Biomass yields, cytogenetics, fertility, and compositional analyses of novel bioenergy grass hybrids (Tripidium spp.)

High biomass yields have been documented for Tripidium spp. (Erianthus spp., Saccharum spp.), but targeted breeding for bioenergy applications has been limited. Advanced, interspecific hybrids between Tripidium ravennae and T. arundinaceum were planted in replicated field plots in 2016. Comparative feedstock evaluations examined biomass yields, cytogenetics, plant fertility, and compositional analyses relative to Miscanthus × giganteus. Dry biomass yields varied as a function of year and accession and increased each year ranging from 3.4 to 10.6, 8.6 to 37.3, and 23.7 to 60.6 Mg/ha for Tripidium hybrids compared to 2.3, 16.2 and 27.9 Mg/ha for M. × giganteus in 2016, 2017, and 2018, respectively. Cytology and cytometry confirmed that Tripidium hybrids were tetraploid with 2n = 4x = 40 (2C genome size = 5.06 pg) and intermediate between T. ravennae with 2n = 2x = 20 (2C genome size = 2.55 pg) and T. arundinaceum with 2n = 6x = 60 (2C genome size = 7.61 pg). Plant fertility characteristics varied considerably with some accessions producing no viable seeds or fewer than that observed for M. × giganteus. Accessions varied significantly for flowering culm number and height and dates of peak anthesis ranging from 14 September to 2 October. Variations in yield and compositional analyses contributed to variations in theoretical ethanol yields ranging from 10,181 to 27,546 L/ha for Tripidium accessions compared to 13,095 L/ha for M. × giganteus. Relative feed value (RFV) indices for winter‐harvested Tripidium accessions varied from 52.8 to 60.0 compared to M. × giganteus with 45.4. RFV for summer‐harvested Tripidium accessions varied from 71.6 to 80.5 compared to M. × giganteus with 61.0. These initial findings for Tripidium hybrids, including high biomass yields, cold hardiness, and desirable traits for multiple markets (e.g., forage, bioenergy, bioproducts), are promising and warrant further development of Tripidium as a temperate bioenergy feedstock.

Less work has been done on other genera in the subtribe Saccharinae, but some of these taxa may have considerable potential as biomass crops. The taxonomic relationships within the Saccharinae remain incomplete despite the greater resolution available with molecular systematics tools (Amalraj & Balasundaram, 2006;Bonnett & Henry, 2011;Irvine, 1999;Whalen, 1991). Though previously recognized as Saccharum (Linnaeus, 1774) and Erianthus (de Beauvois, 1812), Valdés and Scholz (2006) transferred the Old World Erianthus into Tripidium based on molecular and morphological features.
Bioenergy breeding programs at North Carolina State University have focused on developing Tripidium as an alternative bioenergy feedstock. New, advanced (F 2 and F 3 ) interspecific hybrids of T. arundinaceum and T. ravennae have demonstrated cold hardiness and overwintering survival in USDA Zone 6b (T. Ranney, personal observations, 2015) with potentially high biomass yields. The objectives of this study were to further evaluate biomass yields, cytogenetics, fertility, and chemical composition for these advanced hybrids to help advance the development of Tripidium as a biomass/bioenergy crop.

| Field establishment
Field trials were established at Highland Creek Nursery, Hoopers Creek, NC, (35°25′48ʺN, 82°28′35ʺW). Based on preliminary biomass yields, 17 F 3 and 1 F 2 Tripidium hybrids (T. arundinaceum × T. ravennae) and 1 M. × giganteus clone (Illinois) were selected for this study. In March 2016, all selections were propagated from divisions and grown in 2.9 L containers in a 100% pine bark media supplemented with 1.04 kg/m 3 dolomitic lime and 0.74 kg/m 3 micronutrients (Micromax; The Scotts Co.). Field trials were planted in May 2016 with plots arranged in a randomized complete block design with three replicates. Each plot was 2 × 5 m and contained six plants in each of three rows, with plants spaced 1 m apart. Plots were separated by 4 m wide alleys and received drip irrigation, as needed, for the first 3 months of establishment following planting. Plots were treated with S-metolachlor and atrazine (4.7 L/ha, Bicep II Magnum; Syngenta) 2 weeks postplanting and the spot was treated as necessary with glyphosate (12 ml/L; Roundup QuikPRO; Monsanto) or paraquat dichloride (30 ml/L, Gramoxone SL 2.0; Syngenta) to control weeds. Soils were a Comus (colvard) fine sandy loam, coarse-loamy, mixed, active, nonacid, mesic Typic Udifluvents. Soil samples were collected from plots in 2018 and analyzed by the North Carolina Department of Agriculture and Consumer Services Agronomic Division (NCDA&CS) soil testing laboratory (Plank, 1992). Soil pH was 6.2, humic matter was 0.41%, cation exchange capacity (CEC) was 5.1 meq/100 cm 3 , Ca was 57% of CEC, Mg was 24% of CEC, and indexes for P and K were 71 and 31, respectively, and all within recommended levels based on NCDA&CS recommendations for perennial grasses. Yearly temperature and precipitation data were obtained from KAVL weather station at Asheville Regional Airport (5.6 km west of plots) and compiled by the North Carolina Climate Retrieval and Observations Network of the Southeast database (Table 1; NCCRONOS, 2019). Data for the 30 year temperature and precipitation were obtained from data logged at the KAVL weather station and archived with the National Oceanic and Atmospheric Administration (Table 1; NOAA, 2010).

| Biomass yield and morphological characteristics
Annual biomass yields were evaluated following plant senescence and harvest was completed by late December or early January. Six plants from the center row of each plot were harvested approximately 10 cm above the ground level and weighed fresh. A 10-15 L sample (~1 kg) of chopped tissue was collected and weighed fresh. Samples were then oven dried for 7 days at 80°C and reweighed to establish dry harvest weights. Data were analyzed using a repeated measures mixed linear model with SAS PROC GLIMMIX (SAS version 9.4; SAS Institute). Replicate blocks were treated as random effects while accessions and harvest years were treated as fixed effects. Means were separated using Scheffe's multiple comparison  test. Plots were observed regularly to identify the flowering period and duration for all accessions in this study. Peak anthesis was determined when ≥50% of inflorescence culms for a given plot were at or past peak receptivity and pollen dehiscence. Inflorescence culm number and height were determined from three randomly selected plants within each plot following peak anthesis. Total culm counts were determined for each plot by counting the number of culms from three randomly selected plants prior to harvest. Biomass yields as a function of plant morphological characters for Tripidium accessions were analyzed by multiple regression statistical analysis with SAS PROC REG (SAS version 9.4; SAS Institute).

| Cytogenetics
Genome sizes were determined for all taxa in the field study and the hybrid parental species (T. ravennae and T. arundinaceum) using flow cytometry (Doležel et al., 1998). Interior leaf sheath tissues were sampled from nonflowering culms. Three to six subsamples were run per accession using approximately ≈0.5 cm 2 of sample tissue. Tripidium samples were processed together with ≈0.25 cm 2 of an internal standard, Pisum sativum L. "Ctirad" with a known genome size of 2C DNA content = 8.75 pg (Greilhuber, Temsch, & Loureiro, 2007). Miscanthus × giganteus samples were processed together with ≈0.25 cm 2 of an internal standard, Magnolia virginiana "Jim Wilson" with a known genome size of 2C DNA content = 3.75 pg (Parris, Ranney, Knap, & Baird, 2010). Tissues were finely chopped with a razor blade in 0.4 ml of nuclear extraction buffer (CyStain ® PI Absolute P Nuclei Extraction Buffer; Sysmex Partec). Tissue extracts were incubated with propidium iodide (PI) stain buffer at room temperature for 5 min prior to filtration (50-mm nylon) followed by a 90 min (±30 min) incubation in the dark at 4°C. Nuclei were processed on a flow cytometer (Partec PA-II; Sysmex Partec) with three samples per plant accession and a minimum of 3,000 nuclei were counted for each individual assay. Holoploid genome sizes (2C) were calculated as a ratio of the mean fluorescence of the sample to the standard multiplied by the genome size of the standard. Chromosome numbers were determined for the interspecific hybrid (H2012-260-022) and the parental species T. ravennae  to correlate genome sizes with ploidy. Root tips were excised from actively growing containerized plant materials in mid-July. Approximately 5 cm of the tissue was excised and fixed following Tlaskal (1979). Briefly, roots were incubated for 3 hr at 23°C in a solution containing 0.248 mM cycloheximide and 2 mM 8-hydroxyquinoline and, then moved to 4°C for an additional 3 hr. Roots were rinsed in cold distilled water before overnight fixation at 25°C in ~3 ml of Carnoy's solution (six parts 95% ethanol: three parts chloroform: one part glacial acetic acid). Fixed roots were rinsed in several exchanges of ethanol before storage at 4°C in a 70% ethanol solution. Cell wall hydrolysis was facilitated in a 3:1 95% ethanol:12 M HCl solution for ~15 min before chromosome staining in a modified carbol fuchsin solution (Kao, 1975;Singh, 2003). Root tips were placed onto a microscopic slide with excess carbol fuchsin solution and chromosomes were expressed from cells under a cover slip. Chromosomes were viewed and counted at 1,000× magnification.

| Plant fertility
Multiple characteristics of flowering and plant fertility were evaluated. Plant inflorescence numbers were assessed at the onset of flowering. Male fertility was assessed by acetocarmine staining of pollen (Singh, 2003). Three dehiscent inflorescences were randomly sampled from field plots in the morning and transported to the laboratory. Ten freshly dehiscent, randomly selected, anthers were placed on a microscope slide with one drop of a 1% acetocarmine solution. Pollen was expressed from anthers with forceps, anthers were then removed, and a coverslip added. Preparations were sealed with valap (1 petroleum jelly:1 lanolin:1 paraffin by weight) and incubated at room temperature (23°C) at least 30 min prior to assessment. Percent pollen viability was calculated as the number of stained, viable pollen grains divided by the total grains counted and multiplied by 100.
Female fertility was assessed by X-ray photography. Three inflorescences were collected from field plots in November 2018. Each inflorescence was sampled at the upper, middle and lower regions of the inflorescence which were trimmed to fit within image capture window (10-15 cm) of the X-ray image system (Faxitron MX-20; Faxitron Bioptics) and mounted between two pieces of 0.5 mm thick polystyrene plastic. Images were processed individually using ImageJ software (Abràmoff, Magalhães, & Ram, 2004). Images were despeckled as necessary and processed to identify maxima by changing the noise tolerance settings. Image processing calibration was confirmed by manual counts. The percentage of seed viability was calculated as a ratio of filled seeds divided by the total number of fertile florets counted and multiplied by 100.
In January 2019, all inflorescence material collected during the X-ray image capture process was surface sown into 1.3 L pots containing a two-part peat to one-part coarse vermiculite media (v/v). Pots were placed in a greenhouse under intermittent mist (10 s every 15 min) for 6 weeks. Germination was monitored weekly and seedlings were removed upon observation. Additional seed germination was determined following 30 days of cold stratification (4°C). After stratification, pots were returned to the greenhouse with intermittent mist and continued observation for an additional 4 weeks. Percent germination was calculated as the total number of seedlings observed (with or without stratification) divided by the number of seeds counted in image capture process and multiplied by 100. Overall female fertility of each accession was calculated as the percent seed set multiplied by the percent germinated seedlings divided by 100. Fertility data were subject to analysis of variance with SAS PROC GLM (SAS version 9.4; SAS Institute) and means were separated by Fisher's least significant difference. The correlation between pollen viability and overall female fertility and between date of peak flowering and seed set, germination, or overall female fertility for the Tripidium accessions was tested using SAS PROC REG (SAS version 9.4; SAS Institute).

| Compositional analyses
Chemical composition of plant samples (soluble sugars, lignin, cellulose, and hemicellulose) was evaluated using a modified National Renewable Energy Laboratory procedure (Whitfield, Chinn, & Veal, 2016). Approximately 3 L (~1 kg dry weight) of postharvest biomass was collected for continued evaluation and dried at 45°C for 7 days. Samples were ground in a Wiley Mill to pass through a 2 mm mesh screen and stored in sealed polyethylene bags until further processing. Extractable carbohydrates from feedstock materials were quantified by HPLC following Soxhlet extraction, and cellulose and hemicellulose contents were determined by a modified two-stage sulfuric acid hydrolysis protocol following Whitfield et al. (2016). Theoretical maximum ethanol yields were calculated on a dry harvested biomass basis following Kim and Day (2011). A stoichiometric basis of carbohydrate consumption and utilization by Saccharomyces cerevisiae of 51.1% was used in the calculated theoretical ethanol conversion and considered use of both cellulose (C6) and hemicellulose components (C5; Kim & Day, 2011): The utility of the biomass material was further characterized as a forage component for dairy livestock. Green (mid-summer) and dry (winter) harvested material from each replicate was analyzed by North Carolina Department of Agriculture and Consumer Services, Food and Drug Protection Division Laboratory for nutritive qualities. Dry matter (partial and total) were determined following methods of the National Forage Testing Association (Undersander, Mertens, & Thiex, 1993). Samples were ground and prepared in accordance with protocol 922.02 within methods of the Association of Official Analytical Chemists (AOAC; Mertens, 2005). Nitrogen was evaluated following methods of the AOAC (990.03) for total nitrogen or crude protein (AOAC Authors, 2006b). Acid and neutral detergent fiber contents were assessed on the Ankom Technologies A2000 in accordance with AOAC methods 973.18 utilizing the Ankom methods 12 and 13 respectively (Ankom Technology, 2017). Crude fat contents were analyzed by high temperature solvent extraction in accordance with the AOAC standard procedure Am 5-04 (Ankom Technology, 2012). Total ash contents were evaluated by AOAC methods 942.05 (AOAC Authors, 2006a). Data were subject to analysis of variance with SAS PROC GLM (SAS version 9.4; SAS Institute), with cultivar treated as a fixed effect, and means were separated by Fisher's least significant difference.

| RESULTS
Dry biomass yields varied as a function of year, accession, and their interaction (p ≤ .001; Figure 1). At the first harvest (2016), dry matter yields for Tripidium accessions ranged from 3.4 to 10.6 Mg/ha compared to M. × giganteus with 2.3 Mg/ha. Second year dry biomass yields increased for all Tripidium accessions and ranged from 8.6 to 37.3 Mg/ha compared to M. × giganteus with 16.2 Mg/ha. Yields continued to increase for the third year for most accessions and ranged from 23.7 to 60.6 dry Mg/ha for Tripidium compared to 27.9 dry Mg/ha for M. × giganteus. The F 2 Tripidium accession, H2012-260-022, consistently had one of the highest dry yields with 10.6, 37.3, and 60.7 Mg/ha in the first, second, and third seasons, respectively. Yields for Tripidium H2012-260-022 exceed M. × giganteus in each year by 8.3, 21.1, and 44.5 Mg/ha (Figure 1).

F I G U R E 1 Dry biomass yields for
There were significant differences (p ≤ .05) among taxa for date of peak anthesis, flowering culm number, total culm number, and flowering culm height (Table 3). Peak anthesis date varied considerably from 9/14 to 10/05 for Tripidium hybrids, which generally flowered later than M. × giganteus with peak flowering on 9/14. Inflorescence culm number, total culm number, and flowering culm height varied among the Tripidium hybrids, which typically had fewer flowering culms, overlapping total culm numbers, and similar or higher flowering culm height than M. × giganteus. Multiple regression analysis of the data for Tripidium accessions, identified total culm number as the only significant predictor (r = .50; p ≤ .001) of biomass yield among these four traits. Furthermore, there was no significant correlation (p ≤ .05) between date of peak flowering and seed set, germination, or overall female fertility for the Tripidium accessions suggesting that variation in female fertility was not a function of length of seed maturation time.
Compositional analyses for cellulose and hemicellulose varied slightly among Tripidium accessions. Cellulose  was significantly lower in Tripidium accessions than in M. × giganteus and hemicellulose was typically similar to M. × giganteus when compared on a unit weight basis (Table 4). However, the higher dry matter yields achieved by several hybrid accessions would result in greater cellulose production per hectare and increase the supply and value of the feedstock. Free glucose and fructose varied significantly (p ≤ .05) among Tripidium accessions and many were significantly (p ≤ .05) higher than M. × giganteus.
Theoretical ethanol yields calculated based on cellulose, hemicellulose, glucose, and fructose ranged from 10,181 to 27,546 L/ha for Tripidium accessions compared to 13,095 L/ha for M. × giganteus (Table 4). Forage analysis showed Tripidium accessions typically varied among each other and compared to M. × giganteus depending on the specific variable (Tables 5 and 6). Crude protein contents ranged from 111 to 138 g/kg for summerharvested biomass and 53-66 g/kg for winter-harvested Tripidium accessions. The protein contents were close to twice that of M. × giganteus with 75 g/kg for summer and 27 g/kg for winter-harvested materials. Acid detergent fiber, favorable for rumen digestibility, was lower for Tripidium accessions, 375-416 g/kg (summer) and 482-542 g/kg (winter), compared to M. × giganteus with 501 g/kg (summer) and 592 g/kg (winter). Neutral detergent fiber was typically higher in M. × giganteus in both summer (760 g/kg) and winter (878 g/kg) than in most Tripidium accessions (691-750 g/kg in summer and 794-824 g/kg in winter). Nonfiber carbohydrates varied from 103 to 154 g/kg in summer and 89 to 120 g/kg in winter for Tripidium accessions with M. × giganteus falling within the same range at 150 g/kg (summer) and 99 g/kg (winter). Total digestible nutrient content varied from 568 to 685 g/kg in summer and 528 to 576 g/kg in winter-harvested Tripidium accessions which were greater than M. × giganteus with 568 g/kg (summer) and 487 g/kg (winter). Net energy of lactation was also lower for M.  80.5 in summer versus 53 to 60 in winter, which was markedly higher than for M. × giganteus at 61 and 45 respectively.

| DISCUSSION
Prior reports have identified T. arundinaceum as a potential biomass and bioenergy feedstock (Dao et al., 2013;Palmer et al., 2014;Wang et al., 2019;Zhang et al., 2013), though cold hardiness is limited to USDA Zone 7b. The new interspecific Tripidium hybrids examined in this study demonstrated high biomass yields ranging from 24 to 61 dry Mg/ha for the third growing season within the temperate climes of Western North Carolina, USDA Zone 6b/7a (Figure 1). These biomass yields are similar to those for S. spontaneum and other Saccharum hybrids (energy canes) that have yielded between 13 and 67 Mg/ha dry weight in the subtropical climes of the southern United States and Caribbean islands (Alexander, 1985;Bischoff et al., 2008;Matsuoka et al., 2014), but these Saccharum crops are typically limited to USDA Zones 8 and warmer. Yields of many of the interspecific Tripidium hybrids also exceed that of M. × giganteus, a hybrid often found to have one of the highest biomass yields of more temperate grasses Lewandowski et al., 2016). In the third growing season, Tripidium H2012-260-022 had more than twice the biomass yield of M. × giganteus. Considering that one of the Tripidium hybrid's parents, T. ravennae, is cold hardy to USDA Zone 5b, it is likely that further breeding, selection, and evaluation could identify advanced, high-biomass Tripidium hybrids suitable for colder regions than North Carolina. Plant fertility can be important when selecting breeding lines for further crop improvement and/or to avoid reseeding of crops and potential invasiveness. Although all the Tripidium hybrids were confirmed to be tetraploids (isoploids), these types of interspecific, interploid hybrids may have reduced or variable fertility. Accession H2014-228-003 did not flower under our environmental conditions. If this trait is stable under other years/ environments, it may be desirable to prevent reseeding or prevent gene flow in yet-to-be developed transgenic plants. Similarly, we were unable to document any female fertility in Tripidium accessions H2014-230-001, H2014-230-006, and H2014-230-009 that could be valuable in preventing them from naturalizing. There was no correlation between pollen viability and female fertility, indicating that variations in fertility were probably more complex than just failure of homoeologous chromosomes to pair in meiosis. Surprisingly, we did recover seedlings from open-pollinated M. × giganteus. Numerous prior studies have reported no detectable fertility in M. × giganteus, which has been attributed to a combination of it being a triploid cytotype and an interspecific hybrid (Słomka et al., 2012;Yu, Kim, Rayburn, Widholm, & Juvik, 2010). We completed flow cytometry on 30 seedlings from open pollinated M. × giganteus and found that genome sizes varied from 6.81 to 11.48 pg indicating they are aneuploids ranging from ~2.7 to 4.9x (data not shown). The evaluation and development of Tripidium as a biomass/bioenergy/forage crop is still in early stages. However, these initial findings demonstrating high biomass yields, high potential ethanol yields, temperate USDA Zone 6b/7a cold hardiness, and multipurpose applications including forage and bioenergy are promising. Expanded regional evaluation with larger plot sizes would be desirable to better determine cold hardiness limits, performance, biomass yields, and plant fertility in diverse climates. The potential for continued breeding to enhance desirable traits, including cold hardiness and biomass yields, is substantial. The Tripidium accessions in this study were derived from a relatively narrow germplasm base that originated from a single F 1 hybrid plant. Incorporating broader diversity of desirable traits into new breeding lines could greatly expand these opportunities. Due to their considerable high biomass yields and cold hardiness, these hybrids can serve to broaden the diversity of competitive biomass crops. In addition, these hybrids provide new opportunities to be used as foundational crops for bioengineering of other secondary metabolites and coproducts.