Materials and methods

Plants of four grass species (Kentucky bluegrass, smooth brome, western wheatgrass and green needle- grass) were subjected to two treatments (control and

drought) as a 4 9 2 factorial experiment of species and drought treatment. The experiment was con- ducted in 2010 and 2011. The 2010 experiment, in which the species-treatment combination was repli- cated six times (six pots for each species-treatment combination), was used as a test to help establish the number of replications needed to detect the potential statistical differences in biomass allocation, given the observed variability in this experimental system (see later for more about statistical analysis). Based on this initial test, the number of replications for each species -treatment combination in the 2011 experiment was increased to 12. In the remainder of this paper, we mostly discuss the results obtained in 2011, but will mention the 2010 results when necessary. In our 2011 experiment, ninety-six pots, each 907 cm 2 9 241 cm high (bottom area 790 cm 2) in size, were filled with sieved native prairie soil and planted on 10 June 2011 with about 150 seeds per pot. The pots were arranged randomly within the greenhouse and received about 70% of the PAR at mid-day of a typical July clear day with the use of a nylon cloth to prevent excessive temperature increase inside the greenhouse. The pots were rotated once every 2 weeks to minimize the effect of variation in environmental factors of the greenhouse interior. The greenhouse was ventilated with six roof vents open on non-rainy days. In addition, a ventilation fan was triggered when the greenhouse temperature was higher than 30C. The daily average temperature inside the greenhouse was about 1-2C higher than that outside. During establishment, plants were misted with water 3-4 times a week. By June 17, most seeds had started to germinate. On June 22, we planted more western wheatgrass seeds, which started to germinate in a week or so. By June 28, smooth brome was 10 cm tall, green needle- grass was 7-8 cm tall, Kentucky bluegrass 5 cm tall and western wheatgrass <5 cm tall. Plants were thinned to about 30 stems per pot. In early to mid- July, the plants were thinned 2-3 times so that each pot had about 20 plants. The bulk density of the soil used in this experiment (0984 g cm 3) and the average 033 bar and 15 bar volumetric water contents (022 and 032, respec- tively) of prairie soil with the same bulk density (B. Patton, personal communication) were used to estimate the amount of water given to the plants. Assuming daily evapotranspiration of 2 mm from an average pot with plants (Dong et al., 1997, 2010), 18 mL water could evaporate from each pot daily. So, we decided to give 150 mL water to each of the control group pots twice a week and 150 mL water to each of the drought group pots once every 7-10 days, depending on drought stress. The objective was to

allow the drought group plants to be subjected to severe water stress in each watering cycle, but at the same time ensure that they would regain good water condition after water was given. Drought treatment started on 7 July 2011, when plants were in the 2-4- leaf stage. On July 15, all plants were clipped to 5 cm stubble height. The plants were clipped for the second time on August 11 (for the control group plants) and August 16 (for the drought group plants). The removed shoot biomass was dried at 65C in a forced-air oven for 48 h to measure dry mass, and all the removed shoot mass was added to the final harvested shoot mass to obtain the total shoot mass for each pot during the entire experimental period. Leaf water potentials were measured using a pres- sure chamber (PMS-1000, PMS Instrument Company, Corvallis, OR, USA) with its tank filled with high-pres- sure pure nitrogen gas. Because plants were watered regularly, leaf water potentials (predawn values) were only measured on selected days before or after water- ing. For the plants in the control group, leaf water potentials were measured on August 9 (1 d after watering) and August 11 (before watering). Plants of the drought group were measured on August 9 (1 d after watering), August 15 (before watering), as well as August 16 (1 d after watering). The measurement on August 16 for the drought group was to confirm that, despite severe drought stress, plants could restore good water status upon receiving water (Table 1). During the course of the experiment, the plants in the drought group experienced seven cycles of water stress. (The dry mass weights of all leaves used for measuring water potentials were also recorded.) We watered the control group twice a week, instead of once daily as in many other drought studies. While in the field it is still unrealistic to have rainfall twice a week over 3 months of the growing period, we believe that the control treatment in our study provided a contrast in plant water status (compared with the drought-stressed group), which was more or less com- parable with the level of water stress that plants would experience in their natural environments. Throughout the whole growth period, plants were maintained with only vegetative growth. Plants were harvested from September 14 to 23. Harvested plants were rinsed with water to remove soil particles and then separated into four parts: shoots, roots, rhizomes and crowns (shoot base). All plant parts were dried at 65C in a forced-air oven for 48 h and then weighed to measure dry mass. Each data point of the fractional allocation of vege- tative biomass to a particular plant organ was obtained as a ratio of two biomass variables. Thus, the error of the ratio tended to be greater than the error of either of the two original variables due to error propagation

Table 1 Pre-dawn leaf water potentials (MPa) of four grasses grown under a portable greenhouse in summer 2011. Data are leaf water potentials between consecutive watering events in mid-August, 2011 for control group plants (each pot received 150 mL water twice a week) and drought group plants (receiving 150 mL water once a week to once every 10 days). This same watering schedule was applied to the control and drought groups during the whole experimen- tal period from July 7 to September 17. Data are averages and standard deviations (in parentheses, n See main text for detail of pot dimension. Abbreviations: kb: Kentucky bluegrass; sb: smooth brome; ww: western wheatgrass; gn: green needlegrass. Plants received water a day before (8 August 2011). Plants received water immedi- ately following pre-dawn leaf water potential measure- ment.Values within a column followed by the same letter are not significantly different.

(Taylor, 1982). Using a bootstrap resampling procedure of Hesterberg et al. (2006), we estimated from the mea- sured data of 2010 that about 10-11 replications were needed to detect the potential statistical differences in our experiment, given the variability of the measured biomass allocation data. This was accomplished by drawing different number of 'replications' from the six original observations through resampling with replace- ment, followed by the nonparametric Mann-Whitney U-test (Moore and McCabe, 2006) available in MINITAB 13.31 (MINITAB Inc., State College, PA, USA). This result served as the basis for us to increase the number of replications for each species-treatment combination to 12 in our 2011 experiment. The data of biomass partition (in percentages) in different plant parts between the native and invasive species and between drought and control group plants for different species were analysed using the Mann- Whitney U-test. Also, as a comparison, the percentage data were arcsine-transformed and analysed by the one-way ANOVA with Tukey's method (family error rate of 5%) for pairwise comparisons. Pre-dawn leaf water potentials for each species were log-transformed (by first multiplying the measured water potential data by 1) and analysed using one-way ANOVA for the comparisons of water potentials between measure- ments taken on different days.

Abstract | Results

Discussion | References | Abstract | Background | Sustainable intensification - changing paradigm in forage production | Global grasslands under threat | Land-use changes in the steppe ecosystem of Inner Mongolia Autonomous Region, P.R. China | Land-use change in the tropical savannah (Cerrado) of South America | Sustainable intensification of dairy farming - the specific role of grassland- based forage production | References |

© -