Breeding and selection for higher transpiration efficiency (W) has been hampered by tedious and costly methodology. Rapid and less costly methods are needed for screening W in plant improvement programmes. We report the relationship of ash, silicon (Si) concentration, and Si uptake to W in crested wheatgrass (Agropyron desertorum [Fischer ex Link] Schultes), an important C3 range grass in western North America. Clones of crested wheatgrass were grown under three water levels in a field rainout shelter and as potted plants under two water levels in the field and greenhouse. Ash and Si concentrations were compared to previously determined values of shoot mass, transpiration, W, and carbon isotope discrimination (A). Ash and Si concentrations were not consistently related to A and W across all environments; however, ash concentration was positively correlated with A (r 69 • *, df = 22) and negatively correlated with W (r = -0 . 61**, df= 22) in the well-watered field environment. Across all environments and studies, the ranges in the coefficients of variation (CV, %) for clonal means were: W, 4-15; A, 1-4; ash concentration, 6-14; Si concentration, 13-30; and Si uptake, 21-33. The generally lower CV for W, A, and ash concentration suggest that these traits were more repeatable than Si concentration or uptake. Although a consistent relationship was not observed between Si and W and between ash and W, the correlations of ash and W from the well-watered field environment were encouraging. In view of the low cost for ash analysis, we conclude that further research is needed to evaluate the potential of ash as a criterion in selecting for improved W, particularly during the early phases of a breeding programme when large populations are usually involved. Later selections could be based on the more precise and accurate, but costly, A analysis.

Forage improvement programs often select for increased crude protein and dry matter digestibility. Additionally, breeding programs may be interested in selecting for enhanced transpiration effkiency or water use-efticiency. Forage crude protein and dry matter digestibility are commonly determined by near infrared reflectance spectroscopy (NIRS), whereas water use-efficiency is estimated from "C discrimination (A) values obtained from isotope-ratioing mass spectrometers. If NIRS could predict A, then W could be determined simultaneously with quality components at a much lower cost. To test this possibility, leaf samples of alfalfa (ilZe&cugo sativu L.) and several cool-season perennial grasses were analyzed with a dual-inlet, double collector gas isotope mass spectrometer, and values of A were calculated. Subsamples were scanned with monochromators that collected spectra from 400 to 2,500 nm or 1,100 to 2,500 nm, and absorption data were regressed with values of A. Standard errors of calibration for regressing A with NIRS absorption values were higher for grasses than for alfalfa. Coefficients of variation for all validation sample sets used for prediction of A by NIRS were less than 3 % , and NIRS correctly identified 77 to 82% of the samples with the lowest A values as determined by mass spectrometer analysis. This level of predictability may be acceptable for identification of genotypes with high water use-efficiency during the early phases of forage improvement programs.