Validation of the doubly‐labeled water (DLW) method for estimating CO 2 production and water flux in growing poultry chicks

This study is the first validation of the doubly‐labeled water (DLW) method on birds (1) to evaluate the accuracy of 2 points versus multiple points for computing fractional isotopic washout rates (k) and CO 2 production (rCO 2 ), (2) to measure CO 2 production and water flux each day over a 4‐day period, (3) to compare measured fractional evaporative water loss (r G ) with assumed values that provide DLW estimates of rCO 2 with zero error, and (4) to measure the effect of assumed r G on the error of estimating water influx and efflux. Percent error of CO 2 production of six growing poultry chicks estimated by the DLW method was not correlated with mean daily relative growth rates of up to 5% nor with daily rates of energy retained in growth of up to 320 kJ/day/kg, nor was it significantly reduced by using multiple points (5 points) rather than 2 points to compute fractional isotopic washout rates (k) and isotope pool sizes. Its seems clear from our study and the previous 5 validations on growing birds that average relative daily growth rates of up to about 20% do not increase the error of estimating rCO 2 by the DLW method. Arithmetic error was significantly less when using one isotopic pool, rather than two pools, to compute rCO 2 and was less when using an assumed fractional evaporative water loss (r G ) of 0.45 rather than an assumed r G of 0.25 or 0.5 (the two values used predominantly in previous DLW studies). Our study supports Speakman's (1997) suggestion that the one‐pool model is more appropriate than the two‐pool model for birds weighing<1 kg. We recommend using an assumed r G of 0.45 to compute rCO 2 of poultry, which is a compromise between the two schools of r G useage, i.e., r G =0.25 or 0.5, however we hesitate to recommend 0.45 for all birds in all settings. Close agreement between measured r G and an assumed r G that produced zero rCO 2 error supports the validity of using the pooled fractionation correction factors (f pool ) of 0.0339 for tritiated water and 0.0249 for deuterated water. Absolute error decreased with the percent washout of during measurement periods of 1 to 4 days. Accuracy of estimating rCO 2 was not significantly different for durations of 2, 3, and 4 days using either tritiated or deuterated water. The arithmetic error of estimating rCO 2 using a one isotopic pool model, 2 points, an r G of 0.5, and tritiated water was −1.9% (SD=13.5) for the first day of a 4‐day period and −4.0% (SD=8.9) for the entire period. Percent arithmetic error of water influx (rH 2 O inf ) and efflux (rH 2 O eff ) estimated for day 1 from tritiated water washout and an assumed r G of 0.5 was −0.5 (SD=6.4) and 0.1% (SD=11.1), respectively. An r G of 0.5 produced significantly less arithmetic error than an r G of 0.25 or an r G of zero (i.e., no fractionation correction), and less absolute error in rH 2 O inf . Errors were slightly more negative (underestimates) with an r G of 0.25, i.e., −2.2 and −2.0%, respectively and even more negative with no correction for isotopic fractionation (i.e., an assumed r G of zero). Tritiated water estimates of water influx and efflux during the first day had no error when using an r G of 0.57 and 0.48, respectively. With assumed r G s of 0.25 and 0.5, the errors of water influx were −7.8 and −5.9%, and the errors of water efflux were 3.4 and 5.6%, respectively, over 4 days. We recommend using an assumed r G of 0.45 to compute rH 2 O eff for poultry. The error of rCO 2 was about 3 to 4 times more sensitive to values of assumed r G than the error of water flux.