Isotope dilution methodology offers the advantage of an integrative technique for measuring N2 fixation in the field, but may have limitations in grass‐legume mixtures because of the possibility of transfer of fixed N from legume to grass. A study was conducted in the greenhouse and in the field utilizing two soils, Hanford sandy loam and Yolo silt loam (Typic Xerorthents) in which the organic matter was labeled by previous application of either 15N‐depleted or 15N‐enriched compounds. Contributions of atmospheric N2 to total plant N were calculated from the isotopic composition of plant tissue, using the composition of the non‐legume as a reference value. Ladino clover (Trifolium repens L.) and Wimmera ryegrass (Lolium rigidum L.) were grown alone and in mixture for purposes of comparison. In the greenhouse estimation of N2 fixation by the isotope and difference methods were in good agreement. There was essentially no transfer of fixed N2 from the clover to the ryegrass in any of the pot experiments. In the field experiment, clover obtained 85 to 100% of its N from the fixation process. In the clover‐ryegrass mixture, % total N and % 15N excess comparisons both indicated substantial transfer of N from clover to ryegrass after the stand had been established after about 6 months. Up to 79% of the N in ryegrass was calculated to be derived from such transfer. It is concluded that the isotope dilution method gives estimates of N2 fixation at least as good as those obtained by yield‐dependent methods such as the difference method, but it is not suitable for grass‐legume mixtures.

Soil cores were taken from six locations representing three virgin and three cultivated soils in increments of 15 or 30 cm down to depths ranging from 120 to 300 cm. The number of samples of an individual soil profile varied from 20 to 96, though smaller numbers of some subsurface horizons were obtained at two locations where very rocky conditions were encountered. Frequency distribution analysis of δ15N values and of total N showed that they were log‐normally distributed in about half the groups of samples. Assumption of normal distribution in all samples would result in a maximum error of 37% in total N and 0.1% in 15N content if the real distribution were in fact log‐normal. Coefficients of variation of total N were somewhat lower in cultivated soils than in virgin soils, but C.V. values for atom %15N were comparable in the two groups of soils. It is concluded that the magnitude of natural variation in δ15N values at a given sampling site, both laterally and vertically, is so great as to preclude tracing biological events by means of natural abundances.

No abstract

No abstract