly replace silage corn (Zea mays L.). We compared the forage yield, nutritive value, and ensilability of one hybrid of sweet pearl millet and two of sweet sorghum to a locally adapted silage corn hybrid in five Canadian ecozones. Forage yields of sweet pearl millet and sorghum were similar to that of silage corn in the Boreal Shield, Mixedwood Plain, and Atlantic Maritime ecozones, greater in the Prairies, and lower in the Pacific Maritime ecozone. Across sites, forage dry matter concentration was less for sweet pearl millet (289 g kg−1) and sweet sorghum (245 g kg−1) than for silage corn (331 g kg−1). Sweet pearl millet had a lower total digestible nutrient (TDN) concentration (452 g kg−1 DM) and aNDF digestibility (NDFd) than sweet sorghum and silage corn along with greater neutral detergent fibre (aNDF) and water-soluble carbohydrate (WSC) concentrations than silage corn. Sweet sorghum had greater aNDF and WSC, lower starch, and similar TDN (534 g kg−1 DM) concentrations, but greater NDFd compared with silage corn. Sweet pearl millet and sorghum fermented as well as silage corn, reaching low pH values and acceptable concentrations of lactic and volatile fatty acids. Sweet sorghum is therefore a viable alternative to silage corn in Canada except in the Pacific Maritime ecozone, but early-maturing hybrids with acceptable DM concentration at harvest are required.

Because of its high N requirement and low N utilization efficiency, production of silage corn presents a risk of N losses to the environment (Drury et al. 2014). Also corn is increasingly more susceptible to dry conditions (Lobell et al. 2014). Alternative crops are needed for dairy feed that are more drought tolerant and sustainable than silage corn. Because of the likely increase in the frequency of drought events with global warming, and because of the public interest in environmental protection, it seems appropriate to find potential alternatives to silage corn.
Pearl millet and sorghum are annual forage grasses known for their greater resistance to drought and their tolerance to soils with low organic matter (Andrews and Kumar 1992;Bidinger and Hash 2004) as well as for their high N and water use efficiencies (Singh and Singh 1995;Schittenhelm and Schroetter 2014;Thivierge et al. 2015b). These two species have, therefore, valuable characteristics that could be helpful in dealing with certain consequences of global warming, such as increased average temperatures and the heterogeneous distribution of annual precipitations (do Nascimento et al. 2005;Thivierge et al. 2015b).
Sweet pearl millet and sweet sorghum are hybrids developed to contain more fermentable sugars than standard forage types for use in ethanol production. These hybrids can have dual purpose since the press residues (bagasse) from the ethanol industry are suitable as animal feed (dos Passos Bernardes et al. 2016). Whole plants of sweet pearl millet and sweet sorghum could also be used as animal feed because their high sugar concentration contributes to high levels of total digestible nutrients (TDN), that are beneficial in dairy production (do Nascimento et al. 2005;Amer et al. 2012).
Recent studies conducted in Quebec reported forage dry matter (DM) yields for both sweet pearl millet (10 to 20 Mg ha −1 ; Bouchard et al. 2011;Leblanc et al. 2012 al. 2016) that are comparable to the average DM yields of silage corn in Québec (12.9 to 18.9 Mg ha −1 ; La Financière agricole 2018). The nutritive value of sweet pearl millet and sweet sorghum has not been widely studied. Recent studies conducted in eastern Canada (Bouchard et al. 2011;Amer et al. 2012;Bélanger et al. 2018) reported greater fibre and lower total N concentrations for sweet pearl millet and sweet sorghum when compared with recommended average concentrations for a silage corn hybrid (National Research Council 2001), suggesting a lower potential for milk production. In contrast, the same studies indicated greater in vitro neutral detergent fibre digestibility (NDFd) in sweet pearl millet and sweet sorghum, indicating greater fibre digestibility than in silage corn. However, no studies have compared the yield and nutritive value of sweet pearl millet, sweet sorghum, and silage corn under the same growing conditions in Canada. The objective of this study was, therefore, to determine whether sweet pearl millet and sweet sorghum are viable alternatives to silage corn, in terms of yield, nutritive value, and ensilability for feeding dairy cows in five contrasting Canadian ecozones.

Description of sites and treatments
This study was carried out at five experimental sites located in Canadian ecozones where silage corn is grown for dairy feed: 1) Agassiz, BC, Pacific Maritime (lat. 49°14′N, long. year was considered as a random factor. The treatments were replicated four times in a randomized complete block design. The treatments consisted of a locally-adapted silage corn hybrid as control, a sweet pearl millet hybrid, and two sweet sorghum hybrids with the BMR ("brown midrib") genetic mutation from two different regions (Europe and North America) grown under rain-fed conditions.
The sweet pearl millet hybrid (CSSPM7) was developed by AERC Inc. (Delhi, ON). One of the sweet sorghum hybrids (BMR1) was developed in France as X15-06 and the second (BMR2) was developed in the USA as SM#1 (sterile male, catalogue number 327 X 36; Richardson Seeds, Vega, TX). At Lethbridge, only the European BMR sweet sorghum hybrid (BMR1) was tested.
Standard, non-BMR, locally-adapted silage corn hybrids were seeded at each site. The seeding dates are reported in Table 1. Silage corn was planted earlier than the other species due to its lower soil temperature tolerance. Minimum soil temperatures for planting and germination are 6 °C for silage corn and 10 to 12 °C for sweet pearl millet and sweet sorghum. In 2016, the silage corn had to be reseeded in Kentville 12 d after sweet pearl millet and sweet sorghum. Silage corn was seeded at an average depth of about 5 cm and a row spacing of 76 cm, for a plant density of 75 000 to 105 000 plants ha −1 . Sweet pearl millet and sweet sorghum were seeded at a rate of 15 kg ha −1 and 10 kg ha −1 , respectively, at a seeding depth of 1.5 to 2.0 cm, and at a row spacing of 18 cm.
At four sites, part of the N requirements were met with cattle slurry applied in the fall preceding seeding (2014 and 2015) in order to supply about 50 kg available N ha −1 to the crops in the next growing season. At Agassiz, however, cattle slurry was applied 2 to 3 wks before seeding in the spring of 2015 and 2016 according to local practices. The manure was incorporated by ploughing or disking to a depth of 15-20 cm at all sites. In 2015 and 2016, silage corn received an F o r R e v i e w O n l y additional 35 to 100 kg N ha -1 as mineral fertilizer in accordance with local recommendations, taking into account the 50 kg of available N provided by the cattle slurry and assuming no losses.
Sweet pearl millet and sweet sorghum received 65-75% of the amount of mineral N fertilizer applied to silage corn. At Agassiz for example, sweet pearl millet and sorghum received 80 kg mineral N ha -1 which represented 65% of the amount of mineral N fertilizer applied to silage corn.
Half of the mineral N fertilizer was applied at seeding, and the rest at the five-to-six-leaf stage of corn and at tillering of sweet sorghum and sweet pearl millet. At Kentville, mineral N was only applied at seeding because of the low mineral N application rate (35 kg N ha -1 ). Phosphorus and potassium requirements were met in accordance with soil analyses and current recommendations for corn at each site. At all sites, the plots were tilled with a vibrocultivator or a disc harrow before seeding. For controlling broadleaf weeds, sweet pearl millet and sweet sorghum were treated with the herbicide Basagran Forte (active ingredient: bentazon; isopropyl-3 1 H,3 H-benzothiadiazin-2,1,3 one-4 dioxide-2,2) at a rate of 1.08 kg active ingredient ha −1 at the three-to six-leaf crop stage. The corn hybrids at all sites were Roundup Ready® and were treated by hand with glyphosate [N-(phosphonomethyl)glycine] at a rate that varied depending on the site and the weed development stage. Additional weeding was performed by hand when required, especially for the sweet pearl millet and sweet sorghum.

Environmental conditions
Soil characteristics were determined at the beginning of the study. At each site, soil cores were randomly taken prior to seeding to a depth of 15 cm and pooled to make three composite samples. Soil samples were air-dried and sieved to 2 mm prior to analysis. Soil pH was measured F o r R e v i e w O n l y in a 1:1 soil:water ratio (Hendershot et al. 1993). The total organic carbon concentration was determined by dry combustion (Model TruMac, Leco Corp., Loveland, CO). Soil phosphorus and potassium were extracted following the Mehlich III method (Tran and Simard, 1993) and determined by inductively coupled plasma optical emission spectrometry (Optima 4300 DV, PerkinElmer Corp., Norwalk, CT). The characteristics of the soils at each site are presented in Table 1. The long-term average precipitation, cumulative growing degree-days, and corn heat units (CHU) are presented in Table 2.

Forage sampling and yield determination
Each treatment (crop) was replicated four times in a randomized complete block design.   followed with the determination of aNDF concentration of the post-digestion residues. The incubation was done with an ANKOM Daisy II incubator, using the F57 filter bags and the batch incubation procedures of ANKOM Technology Corp.. The rumen fluid was taken from two ruminally fistulated dairy cows fed the same diet constituted of 37% grass silage, 15% corn silage, 8% hay, 30% corn grain, and 10% concentrate mix. This diet was formulated to meet the nutritional requirements of a lactating dairy cow expected to produce 10,200 kg milk yr -1 , according to NRC (2001). Each sample was analyzed in duplicate, with a 5% maximum coefficient of variation between duplicates. The IVTD (g kg -1 DM) and the NDFd (g kg -1 aNDF) were calculated as follows: To determine total N concentration, 0.1 g of each sample was mineralized using a block digester (Digestor 2520; FOSS, Hillerød, Denmark) in accordance with the method of Isaac and Johnson (1976). The solution obtained was then analyzed for total N concentration by a continuous-flow analyzer with automatic injection (QuikChem 8500 Series 2 FIA System, Lachat Zellweger Analytics, Milwaukee, WI) in accordance with method 13-107-06-02-E (Lachat Instruments 2016).
To extract water soluble carbohydrates (WSC), 0.1 g of forage sample was soaked in 25 mL of distilled water for 60 min and the solution was filtered through a Whatman No. 2 filter paper.
The final volume of the liquid fraction was adjusted to 45 mL and five mL of 1 N sulfuric acid (H 2 SO 4 ) was added to the filtrate; the mixture was heated in a water bath at 100 °C for 15 min Suzuki 1971). The WSC were then measured with a spectrophotometer at 415 nm and using a phydroxybenzoic acid hydrazide (PABAH) solution (Blakeney and Mutton 1980). Starch concentration was determined using 0.25 g of ground material. The material was washed three times with ethanol 80%, heated to 60 °C, and centrifuged at 2000 g, and the resulting pellet was then left to air dry overnight. The next day, the samples were diluted in distilled water and heated to 100 °C in order to gelatinize the starch. The starch was hydrolyzed in an alkaline solution (acetate buffer) containing amyloglucosidase (75.12%). The starch concentration of the samples were then measured by a spectrophotometer at 415 nm and using a PABAH solution (Blakeney and Mutton 1980). Crude fat (ether extract) was determined using the Ankom xt15  Table 3. The ratio of prediction

Ensilability
The ensilability of sweet pearl millet, sweet sorghum, and silage corn at harvest time was determined in 2016 at the St-Augustin and Ste-Anne sites as per the method described by Tremblay et al. (2014). Briefly, after the forage from the harvest zone had been ground with a corn chopper, a sufficient quantity of representative material from each plot was used to fill one PVC mini-silo per plot. The mini-silos were 25 cm long with an inner diameter of 7.5 cm. To compact the material, 1200 kPa of pressure was applied to the forage with a hydraulic press. The mini-silos were closed and kept at room temperature (20-23 °C) for 90 d. After this fermentation period, the mini-silos were opened and a silage sample (1/3 of the silo) was stored at −20 °C until analyzed for pH, lactic acid, and volatile fatty acids according to Tremblay et al. (2014).
The pH was measured using an Accumet AR25 pH meter (Fisher Scientific, Fair Lawn, NJ) on 20 g of fresh silage mixed with 200 mL of distilled water. The mixture was then allowed to stand for 24 h at 4 °C, with occasional agitation, for the measurement of lactic acid and volatile fatty acids by ion chromatography using a Dionex ICS-2000 system equipped with an IonPac AS11-HC/AG11-HC column (Dionex Inc., Sunnyvale, CA).

Statistical analyses
The data were analyzed using the PROC MIXED procedure of the SAS software package (SAS Institute Inc., Cary, NC) with sites and treatments (crops) as fixed factors, and years as a F o r R e v i e w O n l y random factor. The homogeneity of variance of the residuals was checked visually using the graphs obtained with PROC GPLOT, and the normality of the residuals was checked using the kurtosis, skewness, and Shapiro-Wilk statistical tests, obtained with PROC UNIVARIATE. The raw data did not need any transformations, and the type I error was set at 0.05. The site × treatment (crops) interaction was significant (P < 0.001) for each of the attributes, with the exception of NDFd and WSC ( Table 4). The data were therefore reanalyzed by site.

Forage yields
The DM yield of silage corn (17.1 Mg DM ha −1 ) was greater than those of sweet pearl millet and the two sweet sorghum hybrids (average of 11.7 Mg DM ha −1 ) at Agassiz, whereas at Kentville, the DM yield of silage corn (15.5 Mg DM ha −1 ) was greater than that of the BMR1 sweet sorghum hybrid (10.6 Mg DM ha −1 ) but not of the sweet pearl millet and BMR2 sweet sorghum hybrids (Fig. 1A). At Lethbridge, however, the DM yield of silage corn was the lowest (15.8 Mg DM ha −1 ), while that of sweet pearl millet was the highest (34.3 Mg DM ha −1 ). The high DM yields of sweet pearl millet and sweet sorghum at Lethbridge compared to that of silage corn can be partly explained by the observations made by Singh and Singh (1995) as well as by Farré and Faci (2006), who concluded that the greater the water stress, the greater the decrease in corn  Table 2).
Sweet pearl millet, the two sweet sorghum hybrids, and silage corn had similar DM yields within sites at St-Augustin and Ste-Anne. At those two sites, the DM yields of sweet pearl millet ( Fig. 1A) were within the range of those reported in previous studies (13.6 to 20.4 Mg DM ha −1 ) conducted at the same sites and with the same hybrid (Bouchard et al. 2011;Leblanc et al. 2012;Thivierge et al. 2015a;dos Passos Bernardes et al. 2015). In some of those studies, however, the DM yield of sweet pearl millet was greater than that of sweet sorghum (Thivierge et al. 2015a;dos Passos Bernardes et al. 2015). Climate variations as well as soil conditions specific to each site may have contributed to the variation in DM yield among sites. These results from five sites across Canada indicate that, in terms of DM yield, sweet pearl millet and sweet sorghum are promising alternatives to silage corn, except at Agassiz.

Forage dry matter concentration
Silage corn DM concentration at harvest varied from 260 to 380 g kg −1 depending on the site with an average across sites of 330 g kg −1 (Fig. 1B; Supplementary Table S1), which is close to our objective of harvesting all crops when silage corn reached a DM concentration of approximately 350 g kg −1 . Within sites at Lethbridge, St-Augustin, and Ste-Anne, the DM concentrations of sweet pearl millet, sweet sorghum, and silage corn differed significantly but there was no difference between the sweet sorghum hybrids at the two latter sites (Fig. 1B).
Average DM concentration across those three sites were 331 g kg −1 for silage corn, 289 g kg −1 for sweet pearl millet, and 245 g kg −1 for the sweet sorghum hybrids. At Agassiz, the silage corn DM concentration (296 g kg −1 ) was not statistically different from the sweet pearl millet DM concentration (264 g kg −1 ), but significantly greater than the two sweet sorghum hybrids (average F o r R e v i e w O n l y of 222 g kg −1 ). Kentville was the only site where sweet pearl millet had a greater DM concentration (300 g kg −1 ) than silage corn (260 g kg −1 ) which was seeded very late in one year (Table 1).
One of the conditions required for ensiling forage is that it must have a DM concentration of 280 to 350 g kg −1 for storage in a bunker silo (Salfer and Linn 1992; Valacta 2017) and 300 to 500 g kg −1 for an unsealed tower silo (Bagg 2013). Low DM concentrations promote fermentation by Clostridia and the occurrence of butyric acid that reduces the palatability of the forage along with increasing the risks of seepage, whereas high DM concentrations lead to the development of moulds (Bagg 2013). In previous studies in Quebec, average DM concentration was 269 g kg −1 for the same sweet pearl millet hybrid and 249 g kg −1 for different sweet sorghum hybrids (Amer and Mustafa 2010;Amer et al. 2012). Overall, sweet pearl millet had a DM concentration at the time of the silage corn harvest that was close to being suitable for storage in a bunker silo. Nevertheless, the silage potential for sweet sorghum was currently constrained by its low DM concentration across sites of 245 g kg −1 .
The two BMR sweet sorghum hybrids had similar ADF and aNDF concentrations except at Ste-Anne. At Lethbridge, the ADF and aNDF concentrations of silage corn and the BMR1 sweet Overall, the sweet sorghum hybrids had greater IVTD (average of 820 g kg −1 DM) than silage corn (average of 778 g kg -1 DM), except at St-Augustin where the IVTD of the sweet sorghum hybrids did not differ from that of the silage corn, and at Kentville where the IVTD of the BMR2 sweet sorghum hybrid and silage corn were similar (Fig. 1E). The IVTD values of sweet sorghum were greater than the average of 790 g kg −1 DM reported by Bélanger et al. (2018) in Quebec with another sweet sorghum hybrid. This generally greater DM digestibility (IVTD) of the two sweet sorghum hybrids could be explained by the presence of the BMR gene, which results in a lower lignification. At all sites where the two sorghum hybrids were studied, the BMR1 had a  (Fig. 1F), with an average of 706 g kg −1 aNDF across sites. The NDFd of BMR2 sweet sorghum hybrid was numerically lower than with BMR1 sweet sorghum hybrid at all sites where they were both studied but the difference was significant only at Ste-Anne (Fig. 1F). The average NDFd for the BMR1 and BMR2 sorghum hybrids (719 and 690 g kg −1 aNDF, respectively) was greater than the average value of 601 g kg −1 aNDF reported by Bélanger et al. (2018) for sweet sorghum. The difference could possibly be, as for IVTD, explained by the presence of the BMR gene in the sweet sorghum hybrids of the present study. Although it is not ideal to compare BMR sweet sorghum hybrids with non-BMR silage corn, our results suggest that BMR sweet sorghum forage could be as good as non-BMR silage corn or even more digestible for dairy cattle, despite greater ADF and aNDF concentrations.

Forage total N, WSC, and starch concentrations
Silage corn, sweet pearl millet, and the two sweet sorghum hybrids did not differ in total N concentration at Agassiz, St-Augustin, and Ste-Anne ( Fig. 2A). At Kentville, silage corn had a significantly greater total N concentration (10 g kg −1 DM) than sweet pearl millet and the two sweet sorghum hybrids (average of 7 g kg −1 DM). At Lethbridge, sweet pearl millet and sweet sorghum had a similar total N concentration (average of 16 g kg −1 DM), which was significantly greater than that of silage corn (11 g kg −1 DM). Total N concentrations from 9 to 14 g kg −1 DM for sweet pearl millet (Bouchard et al. 2011;Leblanc et al. 2012;Bélanger et al. 2018) and from 10 to 19 g kg −1 DM for sorghum (Getachew et al. 2016)  Considering that the recommended average concentrations for a silage corn hybrid are 280 g ADF kg −1 DM, 450 g aNDF kg −1 DM, and 14 g total N kg −1 DM, as well as a NDFd value of 560 g kg −1 aNDF (National Research Council 2001), our corn silage hybrid was in general considered of good maturity with average values across sites of 267 g ADF kg −1 DM, 500 g aNDF kg −1 DM, 10 g total N kg −1 DM, and a NDFd of 589 g kg −1 aNDF. With average values across sites of 392 g ADF kg −1 DM, 644 g aNDF kg −1 DM, 10 g total N kg −1 DM, and a NDFd of 529 g kg −1 aNDF, sweet pearl millet was considered to have greater fibre concentration, but less fibre digestibility than silage corn. And with average values across sites of 347 g ADF kg −1 DM, 592 g aNDF kg −1 DM, 9 g total N kg −1 DM, and a NDFd of 706 g kg −1 aNDF, sweet sorghum forage was considered to have greater fibre concentration and greater fibre digestibility than silage corn.
The sweet pearl millet WSC concentrations were greater than those of the silage corn at Agassiz, St-Augustin, and Ste-Anne but they were similar at Lethbridge and Kentville (Fig. 2B).
The WSC concentrations in sweet pearl millet varied from 100 to 145 g kg −1 DM, in keeping with  (Leblanc et al. 2012;Bouchard et al. 2011;Thivierge et al. 2015a). The WSC concentrations in the BMR sweet sorghum hybrids were greater than in sweet pearl millet and silage corn at all sites (Fig. 2B), and the BMR1 sweet sorghum hybrid had a greater WSC concentration (average across sites of 209 g kg −1 DM) than the BMR2 sweet sorghum hybrid (average of 189 g kg −1 DM). The sweet sorghum WSC concentrations reported in the literature vary from 129 to 281 g kg −1 DM (Thivierge et al. 2015a;dos Passos Bernardes et al. 2016;Saïed et al. 2017). In ruminants, the intake of forage with a high WSC concentration can cause digestive disorders due to the excessive production of volatile fatty acids in the rumen (Owens et al. 1998). On the other hand, a high WSC concentration can promote the fermentation of forage into silage and thus help preserve it (Davies et al. 1998). Sweet sorghum thus has the potential to make good silage.
Silage corn had a much greater starch concentration (110 to 260 g kg −1 DM) than sweet pearl millet and sweet sorghum hybrids (Fig. 2C). The starch concentration of sweet pearl millet hybrids varied from 9 to 62 g kg −1 DM, which was greater than the concentrations in the sorghum

Forage total digestible nutrient concentration
At all sites, sweet pearl millet had the lowest TDN concentrations (429 to 474 g kg −1 DM) (Fig. 2D). The two sweet sorghum hybrids did not differ from each other and had TDN concentrations that were greater (569 g kg −1 DM) than silage corn (533 g kg −1 DM) at Agassiz, similar (525 g kg −1 DM) to silage corn (528 g kg −1 DM) at Lethbridge and Ste-Anne, and lower (525 g kg −1 DM) than silage corn (590 g kg −1 DM) at St-Augustin. At Kentville, the TDN concentration of only the BMR2 sweet sorghum hybrid (513 g kg −1 DM) was lower than that of silage corn (543 g kg −1 DM). The average TDN concentration of silage corn across sites was 544 g kg −1 DM, which is lower than the average value of 688 g kg −1 DM published by the National Research Council (2001). Total digestible nutrients are a summation of four nutritive attributes, namely, non-fibre carbohydrates, crude protein, fatty acids, and aNDF, multiplied by their respective digestibility. It is therefore a very valuable index for comparing the nutritive value of different species. Data on the TDN concentration of sweet pearl millet and sweet sorghum are only available for climates much warmer than those in our study, and that information cannot be used as a valid source of comparison. However, our TDN results suggest that sweet sorghum could be an alternative to silage corn for feeding ruminants but does not support the use of sweet pearl millet (Fig. 2D).

Forage ensilability
At ensiling in 2016 at St-Augustin and Ste-Anne, forage DM concentration was less for sweet pearl millet and sweet sorghum than for silage corn (Table 5). The two sweet sorghum hybrids had a similar DM concentration at St-Augustin, whereas at St-Anne, the DM concentration was less for the BMR2 hybrid than for the BRM1 sweet sorghum. After 90 days of fermentation, however, all silages of sweet pearl millet, sweet sorghum, and silage corn produced in the minisilos were of excellent quality. The silage pH was less than 4.0 after 90 d of fermentation (Table 5) While the effects of forage species were statistically significant for concentrations of lactic, acetic, and propionic acids (Table 5), they were biologically negligible since all of these concentrations were within the range reported for excellent quality silages. Lactic acid is the main acid that lowers the pH during the forage fermentation process. To ensure good palatability of the silage, the lactic acid concentration should be four times greater than the acetic acid concentration (Lafrenière 2008), which should not exceed 20 g kg −1 DM (Leduc and Fournier 1998). Sweet pearl millet and sweet sorghum silages at both sites had three to six times more lactic acid than acetic acid and an acetic acid concentration less than 20 g kg −1 DM (Table 5). Moreover, the propionic and butyric acid concentrations were close to zero for both crop species, which are also indications of good quality and palatable silages. All fermentation parameters measured indicated that sweet pearl millet and sweet sorghum fermented as well as silage corn in mini-silos.

Agronomic and animal nutrition implications
Sweet pearl millet and sweet sorghum offer several agronomic advantages. They would allow a later harvest of winter cover crops or they could be used as a replacement where the planting season for silage corn was missed due to poor weather or other factors. Both species are also more efficient in utilizing fertilizer N as both species yielded as much if not more than silage corn with 65-75% of the amount of fertilizer N. This greater efficiency of N fertilizer utilization, also reported by Thivierge et al. (2015b), would reduce the cost of production, while decreasing the risks of N losses to the environment. Both species performed particularly well in terms of forage yield at Lethbridge, the site with the drier conditions. Hence, they offer interesting alternatives to silage corn in the context of the likely increase in the frequency of drought events in Canada with global warming.
Both species have a greater WSC concentration than silage corn. Forages with a high WSC concentration have the potential to increase N utilization in cows, fat content in milk, and milk yield. The low starch concentrations in sweet pearl millet and sweet sorghum, however, could negatively affect the silage utilization by ruminant animals. Sweet sorghum also has interesting characteristics in terms of digestibility and TDN concentration. The benefits of feeding those two species to dairy cows remain, however, to be demonstrated. Although both species have several advantages over silage corn, seed and hybrid availability remains a challenge.
Our results indicate that the potential of sweet pearl millet and sweet sorghum as silage  also be managed as multi-cut forage crops but this option was not evaluated in our study. More research is needed to evaluate those potential options.

CONCLUSIONS
The sweet pearl millet and sweet sorghum hybrids were adapted to all Canadian regions where silage corn is also adapted, and their forage DM yields were generally similar to or greater than those of silage corn in all regions, except in the Pacific Maritime ecozone. Whereas greater fibre concentrations in sweet sorghum suggests a lower voluntary DM intake in dairy cattle compared to silage corn, this is mitigated by greater digestibility (IVTD and NDFd) and a similar TDN concentration compared to silage corn. By contrast, sweet pearl millet had greater fibre concentration, and lower digestibility (IVTD and NDFd) and TDN concentration than silage corn.
Even though sweet pearl millet and sweet sorghum forages fermented as well as silage corn in laboratory silos, the DM concentrations of both sweet pearl millet (289 g kg −1 ) and sweet sorghum (245 g kg −1 ) were lower than that of silage corn (331 g kg −1 ), suggesting possible problems such as seepage in farm silos. As a result, solely sweet sorghum appeared as a possible alternative to non-BMR silage corn in four of five regions. Early maturing hybrids with a greater DM concentration at harvest are required to ensure good ensilability in most commercial silo types.        Table 5. Concentration of forage dry matter (DM) at ensiling, along with pH and concentrations of lactic, acetic, propionic, and butyric acids after 90 d of fermentation of silage corn, sweet pearl millet (SPM), and two hybrids (BMR1 and BMR2) of sweet sorghum (SS) grown at two sites in QC, Canada (St-Augustin and Ste-Anne) and harvested when the corn reached a dry matter (DM) concentration of around 350 g kg −1 in 2016.    215x279mm (300 x 300 DPI)