DISS. ETH NO. xx Impacts of land use on biodiversity: development of spatially differentiated global assessment methodologies for life cycle assessment A dissertation submitted to ETH ZURICH for the degree of Doctor of Sciences presented by LAURA SIMONE DE BAAN Master of Sciences ETH born January 23, 1981 citizen of Steinmaur (ZH), Switzerland accepted on the recommendation of Prof. Dr. Stefanie Hellweg, examiner Prof. Dr. Thomas Koellner, co-examiner Dr. Llorenç Milà i Canals, co-examiner 2013 In Gedenken an Frans Remarks This thesis is a cumulative thesis and consists of five research papers, which were written by several authors. The chapters Introduction and Concluding Remarks were written by myself. For the sake of consistency, I use the personal pronoun ‘we’ throughout this thesis, even in the chapters Introduction and Concluding Remarks. Summary Summary Today, one third of the Earth’s land surface is used for agricultural purposes, which has led to massive changes in global ecosystems. Land use is one of the main current and projected future drivers of biodiversity loss. Because many agricultural commodities are traded globally, their production often affects multiple regions. Therefore, methodologies with global coverage are needed to analyze the effects of land use on biodiversity. Life cycle assessment (LCA) is a tool that assesses environmental impacts over the entire life cycle of products, from the extraction of resources to production, use, and disposal. Although LCA aims to provide information about all relevant environmental impacts, prior to this Ph.D. project, globally applicable methods for capturing the effects of land use on biodiversity did not exist. The goal of this thesis was thus to develop operational LCA methods for quantifying the effects of land use on biodiversity. The methods needed to provide global coverage and be spatially explicit. One central research question was how we could measure biodiversity, as a very complex and multi-faceted concept, within LCA using available global data. This thesis encompasses four approaches, which were tested in case studies to assess and illustrate their applicability. We began with an evaluation of how biodiversity loss is modeled within LCA (Chapter 2). Two drivers of biodiversity loss have thus far not been quantified: overexploitation and invasive species. The methodologies for assessing the effects of three other drivers (habitat loss, climate change, and pollution) involve several conceptual shortcomings, with scale considerations largely absent. The current practice of aggregating the impacts of multiple drivers of biodiversity loss into a single score is questionable, because species loss refers to different spatial scales of loss (e.g., local, regional, or global loss). Within existing methods, taxonomic and geographic coverage is very limited, globally applicable methods are largely absent, and most methods are not spatially explicit. Based on the identified conceptual shortcomings, we developed four new methods to overcome some of these limitations. The first method (Chapter 3) is based on a global literature review, which collected data on comparative biological surveys (GLOBIO3). The species richness of used and unused (natural) land was compared within one region, and relative local species loss was calculated. The effects on biodiversity were found to depend on the taxonomic group, the biome, the type of land use, and the choice of the biodiversity indicator. In general, a negative impact of land use on biodiversity was observed, but the results showed considerable variation. In Chapter 4, we further developed this method to quantify differences between organic and conventional agricultural products. We collected specific data on vascular plant species’ richness in organic and conventional fields, and developed a regional weighting scheme to assess the values of different ecosystems. Based on a case study of Swedish milk, we showed that while production of organic milk required twice the area, its impact on biodiversity was 5 Summary half that of conventional milk. Feedstuffs imported from tropical countries largely contributed to the overall impact. In Chapter 5, we applied species-area relationship models to quantify regional species losses resulting from accumulated land use for all global ecoregions. We distinguished between the potentially reversible regional losses of non-endemic species due to land occupation and transformation and the irreversible, permanent losses of endemic species. The regions with the highest species losses were similar across taxonomic groups (mammals, plants, birds, reptiles, and amphibians) and overlapped with regions where most natural land had been converted in the past. For regions threatened by future habitat loss, we also illustrated how future impacts could be modeled based on land use scenarios. In Chapter 6, we used habitat suitability models for mammals to model impacts of land occupation and transformation per 900m grid-cell. For all grid-cells, we modeled species richness assuming agricultural use, and calculated species losses compared to the richness under the two reference scenarios of non-use or current land use. The loss of species was weighted by the species’ global rarity and threat status (the latter according to IUCN classification), so that the assessment could serve as a proxy for global species extinction risk. Finally, the method was applied to a case study of the land use impacts of coffee, tea, and tobacco cultivation in East Africa. The results were compared to the results of the methods developed in Chapters 3 and 5. In Chapter 7, we compared the methods developed within this thesis with other methods that have the potential for global application, and offered our recommendations on the application and further development of land use assessment methods. The relative, local method (Chapter 3) should only be applied in conjunction with a regional weighting scheme (Chapter 4). The method developed in Chapter 5 is directly applicable on a global scale and encompasses effects on five taxonomic groups and aspects of reversibility. If the origin of a product is known in detail, the method from Chapter 5 should be complemented with the method from Chapter 6, which captures how the threat of land use affects the survival of single mammal species with a fine spatial resolution. However, the method from Chapter 6 is currently only operational for East Africa, and still needs to be expanded to apply to all world regions. This thesis significantly improves on past land use impact assessment methods. However, modeling spatially differentiated biodiversity impacts within LCA still requires methodological improvements in multiple areas. Further development and testing of these methods will contribute to a better understanding of the impacts of economic activities on biodiversity, and help identify and prevent the shifting of burdens between regions or environmental compartments. 6 Zusammenfassung Zusammenfassung Ein Drittel der weltweiten Landfläche wird heute landwirtschaftlich genutzt, was einen massiven Eingriff in die Ökosysteme bewirkt und stark zum globalem Biodiversitätsverlust beiträgt. Da viele landwirtschaftliche Konsumgüter global gehandelt werden, beeinflusst die Herstellung oft mehrere Weltregionen. Um die Auswirkungen von Landnutzung auf die Biodiversität zu bewerten, reicht daher oft eine regionale Analyse nicht aus. Ökobilanzen setzten hier an. Sie erfassen die Umweltauswirkungen von Produkten über deren gesamten Lebensweg, von der Ressourcenentnahme über die Produktion und Nutzung bis hin zur Entsorgung. Obwohl Ökobilanzen ein möglichst vollständiges Bild aller relevanten Umweltauswirkungen bezwecken, fehlten bisher anwendbare Methoden zur globalen Quantifizierung von Landnutzungseffekten auf die Biodiversität. Das Ziel der vorliegenden Arbeit ist es, global anwendbare und räumlich spezifische Ökobilanz-Methoden zur Bewertung von Biodiversitätsverlust durch Landnutzung zu entwickeln. Eine Hauptfragestellung war, wie anhand der global verfügbaren Daten Biodiversität sinnvoll in Ökobilanzen erfasst werden kann. Wir entwickelten vier neue Methodenansätze und testeten sie in Fallstudien, um Schlussfolgerungen über deren Anwendbarkeit ziehen zu können. Die Arbeit beginnt mit einer Evaluation der existierenden Ansätze zur Bewertung von Biodiversitätsverlust in Ökobilanzen (Kapitel 2). Zwei Ursachen von Artenverlust werden bei diesen Ansätzen nicht berücksichtigt: Übernutzung und invasive Arten. Die bestehenden Bewertungsmethoden für Habitatverlust, Klimawandel und Schadstoffbelastung weisen verschiedene konzeptionelle Mängel auf. Der räumliche Massstab des Artenverlustes ist oft nicht definiert und grösstenteils fehlt eine räumliche Differenzierung. Somit erwies sich die gängige Praxis, die Biodiversitätsverluste von unterschiedlichen Ursachen zu addieren, als wenig sinnvoll, da die Verluste sich auf verschiedene räumliche Massstäbe beziehen (z.B. lokaler, regionaler oder globaler Artenverlust). Nur wenige Artengruppen und Weltregionen sind in den untersuchten Methoden abgebildet, globale Ansätze fehlen weitgehend. Zur Ausbesserung der bestehenden methodischen Mängel wurden vier neue Ansätze entwickelt. Die erste Methode (Kapitel 3) nutzte Daten einer globalen Literaturrecherche biologischer Studien (GLOBIO3). Die Artenzahl von genutzten und ungenutzten (natürlichen) Flächen einer Region wurde verglichen und der relative lokale Artenverlust berechnet. Die Effekte auf die Biodiversität