Temporal variations in plant δ13C values and implications for using the 13C technique in long-term soil organic matter studies

The 13C natural labeling technique in combination with modeling could potentially be used to analyze SOM changes in long-term experiments involving complex cropping systems such as agroforestry systems. It requires information on quantity and 13C abundance of all plant materials entering the soil throughout the experiment. The occurrence of significant year-to-year changes of δ13C values of the plant inputs may complicate the technique, as it would require a systematic isotopic analysis of all plant inputs over the entire life span of the experiment. By analyzing crop and tree samples from different sampling times in a long-term agroforestry experiment under a sub-humid tropical climate we established that year-to-year variations in plant δ13C values were of the order of only 0.4–0.6‰, i.e. only 3–4% of the maximum tracer signal difference between C3 and C4 plants (∼15‰). Hence plant sampling and isotopic analysis can be limited to 2–3 years or even a single representative year in terms of weather. Important and consistent within-year variations were, however, observed for legume tree prunings (up to 2.4‰) and weeds (up to 7‰), pointing to the need for frequent sampling within a single year if such seasonal variation is suspected. Observed δ13C values for the weed vegetation revealed a clear shift in weed composition in terms of C3 and C4 weeds within the season and across treatments, as was confirmed by a visual weed species identification in the field.