Technical sustainability of biological production systems. A model based life cycle assessment of fertilizer management

Accordingly with world’s movement towards sustainable development, agriculture faces mounting pressure to reduce its environmental impact. The present organisation of our food production leads to significant alterations of the global nitrogen cycle causing increased nitrogen emissions into the ecosystem. Aiming at sustainable systems requires to account for long-term implications of practices and the broad interactions and dynamics of agricultural processes. A key goal hereby is to pursue technical sustainability through understanding agriculture from a biophysical perspective in terms of water and nutrient dynamics, and interactions among plant and soil under changing climate conditions and different management strategies. The present work provides an integrated dynamic and process based crop-soil model which is coupled to a life cycle assessment (LCA) to evaluate the technical sustainability of a biological production system. The model simulates crop growth and development as well as soil water and nitrogen dynamics under varying climate conditions and different nitrogen fertilizer application rates. With this dynamic approach to predict the related field nitrogen emissions, a more reliable and realistic assessment of the environmental impact can be obtained to enhance the by default static LCA output. This allowed the assessment of implications of future management scenarios considering potential impact reduction strategies. More specifically, the open field production of a cauliflower leek rotation in Flanders, Belgium was chosen as case study throughout the whole research. It is a commonly applied rotation cycle susceptible for excessive use of inorganic fertilizers.

A life cycle assessment, a widely recognized method within the sustainability assessment, quantifies the environmental impact of a system in terms of impact categories like global warming, eutrophication, toxicity, etc. An LCA provides a comprehensive and objective method of analysis that identifies the environmentally most dominant stage(s) in a product life cycle and allows the comparison of alternative production (sub)systems regarding their environmental burden. A preliminary LCA with commonly used empirical models to estimate the field emissions, showed that increased fertilizer application does not result in a sufficient increase in yield to justify additional emissions and to be environmentally favourable. Application of a lower N dose would benefit the environment, but entails a lower commercial yield. It would be a matter of finding the trade-off between yield and potential environmental costs, as the land occupation favours the higher N doses. Although, the latter is not necessarily true in terms of only the edible part of the crop compared to the commercial yield of the whole crop as functional unit. In any case, besides potential renewable energy sources, efforts should be made to reduce emissions of nitrogen pollutants as they are a major source for climate change, acidification and eutrophication. The empirical approach for their calculation however, is very limited to account for the potential effect of mitigation strategies due to the aggregated estimation level and the lack of predicting their implications on crop growth and soil conditions. As natural variability due to varying biophysical conditions is inherent in agricultural production, future climate and soil conditions could alter the whole nitrogen flow through the crop-soil-air environment and shift the most favourable fertilizer management. Although LCA is praised for its holistic approach, it has an inherent static and linear nature and heavily depends on the quality of input data.

Therefore, driven by meteorological data, soil properties and agricultural management, a crop-soil- climate interaction model was developed which simulates at field scale on a daily basis the soil temperature, crop growth and development, water flow and soil carbon and nitrogen dynamics including emissions of environmental pollutants to the air and ground water. If the soil supply of water and/or nitrogen does not meet the demand of the crop, a deficiency factor is implemented to limit crop growth and actual water and nitrogen uptake. According to the visual match and associated statistical performance indicators, model predictions were fair to very good as well for the calibration as for the validation with three years of observations and different N dose rates. Given the large variability and strict performance rating thresholds, biomass growth, its nitrogen content, the water content and temperature in the different soil layers predicted the observations very well. The soil nitrogen content simulation however suffered from the discrete limited sampling numbers, the lack of detailed knowledge and the complex interaction of different pathways that affect the content simultaneously. Along with the calibration, a local sensitivity analysis of the model responses to changes in model parameters was performed based on the ratio of their coefficients of variation. Certain soil processes, especially runoff, water percolation and nitrogen leaching and the emission of nitrous oxides were found to be sensitive to a 10% change of mainly the runoff curve number for average moisture content and the water content at field capacity of the top layer.

Next, the LCA results were compared with nitrogen field emissions estimated by either the default empirical approach or by the developed dynamic model. Overall, the model based LCA showed a consistently lower impact than the default LCA results of the same crop rotation cycle and fertilizer management. The only exception was the eutrophication potential under the higher N dose application rates for all three years. Differences between the impacts according to both approaches tend to increase with increasing N dose rate besides the impact itself. Changes of impact over the different years were reflected similarly by both outcomes. However, as the empirical approach might look straightforward regarding alternative solutions, they are limited and potentially ineffective. If the LCA needs to support future management decisions, an appropriate choice of approach for estimating field nitrogen emissions is required as it might shift the environmental favourable option to alternative and substantiated solutions, especially considering the timing of reduction strategy implementation. A daily time step and accounting for multiple processes and disturbing factors allows the model based simulation to provide more accurate and efficient adaptations towards sustainable systems. Furthermore, the implications of management adjustments or extreme climate changes on crop yield and nitrogen dynamics cannot be addressed by the default LCA method as the empirical approach depends on standard crop N uptake curves and does not account for precipitation and soil moisture effects for instance. In a dynamic system like the water and nitrogen flow in a crop-soil environment, impact assessment should address ‘when’ even more than ‘which’ potential reduction strategies should be implemented.

Finally, the model based LCA was implemented for a scenario analysis that included potential reduction strategies regarding fractionated fertilizer application and plastic mulching during winter fallow periods. Whereas a fertilizer application distributed in time to meet the crop demand might reduce N stress and increase yield, the winter soil cover could prevent drain and subsequent nitrogen leaching. Although this was to a certain extent reflected in the outcome of the model simulations, the mitigated environmental impact was cancelled by the burden from additional fertilizer equipment and energy use and/or especially from the plastic cover production and disposal. Only the eutrophication potential would be reduced if the strategy would be implemented. It shows that future decisions require a holistic perspective that combines dynamic model predictions and aggregated LCA results, which still would imply a trade-off between different impacts, but would prevent a problem shift. Such scenario analysis is considered less reliable and could be more misleading with the empirical approach as applied in default LCA studies regarding dynamic agricultural systems.

The current implementation of the model based life cycle assessment showed the strength and importance of a system analysis to (i) provide improved process based insight in the agricultural production system with more reliable predictions, (ii) to understand, quantify and optimize the technical sustainability of a product and (iii) to address more complex issues on sustainable production and future decisions.