Clones of tall rescue (Festuca arundinacea Schreb.) regenerated from tissue culture are required for examining interactions between the endophyte, Acremonium ¢oenophialum Morgan-Jones and Gams, and its host. Somaclonal variation is common among regenerated plants, and could limit the utility of this technology to study the interactions of tall rescue and its endophyte. Various methods are used to detect somaclonal variation, but there is little agreement as to the relative advantages of each. Therefore, the objectives of this study were to determine (i) whether somaclonal variation exists among tall rescue plants regenerated from somatic embryos, and (ii) whether pollen viability, yield, morphological tr its, or phenologieal development give consistent estimates of somaclonal variation. Seven regenerants from non-infected tall rescue genotype PDN2 and a nonregenerated PDN2 control were planted in the field in five replications. Plants were screened for somadonal v riation by phenological development, vegetative and flag leaf width/length ratios, total biomass, eed yield, and pollen viability over 2 yr. Although regeneration protocols were designed to minimize somaclonal variation, somaclonal variation of quantitative traits was observed. Total biomass and seed yield of regenerants, used together, were most consistent in detecting somaclonal variation. When using tissue culture to insert endophytes into tall rescue, we found multiple regenerants need to be infected with each isolate of endophyte to ascertain whether differences among response variables are due to cndophyte or somaclonal variation. S STRATEGIES have been used to study interactions between tall fescue and its fungal endophyte. One method is to infect clones of a common tall fescue genotype with different endophyte isolates to assess endophyte-mediated changes in the plant (Hill et al., 1991b). Infection of clones of a common tall fescue genotype with different endophyte isolates necessitates the use of plant tissue culture techniques (Kearney et al., 1991) because the endophyte will not infect mature tall fescue plants (Bacon and Siegel, 1988). A concern of using tissue culture to study tall fescue/endophyte interactions is the possibility that somaclonal variation may confound the effect of an endophyte, making it difficult to assess the endophyte’s contribution to plant responses. Concomitant with concerns that somaclonal variation exists, is the ability to document its existence. Traits that have been evaluated to detect somaclonal variation in plants include pollen viability, plant morphology, phenology, and yield. Eizenga and Dahleen (1990) used pollen viability to identify somaclonal variation in tall fescue, while others have used leaf width/length ratios (Cummings et al., 1976), plant height (Hanna et al., 1984), heading and flowering dates, or seed yield (Stephens et al., 1991) as traits for other species. There is no consensus as to which traits consistently detect somaclonal variation. Possible reasons for the lack of consensus are that (i) the traits chosen by individual J.T. Royiance, N.S. Hill, and W.A. Parrott, Dep. of Crop & Soil Sciences, Univ. of Georgia, Athens, GA 30602-7272. Received 13 Sept. 1993. *Corresponding author (nhill@uga.cc.uga.edu). Published in Crop Sci. 34:1369-1372 (1994). investigators are those which are easiest to measure, (ii) traits may be affected by genotype x environmental interactions, and (iii) selected traits are not always based on a variety of unrelated characters (De Klerk, 1990). Therefore, the objectives of this study were to determine (i) whether quantitative traits were affected by somaclonal variation in tall fescue plants regenerated from somatic embryos; and (ii) examine if pollen viability, yield, morphological traits, or phenological development are the most consistent variables to use when screening for somaclonal variation. MATERIALS AND METHODS Plants regenerated by somatic embryogenesis (Kearney et al., 1991) were used in this experiment. Explants were taken from meristematic regions at the base of elongating tillers from the endophyte-free tall fescue genotype PDN2. The explants were surface-sterilized in a commercial bleach solution (10.5 L-t NaCIO3) for 30 sec and in 700 mL -1 EtOH solution for 3 rain. Explants were rinsed three times with sterile water, cut into 2.5-cm lengths, and placed on a basal medium to induce callus formation. The embryonic medium consisted of MS salts (Murashige and Skoog, 1962), B5 vitamins (Gamborg et al., 1968), 30 g -t sucrose, 3 0 gM 2,4-D, a nd 20 g L-l Gelrite (Gibson Laboratories, Inc., Lexington, KY) as solidifying agent. The pH of the medium was adjusted to 5.8 prior to autoclaving. Cultures were placed in a growth room with 30 gM m-2 sec-~ photon flux density provided by cool white fluorescent tubes; daylength was maintained at 23 h and temperature at 26 + 1 °C. After 28 d on embryonic medium, calli were transferred to auxin-free basal medium to permit growth and development of somatic embryos. Germinating embryos were excised from the calli and placed in GA-7 containers (Magenta Corp. Chicago, IL) with auxin-free basal medium with a sucrose level of 6 g L-1. After 28 d, plants were transferred to soil and allowed to harden. The plants were transferred to the greenhouse and placed in 4-L pots containing a 1:1:1 mixture of Cecil sandy clay loam soil (Typic Kanhapludult, mixed, mesic, thermic)/peat/perlite. Of 77 endophyte-free regenerated plants (Kearney et al., 1991), regenerants R12, R16, R19, R22, R30, R37, and R40 of tall fescue genotype PDN2 were randomly selected for this experiment. Individual tillers were taken from each regenerated plant, cut to 5-cm lengths and screened for a uniform mass of 2 g fresh weight. They were also screened so each contained two visible tiller primordia at the basal node. Fifteen tillers were selected from each regenerant and transplanted into 10-cm pots containing the soil mixture used previously, and grown in the greenhouse for 28 d. The same was done for tillers from the non-regenerated parent plant PDN2. Five plants of each regenerant were selected only if they had developed three fillers with two expanded leaves on each filler. This was conducted to assure that all plant materials were physiologically and developmentally uniform, thereby reducing variability within and among regenerants that could be attributed to the physiological condition of the vegetative propagules. On 1 Oct. 1990, the regenerants were transferred to the Abbreviations: FWL, flag leaf width/length ratio; VWL, vegetative leaf width/length ratio; PCSAS, statistical analysis systems for personal com