The experiment was carried out at Suwanvajokkasikit Research Station, Pak Chong, Nakhonratchasima, Thailand situated at 14˚38'N, 101˚18'E and 388 m above sea level in April 2008-May 2009. The soil was sandy clay loam with moderate fertility and pH of 6.5 (in water). Chemical composition of the top 0 - 15 cm of soil was 13 - 14 ppm of available phosphorus (P), 78 - 160 ppm of available potassium (K) and 1.21% - 1.74% of organic matter, using the in-house method based on AOAC [9] and OMAF [10] methods were determined by the laboratory of the Department of Agriculture, Thailand. Daily rainfall was recorded at the Pak Chong Meteorological Station.

A randomized block design was employed with 8 napiergrass cultivars namely; Bana (BN), Taiwan A148 (TW), Common (CM), Wruk wona (WW), Tifton (TT) and Kampheng San (KS), representing tall types, and Dwarf (DW) and Muaklek (ML), representing short type and four replications.

The area was ploughed and cultivated to produce a firm fine nursery bed before planting on 10 April 2008. Rooted tillers of all cultivars were grown in plastic trays in the nursery for 4 weeks prior to transplanting into the experimental area. The transplanted grasses were grown at 75 × 75 cm spacing. Individual plots consisted of five 6.0 m rows spaced 0.75 m apart, giving 40 plants per plot. An initial basal fertilizer dressing of 15-5-20 (N-P-K, %) at the rate of 160 kg/ha was applied on 10 April 2008. Nitrogen as urea fertilizer was applied after each harvest to provide nitrogen at 375 kg/ha/yr. All plots were cut on 29 May 2008 to commence the observation period and the material discarded. Subsequent harvests were made on 29 August 2008, 29 November 2008, 28 February 2009 and 29 May 2009. In each plot, plant height was measured from an area of 1.50 ´ 2.25 m (6 plants) at the day of cutting. Tiller density was also determined by counting the plant in an area of 1.5 ´ 3.0 m in each plot at the same day of cutting. Forage on a central area of 2.25 × 3.00 m (12 plants) in each plot was cut at 15 cm above ground level and fresh weight recorded. A 500 g sub-sample was taken, separated into leaf and stem components, dried at 60˚C for 72 hours and dry weight was recorded. Samples collected in August 2008 and February 2009 were ground to pass a 1 mm screen and analyzed for, carbon (C), hydrogen (H), oxygen (O), nitrogen (N) and sulfur (S) by using an elemental analyzer (CHNS analyzer, LECO, MI, USA). Cellulose, hemicelluloses and lignin concentrations were determined by the detergent method [11] and heating value by using a bomb calorimeter [9]. Samples from one replication from the August 2008 harvest were analyzed for glucose content. About 100 g dry matter (DM) of sample was cut to an average length of 2.5 - 3.5 cm, and soaked in 125 mL of water for 10 min. Steam explosion was carried out in a 2.5 litre steam explosion apparatus (Nitto Koatsu Co. Ltd, Japan) under pressure of 14 kg/cm² at 200˚C for 5 min. The exploded material was then filtered into solid and liquid components, and kept in sealed plastic bags at 4˚C until glucose analysis. The objective of pretreatment was to remove lignin from cellulose and hemicelluloses prior to glucose analysis. The glucose content was determined by the HPLC system (Shimadzu, Kyoto, Japan) at 80˚C in an Aminex HPX-87P column (Bio-Rad, Sunnyvale, CA, USA), attached with a refractive index detector, with deionized water at a flow rate of 0.6 mL/min. Theoretical ethanol yield was calculated by using the Theoretical Ethanol Yield Calculator software provided by US Department of Energy [12].

All data except for glucose content and ethanol yield were tested for significance at P < 0.05 level by the new Duncan’s Multiple Range Test.

Figure 1 showed a period of low rainfall between November and February and a wet season from March to October with distinct peaks in April-May and September. When the grasses were establishing in April 2008, soil was quite moist owing to moderate rainfall in this month plus 177 mm of irrigation at planting. Sprinkler irrigation