Soil respiration declines with increasing nitrogen fertilization and is not related to productivity in long-term grassland experiments

Abstract

Soil respiration is often used as an index of fertility because the majority of nutrients are cycled through the microbial biomass. We assessed the role of soil respiration as a measure of resource productivity in three long-term grassland experiments near Pietermaritzburg, South Africa. All of these experiments have shown significant changes in grass species composition and productivity. An ongoing Veld (=field) Fertilizer Experiment (VFE) that manipulated the level of nitrogen fertilizer, phosphorus and lime has been running since 1951. A Burning and Mowing Experiment (BME) has been running since 1950. The third experiment is part of the Nutrient Network (NutNet), a global fertilizer experiment that has been manipulating nitrogen, phosphorus, potassium and micronutrients and has been running since 2009. We found that longer-term experiments were more likely to show significant effects on soil respiration. We found several significant effects in the VFE but no significant differences in soil respiration among fertilization treatments in the shortest-term experiment (NutNet). In the VFE, we found significant differences in soil respiration due to levels of nitrogen fertilizer, form of nitrogen fertilizer (limestone ammonium nitrate and ammonium sulphate), phosphorus and lime. We found no significant relationship between above-ground net primary productivity and soil respiration despite the frequent detection of such a pattern due to the link between soil respiration, soil fertility and productivity. We found that, while there was a consistent increase in total soil nitrogen with increasing levels of nitrogen fertilizer applied, there was a consistent decrease in soil microbial respiration. There was a significant positive correlation between soil respiration and pH. Possible mechanisms behind this are unclear but may involve changes in dominant enzymes and possibly switches between dominance of bacteria and fungi. We also found significant effects of the timing of burning in the BME, but not due to the frequency of burning or the occurrence of mowing. Our results suggest that studies may need to be long-term, for example here at least 10 years, before key functional relationships with soil fertility can be reliably understood.