Identifying suitable arable weeds as reference plants for measuring N2 fixation in grain legumes in Zambia using 15N natural abundance

Three options were evaluated for identifying suitable arable weeds as reference plants for estimating N-2 fixation in grain legume symbioses in 3 agro-ecological zones of Zambia using N-15 natural abundance. For each test legume species, 5 different small-farmer fields were randomly selected at each site, and 5 shoots of legume and 5 shoots of at least 3 wild weed species were randomly harvested from each of the 5 fields per site using a paired sampling protocol. The composition of weeds differed from site to site, and the types of weed species were found to increase with annual rainfall. Isotopic analysis of 15N showed that 97% of the weed species were suitable as reference plants for estimating N2 fixation at their sites of origin. No weed species was found with similar 15N content across all sites. Within each site, some species were more dominant than others. The mean delta N-15 values for these dominant species ranged from -0.50 to 3.00 parts per thousand, 4.31 to 5.56 parts per thousand, and 5.39 parts per thousand to 6.64%, respectively, for Kabwe, Kasama, and Magoye, with the differences between species means being significant (P < 0.05) at only Kabwe. When the 5 15N values were averaged across species, the means ranged from 1.06 to 2.6 parts per thousand, 2.86 to 5.34 parts per thousand, and 3.94 to 8.89 parts per thousand, respectively for the Kabwe, Kasama and Magoye sites, with the differences being significant (P <= 0.05) at only Magoye. The combined mean 8 15N value for herbaceous weeds was significantly greater and, with a lower standard error than that of graminaceous species at Kasama and Magoye, suggesting that, as reference plants, the herbaceous species were more likely to provide accurate estimates of N2 fixation for these legume symbioses at the study sites. Selection of reference plants on a per field basis was the only option applicable to all sites, while selection based on suitable weed species across fields was an option for only Kabwe and Magoye, but not for Kasama. However, a 8 15N value for the entire site, obtained from averaging the 8 15N values of all field means of a site, was also as appropriate for estimating N2 fixation at Kabwe and Kasama as the average 8 15N value calculated from means of suitable species at Kasama and Magoye. Based on the mean 8 15N values, both graminaceous and herbaceous weeds were found to be equally suitable as reference plants for estimating N2 fixation at Magoye. However, herbaceous weeds were probably better reference plants for accurately estimating %Ndfa at Kabwe and Kasama.