The big bluestem portion of the study was conducted in three environments using three sets of sward yield trials that were seeded in 1999. The trials were located at the University of Nebraska’s Eastern Nebraska Research, Education, and Extension Center (ENREEC) near Mead, NE (41˚09'16"N; 96˚25'30"W), the South Central Research and Extension Center near Clay Center, NE (40˚34'25"N; 98˚07'46"W), and the Northeast Research and Extension Center near Concord, NE (42˚23'27"N; 96˚57'01"W). The climate at all locations is continental with most of the annual precipitation occurring in the growing season. Weather data were obtained from permanent weather stations at each location [9] . Two trials were planted at each site. One trial had a July pre-heading harvest followed by a late autumn regrowth harvest while the other set of trials had a single post-heading harvest in August. The two-harvest management system would be used for producing high-quality hay for beef cattle while the single August harvest likely would be used for a biomass harvest for bioenergy.

The strains were planted in May 1999 at Mead, Clay Center, and Concord, NE. Soil on the Mead site was a Sharpsburg silty clay loam (fine, smectitic, mesic Typic Argiudoll), a very deep, moderately well-drained soil formed in loess. Soil on the Clay Center site was a Crete silt loam (fine, smectitic, mesic Pachic Argiustoll), a very deep, moderately well-drained soil formed in loess with 0% to 2% slope. Soil on the Concord site was on Alcester silty clay loam (fine-silty, mixed, mesic, Cumulic Haplustoll) soil, a very deep, well and moderately well-drained soil formed in silty colluvial-alluvial sediments with a 0% to 2% slope. The Mead site is 50 km west of Omaha, Clay Center is 155 km southwest of Mead, and Concord is 135 km north of Mead. Plots (1.5 × 3.1 m) were established at each location by seeding 430 pure live seed m⁻² in rows spaced 15 cm apart with a no-till plot drill (Hege Inc., Waldenburg, Germany). Each trial at a location was a randomized complete block with four replicates per strain or cultivar. Each site was uniformly cropped with soybean [Glycine max (L.)] for at least one year prior to establishment. Broadleaf and grassy weeds were controlled with 293 ml ha⁻¹ imazapic [Plateau^®; (±)-2-[4,5-dihydro-4-methyl-4-(1-methylethyl)-5-oxo-1H-imidazol-2-yl]-5-methyl-3-pyridinecarboxylic acid] plus 1.1 kg a.i. ha⁻¹ atrazine [6-chloro-N-ethyl-N’-(1-methylethyl)-1,3,5-triazine-2,4-diamine] applied as a pre-emergent immediately after planting. In the spring of each year, the standing dead material was removed either by burning or by harvesting the standing dead material to a 10 cm cutting height. Fertilizer was applied in late May of each year at 110 kg N ha⁻¹ as ammonium nitrate.

Big bluestem frequency of occurrence was measured in 2000 using the frequency grid [10] . To determine potential differences among big bluestem populations on tiller demographics and morphology, swards of each population at each location were sampled immediately prior to initiation of reproduction in the two-harvest set of plots in July 2002 from 2, 0.1 m² quadrats per replicate and hand-clipped to ground level. Morphology, tiller density, and tiller demographics were determined for each population. Tillers were morphologically classified to quantify MSC and tiller density for each population [11] . Aboveground biomass was determined the week following tiller sampling by cutting and weighing a 0.91-m-wide swath the length of each plot using a flail-type plot harvester (Carter Manufacturing, Brookston, IN) at a 10-cm cutting height. Harvested material was weighed fresh in the field. Subsamples were hand-collected before harvest from each plot, weighed, dried at 50˚C to a constant weight in a forced-air oven, and reweighed to determine dry matter (DM) content. All biomass is reported on a DM basis. Dried samples were ground to pass a 2-mm screen in a Wiley mill and a 1-mm screen in a cyclone mill (Thomas-Wiley Mill Co., Philadelphia, PA).

Switchgrass populations were planted at the University of Nebraska ENREEC near Mead, NE (41˚09'16"N; 96˚25'30"W) on a Sharpsburg silty clay loam soil. Plots (1.5 × 3.1 m) were seeded into a clean seedbed on 7 June 2002 at 430 pure live seed m⁻² in rows spaced 15 cm apart with a no-till plot drill as a randomized complete block with six replicates per population. Plots were separated by a 1.5 m wide alley seeded to switchgrass. The seeded plots were sprayed with a tank mixture of quinclorac (Paramount^®; 3,7-Dichloro-8-quinolinecarboxylic acid) at 0.56 kg·ha⁻¹ and atrazine at 1.1 kg a.i. ha⁻¹ pre-emergence after planting for establishment year weed control. No data was collected in the establishment year. The plots were burned on 11 April 2003 and fertilized in late April with 112 kg N ha⁻¹ as ammonium nitrate. Herbicides for post-establishment year weed control included a single May application of a tank mixture of metolachlor (Dual II^®; 2-chloro-N-(2-ethyl-6-methylphenyl)-N-(2-methoxy-1-methylethyl) acetamide at 1.4 kg a.i. ha⁻¹, atrazine at 1.1 kg a.i. ha⁻¹, and 2,4-D [(2,4-dichlorophenoxy)acetic acid] at 1.1 kg a.i. ha⁻¹. In 2004 and 2005, the plots were mowed in early spring to remove re-growth from the previous year. The same post-establishment herbicide treatment was used in 2004. No herbicides were applied in 2005. Stand frequencies were determined in 2002, 2003, and 2006 using a frequency grid [10] at spring green-up. Plots were harvested on 21 August, 9 August, and 8 August in 2003, 2004, and 2005, respectively. Switchgrass plots were harvested, sampled, and samples processed using the same procedures as for big bluestem.

Big bluestem samples were analyzed in triplicate for in vitro dry matter digestibility (IVDMD) with the ANKOM Rumen Fermenter (ANKOM Technology Corp., Fairport, NY) and in duplicate for neutral detergent fiber (NDF) and acid detergent fiber (ADF) with the ANKOM Fiber Analyzer (ANKOM Technology Corp., Fairport, NY) using the procedures described by Vogel et al. [12] . Crude protein concentration (CP: %N × 6.25) was determined by the LECO combustion method (Model FP 428 and FP 2000, LECO Corp., St. Joseph, MI) [13] [14] .

Switchgrass samples were scanned on a near-infrared reflectance spectrophotometer (NIRS; Model 6500, Silver Spring, MD) and a switchgrass NIRS calibration set was used to predict forage quality parameters. The scans fit within the range of this global calibration set. The calibration samples for the global calibration were chosen by cluster analysis of the reflectance data [15] . Calibration samples were analyzed in triplicate for IVDMD with the ANKOM Rumen Fermenter and in duplicate for NDF and ADF with the ANKOM Fiber Analyzer (ANKOM Technology Corp., Fairport, NY) using the procedures described by Vogel et al. [12] . The CP concentration was determined by the same LECO combustion method. Laboratory means were used to develop NIRS calibration equations by partial least squares [15] .

Data from big bluestem and switchgrass studies were analyzed using PROC GLM [16] . Strains were considered fixed effects while locations and years were considered random effects. Fisher’s protected least significant difference (LSD) was used to separate means of significant effects at α = 0.05.

The biomass yield of all the Pawnee-derived populations selected for seedling traits had significantly lower biomass yields than Pawnee in both the July and August harvests in both sets of experiments averaged over the three locations (Table 1). The Pawnee C3SW and Pawnee ST populations had greater regrowth yields than the parent Pawnee. The Pawnee ST and MT strains did not differ for biomass yield in the two-harvest set of trials for the July harvest or the regrowth harvest. The Pawnee MT strain did have 0.5 Mg ha⁻¹ greater biomass yields than the Pawnee ST population in the single August harvest set of trials averaged over the three locations (Table 1). The base population for the tiller number breeding work, Pawnee C3SW, was intermediate to the MT and ST strains for biomass yield. There were no differences among the strains in the single harvest set of trials harvested in August for forage quality traits (Table 1). In the two-harvest set of trials, the Pawnee ST strain had significantly greater IVDMD than the Pawnee MT strain for both the July and regrowth harvest after a killing frost. In this set of trials, the Pawnee ST strain tended to have lower NDF, ADF, and ADL than the Pawnee MT strain but the differences were not significant. Stand frequency

^†Fisher’s protected least significant difference (LSD). The F test for strains in the three-location analysis was significant at P < 0.05 for all evaluated yield and quality variables except where NS (not significant) is listed for the LSD.

data was not collected on these trials because excellent stands were obtained on all plots and the stand frequency was greater than the 50% threshold levels above which yield responses to increases in stand density do not occur [18] [19] . Tiller demographic analysis at the July harvest indicated that when evaluated in swards, the Pawnee ST and MT strains did not differ for maturity as measured by MSC, in total tiller density, or in vegetative tiller density (Table 2). Although the Pawnee MT population had numerically more of both classes of tillers than the Pawnee ST population, the differences were not significant (Table 2). Both populations bred for differences in seedling tiller number had significantly lower total tiller density and vegetative tiller density than Pawnee from which they were derived averaged over the three locations in 2002 (Table 2).

Breeding for seedling tiller number in the Pawnee C3SW population which changed seedling tiller numbers in the greenhouse and mature plant differences in tiller number in space-transplanted nurseries did not result in consistent differences in biomass yield or quality when evaluated in multi-location sward trials with two different harvest treatments. Combined with previous breeding work for seedling weight in the development of the Pawnee C3SW population, the combined breeding work for greater seedling weight and tiller traits resulted in a significant reduction in biomass yield when evaluated in swards. This demonstrates the importance of sward evaluation to determine the effectiveness of breeding for specific traits in big bluestem and their effect on other traits.

Biomass yields were not different among the three Pathfinder strains developed by breeding for seedling traits for the three years of the study (Table 3). Averaged over the three harvest years, the strains selected for seedling traits had lower biomass yield than Trailblazer but the differences were not consistent or significant. Trailblazer replaced Pathfinder as the recommended upland cultivar for use in pastures in the region at the time this trial was conducted. The Pathfinder ST and MT strains were lower in IVDMD than the Pathfinder C3SW base population from which they were derived (Table 3). The Pathfinder strains selected for seedling tiller number had greater ADF and ADL values than the Pathfinder

^†The LSD values were obtained from the error term for strains from the analysis of variance of the study which was a randomized complete block with six replicates.

C3SW base population which is consistent with their lower IVDMD. In 2002 and 2003, the Pathfinder MT strain had significantly greater stand frequencies than the Pathfinder ST and the base population, Pathfinder C3SW. However, the stand frequency for the strains in both 2002 and 2003 exceeded the threshold stand frequency of 50% above which Schmer et al. [17] and Vogel [18] did not find an association between stand frequency and biomass yield in switchgrass. The results from the switchgrass study are consistent with those from the big bluestem trials.

Das et al. [20] reported that selecting for multiple tillers would be an effective method to increase biomass yield in both lowland and upland switchgrass. These results contrast with the previously discussed report of Smart et al. [8] . Price and Casler [21] reported that in non-competitive space-planted plots, tillering traits in switchgrass can significantly affect biomass yield but that in competitive sward conditions, plant height and leaf area traits are more important determinants of biomass yield. The results of this study support the results of Price and Casler [21] .