Assessing the environmental impacts of soil compaction in Life Cycle Assessment

Maintaining biotic capacity is of key importance with regard to global food and biomass provision. One reason for productivity loss is soil compaction. In this paper, we use a statistical empirical model to assess long-term yield losses through soil compaction in a regionalized manner, with global coverage and for different agricultural production systems. To facilitate the application of the model, we provide an extensive dataset including crop production data (with 81 crops and corresponding production systems), related machinery application, as well as regionalized soil texture and soil moisture data. Yield loss is modeled for different levels of soil depth (0–25cm, 25–40cm and >40cm depth). This is of particular relevance since compaction in topsoil is classified as reversible in the short term (approximately four years), while recovery of subsoil layers takes much longer. We derive characterization factors quantifying the future average annual yield loss as a fraction of the current yield for 100years and applicable in Life Cycle Assessment studies of agricultural production. The results show that crops requiring enhanced machinery inputs, such as potatoes, have a major influence on soil compaction and yield losses, while differences between mechanized production systems (organic and integrated production) are small. The spatial variations of soil moisture and clay content are reflected in the results showing global hotspot regions especially susceptible to soil compaction, e.g. the South of Brazil, the Caribbean Islands, Central Africa, and the Maharashtra district of India. The impacts of soil compaction can be substantial, with highest annual yield losses in the range of 0.5% (95% percentile) due to one year of potato production (cumulated over 100y this corresponds to a one-time loss of 50% of the present yield). These modeling results demonstrate the necessity for including soil compaction effects in Life Cycle Impact Assessment.