Transgenic switchgrass plants (Panicum virgatum L. cv Alamo) expressing OsAT10 and QsuB under control of the maize Ubiquitin1 (pZmUbi-1) and sugarcane O-methyltransferase (pShOMT) promoter, respectively, were generated by Agrobacterium-mediated transformation as previously described (Li et al., 2018; Hao et al., 2021). The plants used in the present study were propagated clonally from the previously described plant lines. The OsAT10 expressing lines used were the two previously described lines designated FT2 and FT8 (Li et al., 2018). For the QsuB expressing switchgrass, lines QsuB10, QsuB13 and QsuB15 were selected from the eight lines previously described, as these all showed significant improvement of saccharification and decrease in lignin content (Hao et al., 2021). Untransformed Alamo plants that had been clonally propagated in the same way as the transgenic plants served as control.

Field experiments were conducted under federal regulation at the University of California Plant Sciences Field Research Facility, Davis, CA under USDA APHIS-BRS release permits (# BRS 16-327-103r). Plants were vegetatively propagated in a greenhouse and were transplanted under permit to a field site at UC Davis. The soil was a Yolo Silt Loam soil series classified as fine-silty, mixed, superactive, nonacid, thermic Mollic Xerofluents (USDA-Natural Resources Conservation Service, https://websoilsurvey.sc.egov.usda.gov). The plots were fully irrigated to match expected crop evapotranspiration at this site so that no water stress occurred. Urea was applied in April and after each harvest at a total level each year of 404, 252, and 269 kg N/ha. The experimental plot design was a randomized complete block design with four replicates for each transgenic line and controls (six cultivars total). Each replicate was planted to 3 rows per plot and 10 plants per row of the experimental plants, which were surrounded by non-transgenic border plants ( Figure 1 ). The rows were spaced 30.5 cm apart, and the plants within rows were also spaced 30.5 cm apart. This field trial was planted using greenhouse transplants on August 15, 2017, and the plants were allowed to establish that fall. Plants were trimmed when dormant in January of 2018. Since transplants were still establishing, no cultivar comparisons were made. Plots were harvested with a cutter-bar type forage harvester three times in each of three consecutive full growing seasons (2018–2020). For each harvest, the wet biomass of each plot was weighed, and a subsample of approximately 500 g was collected for determination of dry weight and compositional analysis. Subsamples were dried in a convection oven at 55°C to a constant weight, and dry matter percentage determined for yield determination and subsequent compositional analysis.

For cell wall analyses and saccharification assays, dry biomass samples were ground using a Model 4 Wiley Mill equipped with a 1-mm mesh (Thomas Scientific, Swedesboro, NJ) and milled into a fine powder using a Mixer Mill MM 400 (Retsch Inc., Newtown, PA) and stainless-steel balls. Ball-milled biomass (1 g) was sequentially extracted by sonication (20 min) with water (twice), 80% (v:v) ethanol (five times), acetone (once), chloroform-methanol (1:1, v/v, once) and acetone (once). Klason lignin was determined using the standard NREL biomass procedure (Sluiter et al., 2008). Monosaccharides in hydrolysates were measured by High Performance Anion-Exchange Chromatography with electrochemical detection as previously described (Scavuzzo-Duggan et al., 2021). Cell-wall-bound p-coumarate and ferulate were released from cell wall residues via mild alkaline hydrolysis as previously described (Eudes et al., 2012) and quantified using HPLC analysis (Rodriguez et al., 2019).

Pyrolysis coupled with gas chromatography mass spectrometry was used to determine lignin S/G ratio on cell wall residues, as previously described (Tian et al., 2021). Specifically, sub-samples of ≈ 0.1 mg were pyrolyzed at 650°C for 30 s using the pyroprobe 6200 (CDS Analytical, Oxford, PA) connected to a gas chromatography system (GC-2010 Plus, Shimadzu Scientific Instruments, Columbia, MD) using a Shimadzu SH-Rxi-5Sil MS column (30 m × 0.25 mm ID × 0.25 DF) attached to a mass spectrometer (GCMS-Q P2010) system operated using helium as carrier (1 ml min⁻¹). The chromatograph was operated at a split ratio of 10 and the program was set at 50°C for 1 min followed by ramping to 300°C at 20°C min⁻¹ and finally maintained at 300°C for 10 min. Released products of S and G origin were identified on the basis of their mass spectra using the NIST08 mass spectrum library and quantified from the chromatogram using the peak area.

For saccharification assays, 10 mg of extracted biomass was pretreated with hot water (1 h, 121 ˚C) followed by a 72-h enzymatic hydrolysis at 50 ˚C (1200 rpm) using 1% w/w Cellic CTec2 enzyme mixture (Novozymes, Denmark) in a total volume of 1 ml as previously described (Li et al., 2018). A 10-µl aliquot of the hydrolysates was collected at 0h, 24h, 48h, and 72h after the addition of enzyme for measurement of reducing sugars using the 3,5-dinitrosalicylic acid assay (Miller, 1959) and glucose solutions as standards.

Photosynthesis rate was measured in the field on July 11, 2018, using a Li-6400/XT Portable Photosynthesis System (LiCor, NE). Chamber CO₂ and temperature were held at 400 ppm and 36 ˚C, respectively, with light provided using a LiCor 6400-40 LCF leaf chamber. Measurements were done on two plants in each replicate plot for Alamo control, FT8 and QsuB13 (total of 8 leaves analyzed for each genotype). The light intensity was set initially at 2000 µmol m^-2 s^-1 and then stepped down to 1000, 300 and 0 µmol m^-2 s^-1. At each point, net CO₂ assimilation and transpiration was logged after a stable reading was observed.

Switchgrass samples were ground to 1 mm cyclone screen and forage quality parameters measured using wet chemistry and Near Infrared Spectroscopy (NIRS) methods from the 2018 full season sample set. First cut samples were measured utilizing wet chemistry methods (National Forage Testing Association, https://www.foragetesting.org/reference-methods) for Neutral Detergent Fiber (NDF) analysis. In vivo (in sacco) measurements utilizing nylon bags inserted into live cows measured disappearance of the NDF fraction in rumen fluid at selected time points. Wet chemistry was conducted at Cumberland analytical Lab (Carlisle, PA). Near-infrared reflectance spectra were collected using a scanning monochromator (Foss model 6500, Foss North America, Eden Prairie, MN). Robust calibrations for prediction of crude protein (CP), NDF, Acid Detergent Fiber (ADF), lignin (Acid Detergent Lignin), digestibility estimates, and forage quality parameters were generated using the 2018 Grass Hay calibration developed by the NIRS Forage and Feed Testing Consortium (Hillsboro, WI). Data were statistically analyzed using SAS to test for the influence of entry (variety), cut, and the interaction between entry and cut.

Plants were harvested with a forage chopper three times each growing season. The dry biomass yields are shown in Figure 2 . The FT8 line showed yield no different from the control, whereas the FT2 line had a lower yield in all years with an average of 16% decrease. These results were in line with the growth deficiency seen for FT2 in the studies conducted in a controlled environment (Li et al., 2018). For the QsuB10 line, a significantly increased yield was observed in the second year, and for the QsuB13 line, a significant increase was seen in the first and second year. Overall, for all three years, the increase was 11% and 16% for QsuB10 and QsuB13, respectively. For QsuB15, a significant increase was not observed. A difference in growth of the QsuB lines had not been observed in the greenhouse study, but the biomass yield was not quantified in that study (Hao et al., 2021). A priori, it had not been expected that QsuB expression would lead to a higher yield, but we recently reported that expression of QsuB under control of a constitutive promoter in sorghum led to an increase in biomass yield of 8-29% in all the five lines tested in the field (Tian et al., 2022).

Overall, the yields were higher in 2019, the second full growth season, than in 2018, as expected with the more established plants. In 2020 a slight decrease was observed, which is not unexpected with switchgrass where yields eventually decrease after multiple seasons. However, variations in weather conditions and other environmental factors can obviously also affect yield in individual years. We did not observe any signs of significant pest or disease impact in any of the growing seasons.

Cell wall monosaccharide compositional analysis was performed on the first and second harvest from the first year. Cell wall hydrolysates from the FT and QsuB lines did not reveal any trends or significant differences from the control ( Figure 3 ). Likewise, lignin content ( Figure 3 ) and wall-bound phenolic content ( Figure 4 ) in FT and QsuB lines were not significantly different from the control. This was surprising since the FT lines had shown a significant decrease in ferulic acid and increase in p-coumaric acid esters in the greenhouse experiments, and since the QsuB lines had shown significant decrease in lignin in the greenhouse studies (Hao et al., 2021). Analysis of lignin composition by pyrolysis coupled with gas chromatography mass spectrometry showed a minor but significant increase in S/G ratio in the QsuB lines, consistent with previous results obtained from QsuB Arabidopsis and poplar ( Figure 5 ).

The FT2 and QsuB13 samples from the first cut in 2018 were selected for standard wet chemistry forage composition analysis that did not reveal significant compositional differences ( Table 1 ). A small but significant improvement in neutral detergent fiber digestibility at 240 hours in situ was observed in the FT2 line ( Table 1 ). Samples from all genotypes and cuts were analyzed with NIRS ( Table 2 ). There were numerically small but significant differences between entries using NIRS, but the differences due to cut (harvest date) were more important ( Table 2 ). There were few interactions between entry and harvest date in forage quality parameters. These quality measurements are in line with harvested switchgrass reported elsewhere (Guretzky et al., 2011) and do not indicate a significant effect of the gene insertions on forage quality of switchgrass.

Biomass from the first cut in 2018 was subjected to saccharification ( Figure 6 ). At the endpoint of 72 h, the QsuB lines showed a 3.6 to 7.6% increase (average 5.0%) in sugar release. The increased sugar yield was small but statistically significant for the QsuB10 and QsuB15 lines (ANOVA and Dunnett test). The FT2 line showed a 5.0% increase in sugar yield that was statistically significant (ANOVA and Dunnett test). For comparison, the same QsuB lines grown in the greenhouse showed an 18-24% increase in sugar release (Hao et al., 2021), and the FT lines grown in the greenhouse showed a 30% increase in senesced tissues (Li et al., 2018). Thus, the improvement of saccharification in the engineered lines was much smaller than expected from the greenhouse studies.

As shown in Figure 2 , the switchgrass QsuB lines exhibited increased yield in the field, similar to what we have observed in sorghum expressing the same transgene. The reason for the increased yield is not understood, but we have observed in Arabidopsis that expression of QsuB is associated with drought tolerance and faster stomatal closure (Yan et al., 2018). We therefore investigated photosynthetic parameters and water use efficiency in the field-grown FT8 and QsuB13 lines. However, under the conditions of this experiment, no difference in CO₂ assimilation or water use efficiency in response to photosynthetically active radiation (PAR) was observed ( Figure 7 ). Variable fluorescence also did not show any differences between the genotypes. Thus, even though the QsuB plants showed more growth, this did not translate to a measurable difference in photosynthesis on a leaf area basis.