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 waren abhängig von der untersuchten Artengruppe, dem Biom, dem Landnutzungstyp und der Wahl des Indikators. Die meisten Landnutzungsformen bewirkten einen überwiegend negativen Einfluss auf die Biodiversität, wobei die Resultate stark streuten.
Kapitel 4 befasst sich mit der Weiterentwicklung dieser Methode um Vergleiche zwischen bio-logischen und konventionellen Produkten zu ermöglichen. Zusätzliche Daten zur Gefässpflanzen-vielfalt biologisch und konventionell angebauter Flächen wurden einbezogen und regionale Unterschiede von Ökosystemen mittels Gewichtungssystem erfasst. Anhand
7 Zusammenfassung einer schwedischen Fallstudie konnten wir zeigen, dass biologische Milch zwar fast doppelt so viel Fläche beanspruchte, der Einfluss auf die Biodiversität aber nur halb so gross war wie der von konventioneller Milch. Aus tropischen Ländern importierte Futtermittel trugen stark zum Biodiversitätsverlust bei.
In Kapitel 5 werden regionale Artenverluste durch die gesamte regionale Landnutzung anhand von Arten-Areal-Kurven und für alle globalen Ökoregionen modelliert. Wir unterschieden dabei potentiell reversible Effekte von Landnutzung und -umwandlung auf nichtendemische Arten und irreversible, permanente Verluste von endemischen Arten. Ähnliche Regionen zeigten den höchsten Artenverlust von verschiedenen Artengruppen (Säugetiere, Pflanzen, Vögel, Reptilien und Amphibien). In diesen Regionen ist bereits heute der grösste Teil der natürlichen Flächen umgewandelt. Für Regionen, die von zukünftigem Habitatverlust bedroht sind, konnten wir aufzeigen, wie Zukunftsszenarien in die Bewertung mit einfliessen können.
In Kapitel 6 werden Informationen zu geeigneten Lebensräumen von Säugetieren genutzt, um die Auswirkungen von Landnutzung pro 900m-Rasterzelle zu erfassen. Wir verglichen dabei Szenarien von landwirtschaftlich genutzten Flächen sowohl mit einem natürlichen Referenzzustand als auch mit heutiger Landnutzung. Der resultierende Artenverlust wurde gewichtet mit der globalen Seltenheit einer Art und deren Gefährdungsstatus gemäss IUCN Einstufung. Dies ermöglichte Aussagen zum Risiko eines globalen Artenverlusts. Schliesslich wurden anhand der Methode die Landnutzungseffekte von Kaffee-, Tee- und Tabakanbau in Ostafrika quantifiziert und mit den Ergebnissen der Methoden aus Kapitel 3 und 5 verglichen.
In Kapitel 7 werden die neuen Methoden mit bestehenden Ansätzen verglichen, die für eine globale Anwendung geeignet wären und Empfehlungen zur Anwendung und Weiterentwicklung der Methoden gegeben. Die relative, lokale Methode aus Kapitel 3 sollte nur in Verbindung von regionalen Gewichtungsfaktoren (Kapitel 4) angewendet werden. Die Methode aus Kapitel 5 ist direkt global anwendbar, umfasst Effekte auf fünf taxonomische Gruppen und Aspekte der Reversibilität. Falls die Herkunft von Produkten genau bekannt ist, sollte die Methode aus Kapitel 5 mit der Methode aus Kapitel 6 ergänzt werden, welche die Gefährdung seltener und bedrohter Säugetierarten durch Landnutzung mit hoher örtlicher Auflösung erfassen kann. Dafür sollte diese Methode, die bisher nur für Ostafrika berechnet wurde, für alle Weltregionen berechnet werden.
Diese Arbeit soll einen massgebenden Beitrag leisten zur Verbesserung der Bewertungsmethoden für Landnutzung innerhalb von Ökobilanzen. Für eine räumlich differenzierte Modellierung von globalen Biodiversitätsverlusten in Ökobilanzen sind weitere Verbesserungen nötig. Die Weiterentwicklung und Anwendung der Methoden sollte schliesslich dazu beitragen, die Auswirkungen von ökonomischen Aktivitäten auf die Biodiversität besser bewerten zu können und eine Verlagerungen von Umweltauswirkungen in andere Weltregionen zu verhindern.
8 Table of contents
Remarks ...... 4
Summary ...... 5
Zusammenfassung ...... 7
1 Introduction ...... 15
1.1 Problem statement ...... 16
1.2 Research gaps ...... 21
1.3 Goal and research questions of thesis ...... 22
1.4 Approach of the thesis ...... 22
1.5 References ...... 25
2 Toward meaningful endpoints of biodiversity in Life Cycle Assessment ...... 31
2.1 Introduction ...... 32
2.2 Biodiversity indicators ...... 33
2.3 Assessing biodiversity loss in LCA ...... 36
2.3.1 Land use ...... 36
2.3.2 Water use ...... 37
2.3.3 Climate change ...... 37
2.3.4 Acidification and eutrophication...... 38
2.3.5 Ecotoxicity ...... 38
2.4 Endpoint unit, scale, and linearity ...... 41
2.5 Use of indicators ...... 41
2.5.1 Genetic component ...... 42
2.5.2 Species and community component ...... 42
2.5.3 Ecosystem and landscape component ...... 42
2.5.4 Biological attributes ...... 43
2.5.5 Multiple impact factors ...... 44
2.6 Taxonomic and geographical coverage ...... 44
2.6.1 Taxonomic coverage ...... 44
2.6.2 Geographic coverage ...... 45
2.7 Research outlook ...... 45
9 2.8 Acknowledgements ...... 46
2.9 References ...... 46
3 Land use impacts on biodiversity in LCA: a global approach ...... 53
3.1 Introduction ...... 55
3.2 Methods ...... 56
3.2.1 Calculation of characterization factors ...... 56
3.2.2 Reference situation ...... 57
3.2.3 Data sources ...... 57
3.2.4 Indicator selection and calculation ...... 58
3.2.5 Statistical analysis ...... 60
3.3 Results ...... 61
3.3.1 Land use impacts on biodiversity ...... 61
3.3.2 Regionalization ...... 63
3.3.3 Indicator comparison ...... 65
3.4 Discussion ...... 67
3.4.1 Choice of indicator ...... 68
3.4.2 Taxonomic coverage ...... 70
3.4.3 Land use classification and regionalization ...... 71
3.4.4 Data limitations and uncertainties ...... 72
3.5 Conclusions and recommendations ...... 73
3.6 References ...... 74
4 Comparing direct land use impacts on biodiversity of conventional and organic milk – based on a Swedish case study ...... 79
4.1 Introduction ...... 81
4.2 Material and Methods ...... 82
4.2.1 Description of compared agricultural systems ...... 82
4.2.2 Land use assessment framework ...... 83
4.2.3 Inventory Analyses ...... 84
4.2.4 Impact Assessment ...... 85
4.3 Results ...... 88
4.3.1 Characterization factors for Biodiversity Damage Potential ...... 88
10 4.3.2 Impact of occupation ...... 89
4.3.3 Impact of transformation ...... 92
4.3.4 Total Biodiversity Damage Potential ...... 93
4.4 Discussion ...... 95
4.4.1 Sensitivity analyses ...... 95
4.4.2 Quantification of biodiversity impacts ...... 98
4.4.3 Calculation of transformed area ...... 98
4.4.4 Indicator for regeneration time ...... 99
4.4.5 Biodiversity weighting system ...... 100
4.4.6 Transferability to other temperate countries ...... 101
4.5 Conclusions ...... 102
4.6 Acknowledgements ...... 102
4.7 References ...... 103
5 Land use in life cycle assessment: Global characterization factors based on regional and global potential species extinction ...... 109
5.1 Introduction ...... 110
5.2 Materials and Methods ...... 112
5.2.1 Modeling species extinction ...... 112
5.2.2 Calculation of characterization factors ...... 113
5.2.3 Input data for model parameters ...... 115
5.2.4 Uncertainty assessment ...... 116
5.2.5 Validation of species extinction ...... 116
5.2.6 Comparison of model choices ...... 117
5.3 Results ...... 117
5.3.1 Regional characterization factors ...... 117
5.3.2 Contribution to uncertainty ...... 119
5.3.3 Model evaluation ...... 119
5.3.4 Comparison of model choices ...... 120
5.4 Discussion ...... 120
5.4.1 Modeling choices ...... 120
5.4.2 Data availability and uncertainty ...... 121
11 5.4.3 Model validity ...... 122
5.4.4 Applicability ...... 123
5.4.5 Implications ...... 124
5.5 References ...... 124
6 Quantifying biodiversity loss due to export-crops in East Africa using life cycle assessment ...... 129
6.1 Introduction ...... 130
6.2 Methods ...... 133
6.2.1 Basic structure of LCA ...... 133
6.2.2 Life cycle inventory analysis ...... 133
6.2.3 Impact assessment methods ...... 133
6.3 Results ...... 136
6.4 Discussion ...... 140
6.5 Conclusions ...... 141
6.6 Acknowledgements ...... 141
6.7 References ...... 141
7 Concluding remarks ...... 145
7.1 Scientific relevance and conclusions ...... 146
7.1.1 Conceptual shortcomings of biodiversity modeling in LCA ...... 146
7.1.2 How should we assess land use impacts on biodiversity within LCA on a global scale? ...... 147
7.1.3 Conclusions and new aspects of the thesis ...... 157
7.1.4 Limitations and uncertainties ...... 158
7.2 Practical relevance ...... 160
7.3 Future research needs ...... 161
7.4 References ...... 164
Appendices ...... 167
Appendix A: Appendix to Chapter 2 ...... 168
A.1 Genetic indicators ...... 168
A.2 Species-based indicators ...... 169
A.3 Community indicators ...... 171
12 A.4 Species or community indicators? ...... 174
A.5 Ecosystem and landscape indicators ...... 174
A.6 Integrative indicators ...... 176
A.7 The use of indicators in biodiversity assessments ...... 177
A.7.1 The Millennium Ecosystem Assessment ...... 177
A.7.2 IMAGE 2.4 and the GLOBIO3 model ...... 179
A.7.3 The Natural Capital Project’s InVEST modelling tool ...... 179
A.7.4 The BioScore tool for European biodiversity assessment ...... 180
A.8 Representing the missing drivers of biodiversity loss ...... 180
A.8.1 Biotic depletion ...... 180
A.8.2 Biotic homogenization ...... 181
A.9 References ...... 181
Appendix B: Appendix to Chapter 3 ...... 189
B.1 References ...... 200
Appendix C: Appendix to Chapter 4 ...... 201
B.1 Full bibliography for characterization factor calculation ...... 222
Appendix D: Appendix to Chapter 5 ...... 226
D.1 Land use assessment framework ...... 226
D.2 Input data ...... 227
D.3 Additional analysis and results ...... 232
D.3.1 Correlation of characterization factors ...... 232
D.3.2 Maps of median characterization factors ...... 234
D.3.3 Contribution to variance (CTV) ...... 239
D.3.4 Comparing model choices ...... 241
D.3.5 Model evaluation ...... 246
D.3.6 Aggregation of characterization factors across taxa ...... 247
D.3.7 Calculation of world average characterization factors ...... 247
D.4 References ...... 248
Appendix E: Appendix to Chapter 6 ...... 249
E.1 Methods ...... 249
E.1.1 Supplementary methods E1: Land use assessment framework in LCA ...... 249
13 E.1.2 Supplementary methods E2: Local, relative method (R-Local) ...... 251
E.1.3 Supplementary methods E3: Regional, absolute method (A-Regional) ...... 251
E.1.4 Supplementary methods E4: Calculation of life cycle inventory data ...... 253
E.1.5 Supplementary methods E5: Calculation of recovery times of biodiversity ...... 259
E.2 Additional results ...... 263
E.2.1 Supplementary results E6: Biodiversity impacts (characterization factors) of all methods ...... 263
E.2.2 Supplementary results E7: Correlation analysis of all methods ...... 265
E.2.3 Supplementary results E8: Unweighted local species loss (UW-Local) results ...... 266
E.3 References ...... 267
Publications and presentations ...... 270
Acknowledgments ...... 272
Curriculum Vitae ...... 273
14
Chapter 1
Introduction
Introduction
1.1 Problem statement
Since the onset of the Industrial Revolution in the mid-18th century, the global human population rapidly increased from about 0.8 billion to 7 billion people and is projected to further increase to 9 billion by 2050. At the same time, the use of fossil fuels increased astronomically, providing the energy for the Industrial Revolution but also allowing mechanization of agriculture to feed the rapidly growing population (Zalasiewicz et al. 2010). Meanwhile, human activities became the main driver of global environmental change, not only changing the climate but also altering nutrient and water cycles and drastically increasing the rate of species extinction (Rockstrom et al. 2009a; Rockstrom et al. 2009b). We are now even considered to live in a new human- dominated geological era: the “Anthropocene” (Crutzen 2002).
Agriculture has largely contributed to this planetary change. About 40% of the global land surface has been altered for agricultural purposes, resulting in loss, modification, and fragmentation of ecosystems. This agricultural land use is thus one of the major causes of global biodiversity loss (Sala et al. 2000). Agriculture is also responsible for about 30% of global greenhouse gas emissions, mainly caused by tropical deforestation, methane emissions from livestock and rice cultivation, and nitrous oxide emissions from fertilized soils (Foley et al. 2011). Of the accessible freshwater, more than 30% is withdrawn for agricultural purposes (Foley et al. 2011). In addition, agriculture strongly altered global nitrogen and phosphorous cycles: More nitrogen is applied as fertilizer than the global natural fixation, and the inflow of phosphorus from rivers to the oceans is eight times higher than the natural background rate (Rockstrom et al. 2009a). With the increasing global population and changing diets (especially increased meat consumption), the environmental impacts related to agriculture are projected to increase even more in the future (Tilman et al. 2001).
Although agriculture dominates large parts of our planet, every eighth person (or 0.87 billion) is chronically undernourished as they do not have sufficient access to food, 25% of all children are malnourished, and more than 30% have a micronutrient deficiency. At the same time, the global prevalence of obesity among adults and children is 12% (FAO 2013) and 3% (Lobstein et al. 2004), respectively. Of the globally produced crops, only 62% are directly allocated to human food; about 35% are used for animal feed (which is a much less efficient way to produce food for humans) and 3% for biofuels and other non-food products (Foley et al. 2011). In addition, about one third of all food is wasted along the value chain (Gustavsson et al. 2011).
In light of these enormous contrasts and challenges, it becomes clear that there needs to be a transition in the way we produce, consume, and distribute food and other agricultural products. In the developed world, the proposed solutions to the environmental problems are manifold. Some opt for reducing external inputs (e.g., organic farming or land-sharing) while others suggest maximizing yields to minimize the area requirements of cultivation (also referred to as land- sparring; e.g., Phalan et al. 2011; Tscharntke et al. 2012). Other proposed solutions aim at increased efficiency (increasing the agricutlural output per external input or sustainable
16 Introduction intensification; e.g., Tilman et al. 2011) or to change consumption patterns (by e.g. avoiding waste or reducing meat consumption (e.g., Meier and Christen 2013)], referred to as sufficiency).
A large body of scientific literature has contributed to better understanding the trade-offs and links between the environmental services we derive from ecosystems (i.e., provisioning ecosystem services, such as food, fuels, fibers, or clean water) and the maintenance of ecosystem functions and processes (i.e., regulating services such as regulation of climate, water and nutrient cycles or diseases; see e.g., Millennnium Ecosystem Assessment 2005; Elmqvist et al. 2011). Most studies analyzing the trade-offs between agricultural production and conservation of ecosystem services or biodiversity focused on regional scales (e.g., Steffan-Dewenter et al. 2007; Polasky et al. 2008; Nelson et al. 2009; Raudsepp-Hearne et al. 2010; Ayanu 2011; Qiu and Turner 2013). However, in our globalized world, environmental impacts often reach beyond single regions, because products are imported from other regions (Fader et al. 2013), where a large share of the impacts occur (referred to as environmental leakage; Ghertner and Fripp 2007). Several studies have shown the important contribution of international trade to deforestation (Mills Busa 2013), biodiversity loss, (Polasky et al. 2004; Lenzen et al. 2012), and land use (Moran et al. 2009).
Several methods have been developed to analyze aspects of environmental leakage (see Kissinger and Rees 2010 for a review of approaches). Material or substance flow analysis assesses the flows and stocks of materials, energy, or substances within a defined space and time and can contribute to recognize depletion of resources or excess environmental loads of substances (Brunner and Rechberger 2004). Virtual land or water flow assessments consider the amount of land or water used to produce products in foreign countries, finally leading to “importing” this land or water via the products (Fader et al. 2011). Ecological footprint analysis goes a step further and assesses the biologically productive land area necessary to produce consumed goods and assimilate waste (Wackernagel and Rees 1996). Local consumption is then compared to local bio-capacity and thus, the dependency of populations on distant ecosystems can be analyzed. The environmental input–output analysis is a top-down technique for analyzing the material and energy inputs and emissions of economic sectors of regions and allows pollution or resource use to be attributed to final demand (Wiedmann 2009). In recent years, multi-region input–output models were developed that couple regional input–output models and analyze the interdependencies with economic sectors of other regions (Wiedmann 2009). In contrast, life- cycle assessment (LCA) is a bottom-up modeling technique, which quantifies all inputs and emissions over the entire life-cycle of products, from the extraction of raw material to the production phase, use, and disposal (i.e., from cradle to grave; ISO 2006). LCA aims for an overall environmental assessment, characterizing the damage caused by inputs and emissions to multiple environmental compartments (such as climate change, water pollution, depletion of non-renewable resources, impacts of land use, acidification, eutrophication, human toxic emissions, etc.). The added value of LCA over other decision-support tools is that it can detect and help to avoid a burden shifting between life-cycle stages, environmental impacts, or regions (Loiseau et al. 2012). In addition, LCA relates impacts to the service provided by the system, i.e., the functional unit (e.g., the environmental impacts of producing one kg of maize, one meal, or
17 Introduction
2000 kcal). This has the advantage, that alternative options are compared on a common basis. However, applying LCA for agricultural products also provides a challenge because a mostly multifunctional agricultural system has to be reduced to one or a few functions (Nemecek et al. 2011; Loiseau et al. 2013).
For application in the agricultural sector, the LCA method should ideally capture all relevant environmental issues related to agriculture, such as greenhouse gas emissions and impacts related to land, water, fertilizer, and pesticide use. However, methods for capturing these environmental impacts in LCA are at different stages of development. For example, no consensus exists on how the impacts of land use on biodiversity loss should be quantified. Many LCA studies therefore assess land use impacts only qualitatively or as simple area use (e.g., Cederberg and Mattsson 2000; Halleux et al. 2008; Ntiamoah and Afrane 2008; Gmuender et al. 2010; Trydeman Knudsen et al. 2010), and do not capture the different impacts caused by the land (e.g., the associated biodiversity, which strongly depends on where the land is located or how the land is managed). Because land use is the main driver of global biodiversity loss (Sala et al. 2000) that should be considered in LCA studies, this thesis aimed at further developing such methods.
Assessing land use impacts in the LCA framework poses challenges. LCA typically has a static and linear structure and is site-generic (Udo de Haes et al. 2004), while impacts of land use change over time and can be non-linear and spatially dependent. A general framework for assessing land use impacts on LCA was proposed in the SETAC (Society of Environmental Toxicology and Chemistry) Working Group on Impact Assessment in 1996 (Finnveden 1996). The framework was further developed in the UNEP/SETAC (United Nations Environmental Program) life-cycle initiative phase I (2002-2007; Milà i Canals et al. 2007a) and phase II (2007-2012; Koellner et al. 2013). The general idea was to assess different types of land use activities, mainly distinguishing between the actual land use phase (i.e., land occupation) and the preparation of the land to suit the activity (i.e., land transformation; see Figure 1.1). The latter can cause permanent impacts, if the impact of the land transformation is not fully reversible. Impacts are assessed on three axes: (i) reduction in ecosystem quality, (ii) duration of the impact, and (iii) area affected by the land use activity. Ecosystem quality was defined by Koellner et al. (2013; p. 1190) as “the capability of an ecosystem (or a mix of ecosystems at the landscape scale) to sustain biodiversity and to deliver services to the human society.” Such services encompass provisioning (e.g., food, fiber, fuel), regulating (e.g., regulation of water cycles), cultural (e.g., aesthetic, recreation), and supporting services (e.g. soil formation; Millennnium Ecosystem Assessment 2005). The decrease in ecosystem quality is assessed by comparing the quality of a land use system with the quality of a reference situation. This reference situation can either be potential natural vegetation (PNV), the (quasi-) natural land cover in each biome/ecoregion or the current mix of land uses (Koellner et al. 2013).
18 Introduction
A" B"
Ecosystem)Quality! Ecosystem)Quality! Permanent)impact!! CF I Permanent)impact'' Perm! Trans!
Qref,1' I Occ! ITrans! Qref,2'
Occupa&on)) Occupa&on)) CFTrans! Transforma&on) impact!! Transforma&on) impact'' CF )impact) Occ! )impact) Qi' Land)use) Recovery) phase) phase) Land'use' Area! Land' Area! change' abandon7 ment'
Time) Time) t ' t t t 0 t1' i,'reg' 2' m' Figure 1.1 A: Land use assessment framework (adapted from Milà i Canals et al. 2007a; Koellner et al. 2013). B: Land use impact calculation. Q: Ecosystem quality; ref: reference situation; i: Land use type: t: time; reg: regeneration; CF: characterization factor; I: Inventory flow; Occ: Occupation; Trans: Transformation; Perm: Permanent
A simplified form of the LCA land use assessment framework is shown in Figure 1.1.A. The orange line indicates changes in ecosystem quality over time. At time t0, the land is transformed from a reference situation Qref,1 to a land use with ecosystem quality Qi. The continued land use activity during the land use phase keeps the ecosystem quality at a reduced level. The framework assumes that at one point in the future (t1) the land will be abandoned. Thereafter, a natural recovery of the ecosystem quality takes place, which gradually increases the ecosystem quality of the land. At t2, a new stable state is reached, Qref,2, which can be of similar or lower quality than the initial reference situation. Based on this framework, the three impact types of land use are calculated as the volume shown in Figure 1.1.B. The inventory flow of occupation, IOcc, is given as area * time, and is multiplied with the characterization factor CFOcc given as the difference between the ecosystem quality of the reference state and the used land. The transformation inventory flow is given as the transformed area and is multiplied by the characterization factor for transformation, CFTrans, which is derived by multiplying the change in ecosystem quality and the time required for the ecosystem to recover. Permanent impacts are calculated based on ITrans multiplied by the characterization factor CFPerm given as the difference between the ecosystem quality of the initial reference situation, Qref,1, and the equilibrium situation after recovery, Qref,2, multiplied by a certain modeling time, tm.
Many studies have suggested approaches for operationalizing this LCA land use assessment framework. The proposed measures for ecosystem quality are spread along the cause-effect chain of land use impacts on biodiversity and ecosystem services (see e.g. Fig. 4 in Koellner et al. 2013). Measures of physical-chemical soil conditions (e.g. soil organic matter or soil organic carbon, SOM/SOC, Milà i Canals et al. 2007b; Brandão and Milà i Canals 2013) are more directly linked to land use than midpoint indicators of soil ecological quality indicators, such as soil erosion, as well as water purification, regulation, and cycling (Baitz et al. 2000; Beck et al. 2010; Saad et al. 2011; Saad et al. 2013). Two studies proposed a set of indicators for soil fertility (Mattsson et al. 2000) and soil quality and quantity (Cowell and Clift 2000). Other measures,
19 Introduction situated more at the midpoint, include carbon sequestration potential (Müller-Wenk and Brandão 2010), net primary productivity (Lindeijer 2000; Weidema and Lindeijer 2001; Pfister et al. 2011), and desertification of dry lands (Núñez et al. 2010; Civit et al. 2013). At the endpoint, emergy, i.e., the solar energy required to build up the soil, has been proposed as a measure of soil erosion caused by land use (Núñez et al. 2013). Cao et al. (2012) suggested the economic value of ecosystem services as an endpoint unit for land use impacts. Achten et al. (2008) proposed six midpoint indicators (soil fertility, soil structure, biodiversity, biomass production, vegetation structure, and on-site water balance) and two endpoint indicators (ecosystem structural quality and ecosystem functional quality).
Many studies focused on the effects of land use on biodiversity in LCA, mostly quantified in terms of species diversity. Some authors also quantified differences in functional diversity (de Souza et al. 2013), naturalness (or hemeroby; Brentrup et al. 2002), and indirect measures of biodiversity (i.e., conditions for maintained diversity, Michelsen 2008). Most studies were developed based on data available in one particular region, and the results therefore cannot be easily extrapolated to other regions. The studies focused on Central and Northern Europe (Koellner 2000; Vogtländer et al. 2004; Kyläkorpi et al. 2005; Koellner and Scholz 2007, 2008; Michelsen 2008; Schmidt 2008; Jeanneret et al. 2009; De Schryver et al. 2010; Urban et al. 2012), but studies also exist for Malaysia/Indonesia (Schmidt 2008), Japan (Itsubo and Inaba 2012), California (Geyer et al. 2010), Canada (Toffoletto et al. 2007), and selected regions in North, Central, and South America (de Souza et al. 2013). The methods of Kyläkorpi et al. (2005) and of Michelsen (2008), both originally developed for a Scandinavian context, have been tested in the Namib desert in Africa (Burke et al. 2008) and in New Zealand (Coelho and Michelsen 2013), respectively. The latter is based on globally available data, implementing an early proposal by Weidema and Lindeijer (2001) to use globally available data on ecosystem vulnerability (i.e., how much natural area of the ecosystem type remains) and scarcity (i.e., how rare is the ecosystem type) to weigh the relative reduction in local biodiversity. Both studies did not quantify the relative reduction in local biodiversity based on empirical data, but relied on rough estimates (Weidema and Lindeijer 2001) and hemeroby values proposed by Brentrup (2002) (Coelho and Michelsen 2013), respectively. Another early idea for a globally applicable method was to use global plant species richness maps as the reference state (see also Koellner and Scholz 2008). Again, quantifying the impacts for specific land use situations remained a challenge and was illustrated only for specific cases in Europe and South America (Lindeijer 2000).
Outside LCA, several authors have developed potentially globally applicable models for assessing land use impacts on biodiversity, which can serve as the basis for developing methods in LCA. The GLOBIO3 model assesses biodiversity loss based on simple cause-effect relationships between environmental drivers and biodiversity impacts (Alkemade et al. 2009). Impacts of land cover change, land-use intensity, fragmentation, climate change, atmospheric nitrogen deposition, and infrastructure development are measured as the relative change in the mean abundance of original species relative to their abundance in undisturbed ecosystems (MSA). Data on cause- effect relationships between drivers and biodiversity loss were derived from a global literature review. The biodiversity intactness index developed by Scholes and Biggs (2005) for southern
20 Introduction
Africa assesses the impacts of different land use activities on the population sizes of species groups, deriving impact scores from expert interviews. The global mammals assessment (GMA) is based on species-specific habitat suitability models, using available knowledge on the species distribution in space and elevation, water and land cover type dependency, and tolerance to human disturbance (Rondinini et al. 2011). The BioScore model applies a similar approach, using species sensitivity scores of selected species (Louette et al. 2010), based on European data. Another modeling approach is based on species-area relationship models (SAR), which assess the decrease in regional species richness based on the regional reduction in natural habitat area. Koh and Ghazoul (2010) and Pereira and Daily (2006) proposed two adaptations of this model, assessing regional species losses due to habitat conversions but also accounting for the fraction of species surviving on human-modified land.
1.2 Research gaps
• Although many studies on measuring land use impacts on biodiversity in LCA exist, a generic and operational method for comparing land use impacts across global regions is still missing. This is especially important in our increasingly globalized economy, where the value chains of products are often globally distributed.
• Spatially differentiated methods, covering multiple taxonomic groups and assessing impacts at different spatial scales, are lacking. Such methods should represent the spatial heterogeneity of biodiversity and provide a broader understanding of how land use affects biodiversity at different spatial scales and how multiple taxonomic groups are affected.
• Although the land use framework conceptually covers irreversible land use impacts, no method has been proposed that quantifies permanent land use impacts on biodiversity. In addition, the representation of non-linear effects on a global scale is still very limited.
• Studies assessing the results of land use based on different reference situations are missing. This would allow a better understanding of how the choice of the reference situation influences results.
• Uncertainties of globally applicable methods can be large, but have thus far not been quantified.
• Finally, case studies illustrating the usefulness of global methods with regional differentiation are missing. Specifically, studies are lacking that inform the debate on the impacts of organic and conventional farming on biodiversity for products, which include life-cycle stages on different continents.
21 Introduction
1.3 Goal and research questions of thesis
The main goal of this thesis was to develop meaningful, operational, and globally applicable life- cycle impact assessment methods for land use impacts on biodiversity. These methods should be spatially differentiated, encompass multiple taxonomic groups, and measure impacts at different spatial scales.
To reach this goal, four main research questions were asked:
1. How can the LCA research community learn from conservation science, and how can we meaningfully transform ecological models into LCA, including the non-linearity and reversibility of impacts?
2. How can we use existing global datasets to meaningfully quantify biodiversity impacts related to land use within life-cycle assessment, and how can we quantify the uncertainties?
3. How can we compare the impacts of different land use types in different world regions, and how can we differentiate the impacts of low- and high-input agricultural systems?
4. At which scale should impacts on biodiversity be assessed and based on which indicators, taxonomic groups, and reference state?
1.4 Approach of the thesis
This thesis is a cumulative Ph.D. thesis, consisting of five research papers, presented in chapters 2 through 6. A graphical overview of the thesis is given in Figure 1.2. This thesis begins with an evaluation of how biodiversity is assessed within LCA and the main conceptual shortcomings (Chapter 2). Chapters 3, 5, and 6 present three novel globally applicable methods, how land use impacts on biodiversity can be quantified within LCA. Chapter 4 further elaborates the method from Chapter 3 to compare organic and conventional production. In Chapter 6, the three methods are compared in three case studies. Chapter 7 summarizes the findings and gives recommendations for future research.
22 Introduction
Critical evaluation of biodiversity assessment in LCA Application to case studies Review of indicators and approaches to model biodiversity loss in LCA Identification of conceptual shortcomings, Chapter 2: Toward meaningful endpoints of biodiversity in Life Cycle Assessment
Development of new land use assessment methods
Weighted relative local Organic / impacts convent. Milk plants, per biome, comparison of Chapter 4: Comparing LU intensity, Chapter 4: direct LU impacts on Comparing direct LU impacts on biodiversity of convent. biodiversity of convent. and and organic milk – organic milk – based on a Swedish based on a Swedish case study case study
Relative local impacts Absolute regional all taxa, per biome, comparison impacts of biodiversity indicators,
data data five taxa, per ecoregion, Chapter 3: Land use impacts on reversible and irreversible biodiversity in LCA: a global impacts, Chapter 5: LU in LCA: approach Global characterization factors Coffee, tea, based on regional and global tobacco from potential species extinction East Africa Chapter 6: Quantifying biodiversity loss due to Weighted absolute local Export-Crops in East Africa using life cycle impacts assessment mammals, per grid-cell, comparison of reference situations, Chapter 6: Quantifying biodiversity loss due to export- crops in East Africa using life cycle assessment
Figure 1.2. Overview of the chapters of this thesis, taxonomic groups considered in the methods, spatial resolution, and special contribution and title of the chapters
In Chapter 2, we review the use of indicators and approaches to model biodiversity loss in LCA. We find serious conceptual shortcomings in the way models are constructed. Considerations of the scale of impacts are largely absent, as well as spatial differentiation of models. Most models are restricted to one or a few geographic regions. Indicators mostly focus on the species richness of one taxonomic group, but the functional and structural attributes of biodiversity are largely neglected. In addition, only three of the five drivers of biodiversity loss as identified by the Millennium Ecosystem Assessment are represented in current impact categories (habitat change, climate change, and pollution), while two are missing (invasive species and overexploitation). We discuss these issues and make recommendations for future research to better reflect biodiversity loss in LCA.
23 Introduction
In Chapter 3, we propose a novel, globally applicable assessment method for land use impacts. We combine data from a quantitative literature review, conducted for the GLOBIO3 model (Alkemade et al. 2009), with national biodiversity monitoring data (BDM 2004). The relative change in local species richness between the land used and a semi-natural reference state in the same region are used to derive biome-specific characterization factors for land occupation. Land use impacts are compared across taxonomic groups, land use types, regions and data sources. In addition, results based on species richness are compared with other biodiversity indicators. The type of land use, the region where the land use is taking place, the taxonomic group used for the analysis, and the selection of the biodiversity indicator influenced the result, and thus the numerical value of the characterization factors finally used for LCA studies.
In Chapter 4, we further develop the methodology presented in Chapter 3 to quantify and compare the direct land use impacts on biodiversity of organic and conventional production of milk. We calculate spatially differentiated land occupation and transformation characterization factors for organic and conventionally farmed land based on the relative difference between plant species richness on agricultural land compared to a (semi) natural regional reference. To account for differences in biodiversity value between regions, we weigh characterization factors based on species richness, vulnerability, and irreplaceability. The method has been applied to a case study of milk production in Sweden (Cederberg and Mattsson 2000). Although 1 liter of organic milk required about twice as much land as conventional milk, the organic milk still had lower direct land use impacts on biodiversity.
In Chapter 5, the impacts of accumulated land use activities within each ecoregion are modeled on a regional scale. An adapted (matrix-calibrated) species-area relationship, developed by Koh and Ghazoul (2010), is used to model the regional impacts of land use on mammals, birds, amphibians, reptiles, and plants. Potentially reversible impacts of land occupation and transformation are quantified as lost non-endemic species. Irreversible, permanent impacts are assessed based on potentially lost endemic species. Species loss was highest in regions where most of the natural habitat had already been converted. In addition to this retrospective approach, alternative modeling options for prospective assessments are illustrated, which are especially important for assessing land use impacts in regions with high current and future land use change rates. The sensitivity and robustness of the results are tested and discussed.
In Chapter 6, habitat suitability models from the global mammals assessment (Rondinini et al. 2011) are used to derive weighted local, absolute occupation, and transformation characterization factors for mammals in East Africa. Species-specific habitat requirements are used to calculate the potential species loss on a grid cell level for agricultural land use, comparing agricultural land with two reference scenarios (i.e., natural land and current land cover). Species loss is weighted by the species global rarity and threat level. Finally, the method is applied to case studies of tea, coffee, and tobacco production in East Africa, and the results are compared to the outcomes for the methods developed in Chapters 3 and 5. The weighted local impacts (Chapter 6) highlight areas of immanent extinction risk, where agricultural production regions overlap with the last remaining habitat of very range-restricted species. The regional method developed in
24 Introduction
Chapter 5 complements this assessment and highlights the impacts on regions with high land use pressure. The local method developed in Chapter 3 has the weakest link to conservation targets and the other two methods (developed in Chapters 5 and 6) should be preferred.
In Chapter 7, we synthesize the main findings, provide guidance on the application of the different methods developed within this thesis, illustrate the practical relevance of the results, and provide an outlook on future research needs.
1.5 References
Achten WMJ, Mathijs E, Muys B Proposing a life cycle land use impact calculation methodology. In: 6th International Conference on LCA in the Agri-Food Sector, Zurich, Nov 1 2008. Alkemade R, van Oorschot M, Miles L, Nellemann C, Bakkenes M, ten Brink B (2009) GLOBIO3: A framework to investigate options for reducing global terrestrial biodiversity loss. Ecosystems 12 (3):374-390. Ayanu YZ, Trung Thanh Nguyen, Carsten Marohn, and Thomas Koellner (2011) Crop Production Versus Surface-Water Regulation: Assessing Tradeoffs for Land-Use Scenarios in the Tat Hamlet Watershed, Vietnam. International Journal of Biodiversity Science, Ecosystem Services & Management 7 (3):231–244. Baitz M, Kreißig J, Wolf M (2000) Methode zur Integration der Naturraum-Inanspruchnahme in Ökobilanzen. Forstwirtschaftliches Centralblatt 119:128-149. BDM (2004) Biodiversity Monitoring Switzerland. Indicator Z9: Species Diversity in Habitats. Bundesamt für Umwelt, BAFU. http://www.biodiversitymonitoring.ch. Accessed 1.2.2011 Beck T, Bos U, Wittstock B, Baitz M, Fischer M, Sedlbauer K (2010) LANCA® Land Use Indicator Value Calculation in Life Cycle Assessment – Method Report. Fraunhofer Verlag, Stuttgart. Brandão M, Milà i Canals L (2013) Global characterisation factors to assess land use impacts on biotic production. The International Journal of Life Cycle Assessment 18 (6):1243-1252. Brentrup F, Kusters J, Lammel J, Kuhlmann H (2002) Life Cycle Impact assessment of land use based on the Hemeroby concept. The International Journal of Life Cycle Assessment 7 (6):339-348. Brunner P, Rechberger H (2004) Practical Handbook of Material Flow Analysis. CRC Press, Boca Raton, Florida Burke A, Kylakorpi L, Rydgren B, Schneeweiss R (2008) Testing a scandinavian biodiversity assessment tool in an African desert environment. Environmental Management 42 (4):698-706. Cao V, Margni M, Favis BD, Descheneŝ L Development of an integrated indicator for land use based on the economic value of ecosystem services. In: Corson MS, van der Werf HMG (eds) 8th International Conference on Life Cycle Assessment in the Agri-Food Sector, Saint Malo, France, 1-4 October 2012. INRA, Rennes, France. Cederberg C, Mattsson B (2000) Life cycle assessment of milk production — a comparison of conventional and organic farming. Journal of Cleaner Production 8 (1):49–60. Civit BM, Arena AP, Núñez M, Muñoz P, Antón A, Rieradevall J (2013) Assessing potential desertification environmental impact in life cycle assessment. Part 2: agricultural case study in Spain and Argentina. The International Journal of Life Cycle Assessment 18 (7):1302-1315. Coelho CRV, Michelsen O (2013) Land Use Impacts on Biodiversity From Kiwifruit Production in New Zealand Assessed with Global and National Datasets. The International Journal of Life Cycle Assessment in press.
25 Introduction
Cowell SJ, Clift R (2000) A methodology for assessing soil quantity and quality in life cycle assessment. Journal of Cleaner Production 8 (4):321-331. Crutzen PJ (2002) Geology of mankind. Nature 415 (6867):23-23. De Schryver AM, Goedkoop MJ, Leuven RSEW, Huijbregts MAJ (2010) Uncertainties in the application of the species area relationship for characterisation factors of land occupation in life cycle assessment. The International Journal of Life Cycle Assessment 15 (7):682-691. de Souza DM, Flynn DFB, Declerk F, Rosenbaum RK, de Melo Lisboa H, Koellner T (2013) Land use impacts on biodiversity: proposal of characterization factors based on functional diversity. The International Journal of Life Cycle Assessment 18 (6):1231-1242. Elmqvist T, Tuvendal M, Krshnaswamy J, Hylander K (2011) Managing Trade-Offs in Ecosystems Services. Implementation Working Paper 4. UNEP, Division of Environmental Policy, Nairobi, Kenya. Fader M, Gerten D, Krause M, Lucht W, Cramer W (2013) Spatial decoupling of agricultural production and consumption: quantifying dependences of countries on food imports due to domestic land and water constraints. Environmental Research Letters 8 (1):014046. Fader M, Gerten D, Thammer M, Heinke J, Lotze-Campen H, Lucht W, Cramer W (2011) Internal and External Green-Blue Agricultural Water Footprints of Nations, and Related Water and Land Savings Through Trade. Hydrology and Earth System Sciences 15 (5):1641–1660. FAO (2013) The State of Food and Agriculture. Food Systems for Better Nutrition. FAO, Rome. Finnveden G (1996) "Resources" and related impact categories. In: Udo de Haes H (ed) Towards a methodology for life cycle impact assessment. SETAC– Europe, Belgium. Foley JA, Ramankutty N, Brauman KA, Cassidy ES, Gerber JS, Johnston M, Mueller ND, OConnell C, Ray DK, West PC, Balzer C, Bennett EM, Carpenter SR, Hill J, Monfreda C, Polasky S, Rockstrom J, Sheehan J, Siebert S, Tilman D, Zaks DPM (2011) Solutions for a cultivated planet. Nature 478 (7369):337–342. Geyer R, Lindner JP, Stoms DM, Davis FW, Wittstock B (2010) Coupling GIS and LCA for biodiversity assessments of land use: Part 2: Impact assessment. The International Journal of Life Cycle Assessment 15 (7):692-703. Ghertner D, Fripp M (2007) Trading away damage: Quantifying environmental leakage through consumption-based, life-cycle analysis. Ecological Economics 63 (2-3):563-577. Gmuender SM, Zah R, Bhatacharjee S, Classen M, Mukherjee P, Widmer R (2010) Life cycle assessment of village electrification based on straight jatropha oil in Chhattisgarh, India. Biomass & Bioenergy 34 (3):347-355. Gustavsson J, Cederberg C, Sonesson U, van Otterdijk R, Meybeck A (2011) Global Food Losses and Food Waste. FAO, Rome. Halleux H, Lassaux S, Renzoni R, Germain A (2008) Comparative life cycle assessment of two biofuels ethanol from sugar beet and rapeseed methyl ester. The International Journal of Life Cycle Assessment 13 (3):184-190. ISO (2006) ISO 14040 International Standard. In: Environmental Management – Life Cycle Assessment – Principles and Framework. International Organisation for Standardization, Geneva, Switzerland. Itsubo N, Inaba A (2012) LIME2. LCIA method based on endpoint modeling. Newsletter. Life- Cycle Assessment Society of Japan. Jeanneret P, Baumgartner DU, Freiermuth-Knuchel R, Gaillard G (2009) Methode zur Beurteilung der Wirkung landwirtschaftlicher Aktivitäten auf die Biodiversität für Ökobilanzen (SALCA-Biodiversität). Agroscope Reckenholz-Tänikon, Zurich. Kissinger M, Rees WE (2010) An interregional ecological approach for modelling sustainability in a globalizing world--Reviewing existing approaches and emerging directions. Ecological Modelling 221 (21):2615–2623. Koellner T (2000) Species-pool effect potentials (SPEP) as a yardstick to evaluate land-use impacts on biodiversity. Journal of Cleaner Production 8:293-311.
26 Introduction
Koellner T, de Baan L, Beck T, Brandão M, Civit B, Margni M, Milà i Canals L, Saad R, de Souza DM, Müller-Wenk R (2013) UNEP-SETAC guideline on global land use impact assessment on biodiversity and ecosystem services in LCA. The International Journal of Life Cycle Assessment 18 (6):1188-1202. Koellner T, Scholz RW (2007) Assessment of land use impacts on the natural environment. Part 1: An analytical framework for pure land occupation and land use change. The International Journal of Life Cycle Assessment 12 (1):16-23. Koellner T, Scholz RW (2008) Assessment of land use impacts on the natural environment. Part 2: Generic characterization factors for local species diversity in Central Europe. The International Journal of Life Cycle Assessment 13 (1):32-48. Koh L, Ghazoul J (2010) A matrix-calibrated species-area model for predicting biodiversity losses due to land-use change. Conservation Biology 24 (4):994-1001. Kyläkorpi K, Rydgren B, Ellegård A, Miliander S (2005) The Biotope Method 2005: a Method to Assess the Impact of Land Use on Biodiversity. Vattenfall, Sweden. Lenzen M, Moran D, Kanemoto K, Foran B, Lobefaro L, Geschke A (2012) International trade drives biodiversity threats in developing nations. Nature 486 (7401):109-112. Lindeijer E (2000) Biodiversity and life support impacts of land use in LCA. Journal of Cleaner Production 8:313–319. Lobstein T, Baur L, Uauy R, IASO International Obesity TaskForce (2004) Obesity in Children and Young People: a Crisis in Public Health. Obesity Reviews 5 (1):4–85. Loiseau E, Junqua G, Roux P, Bellon-Maurel V (2012) Environmental assessment of a territory: An overview of existing tools and methods. Journal of Environmental Management 112 (C):213-225. Loiseau E, Roux P, Junqua G, Maurel P, Bellon-Maurel V (2013) Adapting the LCA framework to environmental assessment in land planning. The International Journal of Life Cycle Assessment 18 (8):1533-1548. Louette G, Maes D, Alkemade J, Boitani L, Knegt Bd, Eggers J, Falcucci A, Framstad E, Hagemeijer W, Hennekens S, Maiorano L, Nagy S, Serradilla A, Ozinga W, Schaminée J, Tsiaousi V, Tol S, Delbaere B (2010) BioScore-Cost-effective assessment of policy impact on biodiversity using species sensitivity scores. Journal for Nature Conservation 18 (2):142-148. Mattsson B, Cederberg C, Blix L (2000) Agricultural land use in life cycle assessment (LCA): case studies of three vegetable oil crops. Journal of Cleaner Production 8:283–292. Meier T, Christen O (2013) Environmental Impacts of Dietary Recommendations and Dietary Styles: Germany as an Example. Environmental Science & Technology 47 (2):877–888. Michelsen O (2008) Assessment of land use impact on biodiversity. The International Journal of Life Cycle Assessment 13 (1):22-31. Milà i Canals L, Bauer C, Depestele J, Dubreuil A, Freiermuth Knuchel R, Gaillard G, Michelsen O, Müller-Wenk R, Rydgren B (2007a) Key Elements in a Framework for Land Use Impact Assessment Within LCA. The International Journal of Life Cycle Assessment 12 (1):5-15. Milà i Canals L, Romanya J, Cowell S (2007b) Method for assessing impacts on life support functions (LSF) related to the use of 'fertile land' in Life Cycle Assessment (LCA). Journal of Cleaner Production, vol 15. Millennnium Ecosystem Assessment (2005) Ecosystems and human well-being: biodiversity synthesis. Washington, DC. Mills Busa JH (2013) Deforestation Beyond Borders: Addressing the Disparity Between Production and Consumption of Global Resources. Conservation Letters 6 (3):192–199. Moran D, Wackernagel M, Kitzes J, Heumann B, Phan D, Goldfinger S (2009) Trading spaces: Calculating embodied Ecological Footprints in international trade using a Product Land Use Matrix (PLUM). Ecological Economics 68 (7):1938-1951. Müller-Wenk R, Brandão M (2010) Climatic impact of land use in LCA-carbon transfers between vegetation/soil and air. The International Journal of Life Cycle Assessment 15 (2):172-182.
27 Introduction
Nelson E, Mendoza G, Regetz J, Polasky S, Tallis H, Cameron DR, Chan KMA, Daily GC, Goldstein J, Kareiva PM, Lonsdorf E, Naidoo R, Ricketts TH, Shaw MR (2009) Modeling multiple ecosystem services, biodiversity conservation, commodity production, and tradeoffs at landscape scales. Frontiers in Ecology and the Environment 7 (1):4-11. Nemecek T, Dubois D, Huguenin-Elie O, Gaillard G (2011) Life cycle assessment of Swiss farming systems: I. Integrated and organic farming. Agricultural Systems 104:217-232. Ntiamoah A, Afrane G (2008) Environmental impacts of cocoa production and processing in Ghana: life cycle assessment approach. Journal of Cleaner Production 16 (16):1735-1740. Núñez M, Antón A, Muñoz P, Rieradevall J (2013) Inclusion of soil erosion impacts in life cycle assessment on a global scale: application to energy crops in Spain. The International Journal of Life Cycle Assessment 18 (4):755–767. Núñez M, Civit B, Muñoz P, Arena A, Rieradevall J, Antón A (2010) Assessing potential desertification environmental impact in life cycle assessment. The International Journal of Life Cycle Assessment 15 (1):67-78. Pereira H, Daily G (2006) Modeling biodiversity dynamics in countryside landscapes. Ecology 87:1877:1885. Pfister S, Bayer P, Koehler A, Hellweg S (2011) Environmental Impacts of Water Use in Global Crop Production: Hotspots and Trade-Offs with Land Use. Environmental Science & Technology 45 (13):5761–5768. Phalan B, Onial M, Balmford A, Green RE (2011) Reconciling food production and biodiversity conservation: land sharing and land sparing compared. Science 333:1289-1291. Polasky S, Costello C, McAusland C (2004) On trade, land-use, and biodiversity. Journal of Environmental Economics and Management 48 (2):911-925. Polasky S, Nelson E, Camm J, Csuti B, Fackler P, Lonsdorf E, Montgomery C (2008) Where to Put Things? Spatial Land Management to Sustain Biodiversity and Economic Returns. Biological Conservation 141 (6):1505–1524. Qiu J, Turner MG (2013) Spatial Interactions Among Ecosystem Services in an Urbanizing Agricultural Watershed. Proceedings of the National Academy of Sciences of the United States of America 110 (29):12149–12154. Raudsepp-Hearne C, Peterson GD, Bennett EM (2010) Ecosystem Service Bundles for Analyzing Tradeoffs in Diverse Landscapes. Proceedings of the National Academy of Sciences of the United States of America 107 (11):5242–5247. Rockstrom J, Steffen W, Noone K, Persson Å, Chapin FS, Lambin EF, Lenton TM, Scheffer M, Folke C, Schellnhuber HJ, Nykvist B, de Wit CA, Hughes T, van der Leeuw S, Rodhe H, Sörlin S, Snyder PK, Costanza R, Svedin U, Falkenmark M, Karlberg L, Corell RW, Fabry VJ, Hansen J, Walker B, Liverman D, Richardson K, Crutzen P, Foley JA (2009a) A safe operating space for humanity. Nature 461 (7):472-475. Rockstrom J, Steffen W, Noone K, Persson Å, Chapin FSI, Lambin E, Lenton TM, Scheffer M, Folke C, Schellnhuber HJ, Nykvist B, de Wit CA, Hughes T, van der Leeuw S, Rodhe H, Sörlin S, Snyder PK, Costanza R, Svedin U, Falkenmark M, Karlberg L, Corell RW, Fabry VJ, Hansen J, Walker B, Liverman D, Richardson K, Crutzen P, Foley J (2009b) Planetary Boundaries: Exploring the Safe Operating Space for Humanity. Ecology and Society 14 (2). Rondinini C, Di Marco M, Chiozza F, Santulli G, Baisero D, Visconti P, Hoffmann M, Schipper J, Stuart SN, Tognelli MF (2011) Global habitat suitability models of terrestrial mammals. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 366 (1578):2633-2641. Saad R, Koellner T, Margni M (2013) Land use impacts on freshwater regulation, erosion regulation and water purification: a spatial approach for a global scale level. The International Journal of Life Cycle Assessment 18 (6):1253-1264. Saad R, Margni M, Koellner T, Wittstock B, Deschênes L (2011) Assessment of land use impacts on soil ecological functions: development of spatially differentiated characterization factors within a Canadian context. The International Journal of Life Cycle Assessment 16:198–211.
28 Introduction
Sala O, Chapin F, Armesto J, Berlow E, Bloomfield J, Dirzo R, Huber-Sanwald E, Huenneke L, Jackson R, Kinzig A, Leemans R, Lodge D, Mooney H, Oesterheld M, Poff N, Sykes M, Walker B, Walker M, Wall D (2000) Global biodiversity scenarios for the year 2100. Science 287 (5459):1770-1774. Schmidt J (2008) Development of LCIA characterisation factors for land use impacts on biodiversity. Journal of Cleaner Production 16:1929-1942. Scholes R, Biggs R (2005) A biodiversity intactness index. Nature 434 (7029):45-49. Steffan-Dewenter I, Kessler M, Barkmann J, Bos MM, Buchori D, Erasmi S, Faust H, Gerold G, Glenk K, Gradstein SR, et al (2007) Tradeoffs Between Income, Biodiversity, and Ecosystem Functioning During Tropical Rainforest Conversion and Agroforestry Intensification. Proceedings of the National Academy of Sciences of the United States of America 104 (12):4973–4978. Tilman D, Balzer C, Hill J, Befort BL (2011) Global Food Demand and the Sustainable Intensification of Agriculture. Proceedings of the National Academy of Sciences of the United States of America 108 (50):20260–20264. Tilman D, Fargione J, Wolff B, D'Antonio C, Dobson A, Howarth R, Schindler D, Schlesinger WH, Simberloff D, Swackhamer D (2001) Forecasting agriculturally driven global environmental change. Science 292 (5515):281. Toffoletto L, Bulle C, Godin J, Reid C, Deschenes L (2007) LUCAS - A new LCIA method used for a Canadian-specific context. The International Journal of Life Cycle Assessment 12 (2):93-102. Trydeman Knudsen M, Yu-Hui Q, Yan L, Halberg N (2010) Environmental assessment of organic soybean (Glycine max.) imported from China to Denmark: a case study. Journal of Cleaner Production 18 (14):1431-1439. Tscharntke T, Clough Y, Wanger TC, Jackson L, Motzke I, Perfecto I, Vandermeer J, Whitbread A (2012) Global food security, biodiversity conservation and the future of agricultural intensification. Biological Conservation 151 (1):53-59. Udo de Haes HA, Heijungs R, Suh S, Huppes G (2004) Three Strategies to Overcome the Limitations of Life‐Cycle Assessment. Journal of Industrial Ecology 8 (3):19-32. Urban B, Haaren Cv, Kanning H, Krahl J, Munack A (2012) Spatially Differentiated Examination of Biodiversity in LCA (Life Cycle Assessment) on National Scale Exemplified by Biofuels. Landbauforschung - vTI Agriculture and Forestry Research 3 (62):65–76. Vogtländer J, Lindeijer E, Witte J, Hendriks C (2004) Characterizing the change of land-use based on flora: application for EIA and LCA. Journal of Cleaner Production 12:47–57. Wackernagel M, Rees WE (1996) Our Ecological Footprint—Reducing Human Impact on the Earth. New Society Publishers, Vancouver Weidema B, Lindeijer E (2001) Physical impacts of land use in product life cycle assessment. Final report of the EURENVIRON-LCAGAPS sub-project on land use. Department of Manufacturing Engineering and Management, Technical University of Denmark, Lyngby. Wiedmann T (2009) A Review of Recent Multi-Region Input–Output Models Used for Consumption-Based Emission and Resource Accounting. Ecological Economics 69 (2):211–222. Zalasiewicz J, Williams M, Steffen W, Crutzen P (2010) The New World of the Anthropocene. Environmental Science & Technology 44 (7):2228-2231.
29 Introduction
30
Chapter 2
Toward meaningful endpoints of biodiversity in Life Cycle Assessment
Michael Curran
Laura de Baan
An M. De Schryver
Rosalie van Zelm
Stefanie Hellweg
Thomas Koellner
Guido Sonnemann
Mark A. J. Huijbregts
Published in Environmental Science and Technology
Volume 45 Issue 1 (2011): 70–7 Toward meaningful endpoints of biodiversity in Life Cycle Assessment
Abstract
Halting current rates of biodiversity loss will be a defining challenge of the 21st century. To assess the effectiveness of strategies to achieve this goal, indicators and tools are required that monitor the driving forces of biodiversity loss, the changing state of biodiversity, and evaluate the effectiveness of policy responses. Here, we review the use of indicators and approaches to model biodiversity loss in Life Cycle Assessment (LCA), a methodology used to evaluate the cradle-to- grave environmental impacts of products. We find serious conceptual shortcomings in the way models are constructed, with scale considerations largely absent. Further, there is a disproportionate focus on indicators that reflect changes in compositional aspects of biodiversity, mainly changes in species richness. Functional and structural attributes of biodiversity are largely neglected. Taxonomic and geographic coverage remains problematic, with the majority of models restricted to one or a few taxonomic groups and geographic regions. On a more general level, three of the five drivers of biodiversity loss as identified by the Millennium Ecosystem Assessment are represented in current impact categories (habitat change, climate change and pollution), while two are missing (invasive species and overexploitation). However, methods across all drivers can be greatly improved. We discuss these issues and make recommendations for future research to better reflect biodiversity loss in LCA.
2.1 Introduction The planet is undergoing extensive changes induced by human appropriation of natural resources. Among the most critical consequences is the stark decline in biological diversity documented over the past decades (CBD 1993; MA 2005b; Alkemade et al. 2009; Sala et al. 2000). The Millennium Ecosystem Assessment (MA 2005b) recently documented a widespread decline in the quality of global ecosystems and biodiversity, stimulating the rapid development of indicators to measure the changing state of nature, the driving pressures behind, and evaluate the effectiveness of policy responses. The MA identified a number of direct drivers of biodiversity loss, of which the most important are i) terrestrial and aquatic habitat change, ii) invasive species, iii) overexploitation of wild populations, iv) pollution, and v) climate change (MA 2005b).
The development and use of assessment tools to track hot-spots of environmental damages in production systems has been of growing interest amongst industry, the public sector, and non- governmental organizations. One such method of environmental assessment at the product level is Life Cycle Assessment (LCA). LCA is used to quantify the potential environmental impacts throughout a product’s life cycle from raw material acquisition, production, use, and finally disposal (Finnveden et al. 2009). The “impact assessment” stage of LCA models impacts along mostly linear, deterministic, cause-effect chains by linking inventory items to so-called midpoint impact categories, such as global warming potential, ecotoxicity and land use. In an optional second step, the cause-effect chain is extended to final endpoints, which express impacts on three areas of protection: natural resources, human health, and ecosystem quality.
The development and inclusion of potential endpoints for biodiversity in LCA has been ongoing for more than a decade (Goedkoop & Spriensma 2001; Lindeijer 2000; Weidema & Lindeijer 2001;
32 Toward meaningful endpoints of biodiversity in Life Cycle Assessment
Blonk et al. 1997; Heijungs et al. 1997). Yet many methods in LCA are still in early stages of development (Koellner & Scholz 2008; Schmidt 2008; Pfister et al. 2009; De Schryver et al. 2009; van Zelm et al. 2007; Larsen & Hauschild 2007; Pennington et al. 2004; Pelletier et al. 2007; Koellner & Scholz 2007). Particular difficulties are posed by the methodological framework of LCA itself, which traditionally required impacts to be generic in space, summed across time horizons, strongly linked to a functional unit, and free of interactions between impact pathways (Udo de Haes 2006). These restrictions are only beginning to be addressed by recent LCA research. If LCA is to be truly informative to decision makers about biodiversity loss, it is important to assess whether current methods are reflecting i) the major drivers of biodiversity loss as identified by the MA and ii) whether they capture the concept of biodiversity adequately both in its inherent variation, and its non-uniform distribution across the planet.
This review article synthesizes how biodiversity indicators are currently employed in LCA during endpoint cause-effect modelling, and how this could be improved in the future by drawing on available methods in biodiversity research. We begin by describing a framework for characterizing biodiversity indicators. We then assess the coverage of biodiversity in relevant impact categories of LCA in light of this framework. We highlight research gaps and offer suggestions for improvement based on a review of a wide range of methods employed in the biodiversity assessment literature, including a comparison of LCA with recent biodiversity assessments such as the MA (see Appendix A.7).
2.2 Biodiversity indicators
Framework We adopt the definition of biodiversity provided in the Convention on Biological Diversity (CBD 1993), as a nested hierarchy of components defined by the level of scale and complexity (Noss 1990). We separate components into four broad levels: gene, species, community and ecosystem (Noss 1990; Niemi & McDonald 2004; differing slightly from the CBD). At each level, components may be characterized in terms of attributes, which reflect composition, the quantity and variety of elements; function, the ecological and evolutionary process acting amongst elements; and structure, the physical organization of elements (Niemi & McDonald 2004). Indicators used to measure biodiversity can be correspondingly described by attribute and component. Table 2.1 illustrates the indicator framework adopted for this review.
33 Toward meaningful endpoints of biodiversity in Life Cycle Assessment Table 2.1. Biodiversity indicators across hierarchical components (gene, species, community and ecosystem) and biological attributes (compositional, structural and functional). The final column on the right illustrates possible assessment methods (adapted from Noss 1990; Niemi & McDonald 2004).
Hierarchical biological attributes assessment tools and methods components composition structure function Genes (biotic) heterozygosity, allelic diversity, % chromosomal or phenotypic mutational diversity, mutation visible polymorphisms, molecular polymorphic loci, genetic polymorphisms, physical rate, duplication rate, selection markers (protein electrophoresis), variance, phylogenetic diversity genetic distance, effective intensity, rate of genetic drift, DNA markers (microsatellites, DNA population size, generation gene flow sequencing), parent-offspring overlap, heritability regression, sibling analysis
Species (biotic) (meta)population size and size, morphological variability, metapopulation dynamics (drift, population censuses, time series number, absolute or relative physiognomy, population bottle necks, inbreeding, analysis, remote sensing and GIS, abundance, frequency, biomass, structure, home range size and outbreeding trends) habitat suitability index, species- cover, intactness, density distribution in space, dispersal demographic processes habitat modeling, population viability patterns, habitat requirements (growth, reproductive, feeding, analysis, species distribution nesting, dispersal rate) modeling
Communities (biotic, species richness, relative habitat structural complexity, nutrient turnover and cascades, remote sensing and GIS, aerial abiotic) abundance, higher taxon diversity, foliage physiognomy and functional group or guild photographs, time series analysis, phylogenetic diversity, number of layering, habitat density, gap diversity, number and strength physical habitat measures, endemics, invasive, threatened or density, volume, surface area, of interspecific interactions observation habitat descriptions, focal species, similarity and slope, aspect, rugosity index, (predator-prey, parasite-host), multi-species, local sampling turnover of species assemblages nearest neighbor distance biomass production, extinction, techniques, multivariate integrative colonization rates indices (Shannon-Wiener index, dispersion, layering, biotic integrity)
Ecosystems and patch diversity, richness, patch shape and configuration disturbance patterns and remote sensing and GIS, aerial landscapes (abiotic) composition, number of (fragmentation, isolation, regimes (frequency, extent, photographs, time series analysis, ecosystems, relative or absolute connectivity, spatial linkage), intensity, seasonality), pattern spatial statistics, mathematical area, area of semi-natural patch size frequency metrics (patch turnover), indices (pattern, connectivity, vegetation in agriculture, distribution, topography, river erosion potential, geomorphic heterogeneity, layering, edge extent, emergent patterns in species and shoreline profile and hydrologic processes, land diversity, fractal measures and distribution (richness, endemism) use patterns and trends autocorrelation)
34
Toward meaningful endpoints of biodiversity in Life Cycle Assessment
Genetic indicators At the genetic level, indicators reflect intraspecific (within-species) or interspecific (between species) diversity (Appendix A.1). Intraspecific indicators include heterozygosity, allelic richness, nucleotide diversity, genetic variance and heritability (Hughes et al. 2008). Interspecific variation is quantified using phylogenetic indicators (Faith 2002), including phylogenetic diversity (Faith et al. 2004).
Species-based indicators Species level indicators describe trends in the abundance or attributes of individual species (Appendix A.2). These include the focal species approach (Lambeck 1997) and species intactness indices (Scholes & Biggs 2005). Methods in modelling species distributions, habitat suitability and sensitivity to environmental stress may take a deductive approach, using expert opinion and meta-analysis to derive cause-effect relationships, or an inductive approach, extracting patterns from empirical data via multivariate statistics with no proposed a priori cause-effect mechanism (Corsi et al. 2000).
Community indices Community level indices describe the emergent patterns in biodiversity resulting from the overlap of individual species ranges (Appendix A.3). They characterize the number and relative abundance of species in a community in a single value. The value of a diversity index may be totally dominated by the number of species in a community (i.e. species richness), or conversely by only the relative abundance of species (i.e. pure evenness indicators). Intermediate points represent well-known indices such as the Shannon-Wiener index and Simpson’s concentration (Jost 2006). Community indices such as the Bray Curtis index or ordination measures also represent beta diversity changes (turnover) between samples or locations. Recent work has focused on modelling emergent community patterns in biodiversity such as richness, turnover and endemism as a substitute for data-demanding species level approaches (Ferrier & Guisan 2006; Appendix A.4). Indices of functional diversity may be derived by grouping species into functional groups, or calculating continuous distance in functional trait space (Petchey & Gaston 2006).
Ecosystem and landscape indicators Indicators of ecosystem diversity are split into those of pattern and process (McGarigal et al. 2009; Appendix A.5). Landscape pattern indicators represent human-perceived patterns in a landscape (diversity, patch size, and configuration of habitat). Their link with biological processes, such as dispersal and persistence of species, is not fully understood (Turner 2005). However, key pattern metrics such as area of native habitat and summed anthropogenic edge length often correlate with processes and patterns in emergent biodiversity (Fischer & Lindenmayer 2007). Variables derived from remote sensing, such as the Normalized Difference Vegetation Index (NDVI), can illustrate disturbance regimes, measure vegetation cover, and chart phenological changes (Foody 2008).
35
Toward meaningful endpoints of biodiversity in Life Cycle Assessment
2.3 Assessing biodiversity loss in LCA In the LCA framework, endpoint impacts on biodiversity resulting from an environmental intervention (e.g. emission of acidifying substances, conversion or occupation of land) have been developed in the impact categories of land use, water use, climate change, acidification and eutrophication, and ecotoxicity. This covers three of the five principal drivers of biodiversity loss as identified by the MA (Table 2.2). Endpoint models generally include three components: i) fate factor, which models the spatial distribution and intensity of pressures induced by an intervention; ii) impact factor (also known as damage or effect factor), which relates the intensity of a unit of pressure to a quantified loss in biodiversity, frequently expressed as the unit-less indicator of “potentially disappeared fraction (PDF) of species” (Goedkoop & Spriensma 2001); and iii) characterization factor, which multiplies impact factors (intensity of the intervention) by fate factor (size and duration of the intervention) to give an endpoint of PDF*mx*years, where the power term, x, equals 2 (for area fate models) and 3 (for volume fate models). Below we briefly summarize the history and state of art of endpoint modelling in each impact category.
2.3.1 Land use The conversion of natural habitat to human use has been the most important driver of biodiversity loss over the past century (MA 2005b). Methods to characterize land use impacts to biodiversity in LCA have mainly used compositional indicators at the level of the local community, primarily species richness (Goedkoop & Spriensma 2001; Lindeijer 2000; Koellner & Scholz 2008; Schmidt 2008; Koellner 2000, 2003; Geyer et al. 2010). Koellner and Scholz (2008) additionally used the number of threatened species as an indirect indicator of ecosystem diversity and land use value. Some studies have included ecosystem level indicators directly, using the relative area of ecosystems (Pelletier et al. 2007; Vogtländer et al. 2004; Kyläkorpi et al. 2005; Müller-Wenk 1998), or by predicting the effect of ecosystem level changes on the regional pool of species (Schmidt 2008; Koellner 2000). Genetic indicators are absent, but Weidema and Lindeijer (2001) proposed calculating the change in community phylogenetic diversity due to land use change.
A first attempt to incorporate the effects of fragmentation at the midpoint level was made by Jordaan et al. (2009). Schenck (2001) presented a range of indicators including the protection of habitats/species, connectivity of habitats, invasive species, and percentage of native-dominated vegetation, but proposed no means to integrate these into an operational framework. Michelson (2008) included a limited set of local indicators of ecosystem function. Net primary productivity (NPP) was proposed early on as a functional indicator (Lindeijer 2000; Weidema & Lindeijer 2001; Blonk et al. 1997). Other approaches used the concept of hemeroby – naturalness of ecosystems – to characterize land use types on a scale of 0, purely artificial, to 1, no human influence (Baitz et al. 2000; Bauer & Zapp 2004; Brentrup et al. 2002).
Vascular plant species richness has been the core taxa for land use assessments, mainly because of data availability and the close associations to specific land uses (Goedkoop & Spriensma 2001; Koellner & Scholz 2008; Schmidt 2008; Koellner 2000). Michelsen (2008) suggested an integrative index, Conditions for Maintained Biodiversity, in order to address taxonomic coverage, but this indicator requires region-specific information, not readily available for many areas or 36
Toward meaningful endpoints of biodiversity in Life Cycle Assessment ecosystems. Koellner and Scholz (2008) included species richness of mollusks and moss in addition to vascular plants. Geyer et al. (2010) used the habitat affinities of vertebrate species to calculate impacts on species richness, abundance, and evenness. Mattsson et al. (2000) recommended using richness of mammals, birds, and butterflies, but no overall assessment framework was provided. A multi-taxa approach to agricultural land use and management regimes using taxon-specific impact scoring was employed in the Swiss Agricultural Life Cycle Assessment (SALCA) method (Jeanneret et al. 2008).
The majority of approaches were developed for specific geographic regions, namely Northern Europe (Koellner & Scholz 2008; Koellner 2000; Michelsen 2008), North America (Toffoletto et al. 2007; Geyer et al. 2010) and South-East Asia (Schmidt 2008). Weidema and Lindeijer (2001) proposed a global approach at a biome scale, but it remains of limited application because of the coarseness of the method (but see Toffoletto et al. 2007 for a regionalized adaptation). The species-area relationship (SAR; Rosenzweig 1995) and sample-based rarefaction (Colwell et al. 2004) have been applied to compare the species richness of standardized sampling areas in different land classes (Koellner & Scholz 2008; Schmidt 2008; Koellner 2000). The SAR also forms the basis for regional damage calculation, estimating the increased risk of regional extinction due to land use change (Koellner & Scholz 2008; Schmidt 2008; Koellner 2000).
2.3.2 Water use Anthropogenic water use reduces regional availability, impairing the functioning and diversity of water-dependent terrestrial (Pfister et al. 2009) and freshwater (Xenopoulos et al. 2005) ecosystems. The assessment of water use impacts in LCA is a relatively new development. Mila i Canals et al. (2009) suggested an indicator, Freshwater Ecosystem Impact, to describe the damage to natural ecosystems resulting from depleted water resources via land occupation and the resulting changes in runoff quality and quantity, and abstractive use for agriculture. Pfister et al. (2009) developed the only existing method to model the impacts of freshwater use on natural ecosystems to the endpoint level. In this work, impacts to terrestrial biodiversity were approximated using water-limited NPP as a functional indicator (based on Nemani et al. 2003). Links to species richness were established via a correlation between NPP and vascular plant species richness at low to medium levels of species richness (Pfister et al. 2009). It was assumed that water-limited terrestrial ecosystems, generally in arid areas, do not display high species richness. Since water availability and vulnerability of ecosystems varies as a function of space, the method of Pfister et al. (2009) was regionalized to provide impact factors for all global watersheds.
2.3.3 Climate change Climate change, driven by anthropogenic greenhouse gas emissions, is expected to cause a large number terrestrial extinction over the next century due to changing temperature, precipitation and seasonality (Thomas et al. 2004; Thuiller et al. 2006). Aquatic effects include extinctions of fish species due to reduced river discharge (Xenopoulos et al. 2005) and mass extinctions of coral reefs due to warming sea temperature (Hughes et al. 2003) and increased ocean acidification (CBD 2009). Within LCA, the only operational impact assessment method for climate change was
37
Toward meaningful endpoints of biodiversity in Life Cycle Assessment restricted to terrestrial biodiversity and is based on the increased extinction risk associated with changes in individual species’ distributions under future climate scenarios (De Schryver et al. 2009). The damage factor was based on the work of Thomas et al. (2004), a review of nine studies linking regional extinction risk with changing climate across a number of species groups.
The method of De Schryver et al. (2009) included assessments of 1084 plant and animal species across five regions (in Europe, Mexico, Australia, South Africa, and Brazil). Climate envelope modelling was used to estimate range area reductions per species and the associated increase in extinction risk (Thomas et al. 2004). Three approaches based on the SAR were used to estimate predicted extinctions. Responses were tested under assumptions of dispersal and non-dispersal ability, and red list species and all species respectively (Thomas et al. 2004). This species level approach was used to approximate the potentially disappeared fraction of species, extrapolated to global terrestrial extent to represent a global impact factor.
2.3.4 Acidification and eutrophication. Acidification and eutrophication leads to a disruption of the natural nutrient balance, altering the species composition of ecosystems, and frequently leading to a loss of biodiversity (Smith et al. 1999; Stevens et al. 2004). The effects of acidification on ecosystems quality have been included in LCA using methods that considered the sensitivity of the receiving ecosystems (Hauschild & Potting 2005; Posch et al. 2008; Krewitt et al. 2001; Seppälä et al. 2006; Potting et al. 1998; Huijbregts et al. 2000), or effects on NPP (Hayashi et al. 2004). Species level impacts for terrestrial acidification have been modelled based on occurrence data for vascular plants (Goedkoop & Spriensma 2001; van Zelm et al. 2007; Goedkoop et al. 2009) and butterflies (Goedkoop & Spriensma 2001). Van Zelm et al. (2007) modelled the probability of occurrence of over 240 forest plant species in Europe. A threshold was constructed for each species as a function of the base saturation of the soil. The PDF was approximated by the percentage of species predicted to be absent because of elevated base saturation. This impact factor was multiplied by the total forest and non-forest area of Europe to approximate overall impacts to the region affected by acidifying emissions.
Freshwater eutrophication was included in the ReCiPe method (Goedkoop et al. 2009) for Europe using the diversity of macro-invertebrate genera as indicators of taxonomic diversity. Environmental modelling was used to predict the absence of over 837 macro-invertebrate genera, approximating PDF, as a function of phosphorus concentration (Goedkoop et al. 2009).
Taxonomic coverage for both acidification and eutrophication remains limited to species groups where pressure and occurrence data exist. Current methods are applicable to Europe, although the impact factors are likely to apply to other temperate climates, but not to tropical and sub- tropical regions (Bobbink et al. 2010). Impact factors for aquatic acidification and marine eutrophication are absent from any proposed method.
2.3.5 Ecotoxicity Chemical emissions to aboveground biomass, air, water, and soil cause toxicity to a variety of organisms. Research on ecotoxicological impacts to biodiversity in the context of LCA has been
38
Toward meaningful endpoints of biodiversity in Life Cycle Assessment ongoing for many years (Larsen & Hauschild 2007). Model species in laboratory settings are used to establish the potentially affected fraction (PAF) of species due to elevated concentrations of a toxin. It is defined as the percentage of species within a community or taxonomic group that is expected to be exposed above a certain effect-related threshold, such as the effect concentration for 50 percent of the population (EC50) or the no observed effect concentration (NOEC; Larsen & Hauschild 2007; de Zwart & Posthuma 2005). This uses species level indicators of abundance and reproductive decline.
The exact relationship between the PAF and species loss from a community is a topic of debate. By comparing laboratory experiments with field data, Van den Brink et al. (2002) showed that direct effects of long-term and acute exposure are generally well reflected by species sensitivity distributions used to calculate the PAF of a community. The study assessed changes in biodiversity across a wide range of animal and plant taxa. Posthuma and de Zwart (2006) showed that in fish species assemblages in North American streams, the observed loss of species ascribed to mixture toxicity closely matched the predicted risks based on EC50. Snell and Serra (2000) modelled reproduction effects on rotifers, and showed that an EC50 will result in population extinction after a long exposure time. The PAF therefore may have the diagnostic properties required to assess ecological responses to ecotoxic stress.
LCA models are available that can be adapted to meet region-specific conditions, but can also provide continental and global factors, such as the USEtox model (Rosenbaum et al. 2008). Freshwater biodiversity responses have received the most attention, and more research is necessary on the response of terrestrial and marine ecosystems. Taxonomic coverage is usually limited to low trophic position, cold-blooded species. The effects of bioaccumulation and biomagnification are only beginning to be investigated in LCA.
39
Toward meaningful endpoints of biodiversity in Life Cycle Assessment Table 2.2. Recent methods in LCA that model endpoints of biodiversity loss cover three of the five drivers identified by the MA. Indicators generally reflect compositional aspects of biodiversity, and are limited both taxonomically and geographically. MA driver and corresponding LCA modeling approach (data type) indicator taxonomic coverage geographic impact category component attribute coverage HABITAT CHANGE Land use Koellner 2000, 2003; Koellner and standardized species richness; standardized number of community composition C. Europe, SE Scholz 2008; Schmidt 2008 threatened species (EDP) (sampled species occurrence data) vascular plants, mollusks Asia (Schmidt Koellner 2000, 2003; Schmidt 2008 SAR-based; proportion of LI land in region and assumed (EDP), moss (EDP) ecosystem composition 2008) species associations (GIS vegetation/LU classes) Michelsen 2008 (cf. Weidema and area index for scarcity; integrative conservation index for weight for IUCN ecosystem integrative global Lindeijer 2001) vulnerability (WWF ecoregion data) threatened species Michelsen 2008 Conditions for Maintain Biodiversity index; decaying wood Scandinavian n/a (multiple proxies (regional estimates), set aside land (area), invasive species ecosystem integrative and Russian used) (percentage community) taiga Water use Pfister et al. 2009 environmental modeling; w-NPP change due to water use n/a (w-NPP proxy for ecosystem function global (remote sensing and GIS) vascular plant richness) CLIMATE CHANGE De Schryver et al. 2009 meta-study-based impact factor; climate-envelope vascular plants, global modeling; 3 SAR-inspired methods (species occurrence species composition mammals, birds, (extrapolated) data) amphibians, insects POLLUTION
Acidification and eutrophication Van Zelm et al. 2007 (acidification) modeled species absence as function of BS (species vascular plants (forest species composition Europe occurrence data); % sp. absent ≈ PDF species) Goedkoop and Spriensma 2001 modeled species absence as function of nitrogen deposition vascular plants, insects species composition Netherlands (acidification and eutrophication) (species occurrence data); % sp. absent ≈ PDF (butterflies) Goedkoop et al. 2009 (eutrophication) modeled genera absence as function of phosphorus conc. insects (macro- species composition N. Europe (genera occurrence data): % gen. absent ≈ PDF invertebrate) Ecotoxicity Van den Brink et al. 2002; Posthuma testing the PAF ≈ PDF relationship with semi-field data cold-blooded model temperate community composition and de Zwart 2006 (community composition and richness) organisms climates Goedkoop and Spriensma 2001 NOEC(SSD)-based PAF ≈ 0.1*PDF; laboratory species global species composition reproductive declines (species abundance data) cold-blooded model (extrapolated) Rosenbaum et al. 2008 USEtox; HC50(SSD)-based PAF; laboratory species organisms reproductive declines (species abundance data) species composition global Endpoint modeling (second column) refers to method and data used to calculate characterization factors. Indicator component and attribute is described in the text. Taxonomic coverage lists species groups assessed. Geographic coverage refers to broad region used in analyis. PDF = potentially disappeared fraction of species; PAF = potentially affected fraction of species; SSD = species sensitivity distribution; HC50 = effect concentration for 50% of the population; NOEC = no observed effect concentration; BS = base saturation; SAR = species area relationship; w-NPP = water limited net primary 40 productivity; EDP = Ecosystem Damage Potential, LI = low intensity; n/a = not applicable.
Toward meaningful endpoints of biodiversity in Life Cycle Assessment
2.4 Endpoint unit, scale, and linearity
Endpoint unit Endpoints of biodiversity in LCA are expressed across all impact categories as an effective loss of habitat, which converts the area/volume partially affected by an intervention (i.e. PDF < 1) to an equivalent area/volume of total loss of habitat value for biodiversity (PDF = 1). The exact proportion is dependent on the intensity of the intervention, which is given by the impact factor. This is a potentially useful approach that has been applied outside of LCA to assess land use impacts using variants of the SAR (Scholes & Biggs 2005; Pereira & Daily 2006) and forms the basis of the GLOBIO3 modelling framework (Alkemade et al. 2009). It has also been expanded to include freshwater impacts across a range of pressures using river section length in place of area (Turak et al. 2011).
Impact scale
Current LCA impact factors estimate species extinctions, in terms of PDF, at largely undefined spatial scales. Biodiversity loss represents a concern over the potential extinction of species at broadly defined scales (sub-national, national, and international; CBD 1993). Likewise, the scale of extinctions in LCA requires standardization at one or multiple scales (i.e. the local community, the ecosystem or landscape, the region, or the globe). Koellner (2003) highlighted the arguable distinction between local impacts, which reflect concerns over the loss of local ecosystem functioning, and regional impacts, which reflect conservation concerns over species loss. Both are valid impacts, but convey very different messages to the end user of an LCA. This raises serious questions about the validity of current aggregative single score assessment tools where damages across impact categories, representing various geographic scales of extinction, are combined by simple summation (e.g. EcoIndicator 99 and ReCiPe 2008).
Current scaling of impact factors to arrive at characterization factors assumes a simple linear damage function (PDF*area). This should be reconsidered given the importance of scale influences in ecology (Wiens 1989) and the presence of non-linearity’s, tipping points, and critical thresholds in biological responses to disturbance (Swift & Hannon 2010). The adoption of the non- linear SAR and its variants (Pereira & Daily 2006; Faith et al. 2008; Koh & Ghazoul 2010; Turak et al. 2011) could instead be used to take advantage of assumption one (endpoint unit) in order to address assumption two (impact scale) and three (linear damage relationship). This would require an extra step in characterization factor development that would consist of expressing effective habitat loss as a reduction in the species pool of the affected ecosystem(s), thereby providing the fraction of species potentially lost at the defined scale. This percentage loss could be related to absolute species losses using widespread regional checklist data for various taxa (e.g. WWF Ecoregions; Olson et al. 2001).
2.5 Use of indicators Indicators in LCA tend to reflect compositional changes in biodiversity, particularly at the species and community level (Table 2.2). Indicators that reflect changes in a variety of components
41
Toward meaningful endpoints of biodiversity in Life Cycle Assessment
(species, communities, ecosystems) and attributes (composition, structure, function) of biodiversity are also often employed to approximate species loss in terms of PDF. For example, the SAR is used to translate ecosystem indicators of habitat area change into predicted species losses. Likewise, single-species indicators are combined to approximate overall impacts across all species, either in a local community (e.g. PAF in ecotoxic impacts) or across an entire region (e.g. climate change impacts). Additionally, the functional indicator of NPP change is used as a proxy for species loss in water use impact assessment. Below we discuss the implications of this approach, and the possibility of developing new impact factors to reflect additional aspects of biodiversity.
2.5.1 Genetic component Genetic indicators are absent from LCA. Developing impact factors based on phylogenetic diversity would enable interspecific genetic diversity to be approximated using existing species data (Faith 2002; Faith et al. 2004). Cadotte et al. (2008) demonstrates an approach to calculate changes in phylogenetic diversity for plant communities using molecular sequence data from GeneBank (www.cnbe.nlm.nih.gov). The impact factor would express changes in the sum of branch lengths linking species from a sample of a community (Faith et al. 2004). Intraspecific (within-species) genetic variation, reflecting impacts to the genetic diversity of single species, such as population declines leading to reduced heterogeneity, will be extremely difficult to incorporate (see Appendix A.1), and is limited by data availability (Laikre 2010).
2.5.2 Species and community component By definition, PDF is a multi-species index, and aggregating single-species indicators assumes that all species react to pressures in the same way as those assessed (Appendix A.4). As the number of assessed species increases, results should be expected to converge (Roberge & Angelstam 2004; MA 2005a). In species level approaches in LCA, the number of species employed to construct impact factors is generally large (e.g. 1084 species for climate change (De Schryver et al. 2009); 240 species for acidification (van Zelm et al. 2007); 837 genera for eutrophication (Goedkoop et al. 2009)). In ecotoxicity the use of model organisms generally does not exceed ten species per substance (van Zelm et al. 2009). This introduces large uncertainties and more studies are needed that test these laboratory results against field data (e.g. Posthuma & de Zwart 2006; Snell & Serra 2000). Inclusion of both species and community approaches across impact categories may offer better estimates of trade-offs and uncertainties associated with different methods.
2.5.3 Ecosystem and landscape component In LCA, ecosystem indicators used to approximate species loss at the endpoint have been employed only in land use, and consider relatively simple effects based on the SAR (Koellner & Scholz 2008; Schmidt 2008; Koellner 2000, 2003). Recent land use methods are including more complex models of habitat area and composition (Geyer et al. 2010). Outside of LCA, the InVEST tool (Nelson et al. 2009) takes a detailed approach in estimating the contribution of each habitat patch to a species’ persistence in the landscape using “countryside SARs” (Pereira & Daily 2006).
42
Toward meaningful endpoints of biodiversity in Life Cycle Assessment
This accounts for patch size, cumulative anthropogenic habitat edge length, configuration, and the habitat requirements and dispersal ability of the assessed species (Nelson et al. 2009).
The current species-level approach toward climate change modelling in LCA could be expanded to employ ecosystem-level indicators by modelling ecosystem area changes and resulting species loss via the SAR. For example, the MA (2005a) and GLOBIO3 (Alkemade et al. 2009) predict biome and vegetation community expansions and contractions under IPCC scenarios using the IMAGE model (MNP 2006), and relates this to species loss predicted by the SAR (Alkemade et al. 2009; Hooper et al. 2005). The ecosystem impact of consumptive water use is modelled to impacts on vascular plant species diversity in LCA through water-limited NPP. Additional, direct impacts of water use on fish species richness of rivers could be captured by the species-discharge relationship (Oberdorff et al. 1995) both for water use and climate change (Xenopoulos et al. 2005). Remote sensing data could help expand this to identify terrestrial drought damages to ecosystems caused by water abstraction or climate change (Foody 2008).
Ecosystem effects of acidification, eutrophication, and ecotoxicity, modelled to the endpoint of species loss, are lacking. Ecosystem impacts have been developed in LCA using critical nitrogen and phosphorous loading values for acidification and eutrophication (Posch et al. 2008; Seppälä et al. 2006), but the consequential effects on species richness were not included. Critical loading has been extensively employed outside of LCA (e.g. Smith et al. 1999; Bobbink et al. 2010). Acidification and eutrophication are covered in both the MA (2005a) and GLOBIO3 model (Alkemade et al. 2009) using meta-analyses of empirical studies documenting the relationship between exceedance of critical load and species loss (e.g. Bobbink 2004). Such a relationship could be adapted to existing LCA methods.
2.5.4 Biological attributes The majority of indicators of biodiversity in LCA measure composition (Table 2.2). Indicators of structure and function are largely absent. There are a range of indicators and methods to infer structural information at the local to ecosystem scale (Table 2.1). Impacts such as eutrophication or land use cause extensive structural alterations to habitats. A meta-analysis of published studies documenting the effect of pressures on the structural diversity of communities could potentially yield impact factors which could be used as a rough proxy for species loss (Gardner et al. 2010).
On an ecosystem level, indicators were used in LCA to reflect fragmentation effects at the midpoint of land use (Jordaan et al. 2009), but no attempt was made to model the damage to species richness. The meta-study of Harper et al. (2005) could be used to complete this process. Outside of LCA, fragmentation effects were included in the GLOBIO3 model using six published datasets that quantify species loss as a function of patch size (Alkemade et al. 2009). The BioScore tool (Louette et al. 2010) uses focal species that are sensitive to fragmentation (e.g. habitat specialists).
Changes in functional diversity are currently considered in LCA only at the ecosystem level using NPP. Abiotic indicators, such as NDVI-based metrics, hydromorphic and geomorphic modelling,
43
Toward meaningful endpoints of biodiversity in Life Cycle Assessment erosion potential, and disturbance indicators could further be used to model impacts to ecosystem functioning (functional diversity; Table 2.1). The relationship between species diversity and ecosystem function is not well enough understood to allow PDF to act as a proxy for functional diversity, or vice-versa (Hooper et al. 2005). New methods in land use developed in the framework of the UNEP-SETAC Life Cycle Initiative will employ abiotic functional indicators to model damages to a separate functional endpoint for ecosystems services (Wittstock et al. 2008). Thuiller et al. (2006) modelled climate change impacts to functional group diversity of plants at the community level. This could be incorporated into LCA directly as a new impact factor for climate change. Interestingly, Cadotte et al. (2008) found phylogenetic diversity to be a better predictor of ecosystem function than functional group diversity. This might indicate how functional and genetic diversity may be incorporated into LCA using a single indicator.
2.5.5 Multiple impact factors In order to better reflect the diverse components and attributes of biodiversity, we see the need to develop multiple impact factors for biodiversity. Currently in LCA, compositional indicators at the species, community, and ecosystem level approximate PDF. Structural indicators at the community and ecosystem level may also be expressed in terms of PDF, such as reductions in habitat complexity, increased fragmentation and habitat patch configuration (Alkemade et al. 2009; Nelson et al. 2009). Genetic diversity will require a separate impact factor of phylogenetic diversity. Phylogenetic diversity may also function as a good proxy for community functional diversity; otherwise functional groups or trait-space distance could be used to create an additional impact factor. Finally, community and ecosystem functional diversity will require an independent impact factor (e.g. damages to ecosystem services; Zhang et al. 2010). Including the genetic component of function and structure, and the species component of structure into LCA is not foreseeable in the near future.
2.6 Taxonomic and geographical coverage
2.6.1 Taxonomic coverage Methodologies for all impact categories (except climate change) were developed using very few taxonomic groups to construct impact factors (Table 2.2). The use of surrogate taxa to reflect the overall response of biodiversity to environmental stress is questionable (Wolters et al. 2006). In a global meta-study of multiple taxon responses to disturbance, Wolters et al. (2006) found a weak average correlation between taxa (r = 0.38). A number of factors influence this including habitat type, taxon, and temporal and spatial scale. Yet precise roles of these factors are poorly investigated and unpredictable in novel situations (Wolters et al. 2006). In the context of LCA, methods should prioritize major trophic or functional groups, taxa which are sensitive to the relevant pressures, and expand coverage based on data availability and feasibility.
The use of deductive methods can aid in overcoming both taxonomic and geographic limitations. Such an approach has been employed in studies such as the BioScore tool to model the response of indicator species to a range of pressures (Louette et al. 2010); the Biodiversity Intactness Index
44
Toward meaningful endpoints of biodiversity in Life Cycle Assessment to model the effects of land degradation (Scholes & Biggs 2005); and the Mean Species Abundance as part of the GLOBIO3 model (Alkemade et al. 2009).
2.6.2 Geographic coverage The geographic coverage of methods in only two impact categories, water use and climate change, is global with respect to the terrestrial environment (Table 2.2). However, current climate change methods cover only 20% of the total terrestrial area across forest, arid/semiarid and mountain regions. Vulnerable terrestrial areas, such as islands and polar areas should be prioritized for future work. The methods in remaining impact categories were developed for use in specific regions or biomes. Detailed biodiversity data on the distribution of species across many taxa is incomplete on a global scale. Worldwide species richness and endemism data is available in equal-area grids and likely to be relatively robust to undersampling only for birds and plants (Orme et al. 2006; Kier et al. 2005; also see IUCN global assessments of other taxa). For other taxa, checklists of predefined terrestrial and aquatic biogeographic regions and expert opinion (e.g. Olson & Dinerstein 2002; Olson et al. 2001; Spalding et al. 2007; Abell et al. 2008) have been used to map and assess biodiversity (Brooks et al. 2006).
Currently no methods exist to quantify aquatic habitat change in the context of land and water use. Meta-analysis or regional case-studies could guide the development of impact factors for lakebed, riverbed and seabed habitats across regions and climates, such as that pursued by the GLOBIO3 model for aquatic and marine environments (http://www.globio.info/). Turak et al. (2011) illustrates how an SAR-based approach can be adapted to freshwater habitats, using river length in place of area, to reflect a range of pressures. Global, spatially differentiated maps of pressures on marine ecosystems exist for 17 anthropogenic pressures at a grid resolution of 1 km2 (Halpern et al. 2008). For climate change, freshwater effects could be included by adopting a similar approach to Xenopolous et al. (2005). For marine climate change impacts, Halpern et al. (2008) provides a spatially resolved global map of climate change impacts on marine biodiversity including sea temperature rise, ocean acidification and UV radiation. This work also provides globally mapped impacts due to pollution runoff into marine waters, information that could be used to regionalize acidification and eutrophication. Ecotoxicity requires further semi-field research to verify the relationship between PAF and PDF across regions and environments (Posthuma & de Zwart 2006; Snell & Serra 2000).
2.7 Research outlook Our review has illustrated the currently poor state of endpoint biodiversity modelling in LCA. The deficiencies across impact categories are not solely due to data limitations or even the inherent complexities of the element under study. They are also conceptual and methodological in nature. In order to meaningfully represent biodiversity in LCA, we present two broad recommendations for future research, presented in order of importance.
Fill the conceptual cracks We see a need to first address the methodological shortcomings of current approaches. Clearly and explicitly defining PDF is an essential and urgently needed first step. Experimenting with
45
Toward meaningful endpoints of biodiversity in Life Cycle Assessment non-linear, potentially unifying relationships (such as the SAR) when scaling impact factors represents another promising area of research that would eliminate the need to derive new impact factors for each scale.
Challenge data limitations We have highlighted a wealth of data on the distribution of pressures (the basis of fate factors), their effects on various taxa (the basis of impact factors), and the global distribution of biodiversity either through raw data, models or surrogate indicators (the basis of characterization factors). Such data should be used to regionalize existing methods in order to capture a representative sample of the Earth’s diverse terrestrial, freshwater, and marine habitats, and to include a range of representative taxa. Following this, integrating new drivers and impact factors reflecting additional attributes of biodiversity could further improve the modelling of biodiversity loss in LCA.
2.8 Acknowledgements We thank the United Nations Environmental Programme and the International Union for the Conservation of Nature for a constructive meeting that led to the concept of the paper. We also thank Rob Leuven for valuable input. The research was partly funded by the European Commission under the 7th framework program on environment; ENV.2009.3.3.2.1: LC-IMPACT - Improved Life Cycle Impact Assessment methods (LCIA) for better sustainability assessment of technologies, grant agreement number 243827; and ENV.2008.3.3.2.1: PROSUITE - Sustainability Assessment of Technologies, grant agreement number 227078. M.C. and L. d. B. were supported by ETH Research Grant CH1-0308-3.
Disclaimer. The views expressed in this text are those of the authors, and are not in any way official policy of their past or present employers such as UNEP.
2.9 References Abell R, Thieme M, Revenga C, Bryer M, Kottelat M, Bogutskaya N, Coad B, Mandrak N, Balderas S, Bussing W, Stiassny M, Skelton P, Allen G, Unmack P, Naseka A, Ng R, Sindorf N, Robertson J, Armijo E, Higgins J, Heibel T, Wikramanayake E, Olson D, Lopez H, Reis R, Lundberg J, Perez M, Petry P (2008) Freshwater ecoregions of the world: A new map of biogeographic units for freshwater biodiversity conservation. BioScience 58 (5):403-414. Alkemade R, van Oorschot M, Miles L, Nellemann C, Bakkenes M, ten Brink B (2009) GLOBIO3: A framework to investigate options for reducing global terrestrial biodiversity loss. Ecosystems 12 (3):374-390. Baitz M, Kreißig J, Wolf M (2000) Methode zur Integration der Naturraum-Inanspruchnahme in Ökobilanzen. Forstwirtschaftliches Centralblatt 119:128-149. Bauer C, Zapp P (2004) Towards generic characterisation factors for land use and water consumption. International Workshop on Life Cycle Assessment of Metals—issues and research directions. SETAC, Pensacola, FL, 96–101. Blonk H, Lindeijer E, Broers J (1997) Towards a Methodology for Taking Physical Degradation of Ecosystems into Account in LCA. The International Journal of Life Cycle Assessment 2 (2):91-98. Bobbink R (2004) Plant species richness and the exceedance of empirical nitrogen critical loads: An inventory. Utrecht University. Netherlands 46
Toward meaningful endpoints of biodiversity in Life Cycle Assessment
Bobbink R, Hicks K, Galloway J, Spranger T, Alkemade R, Ashmore M, Bustamante M, Cinderby S, Davidson E, Dentener F (2010) Global assessment of nitrogen deposition effects on terrestrial plant diversity: a synthesis. Ecological Applications 20 (1):30-59. Brentrup F, Kusters J, Lammel J, Kuhlmann H (2002) Life Cycle Impact assessment of land use based on the Hemeroby concept. The International Journal of Life Cycle Assessment 7 (6):339-348. Brooks T, Mittermeier R, da Fonseca G, Gerlach J, Hoffmann M, Lamoreux J, Mittermeier C, Pilgrim J, Rodrigues A (2006) Global biodiversity conservation priorities. Science 313 (5783):58-61. Cadotte MW, Cardinale BJ, Oakley TH (2008) Evolutionary history and the effect of biodiversity on plant productivity. Proceedings of the National Academy of Sciences 105 (44):17012- 17017. CBD (1993) Convention on Biological Diversity. http://www.cbd.int/ accessed on 13.9.2009 CBD (2009). Scientific Synthesis of the Impacts of Ocean Acidification on Marine Biodiversity. Tech. rept. Secretariat of the Convention on Biological Diversity, Montreal. Cadotte MW, Cardinale BJ, Oakley TH (2008) Evolutionary history and the effect of biodiversity on plant productivity. Proceedings of the National Academy of Sciences 105 (44):17012- 17017. Colwell RK, Mao CX, Chang J (2004) Interpolating, extrapolating, and comparing incidence- based species accumulation curves. Ecology 85 (10):2717-2727. Corsi F, de Leeuw J, Skidmore A (2000) Modeling species distribution with GIS. Research techniques in animal ecology:389-434. De Schryver AM, Brakkee KW, Goedkoop MJ, Huijbregts MA (2009) Characterization factors for global warming in life cycle assessment based on damages to humans and ecosystems. Environmental Science & Technology 43 (6):1689-1695. de Zwart D, Posthuma L (2005) Complex mixture toxicity for single and multiple species: proposed methodologies. Environmental Toxicology and Chemistry 24 (10):2665-2676. Faith DP (2002) Quantifying biodiversity: a phylogenetic perspective. Conservation Biology 16 (1):248-252. Faith DP, Reid C, Hunter J (2004) Integrating phylogenetic diversity, complementarity, and endemism for conservation assessment. Conservation Biology 18 (1):255-261. Faiths DP, Ferrier S, Williams KJ (2008) Getting biodiversity intactness indices right: ensuring that 'biodiversity' reflects 'diversity'. Global Change Biology 14 (2):207-217. Ferrier S, Guisan A (2006) Spatial modelling of biodiversity at the community level. Journal of Applied Ecology 43 (3):393-404. Finnveden G, Hauschild MZ, Ekvall T, Guinee J, Heijungs R, Hellweg S, Koehler A, Pennington D, Suh S (2009) Recent developments in Life Cycle Assessment. Journal of Environmental Management 91 (1):1-21. Fischer J, Lindenmayer DB (2007) Landscape modification and habitat fragmentation: a synthesis. Global Ecology and Biogeography 16 (3):265-280. Foody G (2008) GIS: biodiversity applications. Progress in Physical Geography 32 (2):223-235. Gardner T, Barlow J, Sodhi N, Peres C (2010) A multi-region assessment of tropical forest biodiversity in a human-modified world. Biological Conservation 143:2293-2300. Geyer R, Lindner JP, Stoms DM, Davis FW, Wittstock B (2010) Coupling GIS and LCA for biodiversity assessments of land use: Part 2: Impact assessment. The International Journal of Life Cycle Assessment 15 (7):692-703. Goedkoop M, Spriensma R (2001) Eco-indicator 99, a damage oriented method for Life Cycle Impact Assessment: methodology report. 3rd ed. Tech. rept. Goedkoop M, Heijungs R, Huijbregts M, De Schryver A, Struijs J, van Zelm R (2009) ReCiPe 2008. Main Report, Part 1: Characterization, 1st ed. Tech. rept. VROM, The Netherlands. Halpern BS, Walbridge S, Selkoe KA, Kappel CV, Micheli F, D'Agrosa C, Bruno JF, Casey KS, Ebert C, Fox HE (2008) A global map of human impact on marine ecosystems. Science 319 (5865):948-952.
47
Toward meaningful endpoints of biodiversity in Life Cycle Assessment
Harper KA, Macdonald SE, Burton PJ, Chen J, Brosofske KD, Saunders SC, Euskirchen ES, Roberts D, Jaiteh MS, et al. (2005) Edge influence on forest structure and composition in fragmented landscapes. Conservation Biology 19 (3):768-782. Hauschild M, Potting J (2005) Spatial Differentiation in Life Cycle Impact Assessment – the EDIP2003 Methodology. Tech. rept. The Danish Environmental Protection Agency, Copenhagen. Hayashi K, Okazaki M, Itsubo N, Inaba A (2004) Development of damage function of acidification for terrestrial ecosystems based on the effect of aluminum toxicity on net primary production. The International Journal of Life Cycle Assessment 9 (1):13-22. Heijungs R, Guinée J, Huppes G (1997) Impact categories for natural resources and land use. CML report, p138. Hooper D, Chapin F, Ewel J, Hector A, Inchausti P, Lavorel S, Lawton J, Lodge D, Loreau M, Naeem S, Schmid B, Setala H, Symstad A, Vandermeer J, Wardle D (2005) Effects of biodiversity on ecosystem functioning: A consensus of current knowledge. Ecological Monographs 75 (1):3-35. Hughes AR, Inouye BD, Johnson MT, Underwood N, Vellend M (2008) Ecological consequences of genetic diversity. Ecology Letters 11 (6):609-623. Hughes TP, Baird AH, Bellwood DR, Card M, Connolly SR, Folke C, Grosberg R, Hoegh- Guldberg O, Jackson J, Kleypas J (2003) Climate change, human impacts, and the resilience of coral reefs. Science 301 (5635):929-933. Huijbregts MA, Schöpp W, Verkuijlen E, Heijungs R, Reijnders L (2000) Spatially Explicit Characterization of Acidifying and Eutrophying Air Pollution in Life Cycle Assessment. Journal of Industrial Ecology 4 (3):75-92. Jeanneret P, Baumgartner DU, Knuchel Freiermuth R, Gaillard G (2008) Integration of biodiversity as impact category for LCA in agriculture (SALCA-Biodiversity). In: 6th International Conference on LCA in the Agri-Food Sector, Zurich, November 12–14. Jordaan S, Keith D, Stelfox B (2009) Quantifying land use of oil sands production: a life cycle perspective. Environmental Research Letters 4 (2). Jost L (2006) Entropy and diversity. Oikos 113 (2):363-375. Kier G, Mutke J, Dinerstein E, Ricketts T, Kuper W, Kreft H, Barthlott W (2005) Global patterns of plant diversity and floristic knowledge. Journal of Biogeography 32 (7):1107-1116. Koellner T (2000) Species-pool effect potentials (SPEP) as a yardstick to evaluate land-use impacts on biodiversity. Journal of Cleaner Production 8:293-311. Koellner T (2003) Land Use in Product Life Cycles and Ecosystem Quality, vol 3012. European University Studies, Peter Lang, Berne, Switzerland. Koellner T, Scholz RW (2007) Assessment of land use impacts on the natural environment. Part 1: An analytical framework for pure land occupation and land use change. The International Journal of Life Cycle Assessment 12 (1):16-23. Koellner T, Scholz RW (2008) Assessment of land use impacts on the natural environment. Part 2: Generic characterization factors for local species diversity in Central Europe. The International Journal of Life Cycle Assessment 13 (1):32-48. Koh LP, Ghazoul J (2010) A matrix-calibrated species-area model for predicting biodiversity losses due to land-use change. Conservation Biology 24 (4):994–1001. Krewitt W, Trukenmüller A, Bachmann TM, Heck T (2001) Country-specific damage factors for air pollutants. The International Journal of Life Cycle Assessment 6 (4):199-210. Kyläkorpi K, Rydgren B, Ellegård A, Miliander S (2005) The Biotope Method 2005: A method to assess the impact of land use on biodiversity. Vattenfall, Sweden. Laikre L (2010) Genetic diversity is overlooked in international conservation policy implementation. Conservation Genetics 11 (2):349-354. Lambeck RJ (1997) Focal species: a multi-species umbrella for nature conservation. Conservation Biology 11 (4):849-856. Larsen H, Hauschild M (2007) Evaluation of ecotoxicity effect indicators for use in LCIA. The International Journal of Life Cycle Assessment 12 (1):24-33.
48
Toward meaningful endpoints of biodiversity in Life Cycle Assessment
Lindeijer E (2000) Biodiversity and life support impacts of land use in LCA. Journal of Cleaner Production 8:313–319. Louette G, Maes D, Alkemade J, Boitani L, Knegt Bd, Eggers J, Falcucci A, Framstad E, Hagemeijer W, Hennekens S, Maiorano L, Nagy S, Serradilla A, Ozinga W, Schaminée J, Tsiaousi V, Tol S, Delbaere B (2010) BioScore-Cost-effective assessment of policy impact on biodiversity using species sensitivity scores. Journal for Nature Conservation 18 (2):142-148. MA (2005a) Millennium Ecosystem Assessment, Chapter 10. Tech. rept. World Resource Institute, Washington, DC. MA (2005b) Millennium Ecosystem Assessment. Ecosystems and human well-being: current state and trends. Island Press, Washington, DC Mattsson B, Cederberg C, Blix L (2000) Agricultural land use in life cycle assessment (LCA): case studies of three vegetable oil crops. Journal of Cleaner Production 8:283–292. McGarigal K, Tagil S, Cushman SA (2009) Surface metrics: an alternative to patch metrics for the quantification of landscape structure. Landscape Ecology 24 (3):433-450. Michelsen O (2008) Assessment of land use impact on biodiversity. The International Journal of Life Cycle Assessment 13 (1):22-31. Milà i Canals L, Chenoweth J, Chapagain A, Orr S, Antón A, Clift R (2009) Assessing freshwater use impacts in LCA: Part I—inventory modelling and characterisation factors for the main impact pathways. The International Journal of Life Cycle Assessment 14 (1):28-42. MNP (2006) Integrated Modelling of Global Environmental Change: An Overview of IMAGE 2.4. Bilthoven, The Netherlands: Netherlands Environmental Assessment Agency (MNP). Müller-Wenk R (1998) Land-use - The Main Threat to Species. How to Include Land Use in LCA. IWÖ–Diskussionsbeitrag, 64. Nelson E, Mendoza G, Regetz J, Polasky S, Tallis H, Cameron DR, Chan KMA, Daily GC, Goldstein J, Kareiva PM, Lonsdorf E, Naidoo R, Ricketts TH, Shaw MR (2009) Modeling multiple ecosystem services, biodiversity conservation, commodity production, and tradeoffs at landscape scales. Frontiers in Ecology and the Environment 7 (1):4-11. Nemani RR, Keeling CD, Hashimoto H, Jolly WM, Piper SC, Tucker CJ, Myneni RB, Running SW (2003) Climate-driven increases in global terrestrial net primary production from 1982 to 1999. Science 300 (5625):1560-1563. Niemi G, McDonald M (2004) Application of ecological indicators. Annual Review of Ecology Evolution and Systematics 35:89-111. Noss R (1990) Indicators for monitoring biodiversity - a hierarchical approach. Conservation Biology 4 (4):355-364. Oberdoff T, Guégan J-F, Hugueny B (1995) Global scale patterns of fish species richness in rivers. Ecography 18 (4):345-352. Olson D, Dinerstein E (2002) The Global 200: Priority ecoregions for global conservation. Annals of the Missouri Botanical Garden 89 (2):199-224. Olson D, Dinerstein E, Wikramanayake E, Burgess N, Powell G, Underwood E, D'Amico J, Itoua I, Strand H, Morrison J, Loucks C, Allnutt T, Ricketts T, Kura Y, Lamoreux J, Wettengel W, Hedao P, Kassem K (2001) Terrestrial ecoregions of the worlds: A new map of life on Earth. BioScience 51 (11):933-938. Orme CDL, Davies RG, Olson VA, Thomas GH, Ding T-S, Rasmussen PC, Ridgely RS, Stattersfield AJ, Bennett PM, Owens IP (2006) Global patterns of geographic range size in birds. PLoS biology 4 (7):e208. Pelletier N, Ayer N, Tyedmers P, Kruse S, Flysjo A, Robillard G, Ziegler F, Scholz A, Sonesson U (2007) Impact categories for life cycle assessment research of seafood production systems: Review and prospectus. The International Journal of Life Cycle Assessment 12 (6):414-421. Pennington DW, Payet J, Hauschild M (2004) Aquatic ecotoxicological indicators in life-cycle assessment. Environmental Toxicology and Chemistry 23 (7):1796-1807. Pereira H, Daily G (2006) Modeling biodiversity dynamics in countryside landscapes. Ecology 87:1877:1885. 49
Toward meaningful endpoints of biodiversity in Life Cycle Assessment
Petchey OL, Gaston KJ (2006) Functional diversity: back to basics and looking forward. Ecology Letters 9 (6):741-758. Pfister S, Koehler A, Hellweg S (2009) Assessing the Environmental Impacts of Freshwater Consumption in LCA. Environmental Science & Technology 43 (11):4098-4104. Posch M, Seppälä J, Hettelingh J-P, Johansson M, Margni M, Jolliet O (2008) The role of atmospheric dispersion models and ecosystem sensitivity in the determination of characterisation factors for acidifying and eutrophying emissions in LCIA. The International Journal of Life Cycle Assessment 13 (6):477-486. Posthuma L, De Zwart D (2006) Predicted effects of toxicant mixtures are confirmed by changes in fish species assemblages in Ohio, USA, rivers. Environmental Toxicology and Chemistry 25 (4):1094-1105. Potting J, Hauschild M (2006) Spatial differentiation in life cycle impact assessment - A decade of method development to increase the environmental realism of LCIA. The International Journal of Life Cycle Assessment 11:11-13. Roberge J-M, Angelstam P (2004) Usefulness of the umbrella species concept as a conservation tool. Conservation Biology 18 (1):76-85. Rosenbaum RK, Bachmann TM, Gold LS, Huijbregts MA, Jolliet O, Juraske R, Koehler A, Larsen HF, MacLeod M, Margni M (2008) USEtox-the UNEP-SETAC toxicity model: recommended characterisation factors for human toxicity and freshwater ecotoxicity in life cycle impact assessment. The International Journal of Life Cycle Assessment 13 (7):532-546. Rosenzweig ML (1995) Species diversity in space and time. Cambridge University Press, Cambridge Sala O, Chapin F, Armesto J, Berlow E, Bloomfield J, Dirzo R, Huber-Sanwald E, Huenneke L, Jackson R, Kinzig A, Leemans R, Lodge D, Mooney H, Oesterheld M, Poff N, Sykes M, Walker B, Walker M, Wall D (2000) Global biodiversity scenarios for the year 2100. Science 287 (5459):1770-1774. Schenck R (2001) Land use and biodiversity indicators for Life Cycle Impact Assessment. The International Journal of Life Cycle Assessment 6 (2):114-117. Schmidt J (2008) Development of LCIA characterisation factors for land use impacts on biodiversity. Journal of Cleaner Production 16:1929-1942. Scholes R, Biggs R (2005) A biodiversity intactness index. Nature 434 (7029):45-49. Seppälä J, Posch M, Johansson M, Hettelingh J-P (2006) Country-dependent characterisation factors for acidification and terrestrial eutrophication based on accumulated exceedance as an impact category indicator. The International Journal of Life Cycle Assessment 11 (6):403-416. Smith VH, Tilman GD, Nekola JC (1999) Eutrophication: impacts of excess nutrient inputs on freshwater, marine, and terrestrial ecosystems. Environmental Pollution 100 (1):179-196. Snell TW, Serra M (2000) Using probability of extinction to evaluate the ecological significance of toxicant effects. Environmental Toxicology and Chemistry 19 (9):2357-2363. Spalding MD, Fox HE, Halpern BS, McManus MA, Molnar J, Allen GR, Davidson N, Jorge ZA, Lombana AL, Lourie SA, Martin KD, McManus E, Molnar J, Recchia CA, Robertson J (2007) Marine ecoregions of the world: A bioregionalization of coastal and shelf areas. BioScience 57 (7):573-583. Stevens CJ, Dise NB, Mountford JO, Gowing DJ (2004) Impact of nitrogen deposition on the species richness of grasslands. Science 303 (5665):1876-1879. Swift T, Hannon S (2010) Critical thresholds associated with habitat loss: a review of the concepts, evidence, and applications. Biological Reviews 85 (1):35-53. Thomas CD, Cameron A, Green RE, Bakkenes M, Beaumont LJ, Collingham YC, Erasmus BF, De Siqueira MF, Grainger A, Hannah L (2004) Extinction risk from climate change. Nature 427 (6970):145-148. Thuiller W, Lavorel S, Sykes MT, Araújo MB (2006) Using niche-based modelling to assess the impact of climate change on tree functional diversity in Europe. Diversity And Distributions 12 (1):49-60. 50
Toward meaningful endpoints of biodiversity in Life Cycle Assessment
Toffoletto L, Bulle C, Godin J, Reid C, Deschenes L (2007) LUCAS - A new LCIA method used for a Canadian-specific context. The International Journal of Life Cycle Assessment 12 (2):93-102. Turak E, Ferrier S, Barrett T, Mesley E, Drielsma M, Manion G, Doyle G, Stein J, Gordon G (2010) Planning for the persistence of river biodiversity: exploring alternative futures using process-based models. Freshwater Biology 56 (1):39-56. Turner MG (2005) Landscape ecology: what is the state of the science? Annual Review of Ecology, Evolution, and Systematics:319-344. Udo de Haes H (2006) How to approach land use in LCIA or, how to avoid the Cinderella effect? Comments on 'Key Elements in a Framework for Land Use Impact Assessment Within LCA'. The International Journal of Life Cycle Assessment 11:219 – 221. Van den Brink PJ, Brock TCM, Posthuma L (2002) The value of the species sensitivity distribution concept for predicting field effects:(Non-) confirmation of the concept using semifield experiments. Species sensitivity distributions in ecotoxicology. Boca Raton, FL, USA: Lewis Publishers. p, 155–193. van Zelm R, Huijbregts MA, Posthuma L, Wintersen A, van de Meent D (2009) Pesticide ecotoxicological effect factors and their uncertainties for freshwater ecosystems. The International Journal of Life Cycle Assessment 14 (1):43-51. Van Zelm R, Huijbregts MA, Van Jaarsveld HA, Reinds GJ, De Zwart D, Struijs J, Van de Meent D (2007) Time horizon dependent characterization factors for acidification in life-cycle assessment based on forest plant species occurrence in Europe. Environmental Science & Technology 41 (3):922-927. Vogtländer J, Lindeijer E, Witte J, Hendriks C (2004) Characterizing the change of land-use based on flora: application for EIA and LCA. Journal of Cleaner Production 12:47–57. Weidema B, Lindeijer E (2001) Physical impacts of land use in product life cycle assessment. Final report of the EURENVIRON-LCAGAPS sub-project on land use. Department of Manufacturing Engineering and Management, Technical University of Denmark, Lyngby. Wiens JA (1989) Spatial scaling in ecology. Functional ecology 3 (4):385-397. Wittstock B, Koellner T, Müller-Wenk R, Brandão M, Margni M, Civit B, Milà i Canals L Modeling land use impacts on biodiversity and ecosystem services in LCA. In: Proceeding paper for Ecobalance Conference 2008, Oct 23 2008. p 2. Wolters V, Bengtsson J, Zaitsev AS (2006) Relationship among the species richness of different taxa. Ecology 87 (8):1886-1895. Xenopoulos MA, Lodge DM, Alcamo J, Märker M, Schulze K, Van Vuuren DP (2005) Scenarios of freshwater fish extinctions from climate change and water withdrawal. Global Change Biology 11 (10):1557-1564. Zhang Y, Singh S, Bakshi BH (2010) Accounting for Ecosystem Services in Life Cycle Assessment, Part I: A Critical Review. Environmental Science & Technology 44 (7):2232- 2242.
51
Toward meaningful endpoints of biodiversity in Life Cycle Assessment
52
Chapter 3
Land use impacts on biodiversity in LCA: a global approach
Laura de Baan
Rob Alkemade
Thomas Koellner
Published in The International Journal of Life Cycle Assessment
Volume 18 Issue 6 (2013): 1216–1230
Land use impacts on biodiversity in LCA: a global approach
Abstract
Purpose: Land use is a main driver of global biodiversity loss and its environmental relevance is widely recognized in research on life cycle assessment (LCA). The inherent spatial heterogeneity of biodiversity and its non-uniform response to land use requires a regionalized assessment, whereas many LCA applications with globally distributed value chains require a global scale. This paper presents a first approach to quantify land use impacts on biodiversity across different world regions and highlights uncertainties and research needs.
Materials and methods: The study is based on the UNEP/SETAC land use assessment framework and focuses on occupation impacts, quantified as a Biodiversity Damage Potential (BDP). Species richness of different land use types was compared to a (semi-)natural regional reference situation to calculate relative changes in species richness. Data on multiple species groups were derived from a global quantitative literature review and national biodiversity monitoring data from Switzerland. Differences across land use types, biogeographic regions (i.e. biomes), species groups and data source were statistically analyzed. For a data subset from the biome (Sub-) Tropical Moist Broadleaf Forest, different species based biodiversity indicators were calculated and the results compared.
Results and discussion: An overall negative land use impact was found for all analyzed land use types, but results varied considerably. Different land use impacts across biogeographic regions and taxonomic groups explained some of the variability. The choice of indicator also strongly influenced the results. Relative species richness was less sensitive to land use than indicators that considered similarity of species of the reference and the land use situation. Possible sources of uncertainty, such as choice of indicators and taxonomic groups, land use classification and regionalization are critically discussed and further improvements are suggested. Data on land use impacts were very unevenly distributed across the globe and considerable knowledge gaps on cause-effect chains remain.
Conclusions: The presented approach allows for a first rough quantification of land use impact on biodiversity in LCA on a global scale. As biodiversity is inherently heterogeneous and data availability is limited, uncertainty of the results is considerable. The presented characterization factors for BDP can approximate land use impacts on biodiversity in LCA studies that are not intended to directly support decision-making on land management practices. For such studies, more detailed and site-dependent assessments are required. To assess overall land use impacts, transformation impacts should additionally be quantified. Therefore, more accurate and regionalized data on regeneration times of ecosystems are needed.
Keywords: Biodiversity • Global characterization factors • Land use • LCIA • Regionalization
54 Land use impacts on biodiversity in LCA: a global approach
3.1 Introduction
During the last decades, global biodiversity loss has become a major environmental concern. One of the main drivers of current and projected future biodiversity loss is habitat change or land use (Alkemade et al. 2009; Millennium Ecosystem Assessment 2005; Pereira et al. 2010; Sala et al. 2000). Within research on life cycle impact assessment (LCIA) attempts have been made to quantify the impacts of land use and other important drivers of biodiversity loss, such as climate change and pollution (for a review see Curran et al. 2011). Several approaches on how to quantify land use related biodiversity impacts have been proposed (Achten et al. 2008; Geyer et al. 2010; Kyläkorpi et al. 2005; Koellner 2000; Koellner et al. 2004; Koellner and Scholz 2007; Lindeijer 2000a, b; Michelsen 2008; Müller-Wenk 1998; Penman et al. 2010; Schenck 2001; Schmidt 2008; De Schryver et al. 2010; van der Voet 2001; Vogtländer et al. 2004; Weidema and Lindeijer 2001), of which some have been operationalized in LCA software for broad use by LCA practitioners (e.g. Goedkoop and Spriensma 1999; Goedkoop et al. 2008).
Although the environmental relevance of assessing land use impacts on biodiversity in LCIA is widely recognized, the task remains difficult. Biodiversity is a complex and multifaceted concept, involving several hierarchical levels (i.e. genes, species, ecosystems), biological attributes (i.e., composition, structure, function; Noss 1990) and a multitude of temporal and spatial dynamics (see e.g. Rosenzweig 1995). Biodiversity assessments therefore have to simplify this complexity into a few facets, which are quantifiable with current knowledge and data. Existing land use LCIA methods were mainly developed for one specific region (often Europe) using species richness of vascular plants as an indicator (e.g. Koellner and Scholz 2008; De Schryver et al. 2010). Weidema and Lindeijer (2001) proposed a first approach to assess land use impacts on biodiversity on a global scale, quantifying the biodiversity value of reference habitat of different biomes based on vascular plant species richness, ecosystem scarcity and ecosystem vulnerability. However, the reduction of the biodiversity value of different land use types was estimated based on assumption by the authors and was not supported by empirical data (see Weidema and Lindeijer, 2001, p. 37). To quantify land use impacts across global value chains more accurately a regionalized global method is needed, based on a broader taxonomic coverage. This is required due to the spatial heterogeneity of biodiversity and due to the non-uniform and variable reactions of ecosystems and species to disturbances such as land use. Although plants are important components of terrestrial ecosystems, they only make up an estimated 2 % of all species (Heywood and Watson 1995) and their reaction to land use is not necessarily representative for the impacts on other species groups.
In this paper, we propose a first attempt to quantify biodiversity impacts in LCIA in different world regions based on empirical data, focusing on the facet of species composition. We illustrate how global quantitative analysis of peer-reviewed biodiversity surveys can be combined with national biodiversity monitoring data to assess land use impacts across multiple taxonomic groups and world regions, using a set of species-based biodiversity indicators. The indicator relative species richness is used to calculate characterization factors for occupation impacts of terrestrial ecosystems expressed as a Biodiversity Damage Potential (BDP).
55 Land use impacts on biodiversity in LCA: a global approach
3.2 Methods
This study is based on the framework for Life Cycle Impact Assessment (LCIA) of land use, developed by the UNEP/SETAC Life Cycle Initiative working group (LULCIA; Milà i Canals et al. 2007; Koellner et al. 2013b), which distinguishes three types of land use impacts: transformation impacts (caused by land use change), occupation impacts (occurring during the land use activity) and permanent impacts (i.e. irreversible impacts on ecosystems, which occur when an ecosystem cannot fully recover after disturbance). For calculating transformation and permanent impacts, reliable data on regeneration success and times of the world’s ecosystems is required, which was not available for this study. Therefore, we only focused on occupation impacts and, for modeling purpose, neglected the temporal dynamics of biodiversity by assuming that we can assign a constant “biodiversity score” to occupied land (i.e. no reduction in biodiversity over time) and to a (semi)-natural reference habitat. The impact of land use on biodiversity was assessed by comparing the relative difference of biodiversity of a land use i with a (semi-) natural reference situation. Spatial aspects were considered by using a site-specific reference situation and by calculating impacts per biogeographic region. As proposed in Koellner et al. (2013a) biomes defined by the World Wide Fund For Nature (WWF; see Olson et al. 2001) were used as spatial unit for biogeographic differentiation, which represent the world’s 14 major terrestrial habitat types. Land use was classified based on the UNEP/SETAC LULCIA proposal (Koellner et al. 2013a).
3.2.1 Calculation of characterization factors
Characterization factors of occupation impacts, CFOcc, were calculated according to the UNEP/SETAC framework (Milà i Canals et al. 2007; Koellner et al. 2013b), which is graphically illustrated in the Appendix, Figure B1. CFOcc are given as the difference between the ecosystem quality of a reference situation ref (defined as 100% = 1) and a land use type LUi per region j. In this study, ecosystem quality was expressed as biodiversity, measured as relative species richness Srel (see section 3.2.4).
CF S S 1 S (3.1) Occ,LU = rel,ref ,j − rel,LU = − rel,LU i ,j i ,j i ,j
The numerical value of CFOcc is normally between 0 and + 1 (representing a damaging impact on biodiversity), but negative values are also possible (denoting a beneficial impact). To calculate impact scores for land use occupation, CFOcc is multiplied by the land use occupation flows from a life cycle inventory (given as time tOcc and area AOcc required for a certain land use activity).
Occupation Impact = (Inventory flow land occupation)i CF = (A it )iCF (3.2) Occ Occ Occ Occ Transformation impacts scores are calculated accordingly (Eq. 3.3). Here, the inventory flow is given as a transformed area ATrans and the characterization factor CFTrans is calculated based on Eq.
3.4, with treg being the time required for an ecosystem to recover after a disturbance.
Transformation Impact = (Inventory flow land transform.)iCF = A iCF (3.3) Trans Trans Trans
56 Land use impacts on biodiversity in LCA: a global approach
CF 0.5 (S S ) t 0.5 CF t (3.4) Trans,LU = i rel,ref ,j − rel,LU i reg,LU = i Occ,LU i reg,LU i ,j i ,j i ,j i ,j i ,j
As no reliable data on region and land use type specific regeneration times of biodiversity treg was available for different world regions, CFTrans were not calculated in this study.
3.2.2 Reference situation
Ecosystems and biodiversity are changing over time due to population, succession, and evolutionary dynamics, but also due to intended and unintended human impacts. To quantify land use impacts on biodiversity on a global scale, a temporal baseline or reference situation for biodiversity has to be defined, which lies either in the past, present, or future. Any choice of such a temporal reference involves different degrees of human impacts for different world regions, as the human land use history varies from region to region (see e.g. Ramankutty and Foley 1999). Here we chose the current, late succession habitat stages as reference, which are widely used as target for restoration ecology and serve as a proxy for the Potential Natural Vegetation (PNV), i.e. hypothetical future ecosystems that would develop if all human activities would be removed at once (Chiarucci et al. 2010). Such late succession habitat stages can have experienced different degrees of natural or human disturbances in the past. In many tropical world regions, the past human influence was low, so the chosen reference is to a large extent undisturbed by humans, whereas in many temperate regions few or no undisturbed habitat exists. In Europe, for example, forests currently cover 35% of the surface (SOER Synthesis 2010), whereas the natural post- glacial forest cover (i.e. without human land use) is estimated to be 80 - 90% (Stanners and Philippe 1995). Of the remaining forest area, only 5% is considered as undisturbed forest (SOER Synthesis 2010). Thus, as the reference habitats chosen in this study not necessarily represent pre-human, natural habitats, we use the term “(semi)-natural” to refer to the reference situation. More detail on the data used for quantifying biodiversity of the reference habitat is given in the next section.
3.2.3 Data sources
Two data sources were combined in this study to quantify biodiversity of different land use types and reference situations for different world regions: the GLOBIO3 database, which is based on a quantitative review of literature (Alkemade et al. 2009), and national biodiversity monitoring data of Switzerland (BDM, 2004). The GLOBIO3 database was compiled for the GLOBIO3 model, which aims at assessing impacts of multiple drivers of biodiversity loss at regional and global scales (Alkemade et al. 2009). The database contains datasets extracted from peer-reviewed empirical studies that compare biodiversity of different land use types with an undisturbed or little disturbed reference situation within the same study site. Depending on data provided in each study, the impact of land use is recorded as relative change in species richness or abundance of a range of different taxonomic groups. For each study, we additionally extracted the geographical coordinates of the study site to assign it to the corresponding WWF biome and ecoregion. A total of 195 publications, providing 644 data points on different land use types and 254 data points on
57 Land use impacts on biodiversity in LCA: a global approach
reference situations from a total of 9 out of 14 biomes were included here, but the data was unevenly distributed. Due to publication bias and lack of undisturbed reference habitats in regions with long and intense human land use history, the database contains many studies conducted in tropical regions and less data in temperate and none in boreal zones (for geographical distribution of data see Appendix Figure B1 and Table B1). We therefore complemented our analysis with national biodiversity monitoring data of Switzerland (BDM 2004) used in earlier land use LCIA methods (Koellner and Scholz 2008). The used BDM indicator “species diversity in habitats (Z9)” is based on a grid of 1’600 sampling points evenly distributed over Switzerland, covering two biomes (Temperate Broadleaf & Mixed Forests and Temperate Coniferous Forests). In each of the 10 m2 sampling points, species richness of vascular plants, moss, and mollusks and the corresponding land use type are recorded. To make this dataset comparable to the GLOBIO3 data we first reclassified the land use type of each sampling point based on Koellner and Scholz (2008) into broader land use classes (see Appendix Table B6). We then grouped all sampling points into ecologically similar regions to define regional (semi-) natural reference situations. We split the 10 biogeographic regions of Switzerland defined in BDM (2004) into three altitudinal zones (colline: below 800m a.s.l.; montane: 800 - 1300 m; subalpine: 1300 - 2000 m; see Baltisberger 2009) and excluded the high elevation plots (alpine and nival: above 2000m). This resulted in 26 regions j across Switzerland, as not all altitudinal zones occur in every biogeographic region. For each of the 26 regions and for each of the three sampled species groups, the average species richness of all sampling points per land use type was calculated, resulting in totally 186 averaged data points for different land use types (see also Appendix Table B1). All sampling points in (semi-)natural habitats (forests, grasslands, wetlands, bare areas, and water bodies) were assigned as regional reference situation (for more detail see Appendix Table B6). As for the land use types, the average species richness per region and species group was calculated for the reference, resulting in 72 data points for the reference. To test the sensitivity of choice of reference situations, results were recalculated using an alternative reference habitat containing only forest sampling points.
3.2.4 Indicator selection and calculation
As a primary indicator for biodiversity impacts we chose relative changes in observed species richness Srel between a (semi-)natural reference and a specific land use type i. For each taxonomic group g and region j the species richness of the reference Sref was divided by the species richness of each land use type i, SLUi , (Eq. 3.5). For the BDM dataset, the regionally averaged species richness of the land use types and the reference were used for calculating the relative species richness.
S LU S i ,j ,g (3.5) rel,LU = i ,j ,g S ref j ,g
The selected indicator species richness is a simple and widely applied indicator recording the number of species in a habitat (also referred to as α-diversity or within habitat diversity; Hayek
58 Land use impacts on biodiversity in LCA: a global approach
and Buzas 2010) and data availability is rather high compared to other biodiversity indicators. The disadvantage of using species richness as a proxy for biodiversity is that it only contains limited information on the many facets of biodiversity. It only records the presence or absence of species within a sampling area and gives equal weight to all species recorded in a sample, no matter how abundant or biologically distinct they are (i.e. 10 individuals of an endemic species and 1 individual of an invasive species are both recorded as 1 species). Species richness neither provides information on between-habitat diversity, i.e. species turnover or β-diversity (see Koellner et al. 2004). This indicator is in addition affected by undersampling: the species richness of an ecosystem is often underestimated as the number of species recorded highly depends on sampling efforts.
Besides species richness, a wide range of diversity measures have been developed, each quantifying other aspects of biodiversity (see e.g. Hayek and Buzas 2010; Purvis and Hector 2000). To analyze the influence of choice of indicator on the results, we calculated four additional, commonly used species based biodiversity indicators: Fisher’s α, Shannon’s entropy H, Sørensen
SS and Mean species abundance of original species (MSA) (see formulas in Table 3.1). Fisher’s α (Fisher et al. 1943) is an indicator that corrects for incomplete sampling: it estimates “true” species richness from a sample, fitting the observed values of species richness Sobs and total number of individuals Nobs to a theoretical (empirically derived) relationship between “true” species richness S and “true” number of individuals N. Shannon’s entropy H (Shannon 1948) combines information on species abundance and richness in one number and reaches a maximum when all species occurring in a sample are equally abundant. Sørensen SS (Sørensen 1948) and mean species abundance of original species (MSA, Alkemade et al. 2009) both compare the species composition of two samples (here the reference and a land use type i). Sørensen reports how many reference-habitat species occur in the land use type i and reaches a maximum value of 1 if all of them occur in the land use type i and a minimum value of 0 if none of the reference-habitat species occur in the land use type i. MSA, which has been developed for the GLOBIO3 model (Alkemade et al. 2009), assesses changes in abundance of each reference- habitat species and thus reports changes in species composition earlier than Sørensen, which only indicates a complete absence of a species from a site.
Besides the number of species S, these indicators all require additional information such as species identity (i.e. checklist of species present) and/or abundance (number of individual organisms nk, per species k or total individual organisms N per sample). This additional information complicates the process of data collection and was only available in parts of the studies in the GLOBIO3 database. We therefore performed this indicator comparison with a subset of the data: we chose all those studies from the biome (Sub-) Tropical and Moist Broadleaf Forest (i.e. “tropical rain forest”) in which a full species list indicating the abundance of each species in different land use types and a (semi-)natural reference was provided. The species abundance lists of these studies were extracted to Microsoft Excel to calculate the selected biodiversity indicators (see Table 3.1). Two indicators (Mean species abundance, MSA, and
Sørensen SS) directly calculate the relative change between a land use type i and a reference, for the other three indicators (species richness, Shannon’s entropy and Fisher’s α), the relative
59 Land use impacts on biodiversity in LCA: a global approach
values per land use type LUi and taxonomic group g within each study j were calculated as follows:
I LU I i ,j ,g (3.6) rel,LU = i ,j ,g I ref j ,g
The numerical values range from 0-1 for the two indicators MSA and Sørensen SS, whereas Irel of the other three indicators species richness, Shannon’s entropy and Fisher’s α allow values above 1. For studies containing data from several reference situations, relative indicators were calculated for all possible combinations of references and land use types and also within references, giving an additional estimate of uncertainty. Hence, the reference situation was not fixed at 1 as was the case for the data on Srel from the full dataset (BDM and GLOBIO3 database), where multiple reference plots per study site were averaged before the calculation of the relative indicator. This resulted in a final number of 168 (pairwise) data points for the reference and a total of 337 for all land use types.
Table 3.1 Biodiversity indicators calculated for a subset of studies from the biome (Sub-)Tropical Moist Broadleaf Forest. For the indicators marked with(*), the presented formulas are for calculating the absolute indicator values. Relative values are derived from Eq. 3.5.
Indicator type Name and reference Data requirement Formula
Alpha diversity Species richness S(*) Species numbers n.a.
Sampling corrected Fisher’s α(*) Species numbers ⎛ S ⎞ ⎜e α −1⎟ alpha diversity (Fisher et al. 1943) and total number of N ⎝ ⎠ = individuals S S α
(*) Diversity measure Shannon’s entropy H List of species and H = − p *lnp ∑ k (Shannon 1948) their rel. abundance
Abundance measure Mean species abundance List of species, 1 p *N MSA = k, LUi LUi of original species (MSA) original species and S ∑ p *N ref k k, ref ref (Alkemade et al. 2009) their rel. abundance for all species k ∈ref
Dissimilarity measure Sørensen SS List of species 2c S = (Sørensen 1948) s S + S LUi ref
LUi = land use type i; ref = (semi-)natural reference; S = Number of species; c = number of shared species between two land use types; N = total number of individuals; nk= number of individuals of species k; p = n N = relative abundance k k of species k.
3.2.5 Statistical analysis
Analysis of Variance (ANOVA) was used to analyze the differences in mean relative species richness i, depending on the four factors land use type (LU), taxonomic group (Taxa),
60 Land use impacts on biodiversity in LCA: a global approach
biogeographic region (Biome), and data source (i.e. GLOBIO or BDM; Data), including the interaction of factors. Following model structure was tested:
S = f (LU; Biome; Taxa; Data; LU×Biome; LU× Taxa; Biome× Taxa; LU×Data; rel (3.7) Biome×Data; Taxa×Data; LU×Biome× Taxa; LU×Biome×Data; LU× Taxa×Data)
As the data did not follow the assumption of normal distribution, we additionally applied the
Kruskal–Wallis Test to test the difference of medians of Srel of the four factors (without interaction). Mann–Whitney U Test was conducted for pairwise comparison of median Srel of different land use types.
For each of the five indicators Irel (Table 3.1 and Eq. 3.6) calculated for a subset of data, the differences in means for the three factors LU, Taxa and biogeographic region (Realm) and their interactions were assessed with ANOVA with following model structure:
I = f (LU; Taxa; Realm; LU× Taxa; LU×Realm; Taxa×Realm; LU× Taxa×Realm) (3.8) rel
As with the total dataset, robustness of results was assessed with nonparametric Kruskal-Wallis Tests and Mann-Whitney U Tests. In addition, Pearson’s correlation between indicators was calculated. All data analysis was carried out using R statistical package v2.11 (R Development Core Team 2011).
3.3 Results
3.3.1 Land use impacts on biodiversity
Characterization factors of land occupation CFOcc for BDP were calculated according to Eq. 3.1 and are shown in Table 3.2 and in the Appendix Table B1. For easier interpretation of results, the biodiversity indicator relative species richness Srel is chosen for graphical display (Figures 2.1-
2.3). The CF can be derived by subtracting the median Srel from 1 (see Eq. 3.1).
Averaged across all regions and taxa, relative species richness Srel of all land use types was significantly lower than in the reference, but results varied strongly from negative impacts (Srel <
1) to positive impacts (Srel > 1) (see Figure 3.1). The strongest negative impact was found in annual crops, where Srel was reduced by 60%, followed by permanent crops and artificial areas (40% decreased Srel). In pastures the reduction of Srel was around 30%, in secondary vegetation, used forests and agroforestry around 20%. A pairwise comparison of the difference of median Srel of different land use types is given in the Appendix Table B2.
A significant effect on Srel of land use (LU), taxonomic group (Taxa) and biogeographic region (Biome) and a non-significant effect of the source of data (GLOBIO or BDM) were found for the full dataset both in ANOVA (Table 3.3) and Kruskal-Wallis Test (results not shown). In the ANOVA, land use effects on Srel differed significantly between biomes (LU x Region) and taxa (LU x Taxa), but not between data source (LU x Data). The latter was supported by Mann-Whitney U Tests,
61 Land use impacts on biodiversity in LCA: a global approach
which did not show any significant difference (p<0.05) in Srel between the two data sources for any land use type (results not shown).
Table 3.2 World average and regionalized characterization factors CF (median) and their uncertainties (1. and 3. quartiles) for Biodiversity Damage Potential (BDP) per land use type. CF of four selected biomes are displayed, a full list of CF per biomes and taxonomic groups can be found in the Appendix Table B1. For land use types with less than 5 data points (N), no CF is provided.
areas
not used not
forestry
Forest, Secondary vegetation used Forest, Pasture/ meadow crops Annual Permanent crops Agro Artificial
Total world Median 0 0.18 0.18 0.33 0.60 0.42 0.20 0.44 average 1.Quartile 0 -0.03 -0.05 0.00 0.31 0.06 0.01 -0.01
3.Quartile 0 0.37 0.50 0.55 0.79 0.70 0.48 0.62
N 326 272 148 133 96 52 76 53
Biome 1 (Sub-) Median 0 0.22 0.13 0.45 0.54 0.42 0.18 - Tropical Moist 1.Quartile 0 0.00 -0.09 0.31 0.36 0.18 -0.02 - Broadleaf Forest 3.Quartile 0 0.43 0.45 0.75 0.72 0.70 0.44 -
N 173 172 79 26 46 40 70 1
Biome 4 Median 0 0.08 0.22 0.52 0.76 0.02 - 0.40 Temperate 1.Quartile 0 -0.26 -0.09 -0.35 0.46 -0.11 - -0.10 Broadleaf Forest 3.Quartile 0 0.33 0.43 0.67 0.86 0.69 - 0.58
N 46 20 35 33 24 9 0 24
Biome 5 Median 0 0.17 0.15 0.24 0.54 - - 0.50 Temperate 1.Quartile 0 -0.22 0.02 -0.64 -0.15 - - -0.05 Coniferous Forest 3.Quartile 0 0.30 0.33 0.38 0.87 - - 0.71
N 45 15 7 27 8 3 0 21
Biome 7 (Sub-) Median 0 0.00 0.01 0.12 0.65 - - - Tropical 1.Quartile 0 -0.17 0.00 0.02 0.02 - - - Grassland & Savannah 3.Quartile 0 0.15 0.06 0.27 0.80 - - -
N 21 27 6 8 9 0 0 0
62 Land use impacts on biodiversity in LCA: a global approach
)
rel S
Outlier Upper whisker
75% Median 25%
relative species richness ( relative Lower whisker 0.0 0.5 1.0 1.5 2.0 2.5 3.0 n= 96 n= 52 n= 53 n= 76 n= 326 n= 272 n= 148 n= 133 p<0.01 Pasture Pasture p<0.001 p<0.001 p<0.001 p<0.001 p<0.001 p<0.001 Reference Reference Used forest Used forest Agroforestry Artificial area Annual crops Annual Permanent crops Permanent Second. vegetation Second. vegetation Figure 3.1 Box and whisker plot of relative species richness per land use type, number of data points n per land use type, and test statistics (1-sided Mann-Whitney U Test) of pairwise comparison of each land use with the reference for the full dataset (global averages across all biomes and taxonomic groups)
3.3.2 Regionalization
Data from nine biomes were included in the analysis, but the majority of studies provide data on land use of biome Tropical Moist Broadleaf Forests (Appendix Table B1). For many combinations of land use types and biomes, no or too little data was available to draw conclusions. Due to the inclusion of Swiss BDM data, the number of data points for two temperate biomes (Broadleaf & Mixed Forests and Coniferous Forests) was considerably improved. The relative species richness of the four biomes with the highest data availability is displayed in Figure 3.2. A significantly different land use impact across biomes was only found for three land use types (secondary vegetation, used forest and pasture; Kruskal-Wallis Test, p<0.05). All land use types in all biomes showed a median negative land use impact (Srel < 1), with one exception. Pastures in the biome
“Deserts & Xeric Shrublands” showed a slight positive median land use impact (Srel increased by 8%, Appendix Table B1), but the small number of data points (n=5) and the large variation in data does not allow for strong conclusions here. In general, large within biome variations were observed.
63 Land use impacts on biodiversity in LCA: a global approach
Table 3.3 Results of ANOVA testing the difference in mean Srel of the full dataset depending on land use type (LU), taxonomic group (Taxa), biogeographic region (Region=Biome or Realm) and data sources (Data) and their interactions. Model of full dataset, see Eq. 3.7; model for subsets (GLOBIO, BDM and Tropical data), see Eq. 3.8. MBF: Moist broadleaf forest; Df : degrees of freedom ; *** p-values<0.001; ** p-values<0.01; * p-values<0.05; (*) p-values<0.1; ns p-values > 0.1
Full GLOBIO BDM Subset: Biome (Sub)-Tropical dataset data data MBF data
α s ’
rel rel rel rel rel) Df S Df S Df S Df S MSA Sørensen (rel) Shannon Fisher (
LU 7 *** 7 *** 6 *** 7 *** *** *** *** ***
Region1 8 *** 8 *** 1 ns 4 *** *** *** *** ***
Taxa2 3 *** 3 *** 2 *** 2 * ** *** (*) *
Data 1 ns ------
LU x Region 38 *** 33 ** 6 * 20 *** *** *** * ***
LU x Taxa 20 *** 19 *** 12 *** 11 (*) ** *** ns ns
Region x Taxa 17 (*) 16 *** 2 ns 5 ns ns ns ns ns
LU x Data 4 ns ------
Region x Data 1 (*) ------
Taxa x Data 1 ns ------
LU x Region x Taxa 45 (*) 34 * 12 ns 8 ns ns ns ns ns
LU x Region x Data 1 ns ------
LU x Taxa x Data 2 ns ------
1 Biomes were used as the factor for testing regional differences for all datasets, except for the subset of data from the biome (Sub-)Tropical Moist Broadleaf forest, where biogeographic realms (i.e. continents) were used for regionalization.
2 For BDM data, the factor levels of Taxa were vascular plants, moss and mollusks. For the other datasets, the levels were plants, vertebrates, arthropods, and other invertebrates.
64 Land use impacts on biodiversity in LCA: a global approach
Biome TropMBF )
Biome TropGL
rel Biome TempBLF S Biome TempCF relative species richness ( relative 0.0 0.5 1.0 1.5 2.0 2.5 3.0 (a) n.s. (a) n.s. (a) n.s. (a) n.s. (b) n.s. (a) n.s. (b) n.s. (a) n.s. (b) n.s. (b) n.s. (a) n.a. Pasture Pasture Reference Reference (a) p<0.01 (b) p<0.05 (b) p<0.01 (a) p<0.01 (b) p<0.01 Used forest Used forest Agroforestry Artificial area Annual crops Annual Permanent crops Permanent Second. vegetation Second. vegetation Figure 3.2 Box and whisker plot of relative species richness per land use type and four selected biomes and test statistics of Kruskal-Wallis Test (Srel = f(LU x Biome)) for (a) displayed biomes and (b) all biome in full dataset. n.s.: not significant; n.a.: not applicable (just one biome per land use type); TropMBF: (Sub-)Tropical Moist Broadleaf Forest; TropGL: (Sub-)Tropical Grassland, Savannas and Shrublands; TempBLF: Temperate Broadleaf &Mixed Forest; TempCF: Temperate Coniferous Forest
3.3.3 Indicator comparison
3.3.3.1 Comparing impacts across taxonomic groups Data on different species were aggregated into broad taxonomic groups to get enough data points per group and land use type. The global dataset from GLOBIO3 contains a broad range of taxonomic groups, whereas the Swiss BDM dataset only contains data on plants (vascular plants and moss) and invertebrates (mollusks, see Appendix Table B5). To achieve more or less globally averaged results, we further aggregated groups that were mainly consisting of data from the Swiss BDM. Finally, four classes of taxonomic groups were distinguished: plants, arthropods, other invertebrates and vertebrates. A significantly different (p<0.001) land use impact across taxonomic groups was found in the full dataset for agroforestry and a slight difference (p<0.1) for used forest (Figure 3.3). Overall, plants and invertebrates (excluding arthropods) showed a slightly stronger land use effect than arthropods and vertebrates, but this pattern was not found across all land use types. The variation within taxonomic group was considerable. Therefore, we further separated two groups, neglecting the over-representation of the Swiss data: plants were split into moss and vascular plants and vertebrates were split into birds and other vertebrates, resulting in a total of 6 taxonomic groups (Appendix Figure B3 and Table B1). With this finer scaled classification, significantly different (p<0.05) land use impacts across taxonomic groups were found for all land use types except for secondary vegetation, where only a slight difference
65 Land use impacts on biodiversity in LCA: a global approach
(p<0.1) between taxonomic groups was found (Appendix Figure B3). Here, moss and other invertebrates showed the strongest land use impacts, with roughly a 50-90% median reduction in
Srel in pasture, annual crops and artificial area. A strong positive land use impact (42% increased
Srel) on vascular plants was found in artificial areas. The underlying data (n=16) was purely based on the Swiss BDM, and data from very heterogeneous land use types were included (from dump sites to urban green areas). Therefore, we further split the land use type artificial area in the Swiss BDM data into low and high intensity (Appendix Figure B4). However, no significant differences in Srel between the high and low intensity artificial areas were found for the BDM data with a Mann–Whitney U Test. Within the Swiss BDM data, vascular plants were generally less affected by land use than moss and mollusks (Appendix Figure B4), and showed an increased median relative species richness in pasture, permanent crops and artificial areas. Moss and mollusks showed a decreased relative species richness in all land use types.
Plants ) Arthropods rel
S Other invertebrates
Vertebrates relative species richness ( relative 0.0 0.5 1.0 1.5 2.0 2.5 3.0 n.s. n.s. n.s. n.s. n.s. n.s. n.s. p<0.1 Pasture Pasture p<0.001 Reference Reference Used forest Used forest Agroforestry Artificial area Annual crops Annual Permanent crops Permanent Second. vegetation Second. vegetation Figure 3.3 Box and whisker plot of relative species richness per land use type and taxonomic group and test statistics of Kruskal-Wallis Test (Srel = f(LU x Taxa)) for full dataset. n.s.: not significant
3.3.3.2 Comparing impacts across biodiversity indicator For a subset of data from the biome (Sub-)Tropical Moist Broadleaf forest, four additional indicators were calculated: Fisher’s α, Shannon’s entropy H, Sørensen SS, and Mean species abundance of original species (MSA, see Table 3.1). For all land use types, the impacts varied significantly across indicator (Figure 3.4). Relative species richness was highly correlated with relative Shannon’s H (Pearson’s r=0.79) and relative Fisher’s α (Pearson’s r=0.83, see also Appendix Table B3). This group of indicators showed less negative (or even positive) land use impacts compared to a second group of indicators, Sørensen’s SS and mean species abundance MSA, which were also highly correlated (Pearson’s r=0.81). In Figure 3.4, the reference situation shows a considerable within study variation, calculated as the relative difference in biodiversity
66 Land use impacts on biodiversity in LCA: a global approach
indicators of multiple reference situations given for individual studies. This variation was not calculated for the full dataset (see Figures 3.1-3.3), where the average of multiple references were used to calculate relative changes in species richness.
Species rich. (rel) Shannon (rel) Fishers (rel) Sørensen MSA Indicator value 0.0 0.5 1.0 1.5 2.0 2.5 3.0 n= 29 n= 40 n= 13 n= 35 n= 14 n= 54 n= 168 n= 152 p<0.01 Pasture Pasture p<0.001 p<0.001 p<0.001 p<0.001 p<0.001 p<0.001 p<0.001 Reference Reference Used forest Used forest Agroforestry Annual crops Annual Sec. forest, old Sec. forest, Permanent crops Permanent Sec. forest, young young Sec. forest,
Figure 3.4 Box and whisker plot of different indicators per land use type and test statistics of
Kruskal-Wallis Test (Irel = f(LU x Indicator)) for a subset of data from biome (Sub-)Tropical Moist Broadleaf Forest. Secondary forest was divided into young (<30 years) and old growth forest (>30 years). n: Number of data points (pairwise comparisons) per land use type and indicator.
3.4 Discussion
Biodiversity is a multi-facetted concept and it is difficult to express product related impacts of land use in a single indicator value. Our analysis illustrated the variability of results, ranging from positive to negative impacts of land use, but we also found an overall negative median impact on relative species richness across all analyzed land use types. Land use impacts differed significantly across taxonomic groups and biogeographic regions, but could not be determined for all world regions due to data limitations. The choice of indicator also strongly influenced the outcome, where relative species richness was less sensitive to land use impacts than MSA or Sørensen. In the following sections we want to highlight the uncertainties, limitations and opportunities for improvements of this first attempt to quantify land use impacts on biodiversity in LCIA on a global scale based on empirical data. We discuss the suitability of different indicators for use in land use LCIA, appropriate coverage and classification of taxonomic groups, land use types and biogeographic regions and finally address general uncertainties of the presented approach.
67 Land use impacts on biodiversity in LCA: a global approach
3.4.1 Choice of indicator
From a practical point of view, species richness might be the indicator of choice for biodiversity assessments on species level: data is relatively readily availability, as the data requirements are low compared with other indicators, which need additional information on abundance and/or species composition. However, from a theoretical point of view, the indicator species richness has many shortcomings. Here, we discuss four alternative indicators analyzed in this study which partly overcome some of the disadvantages of species richness, such as (i) high dependence on sampling effort, (ii) missing information on abundance, (iii) no link to conservation targets, and (iv) missing information on species turnover.
First, species richness is highly dependent on sampling size, whereby a non-linear relationship between area sampled and species richness has been observed (due to a species sampling and a species area relationship; Arrhenius 1921; Dengler 2009). To compare absolute species numbers of different land use types, species richness has to be standardized to the same sampling effort (Koellner and Scholz 2008; Schmidt 2008). This proves to be difficult or even impossible when dealing with different taxonomic groups that are surveyed with very different sampling methods (e.g. visual observations of birds along 50m transects; 20 pitfall-traps of arthropods; or plant counts on 10 m2 plots). Therefore, we divided absolute species numbers of each taxon of every land use type i by the absolute species numbers of a regional reference to obtain relative species richness (given that both absolute numbers were obtained with a similar sampling effort). This approach partly circumvents effects of sampling bias. However, a bias remains in cases where the land use type i and the reference show very different species turnover (e.g. homogeneous species composition of arable field vs. heterogeneous rain forest). In such cases, the relative land use impact is underestimated with small sample size, as most species of the arable field are captured, but only a small share of the species richness of the rain forest is recorded.
Several biodiversity indicators have been developed that correct for incomplete sampling (see e.g. Beck and Schwanghart 2010). In our analysis we applied Fisher’s α (Fisher et al. 1943, see formula in Table 3.1) and found a high correlation between land use impacts measured as relative species richness and as relative Fisher’s α (see Figure 3.4). This finding is supported by the empirical study of Kessler et al. (2009), which did not find a strong influence of sampling incompleteness on land use impacts. This indicates that correcting for undersampling might not be the most important aspect to reduce overall uncertainty of biodiversity related land use LCIA.
A second shortcoming of species richness is the missing information on abundance. Shannon’s entropy H, derived from information theory, expresses abundance and richness in one number (see formula in Table 3.1) and reaches a maximum value when all species occur equally abundant. As in our study relative Shannon’s entropy H was highly correlated with relative species richness (see Figure 3.4), the latter might be preferred as indicator, as it is easier to communicate to LCA users or the general public.
Mean species abundance (MSA), an indicator developed for the GLOBIO3 model (Alkemade et al. 2009), is correcting the second and third shortcoming of species richness as it includes abundance and is linked to conservation targets. MSA compares the abundance of “original”
68 Land use impacts on biodiversity in LCA: a global approach
species occurring in natural, undisturbed habitats, in their primary “original” habitat with their abundance in secondary habitats (i.e. on the land use type i). As expected, our results showed that land use impacts were stronger when measured with MSA than with Srel (see Figure 3.4), indicating that the original species adapted to undisturbed habitats are more susceptible to land use changes than species adapted to disturbance. MSA is therefore suitable to report land use impacts in regions, where conservation targets are mainly focusing on protection of primary habitats. In areas such as central Europe, where conservation is mainly targeting at protecting species adapted to traditional land use practices, the definition of “original” species needs to be extended to these species. To base future land use LCIA methods on MSA, the habitats or species relevant for conservation have to be defined for all world regions, whereby the value choices unavoidably involved in this definition need to be critically reflected.
Similar to MSA, Sørensen SS can measure the similarity of the species composition of a land use type and a reference situation but without considering abundance. As they were calculated in this study, both MSA and Sørensen reached a maximum value of 1, when the land use type had a maximum similarity (i.e. the same species composition as the reference) and the two indicators were therefore highly correlated (see Figure 3.4). As expected, the land use impacts measured with Sørensen were smaller than with MSA, as MSA already reports a decreasing abundance of species, whereas Sørensen only reports if a species is not present anymore. Sørensen SS can also be used to calculate similarity of species composition within a land use type or reference, i.e. giving information of species turnover (or β-diversity). In that case, a maximum β-diversity score would be reached within a land use class or reference with minimum average similarity between samples, indicating high rates of turnover. This would require data on species composition of multiple plots of the same land use and multiple plots of a reference within one study site − or studies directly reporting β-diversity. As β-diversity can play a key role in biodiversity conservation (Gardner et al. 2010), this information is increasingly available and might open the way to use this indicator in future biodiversity LCIA.
Biodiversity impacts can be assessed in relative or absolute terms, which finally represents different value choices: if we assess absolute impacts, all species are equally weighted, if we assess relative impacts, all ecosystems get equal weight. The indicators calculated in this study all assess relative impacts. As explained above, this was required to standardize the data from a multitude of studies with different sampling design and species groups. As a consequence, a 40% decrease of relative species richness in a species rich ecosystem (e.g. with 100 species) and in a more species poor ecosystem (e.g. 10 species) are weighed evenly, although the absolute reduction in species richness is much higher in the species rich ecosystem (40 vs. 4 species). To account for regional differences in absolute species richness, a weighing system of land use could be applied as suggested by Weidema and Lindeijer (2001). Global conservation priorities could help to develop such a weighing scheme, using weighing factors such as regional species richness, irreplaceability and vulnerability of ecosystems (Brooks et al. 2006).
To conclude, we think that - in view of current data availability - relative species richness, as an indicator for α-diversity, is a suitable indicator for biodiversity related land use LCIA. If future
69 Land use impacts on biodiversity in LCA: a global approach
research progress allows quantifying land use related impacts on β-diversity or changes in abundance of species important for conservation, indicators such as MSA or Sørensen should be preferred. To also account for regional differences of absolute species richness, a weighing of the presented CFs is required. Here, only a few facets of biodiversity were considered, with a focus on species composition. Including other facets, for example land use impacts on ecosystem functioning (see e.g. Michelsen 2008; Wagendorp et al. 2006) or on ecosystem services (Müller- Wenk and Brandão 2010; Saad et al. 2013; Brandão and Milà i Canals 2013), would be an important complement of this method.
3.4.2 Taxonomic coverage
Attractive species groups, such as mammals, birds, or butterflies are often used as indicator taxa in biodiversity assessments, with the underlying hope “that the known biodiversity is a good surrogate for the unknown” (Rodrigues and Brooks 2007, p. 714). Data availability is therefore biased towards some well-studied species groups. Existing land use LCIA mostly focused on vascular plant species richness as an indicator (e.g. Koellner and Scholz 2008; Schmidt 2008; De Schryver et al. 2010). This makes a method very transparent, but the potential to generalize results from one well-studied species group to biodiversity as whole is questionable (Purvis and Hector 2000). Empirical studies from different world regions found little predictive power of one species group for other species groups (e.g. Billeter et al. 2008; Kessler et al. 2009; Wolters et al. 2006).
In this study, we combined data from global literature review, covering a range of taxonomic groups (see Appendix Table B5), with data from Swiss biodiversity monitoring BDM, containing data on vascular plants, moss, and mollusks. Although being more representative than previous LCIA studies, a publication bias towards some well studied species groups remained. If we compare the share of species groups in our analysis with their estimated global species richness (Heywood and Watson 1995; see Appendix Table B4), we find that vertebrates (26% of data) and plants (43%) were strongly overrepresented, as they only make up 0.4% and 2% respectively of estimated global species richness. With 20% of data points, arthropods were underrepresented in this study as they make up an estimated 65% of global species richness. Some species groups, such as bacteria (7% of estimated global species richness) or fungi (11%) were not at all represented in the used dataset.
Ideally, the impact of different land use types on each target species group in each biogeographic region should be separately assessed. This could later be aggregated into characterization factors for archetypical groups of species, regions and land use types showing similar land use effects. In this study, we present one possible classification, but due to limited data availability we could not make a thorough analysis of different classification options nor recommend an optimal classification, where the variation within each characterization factors is minimized (i.e. representing a homogeneous group). We first divided data into four very broad taxonomic groups (plants, arthropods, other invertebrates and vertebrates), and then further subdivided plants into moss and vascular plants and vertebrates into birds and other vertebrates. A further subdivision
70 Land use impacts on biodiversity in LCA: a global approach
was not possible, as too little data points were then given for each land use type. Especially for plants, the separation into moss and vascular plants revealed very diverging impacts of these subgroups (Appendix Table B1 and Figure B3), suggesting that these groups should be assessed separately. However, it is unclear to which extent this conclusion is valid for other world regions, as for many land use types the data on moss is purely derived from the Swiss BDM (see Appendix Table B5).
Here we chose a classification based on phylogenetic relationship, but closely related species not necessarily show homogeneous reactions to land use (see e.g. Attwood et al. 2008; Blaum et al. 2009; Anand et al. 2010). To find an optimal representation of impacts across species groups other classification criteria, such as functional traits (e.g. morphological, ecophysiological and life history characteristics, see e.g. Vandewalle et al. 2010) or feeding-guilds (see e.g. Scherber et al. 2010) should be tested as alternative grouping factor for species groups.
As data on all species will probably never be available, we need to find the optimal taxonomic coverage for land use LCIA. This requires a clear definition of the target of land use LCIA (i.e. why we want to conserve biodiversity; see also Michelsen 2011). If we aim at conserving biodiversity due to its intrinsic value or due to its potential future economic value (e.g. as medicine), threatened species should get higher weights and species groups selected for LCIA should be proportional to their total richness. If the target is to sustain ecosystem services, we need to conserve functional diversity (and assess land use impacts on important species of ecosystems). However, this requires a more sound understanding of the underlying ecosystem processes, e.g. on how ecosystems react if a certain species occurs more or less abundantly. In addition, better knowledge on vulnerability and potential tipping-points of ecosystems (i.e. non-linear reactions of ecosystems after certain levels of accumulated multiple disturbances, see e.g. Holling 2001; Scholz 2011) is required. Resolving the important normative question of setting appropriate targets for biodiversity assessments within LCA and of finding the right proxy for it remains a challenge for future research.
3.4.3 Land use classification and regionalization
As outlined above, characterization factors ideally should represent archetypical land use impacts on species groups, but also of land use classes and regions showing similar impacts. In our study, very broad land use types were classified showing considerable within class variation of effects. Including further data points would allow to separate intensive and extensive land use (e.g. for agriculture) and could potentially reduce this variation and improve the validity of the characterization factors. However, in the case of artificial area in Switzerland, no significant differences between high and low intensive artificial area were found (Appendix Figure B4). Caution should be taken with extrapolating the findings for artificial area, which are largely based on the Swiss BDM data, to other world regions.
The question of appropriate classification also applies to regionalization. Here, we chose WWF biomes as spatial units as a coarse regionalization scale with ecologically distinct regions. Due to limited data availability, it was not possible to have a more fine scaled regionalization of relative
71 Land use impacts on biodiversity in LCA: a global approach
impacts. However, a weighing of these relative impacts, as suggested above, could be done on ecoregion level, using for example data on species richness of different taxa (see data of Olson et al. 2001; Kier et al. 2005). As significant differences in land use impacts were not only found across biomes (full dataset), but also across biogeographic realms (subset (Sub-)Tropical Moist Broadleaf Forest, see Table 3.3), a further distinction of biomes across realms might better reflect differences in relative impacts. The analysis of the Swiss BDM data, covering two biomes, showed no significant difference between their reactions to land use. This suggests that not only the broad ecosystem type is important to determine land use impacts, but also the geographical proximity or similarity of land use history. Of course, aspects of practicality also need to be considered when choosing an optimal scale of regionalization. To finally assess land use impacts in LCA, not only the characterization factors have to be regionalized, but also the inventory data. How the presented CFs can be applied is illustrated in a case study on margarine by Milà i Canals et al. (2013).
As for taxonomic groups, data availability of land use impacts on biodiversity is biased towards some biogeographic regions, with data dominantly derived from a few well-studied research stations in tropical regions (see Gardner et al. 2009; 2010). In addition, some ecosystem types, such as grassy ecosystems, received less attention of researchers than forest ecosystems (Bond and Parr 2010). The uneven regional distribution is also visualized in the data distribution of this study (see Appendix Figure B2). Very little or no data was available for following five out of fourteen biomes: (Sub)-Tropical Coniferous Forests, Boreal Forests/Taiga, Flooded Grasslands & Savannas, Tundra, and Mangroves (see Appendix Table B1 and Figure B2). For three biomes, enough data was only provided for pastures (Temperate Grassland & Savannah, Mediterranean Forests, Woodlands & Scrub, and Deserts & Xeric Shrublands). For permanent crops, agroforestry and artificial areas, data was only available from two biomes. In general, the biome (Sub)-Tropical Moist Broadleaf Forest had the highest data availability. The two temperate biomes Mixed & Broadleaf and Coniferous Forest also showed a reasonable amount of data, but as this was mainly derived from Swiss BDM data, results are highly biased towards the European context. To which extent these results are valid for temperate forest biomes in other world regions remains a question for future research.
3.4.4 Data limitations and uncertainties
In this study we combined global literature data with national biodiversity monitoring data. Both datasets have different sources of uncertainties. Summarizing data from multiple studies involves consideration of within and between study variance (Gurevitch and Hedges 1999). As it was beyond the scope of this study to perform a full statistical meta-analysis, only between study variation was considered. Therefore, the overall assessment on relative species richness suggests no variation of the reference habitat (see Figures 3.1 – 3.3), which does not reflect reality. For the subset of data used to compare biodiversity indicators, the within-study variance was included when studies reported data on multiple reference habitats. The considerable variation of indicator values of reference habitats observed in the subset of data (see Figure 3.4)
72 Land use impacts on biodiversity in LCA: a global approach
suggests that variation of results (including within-study variance) of the full dataset was underestimated.
For the Swiss BDM data, the main sources of uncertainties are the definition of ecologically similar regions (see section 3.2.3) and the definition of reference habitat. It was beyond the scope of this study to test the sensitivity of results to choice of boundary of regions. However, for the definition of reference habitat per region, the sensitivity of results to selection of two different reference situations was tested. We compared the outcomes for using (i) a combination of all potential natural habitats (n=305 monitored sampling points in forests, grasslands, wetlands, bare areas and water bodies) and (ii) only forest sampling points (n=221). No significantly different result of any land use type was found between the two alternative reference situations. Although there is a large overlap of data points between the two alternatives, it indicates that results are not very sensitive to choice of reference habitat. However, in both alternatives the reference habitat experienced considerable past (and present) human disturbance, as no pristine areas exist in Switzerland, whereas more pristine reference habitat was included for other biomes in the GLOBIO3 database. However, this inconsistency is unavoidable when a globally valid reference situation has to be defined, as different world regions show different land use history.
In our study, we found a median reduced relative species richness across all globally averaged land use types. However, we cannot rule out that other factors, such as changes in overall landscape composition or pollution might also have contributed to the result. A meta-study across multiple taxonomic groups in the Western Ghats, India, for example found no significant effect of land use on species richness, but a significant effect of native forest cover within the landscape (Anand et al. 2010). Besides the necessity to understand cause-effect chains of biodiversity loss, this illustrates the importance of spatial context of land use (i.e. in what landscape a land use occurs). Despite their importance, it was beyond the scope of this study to include spatial and temporal effects. To improve the assessment of biodiversity loss related to land use or other drivers of biodiversity loss, better concepts including these temporal and spatial aspects are required for LCIA (see also Curran et al. 2011).
3.5 Conclusions and recommendations
Although uncertainties and data and knowledge gaps are considerable, human impacts on biodiversity are ongoing. Decisions how to adapt production towards being less harmful for biodiversity need to be taken urgently, and cannot wait until all data and knowledge gaps are filled. Based on empirical data, this study provides a first attempt to quantify land use impacts on biodiversity within LCA across world regions to support such decisions. Due to the mentioned challenges to quantify biodiversity impacts, the presented characterization factors (CF) should be used with caution and remaining uncertainties should be considered when LCA results are interpreted and communicated. In LCA studies, where the “user may not directly decide on the land management practices” (Milà i Canals et al. 2007, p. 13), our CF can serve as a first screening of potential land use impacts across global value chains. For LCA studies aiming to
73 Land use impacts on biodiversity in LCA: a global approach
support decisions of specific land management, a more detailed, site-dependent assessment, including additional region- or site-specific data, is indispensable (see e.g. Geyer et al. 2010).
In this paper, occupation impacts of a range of land use types in many world regions could be assessed, but some data gaps remain. Research priorities should be set to first close data gaps for environmentally important land use activities (such as agri- and silviculture, construction, mining and land filling) in economically important world regions (e.g. by using regionalized global inventories such as the inventory of global crop production from Pfister et al. 2011). To assess total land use impacts on biodiversity, we need to complement the presented CF of occupation with regionalized global estimates of transformation impacts. This requires more reliable information on regeneration times of ecosystems across the world, as transformation impacts (calculated according to the UNEP/SETAC framework; Milà i Canals et al. 2007; Koellner et al. 2013b) are highly sensitive to this parameter and currently available estimates vary considerably (Schmidt 2008). Estimates of regeneration times should ideally be based on empirical data, for example derived through meta-analysis of ecosystem regeneration studies.
In view of current data availability, the applied indicator relative species richness is suitable for biodiversity related global land use LCIA. As ecological research evolves, LCIA methods should be complemented with indicators measuring other facets of biodiversity, such as conservation value, species abundance or turnover. This applies not only to land use impacts, but also to other drivers of biodiversity loss, such as climate change, eutrophication, acidification or ecotoxicity. To inform decision-makers about potential trade-offs of different drivers of biodiversity loss along the life cycle, indicators need to be comparable across impact pathways (see also Curran et al. 2011). Finding a measure to quantify impacts of concurrent multiple drivers of biodiversity loss in a globally applicable and spatially differentiated way will be a challenge for future LCA research. As the importance of halting global biodiversity loss is increasingly recognized in research, industry and policy (e.g. formulated as the 2020 targets of the Convention on Biological Diversity; CBD 2010), increased research efforts are made to close some of the mentioned knowledge and data gaps. This will also open the way to improve the accuracy of biodiversity assessments within LCA and allow for more robust and credible decision-support.
Acknowledgements: The authors wish to thank Biodiversity Monitoring Switzerland (BDM) and the team of GLOBIO for providing data. The research was funded by ETH Research Grant CH1-0308-3 and by the project “Life Cycle Impact Assessment Methods for Improved Sustainability Characterisation of Technologies” (LC-IMPACT), Grant Agreement No. 243827, funded by the European Commission under the 7th Framework Programme. We appreciate helpful comments by M. Curran, S. Hellweg, J.P. Lindner, R. Müller-Wenk, A. Spörri and two anonymous reviewers.
3.6 References
Achten WMJ, Mathijs E, Muys B Proposing a life cycle land use impact calculation methodology. In: 6th International Conference on LCA in the Agri-Food Sector, Zurich, Nov 1 2008
74 Land use impacts on biodiversity in LCA: a global approach
Alkemade R, van Oorschot M, Miles L, Nellemann C, Bakkenes M, ten Brink B (2009) GLOBIO3: A Framework to Investigate Options for Reducing Global Terrestrial Biodiversity Loss. Ecosystems 12 (3):374–390 Anand M, Krishnaswamy J, Kumar A, Bali A (2010) Sustaining biodiversity conservation in human-modified landscapes in the Western Ghats: Remnant forests matter. Biological Conservation 143:2363–2374 Arrhenius O (1921) Species and area. J Ecol 9 (1):95-99. Attwood SJ, Maron M, House APN, Zammit C (2008) Do arthropod assemblages display globally consistent responses to intensified agricultural land use and management? Global Ecology and Biogeography 17(5):585-599 Baltisberger M (2009) Systematische Botanik. Einheimische Farn- und Samenpflanzen. 3. edn. vdf Hochschulverlag AG, ETH, Zuerich BDM (2004) Biodiversity Monitoring Switzerland. Indicator Z9: Species Diversity in Habitats. Bundesamt für Umwelt, BAFU. http://www.biodiversitymonitoring.ch Beck J, Schwanghart W (2010) Comparing measures of species diversity from incomplete inventories: an update. Methods in Ecology and Evolution 1(1):38-44 Billeter R, Liira J, Bailey D, Bugter R, Arens P, Augenstein I, Aviron S, Baudry J, Bukacek R, Burel F, Cerny M, De Blust G, De Cock R, Diekotter T, Dietz H, Dirksen J, Dormann C, Durka W, Frenzel M, Hamersky R, Hendrickx F, Herzog F, Klotz S, Koolstra B, Lausch A, Le Coeur D, Maelfait JP, Opdam P, Roubalova M, Schermann A, Schermann N, Schmidt T, Schweiger O, Smulders MJM, Speelmans M, Simova P, Verboom J, van Wingerden WKRE, Zobel M, Edwards PJ (2008) Indicators for biodiversity in agricultural landscapes: a pan- European study. Journal of Applied Ecology 45 (1):141-150 Blaum N, Seymour C, Rossmanith E, Schwager M, Jeltsch F (2009) Changes in arthropod diversity along a land use driven gradient of shrub cover in savanna rangelands: identification of suitable indicators. Biodiversity and Conservation 18(5):1187-1199 Bond WJ, Parr CL (2010) Beyond the forest edge: ecology, diversity and conservation of the grassy biomes. Biological Conservation 143:2395–2404 Brandão M, Milà i Canals L (2013) Global characterisation factors to assess land use impacts on biotic production. The International Journal of Life Cycle Assessment 18(6):1243-1252 Brooks T, Mittermeier R, da Fonseca G, Gerlach J, Hoffmann M, Lamoreux J, Mittermeier C, Pilgrim J, Rodrigues A (2006) Global biodiversity conservation priorities. Science 313(5783):58-61 CBD (2010) Aichi Biodiversity Targets. Convention on Biological Diversity. http://www.cbd.int/sp/targets/. Accessed 26 October 2011 Chiarucci A, Araujo MB, Decocq G, Beierkuhnlein C, Fernandez-Palacios JM (2010) The concept of potential natural vegetation: an epitaph? Journal of Vegetation Science 21(6):1172–1178 Curran M, de Baan L, De Schryver A, van Zelm R, Hellweg S, Koellner T, Sonnemann G, Huijbregts MAJ (2011) Toward Meaningful End Points of Biodiversity in Life Cycle Assessment. Environmental Science and Technology 45(1):70-79 De Schryver AM, Goedkoop MJ, Leuven RSEW, Huijbregts MAJ (2010) Uncertainties in the application of the species area relationship for characterisation factors of land occupation in life cycle assessment. The International Journal of Life Cycle Assessment 15(7):682-691 Dengler J (2009) Which function describes the species-area relationship best? A review and empirical evaluation. Journal of Biogeography 36 (4):728-744 Fisher R, Corbet A, Williams C (1943) The relation between the number of species and the number of individuals in a random sample of an animal population. Journal of Animal Ecology 12(1):42-58 Gardner T, Barlow J, Sodhi N, Peres C (2010) A multi-region assessment of tropical forest biodiversity in a human-modified world. Biological Conservation 143:2293-2300 Gardner TA, Barlow J, Chazdon R, Ewers RM, Harvey CA, Peres CA, Sodhi NS (2009) Prospects for tropical forest biodiversity in a human modified world. Ecology Letters 12(6):561-582
75 Land use impacts on biodiversity in LCA: a global approach
Geyer R, Lindner JP, Stoms DM, Davis FW, Wittstock B (2010) Coupling GIS and LCA for biodiversity assessments of land use: Part 2: Impact assessment. The International Journal of Life Cycle Assessment 15 (7):692-703 Goedkoop M, Heijungs R, Huijbregts MAJ, De Schryver A, Struijs J, van Zelm R (2008) ReCiPe 2008. A life cycle impact assessment method which comprises harmonised category indicators at the midpoint and the endpoint level; First edition Report I. Den Haag Goedkoop M, Spriensma R (1999) The Eco-Indicator 99. A Damage Oriented Method for Life Cycle Impact Assessment. Methodology Report. PRé Consultants, Amersfoort Gurevitch J, Hedges L (1999) Statistical issues in ecological meta-analyses. Ecology 80 (4):1142-1149 Hayek L-AC, Buzas MA (2010) Surveying Natural Populations. Quantitative Tools for Assessing Biodiversity. 2nd edn. Columbia University Press, New York Heywood VH, Watson RT (1995) Global Biodiversity Assessment. Cambridge University Press, Cambridge Holling CS (2001) Understanding the complexity of economic, ecological, and social systems. Ecosystems 4 (5):390-405 Kessler M, Abrahamczyk S, Bos M, Buchori D, Putra DD, Gradstein SR, Hoehn P, Kluge J, Orend F, Pitopang R, Saleh S, Schulze CH, Sporn SG, Steffan-Dewenter I, Tjitrosoedirdjo S, Tscharntke T (2009) Alpha and beta diversity of plants and animals along a tropical land-use gradient. Ecological Applications 19(8):2142-2156 Kier G, Mutke J, Dinerstein E, Ricketts T, Kuper W, Kreft H, Barthlott W (2005) Global patterns of plant diversity and floristic knowledge. Journal of Biogeography 32(7):1107-1116 Koellner T (2000) Species-pool effect potentials (SPEP) as a yardstick to evaluate land-use impacts on biodiversity. Journal of Cleaner Production 8:293-311 Koellner T, de Baan L, Beck T, Brandão M, Civit B, Goedkoop MJ, Margni M, Milà i Canals L, Müller-Wenk R, Weidema B, Wittstock B (2013a) Principles for Life Cycle Inventories of land use on a global scale. The International Journal of Life Cycle Assessment 18(6): 1203-1215 Koellner T, de Baan L, Beck T, Brandão M, Civit B, Margni M, Milà i Canals L, Saad R, de Souza DM, Müller-Wenk R (2013b) UNEP-SETAC Guideline on Global Land Use Impact Assessment on Biodiversity and Ecosystem Services in LCA. The International Journal of Life Cycle Assessment 18(6): 1188-1202 Koellner T, Hersperger A, Wohlgemuth T (2004) Rarefaction method for assessing plant species diversity on a regional scale. Ecography 27:532-544 Koellner T, Scholz RW (2007) Assessment of land use impacts on the natural environment. Part 1: An analytical framework for pure land occupation and land use change. The International Journal of Life Cycle Assessment 12(1):16-23 Koellner T, Scholz RW (2008) Assessment of land use impacts on the natural environment. Part 2: Generic characterization factors for local species diversity in Central Europe. The International Journal of Life Cycle Assessment 13(1):32-48 Kyläkorpi K, Rydgren B, Ellegård A, Miliander S (2005) The Biotope Method 2005: A method to assess the impact of land use on biodiversity. Vattenfall, Sweden Lindeijer E (2000a) Biodiversity and life support impacts of land use in LCA. Journal of Cleaner Production 8:313–319 Lindeijer E (2000b) Review of land use impact methodologies. Journal of Cleaner Production 8:273–281 Michelsen O (2008) Assessment of land use impact on biodiversity. The International Journal of Life Cycle Assessment 13(1):22-31 Michelsen O Impacts On Biodiversity From Land Use And Land Use Changes – Did We Forget The First Fundamental Question? In: ISIE Conference, University of California, Berkeley, June 7-10 2011 Milà i Canals L, Bauer C, Depestele J, Dubreuil A, Freiermuth Knuchel R, Gaillard G, Michelsen O, Müller-Wenk R, Rydgren B (2007) Key Elements in a Framework for Land
76 Land use impacts on biodiversity in LCA: a global approach
Use Impact Assessment Within LCA. The International Journal of Life Cycle Assessment 12(1):5-15 Milà i Canals L, Rigarlsford G, Sim S (2013) Land use impact assessment of margarine. The International Journal of Life Cycle Assessment 18(6): 1265-1277 Millennium Ecosystem Assessment (2005) Millennnium Ecosystem Assessment. Ecosystems and Human Well-being: Biodiversity Synthesis. World Resources Institute, Washington DC Müller-Wenk R (1998) Land Use – The Main Threat to Species, How to Include Land Use in LCA. IWÖ - Diskussionsbeitrag Nr. 64. Institut für Wirtschaft und Ökologie, Universität St.Gallen, St. Gallen Müller-Wenk R, Brandão M (2010). Climatic impact of land use in LCA – carbon transfers between vegetation/soil and air. The International Journal of Life Cycle Assessment 15(2):172-182 Noss R (1990) Indicators for monitoring biodiversity - a hierarchical approach. Conservation Biology 4(4):355-364 Olson D, Dinerstein E, Wikramanayake E, Burgess N, Powell G, Underwood E, D'Amico J, Itoua I, Strand H, Morrison J, Loucks C, Allnutt T, Ricketts T, Kura Y, Lamoreux J, Wettengel W, Hedao P, Kassem K (2001) Terrestrial ecoregions of the worlds: A new map of life on Earth. BioScience 51(11):933-938 Penman TD, Law BS, Ximenes F (2010) A proposal for accounting for biodiversity in life cycle assessment. Biodiversity and Conservation 19(11):3245-3254 Pereira H, Leadley P, Proença V, Alkemade R, Scharlemann JPW, Fernandez-Manjarrés JF, Araújo MB, Balvanera P, Biggs R, Cheung WWL, Chini L, Cooper HD, Gilman EL, Guénette S, Hurtt GC, Huntington HP, Mace GM, Oberdorff T, Revenga C, Rodrigues P, Scholes RJ, Sumaila UR, Walpole M (2010) Scenarios for Global Biodiversity in the 21st Century. Science 330(6010):1496-1501 Pfister S, Bayer P, Koehler A, Hellweg S (2011) Environmental Impacts of Water Use in Global Crop Production: Hotspots and Trade-Offs with Land Use. Environmental Science and Technology 45:5761–5768 Purvis A, Hector A (2000) Getting the measure of biodiversity. Nature 405(6783):212-219 R Development Core Team (2011) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria Ramankutty N, Foley JA (1999) Estimating historical changes in global land cover: Croplands from 1700 to 1992. Global Biogeochemical Cycles 13(4):997-1027 Rodrigues ASL, Brooks TM (2007) Shortcuts for biodiversity conservation planning: The effectiveness of surrogates. Annual Review of Ecology, Evolution, and Systematics 38:713-737 Rosenzweig ML (1995) Species diversity in space and time. Cambridge University Press, Cambridge Saad R, Koellner T, Margni M (2013) Land use impacts on freshwater regulation, erosion regulation and water purification: a spatial approach for a global scale level. The International Journal of Life Cycle Assessment 18(6): 1253-1264 Sala O, Chapin F, Armesto J, Berlow E, Bloomfield J, Dirzo R, Huber-Sanwald E, Huenneke L, Jackson R, Kinzig A, Leemans R, Lodge D, Mooney H, Oesterheld M, Poff N, Sykes M, Walker B, Walker M, Wall D (2000) Global biodiversity scenarios for the year 2100. Science 287(5459):1770-1774 Schenck R (2001) Land use and biodiversity indicators for Life Cycle Impact Assessment. The International Journal of Life Cycle Assessment 6(2):114-117 Scherber C, Eisenhauer N, Weisser WW, Schmid B, Voigt W, Fischer M, Schulze E-D, Roscher C, Weigelt A, Allan E, Beßler H, Bonkowski M, Buchmann N, Buscot F, Clement LW, Ebeling A, Engels C, Halle S, Kertscher I, Klein A-M, Koller R, Konig S, Kowalski E, Kummer V, Kuu A, Lange M, Lauterbach D, Middelhoff C, Migunova VD, Milcu A, Muller R, Partsch S, Petermann JS, Renker C, Rottstock T, Sabais A, Scheu S, Schumacher J,
77 Land use impacts on biodiversity in LCA: a global approach
Temperton VM, Tscharntke T (2010) Bottom-up effects of plant diversity on multitrophic interactions in a biodiversity experiment. Nature 468:553–556 Schmidt J (2008) Development of LCIA characterisation factors for land use impacts on biodiversity. Journal of Cleaner Production 16:1929-1942 Scholz RW (2011) Environmental Literacy in Science and Society. From Knowledge to Decisions. Cambridge University Press, Cambridge Shannon CE (1948) A mathematical theory of communication. Bell System Technical Journal 27: 379-423 SOER Synthesis (2010) The European environment – state and outlook 2010: synthesis. European Environment Agency, Copenhagen. Sørensen T (1948) A method of establishing groups of equal amplitude in plant sociology based on similarity of species content. K dan Vidensk Selsk Biol Skr 5:1-34 Stanners D, Philippe B (1995) Europe’s environment—the Dobris assessment. European Environment Agency, Copenhagen van der Voet E (2001) Land use in LCA. CML-SSP Working Paper 02.002, Leiden Vandewalle M, de Bello F, Berg MP, Bolger T, Doledec S, Dubs F, Feld CK, Harrington R, Harrison PA, Lavorel S, da Silva PM, Moretti M, Niemela J, Santos P, Sattler T, Sousa JP, Sykes MT, Vanbergen AJ, Woodcock BA (2010) Functional traits as indicators of biodiversity response to land use changes across ecosystems and organisms. Biodiversity and Conservation 19(10):2921-2947 Vogtländer J, Lindeijer E, Witte J, Hendriks C (2004) Characterizing the change of land-use based on flora: application for EIA and LCA. Journal of Cleaner Production 12:47–57 Wagendorp T, Gulinck H, Coppin P, Muys B (2000) Land use impact evaluation in life cycle assessment based on ecosystem thermodynamics. Energy 31:112-125 Weidema B, Lindeijer E (2001) Physical impacts of land use in product life cycle assessment. Final report of the EURENVIRON-LCAGAPS sub-project on land use. Department of Manufacturing Engineering and Management, Technical University of Denmark, Lyngby Wolters V, Bengtsson J, Zaitsev AS (2006) Relationship among the species richness of different taxa. Ecology 87(8):1886-1895
78
Chapter 4
Comparing direct land use impacts on biodiversity of conventional and organic milk – based on a Swedish case study
Carina M. Mueller
Laura de Baan
Thomas Koellner
Published in The International Journal of Life Cycle Assessment
(2013)
Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
Abstract
Purpose: Halting the loss of biodiversity while providing food security for a growing and prospering world population is a challenge. One possible solution to this dilemma is organic agriculture, which is expected to enhance biodiversity on the farmland. However, organic products often require larger areas. This study demonstrates how we can quantify and compare the direct land use impacts on biodiversity of organic and conventional food products such as milk.
Material and Methods: This study assessed direct land use impacts of one litre of milk leaving the farm gate. Inventory data on land occupation were extracted from a life cycle assessment study of 15 farms in southern Sweden. Direct land use change data was derived from the FAO statistical database. Spatially differentiated characterization factors of occupation (CFOcc) and transformation (CFTrans) were calculated based on the relative difference of plant species richness on agricultural land compared to a (semi) natural regional reference. Data on plant species richness and regeneration times of ecosystems (for calculating transformation impacts) was derived from a literature review. To account for differences in biodiversity value between regions a weighting system based on absolute species richness, vulnerability and irreplaceability was applied.
Results and discussion: Organic milk had a lower direct land use impact than conventional milk, although it required about double the area. Occupation impacts dominated the results and were much smaller for organic than conventional milk, as CFOcc of organic land uses were considerably smaller. For transformation impacts, differences between the two farming practices were even more pronounced. The highest impacts were caused by soymeal in concentrate feeds (conventional milk) due to large-scale deforestation in its country of cultivation (i.e. Brazil and Argentina). However, lack of reliable data posed a challenge in the assessment of transformation impacts. Overall, results were highly sensitive to differences in land occupation area between farms, the CFOcc and assumptions concerning transformed area. Sensitivity and robustness of results were tested and are discussed.
Conclusions: Although organic milk required about twice as much land as conventional, it still had lower direct land use impacts on biodiversity. This highlights the importance of assessing land use impacts not only based on area, but also considering the actual impacts on biodiversity. The presented approach allows to quantify and compare hot- and coldspots in the agricultural stage of milk production and could potentially also be applied to other agricultural products. However more research is needed to allow quantification of indirect land use impacts.
Keywords: Biodiversity, land use, organic agriculture, life cycle impact assessment, milk
80 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
4.1 Introduction One of the most important drivers of global biodiversity loss is land use change or habitat destruction (Sala et al. 2000; Foley et al. 2005; Pereira et al. 2010), which is indirectly linked to increased food and energy demand by a growing and prospering world population (Tilman 2001). Providing food security for the expected future human population of 9 billion people (Zhang 2008) while halting the loss of biodiversity is a challenge. Over the past years there has been a growing debate about whether land sparing (i.e. agricultural intensification to minimize pressure to convert wild nature) or wildlife-friendly farming (i.e. increasing biodiversity on farmland which might reduce agricultural yields) is the more promising solution (Tscharntke et al. 2012; Chappell and LaValle 2011; Ewers et al. 2009). Organic farming practices are found to increase species abundance (Crowder et al. 2010; Rahmann 2011) and species richness of many taxa (Bengtsson et al. 2005; Fuller et al. 2005; Gomiero et al. 2011) as well as the presence of endangered species (van Elsen 2000) compared to conventional farming. This positive effect on biodiversity points to the management principles of organic agriculture, which prohibit the use of chemical fertilizer or inorganic pesticides (Hole et al. 2005) and are characterized by a lower stocking rate (Hansen et al. 2001) and a more varied crop rotation (Maeder et al. 2002).
Especially in Europe, efforts to preserve the high biodiversity value of long-established agricultural ecosystems are high (Green 2005). Therefore, subsidies for organic farming were introduced in the early 1990s (Stolze and Lampkin 2009). In 2005, the European Union´s (EU plus national funds) support of organic farming increased to 17 % of the yearly expenditures (or a total of 0.6 billion Euro) dedicated to agri-environment measures (European Union 2007). In order to assess the usefulness of these large amounts of subsidies for biodiversity conservation, the overall impacts of organic and conventional products on biodiversity need to be quantified.
An important decision support tool in this context is life cycle assessment (LCA), which allows the quantification of a range of environmental impacts throughout all stages of a product's (often globally distributed) life cycle (e.g. Milà i Canals et al. 2007). Several approaches towards the assessment of direct land use impacts on biodiversity have been published. However, these studies are either only applicable to one or two world regions, and thus cannot capture land use impacts occurring in many world regions (e.g. Koellner 2000; Koellner and Scholz 2007; Michelsen 2008; Schmidt 2008a), or do not capture differences among different farming systems (e.g. Weidema and Lindeijer 2001; de Baan et al. 2013; Mattila et al. 2012).
In this study, we quantify and compare the direct land use impacts on biodiversity of organic and conventional milk. We therefore further develop the methodology presented in de Baan et al. (2013) and apply it to a case study of milk production in Sweden, where organic milk was shown to require considerably larger areas compared to conventional milk (Cederberg and Flysjoe 2004). Milk is an interesting example, because its land use is distributed over many world regions via the supply chain of concentrate feed (Cederberg and Mattson 2000). Thus, a regionalized global approach (as suggested in de Baan et al. 2013) is needed to accurately assess biodiversity impacts, due to its heterogeneous distribution and variable response to land use. In addition, the composition and origin of feedstock for organic and conventional milk varies substantially. For
81 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
producing organic milk in the EU, more than 60 % of the daily intake of cows has to be roughage such as silage, hay or pasture and more than 50 % of the feed has to be produced on the farm itself (EC 2007). Conventional dairy cows are fed more on concentrate feed compared to organic cows, and the feedstock are to a larger degree imported from sub-/tropical countries (Cederberg and Mattson 2000). In these countries, large areas of natural vegetation of high species richness and endemism (Brooks et al. 2006; Kier et al. 2005; Myers et al. 2000) have been transformed in the last decades (FAOSTAT 2012a; Mayaux et al. 2005).
Milk is not only an interesting case to illustrate the trade-offs between area requirement and impacts on biodiversity, but also an important commodity. The consumption of milk products is projected to rise considerably in the next decades (Delgado 2003). Many LCA studies have been conducted on the comparison on organic and conventional milk. In terms of energy use, eutrophication and global warming potential no clear result of organic compared to conventional milk can be found (Thomassen et al. 2008; Flysjoe et al. 2012). In contrast, land use (in terms of area) was found to be considerably larger for organic milk than for conventional milk (Cederberg and Mattson 2000; Thomassen et al. 2008; van der Werf et al. 2009). However, the impacts of land use on biodiversity have not yet been incorporated into LCA studies of milk (Yan et al. 2011).
4.2 Material and Methods
4.2.1 Description of compared agricultural systems Data on land occupation area and country of cultivation of fodder crops were taken from a study on dairy farms in south-west Sweden (Cederberg and Flysjoe 2004), following the same approach as Cederberg and Mattson (2000). In this study, we considered nine high intensity conventional farms and six organic farms. In terms of total farm area the two agricultural systems were comparable, however the conventional farms had to purchase more concentrated feed as they had more cows on their farm (Table 4.1). The higher share of roughage feed on the diets of organic cows results in lower milk yields of organic compared to conventional cows. The functional unit (FU) of the study was "one kg of energy corrected milk" leaving the farm-gate, i.e. transportation and processing of raw milk was excluded (Cederberg and Flysjoe 2004). Partitioning of environmental impacts (i.e. allocation) between meat and milk as well as co- products in purchased concentrate feed, such as meal and oil from rapeseed, was based on their economic value (Cederberg and Flysjoe 2004). Additionally, direct land use impacts were ascribed to the region in which the crop was most likely cultivated according to Cederberg and Flysjoe (2004) and for organic soy according to FiBL (2012). In this study, it was assumed that the land was occupied for one whole year for most crops, as in temperate latitudes only one fodder crop can be grown per year and oil palm fruit, meadows and pastures are cultivated permanently (Milà i Canals et al. 2013). Only after the harvest of soy another fodder crop can be grown in the same year. Thus, about 25 % double-cropping of soy was assumed (suggested in Dalgaard et al. 2008) which was in contrast to Cederberg and Flysjoe (2004) who did not account for any double- cropping.
82 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
Table 4.1 General characteristics of the two farm types Organic Conventional Number of studied farms 6 9 Arable land (ha farm-1) 63 70 Natural meadows (ha farm-1) 19 11 Number of cows per farm 39 65 Stocking rate (LUa ha-1year-1) 0.9 1.2 Total purchased feed (kg cow-1year-1) 1457 2951 Milk yield (kg ECMb/cow × year) 9400 10100 a lifestock unit: corresponds to one dairy cow including a calf younger than one month, or six calves aged between one and six months, or three heifers older than half a year b Energie corrected milk
4.2.2 Land use assessment framework Following the framework of the United Nations Environment Programme (UNEP)/Society of Environmental Toxicology and Chemistry (SETAC) Life Cycle Initiative (Milà i Canals et al. 2013; Koellner et al. 2013), we distinguished two land use impacts: land occupation (using land) and land transformation (changing the land use) in this study .(Milà i Canals et al. 2013; Koellner et al. 2013). Permanent impacts (i.e. irreversible damages to ecosystems) were not considered here.
Characterization factors of land occupation, CFOcc, were calculated as the difference in relative plant species richness (Srel) of a reference system and a land use type (LU) i per biome j (de Baan et al., 2013; Koellner et al. 2013). As globally available biodiversity data focus on the species level and species numbers (Curran et al. 2011), we chose relative species richness of a certain land use type i compared to a reference as biodiversity indicator. Srel of the reference system (ref) was set to 100 % =1 (see Section Biodiversity indicator).
CF = S − S = 1− S (4.1) Occ,LUi ,j rel,Ref ,j rel,LUi ,j rel,LUi ,j Natural vegetation was chosen as reference. In many European countries, only few anthropogenic undisturbed habitats remain, so they can not be considered natural habitats (SOER Synthesis 2010). Consequently in the following, we will refer to the reference situations as "(semi)-natural", as proposed in de Baan et al. (2013).
The numeric values of CFOcc normally take on values between zero and one (expressing detrimental impacts), but can sometimes also be negative (expressing beneficial impacts, de
Baan et al. 2013). The occupation impact can then be calculated as the product of CFOcc, the cultivated area AOcc, and the duration of the occupation process tOcc (de Baan et al. 2013; Koellner et al. 2013).
Occupation impact = A it iCF (4.2) LU Occ Occ Occ
Transformation impacts are assessed accordingly with the transformed area ATrans (see Section
2.2) and the time a system requires to regain the ecosystem quality of the reference (treg) after an anthropogenic disturbance (de Baan et al. 2013; Koellner et al. 2013).
83 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
1 Transformation impact A CF A CF t (4.3) Ref →LU = Trans i Trans = Trans i Occ i reg 2 The Biodiversity Damage Potential (BDP) of land use can then be calculated as the sum of the transformation impact and the occupation impact over all land use types i and biomes j (Koellner and Scholz 2007).
BDP = Transformation impact + Occupation impact ∑ Ref →LUi ,j ∑ LUi ,j (4.4)
4.2.3 Inventory Analyses In the study of Cederberg and Flysjoe (2004), transformation impacts are not considered. Thus, we calculated the inventory data for transformed area as proposed by Milà i Canals et al. (2013). This approach only associates direct land transformation with a fodder crop if (1) in its country of origin the harvested area of this specific crop increased in the last 20 years and if additionally (2) the area of its land use type (i.e. arable land, permanent crops or meadows and pastures) increased. In case these two conditions applied, the transformed area for every occupied hectare and year was calculated by dividing the increase in land use type area over the last 20 years by the current area of this land use type (as proposed in Milà i Canals et al. 2013). The type of land that was transformed was proportionally assigned based on the decrease in area of all land use types. Historical data on country specific developments of agricultural production areas were extracted from the statistical database of the Food and Agriculture Organisation (FAOSTAT 2012a; b). For grass pellets, data were extracted from Statistics Denmark (StatBank Denmark 2012) as no information was given in the FAOSTAT database. For organic land use types, no specific information on land use change was available and standards of the International Federation of Organic Agriculture Movements (IFOAM) only prohibit the conversion of areas with high conservation value five years before organic farming starts (IFOAM 2012). Therefore we assumed organic farming to result in the same amount of transformed area per cultivated hectare as conventional farming. To account for yearly variation in land use, 5-year averages around the years 1987 and 2007 were calculated, respectively (as suggested in Milà i Canals et al. 2013). As data on forest area before 1990 was not available for all countries, pre-1990 areas were linearly extrapolated using data from the years 1990 to 2009. This seems a feasible estimate, as the forest conversion rate after 1990 followed a linear relationship for the countries considered in this study. Furthermore within the timeframe in question (1985-2009) the same drivers were prevalent, as demonstrated in various studies (e.g. Fearnside 2005; Gasparri et al. 2008; McMorrow and Talip 2001).
Forest is the only natural terrestrial land cover type included in the FAOSTAT database (2012a). Most Brazilian soy is cultivated on land where naturally savannahs prevail (Fearnside 2001). Savannahs still fall under the definition of "forest" according to the FAOSTAT database, as they contain vegetation with trees exceeding 5 m and a canopy cover denser than 10 %. Consequently, these data were seen as appropriate for estimating conversion from natural vegetation within the biome sub-/tropical grass-/shrublands and savannahs.
84 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
4.2.4 Impact Assessment To account for spatial differences in ecological responses to land use, region-specific reference situations and regeneration times (see Eq. 4.3) were used and impacts were calculated separately for each biogeographic region (as suggested in de Baan et al. 2013). Biomes defined by Olson et al. (2001) and recommended to use for land use impacts in LCA on global scales by Koellner et al. (2013) were used as biogeographic units. CORINE Plus (elaborated by Koellner and Scholz 2008) was used for land use classification as it allowed the discrimination of organic and conventional land use types (Koellner et al. 2013).
Biodiversity indicator The analysis of a biodiversity measure was restricted to vascular plants because data availability for organic land use types is relatively good for this taxon (Hole et al. 2005). Several data sources were combined here to quantify plant species richness of relevant land use types and reference situations: the Countryside Survey of the United Kingdom (Carey et al. 2007), the report "Biodiversity of Saxonian arable land" (Kreuter 2005), the biodiversity monitoring data of Switzerland (BDM 2004) and the databases used in Koellner (2000) and de Baan et al. (2013). As these data sources only contained little data on organic land uses or data from the biome sub- /tropical grass-/shrubland and savannahs, a literature search in the Web of Science database was performed. Overall, this search resulted in 66 studies (see Appendix C for a full bibliography) providing 111 data points for the different land use types and 53 data points for the reference situations in three different biomes of feedstock production for Swedish milk. As sampling area varied strongly among studies, sampled species richness (S) was standardized to an area (A) of 100 m² using the transformed power-model of the species-area relationship proposed in Kier et al. (2005):
z ⎛ ⎞ A 2 S = S i⎜ 100m ⎟ 100m2 Sampled ⎜ A ⎟ ⎝ Sampled ⎠ (4.5) where z is the species accumulation factor (see Appendix Table C1 for results of each data point). For each biome, z-values were obtained from Kier et al. (2005).
Regeneration time The time required for the regeneration after human interventions stop is used for the calculation of transformation impacts (Eq. 4.3). To derive regeneration times of different land use types, a literature search of peer-reviewed studies was conducted. Only investigations located below 1000 m a.s.l. analyzing regeneration of plant species richness after human disturbances (e.g. logging or agricultural use) were included. However for the regeneration times needed for this study only very few data points (1 to 5 per land use type and biome) could be found (Table 4.2). Therefore, alternative regeneration times based on Dobben et al. (1998) were calculated as well for a comparison. It has to be noted that regeneration times of tropical biomes are not directly comparable to the numbers found for temperate biomes as the compared land use types for the biomes differ. Regeneration of species richness on arable land is expected to take much longer
85 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
than on pastures/meadows or permanent crops as the human intervention is much more intense due to for example regular plowing.
Dobben et al. (1998)´s regeneration times have been used in previous LCA studies on land use (e.g. Milà i Canals et al. 2013), and are based on broad estimates of biomass regeneration dependent on altitude and latitude after clear-cutting. To make this estimate comparable to the empirical data, biome-specific regeneration times were calculated depending on each biome's area percentage (quantified with ArcGIS) in three different altitudinal and six different latitudinal zones. These values were considered to represent regeneration times of arable land
(treg,arable,Dobben). Thus, the regeneration time (treg,LUi,Dobben) of every land use type i was then calculated using the ratio of the CFOcc per biome j (see Section 2.1) as follows:
CF t = Occ,LUi ,j it reg,LUi ,j ,Dobben CF reg,arable,i ,Dobben Occ,arable (4.6)
It was assumed that the regeneration rate, defined as the change of CFOcc, LUi with treg, of each land use was constant within each biome.
Table 4.2 Regeneration times by land use type Based on Dobben Mean of Empirical n Source et al. (1998) studies treg (years) treg (years) Biome TropMBF Pasture/meadow 45 33 a 5 (Aide et al. 1995; Aide et al. 1996; Guariguata et al. 1997; Letcher and Chazdon 2009) Permanent crops 42 35a 3 (Grau et al. 1997; Pascarella et al. 2000; Zimmermann et al. 2007)
Biome TempGL Arable - 45 b 1 (Scott and Morgan 2012)
Biome TropGL Arable 57 - n number of data points per biome and land use type aall studies were conducted in Middle or South America b from Australia
Statistical Analysis No studies were found which had investigated the influence of organic and conventional farming practice and the species richness on comparable (semi-)natural references together. Therefore, we performed a resampling procedure with 1000 bootstrap samples (with replacement) to obtain several statistical parameters (e.g. median, quartiles, 95 % confidence intervals) of the relative species richness Srel. For each land use type, bootstrapping combined all plant species richness data points randomly with (semi-) natural reference data points from the same biome.
86 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
Analyses of Variance (ANOVA) was applied to test differences in means of Srel among the factors land use type, biome and farming practice as well as their interactions. Prior to statistical analyses bootstrapping results were log-transformed to improve fit to a normal distribution.
As data did not satisfy the assumptions of the ANOVA (i.e. normal distribution and homogenous residual variance), robustness of results was evaluated by additionally conducting the Kruskal-
Wallis test to analyze the influence of the three factors (without interaction) on medians of Srel.
The Mann-Whitney U test revealed significant differences in median Srel between pairs of land use types. All data were performed using R 2.13.0 (R Development Core Team 2011) and the R package “boot” (Angelo and Ripley 2012) for the bootstrapping.
Biodiversity weighting factor Expressing species richness in relative instead of absolute terms allowed the comparison of land use impacts in various ecoregions with different data qualities (Kier et al. 2005; Koellner et al. 2013). To account for differences in absolute species numbers and conservation value between ecoregions, a weighting system was applied, as proposed in Weidema and Lindeijer (2001), which was calculated separately for each ecoregion. Following global biodiversity prioritization concepts the biodiversity weighting factor (BWF) was based on absolute species richness, irreplaceability and vulnerability (see review by Brooks et al. 2006).
The three indices to quantify biodiversity value of each ecoregion were calculated as follows. Absolute species richness (S) was calculated as area-corrected total number of amphibian, reptile, mammal and bird species per ecoregion (derived from Kier et al. 2005). Irreplaceability was quantified as the area-corrected number of strict endemic species of amphibians, reptiles, mammals and birds (EndS). For endemism, these are the only taxonomic groups where data per ecoregion is available. For consistency, the same selection of taxonomic groups was also chosen for species richness, and data on plants were excluded. Vulnerability was expressed as the 'Conservation Risk Index' (CRI), which is calculated as the ratio of converted ecoregion area in percent to protected ecoregion area in percent (developed by Hoekstra et al. 2005). The latter concept assumes that the more area is occupied the more damaging an occupation or transformation will be for the remaining ecosystem (Koellner 2000). To prevent division by zero, all values below 1 % were set to 1 % as such low habitat proportions are likely below the threshold necessary for long-term species conservation (Swift and Hannon 2010). The same procedure was performed for converted area as the impact of converting 1 % of a habitat is assumed to be negligible in terms of species conservation.
Data to calculate all indices were extracted from the Terrestrial Ecoregions Base Global Dataset (Olson et al. 2001). As suggested by Weidema and Lindeijer (2001), these three measures of conservation prioritization were linked by multiplication for each ecoregion k. To give each measure equal weight in the multiplication, they were all normalized (expressed with an asterisk * in equation) to range between 1 and 10 (Table 4.3):
87 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
⎛ S i9 ⎞ ⎛ EndS i9 ⎞ ⎛CRI i9 ⎞ BWF k 1 k 1 k 1 S ∗ EndS CRI ∗ k = ⎜ + ⎟ i⎜ + ⎟ i⎜ + ⎟ = k i k i k ⎝ S ⎠ ⎝ EndS ⎠ ⎝ CRI ⎠ max max max (4.7)
Overall, the weighted BDP was calculated as follows: weighted BDP = TI + OI iBWF k (∑ Ref →LUi ,k ∑ LUi ,k ) k (4.8)
4.3 Results
4.3.1 Characterization factors for Biodiversity Damage Potential
For almost all land use types medians of CFOcc were significantly greater than zero, indicating detrimental impacts on biodiversity (see Figure 4.1, Appendix Table C2 and Table C6). Only the median of organic pasture/meadow was not significantly different from zero and showed a trend towards a negative value (i.e. beneficial impact). Overall, meadows/pastures were least harmful to biodiversity, followed by permanent crops. Arable land had the highest impact on biodiversity, which was most pronounced in the biome sub-/tropical moist broadleaf forests. A significant effect of land use type, biome and farming practice on CFOcc was shown with ANOVA and Kruskal- Wallis test (results not shown). The ANOVA also revealed that the influence of land use type on
CFOcc was significantly affected by biome and farming practice.
Biome TempBMF Biome TropGL Biome TropMBF 1.0 Max Conv * Org * 75% 0.5 * * Median
* * 25% a ) 2 0.0
Min -0.5 (PDF per m per (PDF
Occ -1.0 CF
-1.5
-2.0
Pasture /meadow Arable Arable Permanent crops
n=14x33 n=14x33 n=28x33 n=26x33 n=15x6 n=2x6 n=12x14
Figure 4.1 Box and Whisker plot of bootstrapped characterization factors of occupation (CFOcc) for each land use type, farming practice and biome. TempBMF: broadleaf and mixed forests; TropGL: sub-/tropical grass-/shrublands and savannahs; TropMBF: sub-/tropical moist broadleaf forests. Asterisks mark significant differences in medians compared to the reference which is defined as
CFOcc = 0 (Mann-Whitney U test: p < 0.05). n is the total number of data points used for the bootstrap of CFOcc; the first number stands for the number of species richness data points per
88 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
land use type and the second for the data points per reference. n = 14 x 33, e.g., means that 14 data points on the species richness of the land use type were combined with 33 data points for the reference situation, which resulted in a total of 462 data points for CFOcc.
Biome TropGL Biome TropMBF
20 Conv Org
15 a ) 2
10 (PDF per m per (PDF
Trans 5 CF
0
Transformation from: Forest to arable Permanent crops Forest to to arable permanent crops
n=15x6x1 n=2x6x1 n=12x14x3
Figure 4.2 Box and Whisker plot of characterization factors of transformation (CFTrans) for each farming practice, biome and land use type regenerating to after human abandonment. TropGL: sub-/tropical grass-/shrublands and savannahs; TropMBF: sub-/tropical moist broadleaf forests. Shown is only the variation due to differences in species richness between data points. See Appendix Figure C1 for variation in species richness per land use type and biome.
In all studied land use types and biomes, organic farming practices were less detrimental for biodiversity than conventional farming practices. Medians of CFOcc of conventional compared to organic pasture/meadow showed a 30 % higher potentially disappeared fraction of species (PDF) per m² and year. In the median, 45 % less species were found on conventional arable land in the biome temperate broadleaf and mixed forests than on the organic equivalent. In the biome sub- /tropical grass-/shrublands and savannahs this difference amounted to about 40 %.
4.3.2 Impact of occupation Although organic milk required about double the amount of agricultural land to produce one kg of milk (Figure 4.3a), the occupation impact of organic milk was only half the one of conventional milk (Figure 4.3b). CFOcc of organic land use types were always considerably lower than the conventional ones thus leading to smaller occupation impacts. In addition, the different composition of the feedstock, with larger shares of roughage feed and grazing for the organic cows and larger shares of concentrate feed for the conventional cows, considerably influenced the result (Appendix Table C3).
89 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
Biome Biome Biome TempBMF a) TropGL TropMBF 3.0 Conv Org 2.5
2.0
1.5 a per FU per a ² m
1.0
0.5
0.0
/ /
Total Grains Oil palm Oil Pasture meadow Soybean Rapeseed Legumes temporary grasses Sugarbeet
Biome Biome b) Biome TempBMF TropGL TropMBF 1.0 Conv Org
0.8
0.6
PDF per FU per PDF 0.4
0.2
0.0
Total Grains Oil palm Oil Pasture/ meadow Soybean Legumes/ Legumes/ temporary grasses Rapeseed Sugarbeet Figure 4.3 Contribution to (a) land occupation and (b) occupation impact of different fodder crop types per biome TempBMF broadleaf and mixed forests, TropGL sub-/tropical grass-/shrublands and savannahs and TropMBF sub-/tropical moist broadleaf forests, calculated with non-weighted characterization factors for biodiversity impact, which is expressed as potentially disappeared fraction of species (PDF) per functional unit (FU).
90 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
Arable land and more precisely grains and legumes/temporary grasses in the biome temperate broadleaf and mixed forest largely dominated the impact for both farming practices (Figure 4.4). For the fodder crop category legumes/temporary grasses 100 % and 80 % of the impact was due to ley for conventional and organic milk, respectively. Both conventional and organic farms cultivated ley on about 60 % of their on-farm arable land (Cederberg and Flysjoe 2004). Grains accounted for more than 75 % in the organic and about 50 % in the conventional concentrate feed (Appendix Table C3). This explains the large land use and occupation impact allocated to this crop type. In contrast, rapeseed, sugar beet and palm oil have relatively small shares in the occupation impact of conventional milk, although they constitute about 37 % of the concentrate feed. Due to their form as co-products of the sugar/oil/starch production only a small part of the environmental burden was allocated to them, minimizing their impact. Although organic milk required a relatively large area of pastures and meadows, the occupation impact was slightly negative (i.e. beneficial) due to their negative median CFOcc value. In the organic feedstock, some conventionally produced grains are included. This is the reason, why grains and legumes have a similar occupation impact (Figure 4.3b), although larger areas are required for legumes production than for grain production (Fig 4.3a).
91 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
Biome TempBMF BiomeTropGL Biome TropMBF 1.0 Conv Org
0.8
0.6
PDF per FU per PDF 0.4
0.2
0.0
Pasture/ Arable Arable Permanent Total meadow crops
Figure 4.4 Occupation impact of conventional and organic milk illustrated per land use type and biome TempBMF broadleaf and mixed forests, TropGL sub-/tropical grass-/shrublands and savannahs and TropMBF sub-/tropical moist broadleaf forests, calculated with non-weighted characterization factors for biodiversity impact, expressed as potentially disappeared fraction of species (PDF) per functional unit (FU).
4.3.3 Impact of transformation Conventional milk had a more than three-fold larger transformation impact than the same amount of organic milk, regardless of the method used for estimating regeneration times (Figure 4.5). In this study soy and palm oil were the only fodder crops originating from countries, where in the last 20 years the harvested crop area as well as the area of the corresponding land use type (i.e. arable land, pastures/meadows or permanent crops) increased. Soy was responsible for the majority of the transformation impact (99 % for conventional milk and 98 % for organic milk). In contrast, palm oil caused only a very small impact because palm kernel expels have only little economic value and the quantities in the concentrate feeds were small (3 % for conventional milk and 4 % for organic milk in concentrates). The land use change from natural vegetation to each harvested hectare of crop was the same for the conventional soy from Brazil (0.20 ha converted/harvested ha per year) and palm oil from Malaysia (0.20 ha converted/harvested ha per year), but slightly smaller for the organic soy from Argentina (0.14 ha converted/harvested ha per year). Argentina was the only country in the milk value chain, where one agricultural land use type, permanent crops, was converted into another agricultural land use type (Appendix Table C4). Thus the transformation impact of Argentinian soy is lower than Brazilian soy as the regeneration time of arable land to permanent crops is smaller than from arable land to secondary vegetation. Also the much higher yields of conventional compared to organic soy were
92 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
counterbalanced by the lower CFTrans of organic arable land in this biome. As a consequence, the difference in the transformation impact between organic and conventional milk can largely be explained by the fact that conventional cows are fed about three times more soy products than organic cows to produce the same amount of milk.
0.7 a) b)
Biome TropMBF 0.6 Biome TropGL
0.5
0.4
PDF per FU per PDF 0.3
0.2
0.1
0.0 Conv Org Conv Org
Figure 4.5 Transformation impact calculated with regeneration times based on a) empirical data and b) expert estimates of biomass regeneration by Dobben et al. (1998), calculated with non- weighted characterization factors for biodiversity impact, expressed as potentially disappeared fraction of species (PDF) per function unit (FU). TropGL: sub-/tropical grass-/shrublands and savannahs; TropMBF: sub-/tropical moist broadleaf forests.
4.3.4 Total Biodiversity Damage Potential The total BDP is about 60 % lower for organic compared to conventional milk. The share of the transformation impact on the BDP amounts to about 30 % and 20 % for conventional and organic milk, respectively (Figure 4.6a). From all feedstock, soy showed the largest impacts on biodiversity. Although soymeal comprised only about 10 % of the conventional concentrate feed, it caused almost 40 % of the BDP. In the case of organic milk, soy constituted only 4 % in the organic concentrate feed, but was responsible for 25 % of the BDP.
93 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
a) 20 b) 1.2 Pattern: Transformation impact 1.0 15 Occupation impact
Colour: 0.8 Cerrado
10 Sarmatic mixed forest 0.6
PDF per FU per PDF Atlantic mixed forest Baltic mixed forest 0.4 Peninsular Malaysian rain forests 5
0.2
0.0 0 Conv Org Conv Org
Figure 4.6 Biodiversity Damage Potential (sum of occupation and transformation impact) calculated a without and b with the biodiversity weighting factor per ecoregion, expressed as potentially disappeared fraction of species (PDF) per functional unit (FU). Be aware of the different scales of the axes.
To account for differences in absolute species numbers and conservation value between ecoregions, we calculated a biodiversity weighing factor. For the ecoregions which are relevant for the production of Swedish milk, the normalized weighing factor ranged between 4.3 for the Sarmatic mixed forest and 31.7 for the Cerrado (Table 4.3). The ratio in the weighted BDP of organic to conventional milk was slightly decreased from 43 % to 36 %. The impact of rapeseed from the Baltic mixed forest, which was negligible without the weighting system, became obvious. After weighting, soy largely dominated results of the BDP. This is due to the relatively high weighting factor of the Cerrado ecoregion where most of the soy in this study was cultivated. 50 % of the Cerrado has been converted whereas only 1.1 % of its area is protected, resulting in a high CRI value.
94 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
Table 4.3 Results of normalisations (*) of species richness (S), endemic species richness (EndS) and Conservation Risk Index (CRI) and their product, the Biodiversity Weighting factor (BWF) per ecoregion Biome Ecoregion S* EndS* CRI* BWF
Sub-/tropical moist Peninsular Malaysian rain forests 4.6 1.5 3.0 18.1 broadleaf forest Temperate broadleaf & Atlantic mixed forests 2.4 1.0 7.8 18.6 mixed forests Baltic mixed forests 2.4 1.0 5.4 12.8
Sarmatic mixed forests 2.1 1.0 1.6 3.4
Sub-/tropical grass- Cerrado 3.3 1.8 5.4 31.7 /shrublands and savannahs Sources: Olson et al. (2001) and own calculations. A full list of BWF per ecoregion can be found in Appendix Table C6
4.4 Discussion For some, organic agriculture is seen a solution for environmentally friendly food production, while for others, it is an inefficient and resource-intensive production system. To better understand the involved trade-offs, quantitative comparisons are needed. To our knowledge, this is the first study, trying to quantify and compare the land use impacts on biodiversity of organic and conventional farming on a product level, including impacts of imported feedstock. While organic milk required about double the area than conventional milk, the impacts on biodiversity were less than half for organic compared to conventional milk. The impacts on biodiversity per organically farmed area were considerably smaller for conventional farming and less fodder crops originated from sub-/tropical countries where forest conversion to agricultural land uses is prevalent. These reduced impacts of organic milk could by far outweigh the larger area requirements.
Trying to express such complex issues as global land use impacts on biodiversity in one number is inherently accompanied by large natural variation and uncertainties. In the following, we want to discuss the sensitivity and robustness of results. In addition, we will discuss methodological choices made in this study, such as how to quantify biodiversity impact and the amount of transformed area, the selection of the indicator for regeneration times and the appropriateness of the weighing factor. Finally, we will reflect on the transferability of results of this study to milk produced in other countries.
4.4.1 Sensitivity analyses This paper combined data from a Swedish LCA study, global literature on plant species richness and FAO statistics. All databases exhibit diverse sources of uncertainty. Conducting a full Monte Carlo uncertainty analyses was not possible in this study as information on uncertainty or variability was not available for all parameters. Therefore, to assess how the uncertainty of selected parameters influenced the results, a sensitivity analyses was performed (see Figure 4.7 for an overview).
95 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
Conv 1.5 Org
1.0 PDF perFU PDF
0.5
0.0 S0 S1 S2 S3 S4 S5 Figure 4.7 Sensitivity of Biodiversity Damage Potential (sum of occupation and transformation impact) per FU for scenario S0 to S5 (S0 = standard scenario used in this study; S1 = comparing organic farm with highest land use with conventional farm with lowest land use; S2 = two times lower organic yields than used here; S3 = South American soybean meal substituted with North American soybean meal; S4 = ley classified as pastures and meadows instead of arable land; S5 = Using 3rd quartiles instead of medians of Characterization factors of occupation.
Inter-farm variation is rarely quantified in LCA studies on agricultural products, but can be considerable. Land use inventory data for this case study came from a survey of nine conventional and six organic dairy farms from the same region within Sweden. For these farms, the variability within the land occupation of both farming practices is remarkable; the farm with highest land use needs twice the area per functional unit than the one with lowest, in both farming practices, respectively. To test the robustness of the results, we compared milk from the organic farm with the largest land use and milk from the conventional farm with the smallest land use (see Figure 4.7 scenario S1). In this latter case, the organic farm had a five times larger land use than the conventional farm. This represents an extreme case for Europe, as in the published studies of other countries differences in land requirements between organic and conventional milk were less pronounced (Thomassen et al. 2008). In this case, the direct land occupation impact of organic milk from the farm with the largest land use was slightly higher than conventional milk from the farm with the smallest land use. Understanding the factors causing this inter-farm variability would be an important topic for future research as it might offer optimization potential for both farming practices.
In agricultural LCA studies, yields are often one of the determining factors. In this current study, organic compared to conventional yields were estimated to be 20 to 50 % lower per crop.
96 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
Compared to yield differences ranging between 5 and 34 % found in a global meta-study (Seufert et al. 2012), our yield estimates are conservative. Even if the yields of organic crops were on average about two times lower than adopted in this study (assuming a constant composition and origin of feedstock for both alternatives), the organic milk would still be favorable in terms of impacts on biodiversity (see Figure 4.7 scenario S2). Purchased feedstock is often sourced from various countries. As we did not have data for all world regions on plant species richness of organic and conventional agriculture, we could not directly test how the results would change assuming a different origin. As soy meal was responsible for about 25 % of the total BDP of organic and about 40 % of conventional milk, the total direct land use impacts could change significantly, when sourcing soy meal from another country, such as the United States of America, China or India. According to the applied methodology, in those countries, no transformation impact would have been allocated to soy, as in these countries the area of arable land remained stable for the last two decades (FAOSTAT 2012a) therefore decreasing the BDP (see Figure 4.7 scenario 3).
Uncertainty also lies in the classification of land uses. Leys cultivated with temporary grasses and legumes grown on areas set-aside were classified as arable land in this study. The reason for this decision was that leys were cultivated on average for 3 to 8 years in a crop rotation system with regular plowing. According to Koellner et al. (2013), the latter is an important differentiating factor from pastures/meadows in the land use and land cover classification used in this study. A Norwegian study found that the species richness of leys reached levels comparable to semi- natural grassland not before 30 years after last plowing (Austrheim and Olsson 1999). However, ley grasslands might have a higher species richness than average arable fields cultivated as monocultures (Andreasen and Stryhn 2008). Thus, we performed a sensitivity analyses classifying all the ley area as meadow/pasture. This decreased the occupation impact of conventional and organic milk about 22 % and 46 %, respectively and resulted in even more pronounced differences between organic and conventional milk (see Figure 4.7 scenario S4). However up to now almost no studies were conducted on the plant species richness of leys. Thus more research is needed to improve reliability of biodiversity impact assessments dealing with land use on leys.
An in-depth uncertainty assessment of direct land use impacts on biodiversity of globally traded cocoa products showed, that the CFOcc were by far the largest contributor to variance (Mutel et al.
2013). In this study, the variance of CFOcc was also considerable, even ranging from positive to negative impacts for pasture and arable land in the temperate biome and for permanent crops in the tropical biome (see Figure 4.1). Part of this variation might be due to the fact that different data sources had to be used for the plant species richness of the different land use types and of the reference. Studies directly comparing the species richness of organic agriculture and a reference within the same region were not available. Consequently, data points of land use types had to be randomly combined with (semi-) natural reference data points from the same biome to generate CFOcc, possibly resulting in plant species richness data from species-poor regions being combined with (semi-) natural references from species-rich regions in the same biome.
97 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
We tested the sensitivity to CFOcc by recalculating the BDP results based on the third quartiles of
CFOcc Using the third quartiles instead of medians of CFOcc increased the BDP for conventional and organic milk by about 120 % and 230 %, respectively (see Figure 4.7 Scenario S5). In this case, organic milk still had only half of the impact of conventional milk. When using the first quartile of
CFOcc for conventional land use types and the third quartile of CFOcc for organic land use types, the result changes and the BDP of conventional milk would be lower than the one of organic milk.
Thus, future research should aim to decrease the uncertainty of CFOcc by possibly combining data points on species richness of land use types with data points on (semi-) natural references from the same region. However, a large variation will possibly still remain, as biodiversity shows a strong natural heterogeneity.
4.4.2 Quantification of biodiversity impacts The strong difference in plant species richness between organic and conventional farmed land found in this study has also been found in other studies (e.g. Bengtsson et al. 2005; Fuller et al. 2005; Gomiero et al. 2011). A meta-analyses by Hole et al. (2005) on the influence of organic agriculture on biodiversity reported that differences between the two farming practices were especially pronounced for plants, birds and insects. However for other taxonomic groups such as earthworms or beetles the influence of organic farming on species richness was less clear, as studies reported both positive and negative effects (Hole et al. 2005). Another meta-analysis by Bengtsson et al. (2005) found that the effect of organic agriculture was largest on studies on the plot-scale, whereas on the farm-scale results were still significant, however less pronounced. This might suggest that the kind of data that were chosen in this study (i.e. plant species richness on field-scale or smaller) made results more clear than it could have been in the case, if all organism groups had been studied. However, plants were the only taxon for which sufficient data were available for all relevant biomes and land use types. Moreover, two studies (Haas et al. 2001; Schader et al. 2010), which compared biodiversity impacts of organic and conventional milk production systems on farm-scale, also found lower impacts of organic dairy farms. However their approach did not allow the quantification of impacts related to off-farm land use which this study showed to be responsible for a considerable part of the BDP of milk.
Another factor which seems to have an influence on the species richness of agricultural land, is the surrounding landscape. In their meta-analyses, Bengtsson et al. (2005) concluded that positive effects of organic farming will more likely occur in intensively used agricultural land, but not inevitably in small-scaled mosaic landscapes with a high share of natural vegetation. Thus, the results found here seem to be more representative for the highly agriculturally used landscapes found in middle Europe.
4.4.3 Calculation of transformed area The quantification of area transformed for land use is largely based on FAO statistics. The database has the advantage of being easily applicable and consistent across countries (Milà i Canals et al. 2013). However, one has to be aware of several shortcomings of the database in this context. First, the lack of forest area data for the last 20 years poses a problem for studies like this one. Calculations can either be based on extrapolations of existing data or periods shorter
98 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
than 20 years. This influences the results (compare with Milà i Canals et al. 2013). A second shortcoming is that the only natural terrestrial vegetation type included is "forest" (see section 2.2 for the definition). All areas which are not classified as agricultural land or forest area are compiled in "other land" which e.g. sums up to 37 % of the country area in the case of Argentina. Hence the assessment of conversion from more grassland-type vegetation is difficult. This study tried to overcome this problem by assuming that the entire increase of the land use type of a crop was due to direct land transformation from natural vegetation in case no other agricultural land use type decreased in area during the respective time period in the country. Furthermore, uncertainty of the FAO database is high as many data are "FAO estimates" or "unofficial figures". Thus, in some cases it could be appropriate to compare FAO data with national inventory data, as performed for land use estimates by Ramankutty et al. (2008). Moreover the method does not consider indirect land use change (iLUC) in other countries (Schmidt 2008b; Milà i Canals et al. 2013) which might be relevant in comparing different farming practices. It is possible that organic products have a higher iLUC as their yields are lower than conventional products (Seufert et al. 2012). However in the scope of this study it was not possible to model world-wide socio-economic impacts of lower organic yields compared to conventional fodder crops on iLUC in other countries. Including iLUC into environmental impact assessments has only developed in the last years and was mainly limited to calculations of carbon emissions due to biofuel cultivation (e.g. Kloverpris and Mueller 2013). Methodological considerations and assumptions concerning iLUC are highly debated (Kloverpris and Mueller 2013; Hertel et al. 2010; Mathews and Tan 2009) and estimates vary considerably (Plevin et al. 2010). Nevertheless it is expected that parameter values used to model iLUC of biofuels are not comparable to the ones that would be needed to model iLUC of organic crops. One parameter which is different is price: organic milk is about 15-83 % more expensive in the European Union (Austrian Federal Ministry of Agriculture, Forestry, Environment and Water Management 2011) which is known to decrease milk consumption (Andreyeva et al. 2010) and thus expected iLUC. However, the socio-economic cause-effect chain of converting agricultural land to organic land is complex and potential impacts on land management in other countries are barely understood. Moreover not enough data to calculate
CFOcc or regeneration times for other biomes where natural ecosystems are converted for agricultural use were available. Nevertheless including iLUC into the assessment of greenhouse gas emissions of biofuels was shown to cancel out their benefit (e.g. Hertel et al. 2010). Therefore in future comparisons of different farming practices related to land use change should include this aspect.
4.4.4 Indicator for regeneration time Up to now biomass estimates (e.g. Milà i Canals et al. 2013; Schmidt 2008a) were used in LCIA as indicator for the regeneration of biodiversity as a whole, in contrast this study also used empirical data of plant species richness. The similarity of the results for transformation impacts (see Figure 4.5) calculated with empirical data in comparison with biomass estimates by Dobben et al. (1998) is astonishing. Dobben et al. (1998)´s estimates depend only on latitude and altitude whereas the regeneration of species richness is also modified by factors such as the natural disturbance regime (Brown and Lugo 1990), soil nutrients or the proximity to patches of natural
99 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
vegetation in the surrounding landscape (Prach and Rehounkova 2006). But what takes much longer than the regeneration of biomass or vascular plant species richness are the regeneration of species composition (Aide et al. 2000; Pascarella et al. 2000; Dunn 2004), species richness of other taxa (Barlow et al. 2007; Berry et al. 2010) or endemic species richness (Liebsch et al. 2008). A recent meta-analyses of empirical studies showed that primary tropical forests are irreplaceable in terms of biodiversity (Gibson et al. 2011). A practical solution for this problem in LCA applications could be to assume a very long regeneration phase (Koellner and Scholz 2007). Koellner et al. (2013) recommended to limit the modeling period in land use impact assessments to 500 years. Incorporating such long regeneration times would increase the share of transformation impacts on the overall BDP considerably. Ultimately the choice of indicator depends on the aim of biodiversity LCIA (see Michelsen 2011). Depending on whether we focus on ecosystem services, functional diversity should be protected (see also de Baan et al. 2013) which correlates well with plant species richness (Petchey et al. 2004). However if we focus on the conservation or the potential prospective economic value (e.g. medicinal plants) of biodiversity, endemic species richness or species composition of primary habitat should be chosen. A meta- analyses on empirical regeneration times of species richness for different world regions and taxa will become available soon (Michael Curran, personal communication) which will improve reliability of results as uncertainty of regeneration times used in this study is high due to few available data points (Table 4.2).
4.4.5 Biodiversity weighting system To account for differences in absolute species numbers as well as conservation value between ecoregions, a weighting factor was introduced, as recommended in de Baan et al. (2013). The factor proposed here is easy to apply and consistent over ecoregions on a global level. However, up to now only information for amphibians, reptiles, mammals and birds could be included. Thus as soon as data on endemic plant species or on other important species groups such as arthropods become available, additional taxa should be included in the calculation of a weighting system to improve the information value.
Here, vulnerability was assessed as the ratio of converted to protected ecoregion area. Yet, converted areas can also be a main target of nature conservation for example the species-rich grasslands developed after deforestation and subsequent extensive grazing and mowing in Europe (Poschlod and Wallis de Vries 2002). Therefore, the vulnerability assessment might be improved by calculating the ratio of natural and extensive land uses to protected ecoregion area. Including the areas of extensively used land might have resulted in a higher biodiversity weighting factor of the Sarmatic mixed forest. Currently, data on areas of extensive or traditional land uses are not available on ecoregion scales. Nevertheless, results of the introduced weighting scheme in this study showed that the Cerrado exhibits a considerably more valuable biodiversity than European ecoregions. Accounting for this remarkable richness is essential in the assessment of biodiversity impacts in LCAs.
100 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
4.4.6 Transferability to other temperate countries Results for occupation impacts of milk found in this study cannot be transferred easily to milk produced in other countries. When comparing the land use inventories of various European milk LCA studies, the variation in allocation procedures and assumptions is posing a difficulty (De Boer 2003; De Vries and De Boer 2010; Yan et al. 2011). The sources of feedstock vary, where some studied farms source concentrate feed components from countries such as India (De Boer et al. 2012) or Australia (Thomassen et al. 2008) The biggest difference across countries seems to be their share of pastures/meadows on the total land use. As CFOcc of pasture/meadow are relatively low, high proportions of this land use decrease the occupation impact. In Sweden, land use relies to only 12 % for conventional and 20 % for organic milk on permanent pastures/meadows (calculation based on data in Cederberg and Flysjoe 2004). These percentages do not include the ley areas which seem to be special for Sweden but are classified as arable due to their regular plowing. Thus land use for Swedish milk production is based less heavily on meadows/pastures compared to Belgium with 42 % for zero-grazing (i.e. no grazing on pastures but roughage from meadows fed) conventional milk (Meul et al. 2012), Germany (Mueller-Lindenlauf et al. 2010) with 45-89 % for organic milk or the Netherlands (Thomassen et al. 2008) with 43 % for conventional and 54 % for organic milk. The highest share of permanent pastures/meadows on the overall land use for milk seems to be found New Zealand (Basset-Mens et al. 2009) where 95 % of the land use for high-input milk and 100 % of the land use for low-input milk (i.e. no purchased feed outside the farm) is based on permanent mixed pastures. However, it has to be investigated whether CFOcc generated in this study are transferable to other European countries or regions in the biome temperate broadleaf and mixed forest (e.g. New Zealand, China or USA) or to which extent other factors such as agricultural history or sensitivity of native species to agriculture cause significantly different results.
Transformation impacts found in this study seem to be transferable to milk produced in other European countries. Most middle and northern European dairy farms source their feed largely from countries where the area of agricultural production remained stable for decades (FAOSTAT 2012a). Therefore the share of soy meal (from South America) and palm kernel expels (from South-East Asia) in the concentrate dairy feed might be a good proxy for transformation impacts. The percentage and origin of soy and palm kernel expels in conventional concentrate feed in this case study (about 10 % of 0.32 kg concentrate feed per FU) is comparable to other European countries. A Dutch study listed 12 % Brazilian soy of 0.26 kg concentrate feed per FU (De Boer et al. 2012), a French study 13.8 % soy from Brazil (Lehuger et al. 2009) and in the regional estimate of the FAO (2010) 13 %. In contrast percentages in organic concentrate feed (5 %) in this study seem to represent values of the upper range. In Austria (Hoertenhuber et al. 2010) no soy meal or palm kernel expels were included in organic concentrates and almost none (accounting for economic allocation) in the Netherlands (Thomassen et al. 2008). Thus, transformation impact results calculated for conventional milk seem to be representative for other European countries, whereas results for organic milk seem to demonstrate values on the upper margin. Finally, the statement that organic milk has lower transformation impacts than conventional milk seems to be valid for the studies mentioned here.
101 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
4.5 Conclusions Although organic land requires about double the area than conventional milk, the direct impacts on biodiversity was less than half. This illustrates the importance of differentiating CFOcc depending on the land use intensity (e.g. organic versus conventional). However as including indirect land use change in assessments of greenhouse gas emissions of biofuels was shown to be crucial, future research on the biodiversity impact of organic products should study socio- economic effects of lower organic compared to conventional yields on indirect land use change. For an application to other studies, a more complete data set on species richness of organic and conventional farms needs to be compiled, to cover more world regions. However, for studies concerned with similar countries of origin, the presented CFs can be used to approximate impacts, considering the underlying uncertainties as discussed above. To increase validity of results, more research is required especially on the impact of land use on biodiversity compared to a (semi-) natural reference in grassland biomes, which are currently gaining in agricultural importance. Future research should assess the relevance of a more fine-scaled regionalization and develop concepts on incorporating aspects such as type of surrounding landscape or history of land use, which have a considerable influence in some regions (Bengtsson et al. 2005).
In the present study, we just compared the direct land use impacts to the endpoint Biodiversity Damage Potential (Koellner et al. 2013) and did not consider other midpoint oriented impact indicators (e.g. climate regulation, biotic production, freshwater regulation). To better understand the role of organic agriculture in an environmentally friendly solution to feed the world, other environmental as well as socio-economic impacts should be considered. As was shown in this study, further methodological developments might be needed to capture the differences between organic and conventional farming systems within LCA.
Another finding of this study is that meadows and pastures are the most biodiversity-friendly feedstock for milk production. Meadows and pastures generally will have a higher species richness than arable land, assuming stocking rates are not detrimental to biodiversity. The larger the share of pastures and meadows on the land occupation of milk, the lower the impact to biodiversity will be. As a consequence, politics should create incentives for farmers to maintain pastures and meadows, if they contain high species richness. Further improvement potential exists by sourcing soy from countries without deforestation or to replace soy with another, more biodiversity-friendly crop. Results found here also stress the importance of subsidies for organic agriculture as this type of farming practice makes an important contribution to the maintenance of species richness in the agricultural landscape.
4.6 Acknowledgements We are thankful to Anna Flysjoe (Arla Foods, Denmark) who explained the allocation procedure performed in Cederberg and Flysjoe (2004). We are also grateful to Manuel Steinbauer, University of Bayreuth for statistical advice and Christel Cederberg (Arla Foods, Denmark) and Anna Flysjoe, the Countryside Survey of the United Kingdom and the Biodiversity Monitoring Switherland (BDM) for providing data. We are grateful for helpful comments by Matthias Meier, Anna Flysjoe and
102 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
three anonymous reviewers. Finally, we would like to thank Daniel Kachelriess, University of Bayreuth for revising the English.
4.7 References Aide T, Zimmerman JK, Herrera L, Rosario M, Serrano M (1995) Forest recovery in abandoned tropical pastures in Puerto Rico. For Ecol Manag 77:77–86 Aide TM, Zimmerman JK, Pascarella JB, Rivera L, Marcano-Vega H (2000) Forest regeneration in a chronosequence of tropical abandoned pastures: Implications for restoration ecology. Restor Ecol 8(4):328–338 Aide TM, Zimmerman JK, Rosario M, Marcano H (1996) Forest recovery in abandoned cattle pastures along an elevational gradient in northeastern Puerto Rico. Biotropica 28(4):537– 548 Andreasen C, Stryhn H (2008) Increasing weed flora in Danish arable fields and its importance for biodiversity. Weed Res 48:1–9 Andreyeva T, Long MW, Brownell KD (2010) The impact of food prices on consumption: a systematic review of research on the price elasticity of demand for food. Gov Politi Law 100(2):216-222 Angelo C, Ripley B (2012) boot: Bootstrap R (S-Plus) Functions. R package version 1.3-6. from http://cran.r-project.org/web/packages/boot/citation.html, Accessed July 2012 Austrheim G, Olsson EG (1999) How does continuity in grassland management after ploughing affect plant community patterns? Plant Ecol 145:59-74 Austrian Federal Ministry of Agriculture, Forestry, Environment and Water Management (2011) Production and commercialization of organic milk in Europe, Austrian Federal Ministry of Agriculture, Forestry, Environment and Water Management, Department III/6, Vienna Barlow J, Gardner TA, Araujo IS, Avila-Pires TC, Bonaldo AB, Costa JE, Esposito MC, Ferreira LV, Hawes J, Hernandez MIM, Hoogmoed MS, Leite RN, Lo-Man-Hung NF, Malcolm JR, Martins MB, Mestre LAM, Miranda-Santos R, Nunes-Gutjahr AL, Overal WL, Parry L, Peters SL, Ribeiro-Junior MA, da Silva MNF, da Silva Motta C, Peres CA (2007) Quantifying the biodiversity value of tropical primary, secondary, and plantation forests. Proc Natl Acad Sci USA 104(47):18555–18560 Barthlott W, Mutke J, Rafiypoor D, Kier G, Kreft H (2005) Global centers of vascular plant diversity. Nova Acta Leopold 92:61–83 Basset-Mens C, Ledgard S, Boyes M (2009) Eco-efficiency of intensification scenarios for milk production in New Zealand. Ecol Econ 68(6):1615–1625 BDM (2004) Biodiversity monitoring Switzerland. Indicator Z9: species diversity in habitats. Bundesamt für Umwelt, BAFU. http://www.biodiversitymonitoring.ch Bengtsson J, Ahnstroem J, Weibull AC (2005) The effects of organic agriculture on biodiversity and abundance: a meta-analysis. J Appl Ecol 42(2):261–269 Berry NJ, Phillips OL, Lewis SL, Hill JK, Edwards DP, Tawatao NB, Ahmad N, Magintan D, Khen CV, Maryati M, Ong RC, Hamer KC (2010) The high value of logged tropical forests: lessons from northern Borneo. Biodivers Conserv 19(4):985–997 Brooks TM, Mittermeier R, da Fronseca G, Gerlach J, Hoffmann M, Lamoreux JF, Mittermeier C, Pilgrim J, Rodrigues A (2006) Global biodiversity conservation priorities. Science 313(5783):58–61 Brown S, Lugo AE (1990) Tropical secondary forests. J Trop Ecol 6:1-32 Carey PD, Wallis S, Chamberlain P, Cooper A, Emmett BA, Maskell LC, McCann T, Murphy J, Norton LR, Reynolds B, Scott W, Simpson I, Smart S, Ullyett J (2007) 2008 Countryside survey: UK results from 2007, NERC/Centre for Ecology and Hydrology. http://www.countrysidesurvey.org.uk. Accessed June 2012 Cederberg C, Flysjoe A (2004) Life cycle inventory of 23 dairy farms in south-western Sweden. SIK-report Nr 728. The Swedish Institute for food and biotechnology, Goeteborg
103 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
Cederberg C, Mattson B (2000) Life cycle assessment of milk production — a comparison of conventional and organic farming. J Clean Prod 8:49–60 Chappell MJ, LaValle LA (2011) Food security and biodiversity: can we have both? An agroecological analysis. Agric Hum Values 28(1):3–26 Crowder DW, Northfield TD, Strand MR, Snyder WE (2010) Organic agriculture promotes evenness and natural pest control. Nature 466(7302):109–112 Curran M, de Baan L, De Schryver AM, van Zelm R, Hellweg S, Koellner T, Sonnemann G, Huijbregts MAJ (2011) Toward meaningful end points of biodiversity in life cycle assessment. Environ Sci Technol 45(1):70–79 Dalgaard R, Schmidt J, Halberg N, Christensen P, Thrane M, Pengue WA (2008) LCA of soybean meal. Int J Life Cycle Assess 13(3):240–254 de Baan L, Alkemade R, Koellner T (2013) Land use impacts on biodiversity in LCA: a global approach. Int J Life Cycle Assess 18(6): 1216-1230. de Boer IJ (2003) Environmental impact assessment of conventional and organic milk production - Review. Livest Prod Sci 80(1-2):69–77 de Boer IJM, Hoving IE, Vellinga TV, Ven GWJ, Leffelaar PA, Gerber PJ (2013) Assessing environmental impacts associated with freshwater consumption along the life cycle of animal products: the case of Dutch milk production in Noord-Brabant. Int J Life Cycle Assess 18:193–203 de Vries M, De Boer I (2010) Comparing environmental impacts for livestock products: A review of life cycle assessments. Livest Sci 128:1–11 Delgado CL (2003) Rising consumption of meat and milk in developing countries has created a new food revolution. J Nutr 133:3907S-3910S Dobben HF, Schouwenberg EPAG, Nabuurs GJ, Prins AH (1998) Biodiversity and productivity parameters as a basis for evaluating land use changes in LCA. Annex 1 In: Lindeijer EW, van Kampen M, Fraanje PJ, van Dobben HF, Nabuurs GJ, Schouwenberg EPAG, Prins AH, Dankers N, Leopold MF (1998) Biodiversity and life support indicators for land use impacts in LCA. Delft: Rijkswaterstaat, Dienst Weg-en Waterbouwkunde. (Publication series raw materials 1998/07) Dunn R (2004) Recovery of faunal communities during tropical forest regeneration. Conserv Biol 18(2):302–309 EC (2007) Council Regulation (EC) No. 834/2007 of 28 June 2007 on organic production and labeling of organic products and repealing Regulation (EEC) No 2092/91. Official Journal of the European Union L 189/1, Luxembourg European Union (2007) Rural development in the European Union. Statistical and economic information report. Directorate-General Agriculture and Rural Development, Brussel Ewers RM, Scharlemann JPW, Balmford A, Green RE (2009) Do increases in agricultural yield spare land for nature? Glob Chang Biol 15(7):1716–1726 FAO (2010) Greenhouse gas emissions from the dairy sector. A life cycle assessment, Food and Agriculture Organisation of the United Nations, Rome, Italy FAOSTAT (2012a) Land Resources data 1985-2009 from http://faostat.fao.org/site/377/default.aspx#ancor. Accessed April 2012 FAOSTAT (2012b) Crop production data 1985-2009 from http://faostat.fao.org/site/567/default.aspx#ancor. Accessed March 2012 Fearnside PM (2001) Soybean cultivation as a threat to the environment in Brazil. Environ Conserv 28(1):23–38 Fearnside PM (2005) Deforestation in Brazilian amazonia: history, rates, and consequences. Conserv Biol 19(3):680–688 FiBL (2012) Organic soy area: organic area, fully converted area and area under conversion 2010 (conversion and in-conversion not necessarily complete, both at region or country level). Research Institute of Organic Agriculture FiBL, Frick, Swiss. http://www.organic- world.net/fileadmin/documents/statistics/data-tables/crops-statistics/. Accessed June 2012
104 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
Flysjoe A, Cederberg C, Henriksson M, Ledgard S (2012) The interaction between milk and beef production and emissions from land use change - critical considerations in life cycle assessment and carbon footprint studies of milk. J Clean Prod 28:134–142 Foley JA, Defries R, Asner G, Barford C, Bonan G, Carpenter S, Chapin F, Coe M, Daily G, Gibbs H, Helkowski J, Tracey H, Howard E, Kucharik CMC, Patz J, Prentice C, Ramankutty N, Snyder P (2005) Global consequences of land use. Science 309(5734):570–574 Fuller R, Norton L, Feber R, Johnson P, Chamberlain D, Joys A, Mathews F, Stuart R, Townsend M, Manley W, Wolfe M, MacDonald D, Firbank L (2005) Benefits of organic farming to biodiversity vary among taxa. Biol Lett 1:431–434 Gasparri NI, Grau HR, Manghi E (2008) Carbon pools and emissions from deforestation in extra-tropical forests of northern Argentina between 1900 and 2005. Ecosystems 11(8):1247–1261 Gibson L, Lee TM, Koh LP, Brook BW, Gardner TA, Barlow J, Peres CA, Bradshaw CJA, Laurance WF, Lovejoy TE, Sodhi NS (2011) Primary forests are irreplaceable for sustaining tropical biodiversity. Nature 478(7369):378–381 Gomiero T, Pimentel D, Paoletti MG (2011) Environmental impact of different agricultural management practices: conventional vs. organic agriculture. Crit Rev Plant Sci 30(1- 2):95–124 Grau H, Arturi M, Brown A, Aceñolaza P (1997) Floristic and structural patterns along a chronosequence of secondary forest succession in Argentinean subtropical montane forests. For Ecol Manag 95(2):161–171 Green RE (2005) Farming and the fate of wild nature. Science 307(5709):550–555 Guariguata MR, Chazdon RL, Denslow JS, Dupuy JM, Anderson L (1997) Structure and floristics of secondary and old-growth forest stands in lowland Costa Rica. Plant Ecol 132:107–120 Haas G, Wetterich F, Kopke U (2001) Comparing intensive, extensified and organic grassland farming in southern Germany by process life cycle assessment. Agric Ecosyst Environ 83:43–53 Hansen B, Alroe HF, Kristensen E (2001) Approaches to assess the environmental impact of organic farming with particular regard to Denmark. Agric Ecosyst Environ 83:11–26 Hertel TW, Golub AA, Jones AD, O´Hare M, Plevin RJ, Kammen DM (2010) Effects of US Maize Ethanol on Global Land Use and Greenhouse Gas Emissions: Estimating Market- mediated Responses. Bioscience 60:223–231 Hoekstra JM, Boucher TM, Ricketts TH, Roberts C (2005) Confronting a biome crisis: global disparities of habitat loss and protection. Ecol Lett 8(1):23–29 Hoertenhuber S, Lindenthal T, Amon B, Markut T, Kirner L, Zollitsch W (2010) Greenhouse gas emissions from selected Austrian dairy production systems—model calculations considering the effects of land use change. Renew Agric Food Syst 25(04):316–329 Hole D, Perkins A, Wilson J, Alexander I, Grice P, Evans A (2005) Does organic farming benefit biodiversity? Biol Conserv 122(1):113–130 IFOAM (2012) The IFOAM norms for organic production and processing. International Federation of Organic Agriculture Movement, Bonn, Germany Kier G, Mutke J, Dinerstein E, Ricketts TH, Kueper W, Kreft H, Barthlott W (2005) Global patterns of plant diversity and floristic knowledge. J Biogeogr 32(7):1107–1116 Kloverpris JH, Mueller S (2013) Baseline time accounting: Considering global land use dynamics when estimating the climate impact of indirect land use change caused by biofuels. Int J Life Cycle Assess 18:319-330 Koellner T (2000) Species-pool effect potentials (SPEP) as a yardstick to evaluate land-use impacts on biodiversity. J Clean Prod 8:293–311 Koellner T, de Baan L, Beck T, Brandão M, Civit B, Margni M, Milà i Canals L, Saad R, de Souza DM, Mueller-Wenk R (2013) UNEP-SETAC guideline on global land use impact assessment on biodiversity and ecosystem services in LCA. Int J Life Cycle Assess 18(6): 1188-1202
105 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
Koellner T, de Baan L, Beck T, Brandão M, Civit B, Goedkoop M, Margni M, Milà i Canals L, Mueller-Wenk R, Weidema B, Wittstock B (2013) Principles for life cycle inventories of land use on a global scale. Int J Life Cycle Assess 18 (6): 1203-1215 Koellner T, Scholz RW (2007) Assessment of land use impacts on the natural environment. Part 1: An analytical framework for pure land occupation and land use change. Int J Life Cycle Assess 12(1):16–23 Koellner T, Scholz RW (2008) Assessment of land use impacts on the natural environment - Part 2: Generic characterization factors for local species diversity in central Europe. Int J Life Cycle Assess 13(1):32–48 Kreuter TNO (2005) Biodiversität sächsischer Ackerflächen. Schriftenr Sächs Landesanst Landwirtsch 9(10):1–97 Lehuger S, Gabrielle B, Gagnaire N (2009) Environmental impact of the substitution of imported soybean meal with locally-produced rapeseed meal in dairy cow feed. J Clean Prod 17(6):616–624 Letcher SG, Chazdon RL (2009) Rapid recovery of biomass, species richness, and species composition in a forest chronosequence in northeastern Costa Rica. Biotropica 41(5):608– 617 Liebsch D, Marques MC, Goldenberg R (2008) How long does the Atlantic Rain Forest take to recover after a disturbance? Changes in species composition and ecological features during secondary succession. Biol Conserv 141(6):1717–1725 Loreau M, Naeem S, Inchausti P, Bengtsson J, Grime J, Hector A, Hooper D, Huston M, Raffaelli D, Schmid B, Tilman D, Wardle D (2001) Biodiversity and Ecosystem Functioning: Current Knowledge and Future Challenges. Science 294(5543):804–808 Maeder P, Fließbach A, Dubois D, Gunst L, Fried P, Niggli U (2002) Soil fertility and biodiversity in organic farming. Science 296(5573):1694–1697 Mathews JA, Tan H (2009) Biofuels and indirect land use change effects: the debate continues. Biofuels, Bioprod Bioref 147:1-13 Mattila T, Helin T, Antikainen R (2012) Land use indicators in life cycle assessment: a case study on beer production. Int J Life Cycle Assess 17(3):277–286 Mayaux P, Holmgren P, Achard F, Eva H, Stibig H, Branthomme A (2005) Tropical forest cover change in the 1990s and options for future monitoring. Philos Trans R Soc Br: Biol Sci 360(1454):373–384 McMorrow J, Talip MA (2001) Decline of forest area in Sabah, Malaysia: Relationship to state policies, land code and land capability. Glob Environ Chang 11:217–230 Meul M, Passel S, Fremaut D, Haesaert G (2012) Higher sustainability performance of intensive grazing versus zero-grazing dairy systems. Agron Sustain Dev 32(3):629–638 Michelsen O (2008) Assessment of land use impact on biodiversity. Proposal of a new methodology exemplified with forestry operations in Norway. Int J Life Cycle Assess 13(1):22–31 Michelsen O (2011) Impacts on biodiversity from land use and land use changes — did we forget the first fundamental question? In: ISIE Conference, University of California, Berkeley, June 7-10 Milà i Canals L, Bauer C, Depestele J, Dubreuil A, Freiermuth KR, Gaillard G, Michelsen O, Mueller-Wenk R, Rydgren B (2007) Key elements in a framework for land use impact assessment within LCA. Int J Life Cycle Assess 12(1):5–15 Milà i Canals L, Rigarlsford G, Sim S (2013) Land use impact assessment of margarine. Int J Life Cycle Assess 18 (6):1265-1277 Mueller-Lindenlauf M, Deittert C, Koepke U (2010) Assessment of environmental effects, animal welfare and milk quality among organic dairy farms. Livest Sci 128:140–148 Mutel C, de Baan L, Hellweg S (2013) Two-step sensitivity testing of parameterized and regionalized life cycle assessments: methodology and case study. Environ Sci Technol 47:5660-5667 Myers N, Mittermeier R, Mittermeier C, da Fronseca G, Kent J (2000) Biodiversity hotspots for conservation priorities. Nature 403:853–859
106 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
Olson DM, Dinerstein E, Wikramanayake ED, Burgess ND, Powell GV, Underwood EC, D’Amico JA, Itoua I, Strand HE, Morrison JC, Loucks CJ, Allnutt TF, Ricketts TH, Kura Y, Lamoreux JF, Wettengel WW, Hedao P, Kassem KR (2001) Terrestrial ecoregions of the worlds: A new map of life on Earth. Bioscience 51(11):933–938 Pascarella JB, Aide TM, Serrano MI, Zimmerman JK (2000) Land-use history and forest regeneration in the Cayey Mountains, Puerto Rico. Ecosystems 3:217–228 Pereira HM, Leadley PW, Proenca V, Alkemade R, Scharlemann JPW, Fernandez-Manjarres JF, Araujo MB, Balvanera P, Biggs R, Cheung WWL, Chini L, Cooper HD, Gilman EL, Guenette S, Hurtt GC, Huntington HP, Mace GM, Oberdorff T, Revenga C, Rodrigues P, Scholes RJ, Sumaila UR, Walpole M (2010) Scenarios for global biodiversity in the 21st century. Science 330(6010):1496–1501 Petchey OL, Hector A, Gaston KJ (2004) How do different measures of functional diversity perform? Ecology 85(3):847–857 Plevin RP, O´Hare M, Jones AD, Torn MS, Gibbs HK (2010) Greenhouse gas emissions from biofuels´ indirect land use change are uncertain but may be much greater than previously estimated. Environ Sci Technol 44:8015-8021 Poschlod P, WallisdeVries M (2002) The historical and socioeconomic perspective of calcareous grasslands—lessons from the distant and recent past. Biodivers Conserv 104:361–376 Prach K, Rehounkova K (2006) Vegetation succession over broad geographical scales: which factors determine the patterns? Preslia 78:469–480 R Development Core Team (2011) R: A language and environment for statistical computing; R Foundation for Statistical Computing, Vienna. Rahmann G (2011) Biodiversity and organic farming: What do we know? Agric For Res 3(61):189–208 Ramankutty N, Evans A, Monfreda C, Foley JA (2008) Farming the planet: 1. Geographic distribution of global agricultural lands in the year 2000. Glob Biogeochem Cycles, 22:1– 19 Sala OE, Chapin F, Armesto JJ, Berlow E, Bloomfield J, Dirzo R, Huber-Sanwald E, Huenneke LF, Jackson R, Kinzig A, Leemans R, Lodge D, Mooney H, Oesterheld M, Poff N, Sykes MT, Walker BH, Walker M, Wall D (2000) Global biodiversity scenarios for the year 2100. Science 287(5459):1770–1774 Schader C, Drapela T, Markut T, Hoertenhuber S, Lindenthal T, Meier M, Pfiffner L (2010) Biodiversity impact assessment of Austrian organic and conventional dairy products. In: LCA Discussion Forum: Integrating biodiversity in LCA, Lausanne, Nov 19, 2010 Schmidt J (2008a) Development of LCIA characterisation factors for land use impacts on biodiversity. J Clean Prod 16(18):1929–1942 Schmidt J (2008b) System delimitation in agricultural consequential LCA Outline of methodology and illustrative case study of wheat in Denmark. Int J Life Cycle Assess. 13:350–364 Scott AJ, Morgan JW (2012) Resilience, persistence and relationship to standing vegetation in soil seed banks of semi-arid Australian old fields. Appl Veg Sci 15(1):48–61 Seufert V, Ramankutty N, Foley JA (2012) Comparing the yields of organic and conventional agriculture. Nature 485(7397):229–232 SOER Synthesis (2010) The European environment — state and outlook 2010: synthesis. European Environment Agency, Copenhagen StatBank Denmark (2012) Crops by crop, unit and area from http://www.statbank.dk/statbank5a/SelectVarVal/Define.asp?MainTable=AFG1&PLangua ge=1&PXSId=0. Accessed June 2012 Stolze M, Lampkin N (2009) Policy for organic farming: Rationale and concepts. Food Policy 34(3):237–244 Swift TL, Hannon SJ (2010) Critical thresholds associated with habitat loss: a review of the concepts, evidence, and applications. Biol Rev 85(1):35–53
107 Comparing direct LU impacts on biodiversity of conventional and organic milk –a Swedish case study
Thomassen M, Vancalker K, Smits M, Iepema G, De Boer I (2008) Life cycle assessment of conventional and organic milk production in the Netherlands. Agric Syst 96:95–107 Tilman D (2001) Forecasting agriculturally driven global environmental change. Science 292(5515):281–284 Tscharntke T, Clough Y, Wanger TC, Jackson L, Motzke I, Perfecto I, Vandermeer J, Whitbread A (2012) Global food security, biodiversity conservation and the future of agricultural intensification. Biol Conserv 151:53–59 Van der Werf HM, Kanyarushoki C, Corson MS (2009) An operational method for the evaluation of resource use and environmental impacts of dairy farms by life cycle assessment. J Environ Manag 90(11):3643–3652 Van Elsen T (2000) Species diversity as a task for organic agriculture in Europe. Agric Ecosyst Environ 77:101–109 Weidema B, Lindeijer E (2001) Physical impacts of land use in product life cycle assessment. Final report of the EURENVIRON-LCAGAPS sub-project on land use. Department of Manufacturing Engineering and Management, Technical University of Denmark, Lyngby Yan M, Humphreys J, Holden NM (2011) An evaluation of life cycle assessment of European milk production. J Environ Manag 92(3):372–379 Zhang W (2008) A forecast analysis on world population and urbanization process. Environ Dev Sustain 10(6):717–730 Zimmermann JK, Aide TM, Lugo AE (2007) Old fields: dynamics and restoration of abandoned farmland. In: Chazdon RL (2008) Beyond Deforestation: Restoring Forests and Ecosystem Services on Degraded Lands. Science 320(5882):1458–1460
108
Chapter 5
Land use in life cycle assessment: Global characterization factors based on regional and global potential species
extinction
Laura de Baan
Christopher L. Mutel
Michael Curran
Stefanie Hellweg
Thomas Koellner
Published in Environmental Science and Technology
Volume 47 Issue 46 (2013): 9281–9290
Land use in LCA: Global CFs based on regional and global potential species extinction
Abstract
Land use is one of the main drivers of biodiversity loss. However, many life cycle assessment studies do not yet assess this effect because of the lack of reliable and operational methods. Here, we present an approach to modeling the impacts of regional land use on plants, mammals, birds, amphibians, and reptiles. Our global analysis calculates the total potential damage caused by all land uses within each WWF ecoregion and allocates this total damage to different types of land use per ecoregion. We use an adapted (matrix-calibrated) species-area relationship to model the potential regional extinction of non-endemic species caused by reversible land use and land use change impacts. The potential global extinction of endemic species is used assess irreversible, permanent impacts. Model uncertainty is assessed using Monte Carlo simulations. The impacts of land use on biodiversity varied strongly across ecoregions, showing the highest values in regions where most natural habitat had been converted in the past. The approach is thus retrospective and was able to highlight the impacts in highly disturbed regions. However, we also illustrate how it can be applied to prospective assessments using scenarios of future land use. Uncertainties, modeling choices, and validity are discussed.
5.1 Introduction
Life cycle assessment (LCA) is a methodology for quantifying the environmental impacts of products and processes over their entire life cycle (ISO 2006). LCA results can highlight reduction potentials of environmental impacts and help to avoid solutions that simply shift burdens between different environmental compartments (e.g., reducing CO2 emissions but increasing pressure on biodiversity) or between different life cycle stages of products. LCA studies should ideally capture all environmentally relevant aspects of economic activities, but for some aspects, such as land use impacts on biodiversity, reliable methods are still lacking (Milà i Canals et al. 2007; Koellner et al. 2013b). Because land use is a main driver of global biodiversity loss (Sala et al. 2000; Millennnium Ecosystem Assessment 2005; Alkemade et al. 2009; Pereira et al. 2010), it cannot be ignored in environmental decision-making.
Biodiversity is a multifaceted concept that encompasses different hierarchical levels of life (genes, species, populations, and ecosystems) and their various attributes (composition, function, and structure; Noss 1990), including strong spatial and temporal dynamics. Different indicators have been proposed for measuring land use impacts on biodiversity in life cycle impact assessment (LCIA; Lindeijer 2000b; Curran et al. 2011). Many existing methods quantify biodiversity loss based on differences in local species richness between land use types; however, some studies included other biodiversity indicators, such as ecosystem scarcity and vulnerability (Weidema and Lindeijer 2001; Michelsen 2008), functional species diversity (de Souza et al. 2013), and the bio-physical impacts of land use on ecosystem services (Müller-Wenk and Brandão 2010; Brandão and Milà i Canals 2013; Saad et al. 2013). For assessing impacts on species richness, most methods were developed for specific world regions, such as Central and Northern Europe (Koellner 2000; Vogtländer et al. 2004; Koellner and Scholz 2007, 2008; Schmidt 2008; Jeanneret
110 Land use in LCA: Global CFs based on regional and global potential species extinction et al. 2009; De Schryver et al. 2010), Malaysia/Indonesia (Schmidt 2008), Japan (Itsubo and Inaba 2012), and California (Geyer et al. 2010). Extrapolating results from one region to others poses a challenge because biodiversity varies strongly across regions, and the required input data are often only available for one or a few countries or regions. Data availability also limits the range of taxonomic groups that are evaluated within LCA. Although results have been shown to depend on the assessed taxonomic groups (Koellner and Scholz 2008; de Baan et al. 2013a), most methods are based on a single taxon, mainly vascular plants (Koellner 2000; Lindeijer 2000a; Weidema and Lindeijer 2001; Vogtländer et al. 2004; Schmidt 2008; De Schryver et al. 2010; Itsubo and Inaba 2012), but also vertebrates (Geyer et al. 2010). Only a few authors proposed globally applicable land use LCIA methods. Two early studies combine global plant species richness maps with information on the impact of different land use types (Lindeijer 2000a; Weidema and Lindeijer 2001). One recent study used empirical data for multiple taxonomic groups to calculate the relative difference in species richness of different land use types and regional reference habitats for various biomes (de Baan et al. 2013a). However, this previous study considered only local land use impacts, and did not consider impacts of land use change and results remained at a coarse spatial resolution (biomes).
In the present study, we focus on quantifying potential species extinction due to land use. The concept of extinction is highly relevant to public debates about biodiversity loss and is clearly communicable to decision-makers. Land use activities can contribute to local, regional or even global species extinction, which show different degrees of reversibility. For example, converting one hectare of rain forest to cropland can lead to a local displacement of about 50% of all species (de Baan et al. 2013a). If viable populations of these locally displaced species still exist in adjacent rain forest areas, they can gradually recolonize the cleared field after it is abandoned. In this case, local impacts are (to a large extent) reversible, although decades or centuries might be needed for full biodiversity recovery (Curran et al. in press). The risk of regional extinction of species increases if only small remnants of rain forest habitat remain (Swift and Hannon 2010). Reintroducing regionally extinct species might still be possible if sufficient suitable habitat can be provided in the region and species have the ability to recolonize the area. However, if the region contains the full geographic range of a species (i.e., it is endemic), there is a high risk that it will become globally extinct (Purvis et al. 2000), which is totally irreversible.
When species extinction is used in LCA as a measure of biodiversity loss, it is therefore essential to define clearly the spatial scale of impacts. Existing LCIA methods of land use differ in the spatial scales at which they assess impacts. Some studies address both local and regional impacts (Koellner 2000; Weidema and Lindeijer 2001; Vogtländer et al. 2004; Schmidt 2008; De Schryver et al. 2010), whereas others consider only local (Lindeijer 2000a; Koellner and Scholz 2007, 2008; de Baan et al. 2013a) or regional impacts (Geyer et al. 2010; Itsubo and Inaba 2012). To our knowledge, no existing land use LCIA method assessed global species extinction, despite its irreversible nature and high political relevance (Convention on Biological Diversity 2013).
We present a spatially explicit approach to assess the impacts of land use on biodiversity at both regional and global scales. We model the potential regional species loss caused by total
111 Land use in LCA: Global CFs based on regional and global potential species extinction accumulated land use activities within all global ecoregions (Olson et al. 2001) and use this data to calculate damage scores (so-called characterization factors) per land use activity, ecoregion and taxon (mammals, birds, amphibians, reptiles, and plants). Following a convention of LCA (Milà i Canals et al. 2007; Koellner et al. 2013b), we calculate characterization factors for reversible impacts caused by occupation (land use), transformation (land use change) as well as permanent impacts (further explained in Section 5.2). Finally, using Monte Carlo calculations, we analyze and quantify the uncertainties of the characterization factors.
5.2 Materials and Methods
5.2.1 Modeling species extinction
A species-area relationship (SAR) model was used to assess the number of species that might be driven to extinction because of land use. This model is derived from island biogeography theory (MacArthur and Wilson 1963), which describes a power relationship between the area A of an ecosystem and the number of species S it contains, as follows
z S =cA (5.1) where c and z are constants (Arrhenius 1921).
SAR models are commonly used to predict species extinction caused by habitat loss (Brooks et al.
2002; Millennnium Ecosystem Assessment 2005; Pimm et al. 2006). The number of species Snew of an area Anew is then calculated as a function of the number of species Sorg occurring in the original habitat area Aorg (Koh and Ghazoul 2010):
z S ⎛ A ⎞ new = ⎜ new ⎟ (5.2) S ⎜ A ⎟ org ⎝ org ⎠
A shortcoming of the SAR model is that it traditionally focuses on natural habitats and assumes that no species persist on human-modified land (the so-called “matrix”), although in reality this land provides habitat for some species groups (e.g., farmland birds; Pereira and Daily 2006; de Baan et al. 2013a). We therefore used the adapted, matrix-calibrated species-area relationship model (hereafter, matrix SAR) developed by Koh and Ghazoul (2010). This model predicts lower species extinction risks due to habitat conversion when the converted region contains suitable habitat for some species groups. Mathematically, it lowers the curve of the species-area relationships (Koh and Ghazoul 2010) by adapting the z value (Eq. 3.3; Koh and Ghazoul 2010) of the power model (Eq. 5.2). The suitability of the matrix depends on the sensitivity σ of the species group to all land use types, i to n, and on the composition of the matrix, expressed as the relative area share p of each land use type i from the total converted land area:
n z ' = z p σ (5.3) ∑i i i
112 Land use in LCA: Global CFs based on regional and global potential species extinction
The sensitivity σ is quantified as the relative decrease in species richness (S) between a land use type i (Si) and a (natural) reference habitat (Sref). This equals local land occupation characterization factors CFloc, as calculated by de Baan et al. (2013a) based on median values of CFloc per biome.
S S ref − σ =CF = i (5.4) i loc,Occ,i S ref
The species lost Slost per taxonomic group g due to cumulative land use in a region is thus given by substituting Eqs. 5.3 and 5.4 into Eq. 5.2 (Koh and Ghazoul 2010).
n z p CF ∑i i loc ,i ⎛ A ⎞ S = S − S = S − S i⎜ new ⎟ (5.5) lost,g org,g new ,g org,g org,g ⎜ A ⎟ ⎝ org ⎠
Following the suggestions of Koellner et al. (2013a), we chose ecoregions delineated by the World Wide Fund for Nature (WWF; Olson et al. 2001) as spatial units for calculating species loss caused by land use. Ecoregions contain distinct communities of species, and their boundaries approximate the original extent of natural ecosystems prior to major land use change (Olson et al. 2001).
Because CFloc,i ranges from positive to negative values (representing a higher species richness on used land than on the reference habitat; de Baan et al. 2013a), the exponential term in Eq. 5.5 could be negative, resulting in a gain in regional species richness (negative Slost). Although human land use can sometimes increase the regional species pool (Zobel 1997; Zobel 2001), the matrix SAR was not developed to model this aspect. We thus did not allow an overall regional species gain (negative Slost) by setting ΣpiCFloc,i <0 to ΣpiCFloc,i=0 resulting in Slost,g=0. The beneficial impacts of individual land use types (i.e., CFloc,i < 0) on species richness were permitted, which resulted in negative characterization factors (see below).
5.2.2 Calculation of characterization factors
According to the UNEP/SETAC Life Cycle Initiative (Milà i Canals et al. 2007; Koellner et al. 2013b), three types of land use impacts can be distinguished in LCA, which are quantified as the reduction of biodiversity over time and space (see Appendix Figure D1). The transformation impact quantifies the original change in diversity due to natural habitat conversion and additionally includes the time lag in the eventual recovery of the site back to a natural state (at some undetermined point in the future). The occupation impact captures the biodiversity loss attributed to preventing this recovery from taking place (i.e., because the site is occupied for human land use, it is unavailable for a proportion of species to occupy). These two impacts are considered fully reversible, given sufficiently long time horizons. Permanent impacts account for the irreversible damages due to incomplete biodiversity recovery. Here, we consider the global extinction of endemic species irreversible (i.e. permanent impact), whereas the regional extinction of non-endemic species is treated as potentially reversible (i.e. transformation and occupation impacts).
113 Land use in LCA: Global CFs based on regional and global potential species extinction
We used Eq. 5.5 to calculate the total number of non-endemic species lost per ecoregion j and taxonomic group g because of the cumulative land use within the ecoregion. This total regional damage was then allocated to the different land use types i according to their relative frequency pi,j in the region j and their habitat quality CFloc,i,j. The allocation factor a is then calculated for each land use type i and ecoregion j as follows (Eq. 5.6):
p CF i ,j loc,i ,j a (5.6) i ,j = n p CF ∑i i ,j loc,i ,j
Regional characterization factors for occupation of each land use type, CFOcc,reg,i,j,g, were calculated by multiplying the potentially lost non-endemic species per region j with the corresponding allocation factor ai,j and dividing this by the area occupied by the land use type, Ai,j (Eq. 5.7). This finally gives us the unit [potentially lost non-endemic species] for occupying 1 square meter for 1 year.
S a lost,nonend ,j ,g i i ,j CF = (5.7) Occ,reg,i ,j ,g A i ,j
Regional characterization factors for transformation, CFTrans,reg,i,j,g, were calculated as a multiplication of CFOcc,reg,i,j,g with half the regeneration time (Milà i Canals et al. 2007; Koellner et al. 2013b), treg (Eq. 5.8, see Appendix Figure D1). Here, the unit is [potentially lost non-endemic species · years] for transforming 1 square meter.
S a 1 1 lost,nonend ,j ,g i i ,j CF = it iCF = it (5.8) Trans,reg,i ,j ,g 2 reg,i ,j ,g Occ,reg,i ,j ,g 2 reg,i ,j ,g A i ,j
Finally, regional characterization factors for permanent impacts, CFPerm,reg,i,j,g, were calculated based on the total number of potentially lost endemic species, Slost,end,j,g, per ecoregion j and taxonomic group g caused by all the accumulated land uses within the region (Eq. 5.9). This total damage was allocated to the different types of land use within the region (multiplication by ai,j) and divided by the area of each land use. Because global extinction is fully irreversible, and the impacts potentially continue during an infinite time, we calculated the impacts only over the modeling
6 period, tm = 10 years, representing the estimated time for a new species to evolve (Weir and Schluter 2007; Smith et al. 2012). The unit is [potentially lost endemic species · years] for transforming 1 square meter of land.
S a lost,end ,j ,g i i ,j CF =t i (5.9) Perm,reg,,i ,j ,g m A i ,j
For application in an LCA study, the CFOcc are multiplied by the inventory flow of occupation, that
2 is, the land requirements of a product given in [m ·years]. The CFTrans and CFPerm are multiplied by the inventory flow of transformation, that is, the amount of land use change per product in [m2]. The three impacts can be summed up into the total regional biodiversity depletion potential for
114 Land use in LCA: Global CFs based on regional and global potential species extinction
2 each taxonomic group g (rBDPg) expressed in the unit [potentially lost species · m · year]. By choosing the modeling time for the permanent impacts, tm, a weighting can be performed to indicate the relative importance of global species extinction relative to regional extinction. The Appendix includes a proposed method for aggregating CFs across taxa (Appendix section D.3.6) and deriving world average CFs for land use flows with unspecified locations (Appendix section D.3.7).
5.2.3 Input data for model parameters
For each of the aforementioned parameters, we used globally available data and, where possible, created an uncertainty distribution for each parameter (see overview in Appendix Table D1). CFloc
(Alkemade et al. 2009; de Baan et al. 2013a), z (Drakare et al. 2006) and treg (Curran et al. in press) were derived from global meta-studies and the data were subset into various configurations based on data availability and relevance. The CFloc were shown to differ significantly across biome and land use types (de Baan et al. 2013a). Thus, we split the data into CFloc specific to land use type and biome. If less than five data points per land use type and biome were available, world average CFloc was used for the land use type. The z-values strongly differed between broad habitat types (Drakare et al. 2006) and were calculated specifically for islands, forests, and non-forest ecoregions (see spreadsheet EcoregionDescription in online supporting information SI2 of de Baan et al. (2013b) for how ecoregions were assigned to these habitat types and Appendix Table D2 for the applied z-values).
Recovery times, treg, were estimated based on a meta-analysis by Curran et al. (in press). This study reviewed 108 peer-reviewed publications that compared the diversity of old growth (OG) and secondary growth (SG) habitats of different ages, using the occurrence-based Sørenson similarity index as a measure of diversity. Curran et al. (in press) used generalized linear models to predict the time it would take for SG–OG similarity to reach average OG–OG values, based on within-study comparisons. Predictors of recovery included habitat age, taxon, latitude, altitude, previous disturbance intensity, biogeographic realm, and a simplified biome classification. Based on the model parameters for these predictors, we calculated land use-, taxon- and region-specific recovery times for 520 archetypical situations (parameter combinations): recovery after “intensive” land use (agriculture and urban land, pasture in forest ecosystems, and managed forests in non-forest ecosystems) or “extensive” land use (pasture in open ecosystems and managed forests in forest ecosystems) for each taxonomic group (plants, birds, mammals, and herpetofauna) in 65 world regions (WWF biomes-realm combinations). For each region, the median distance from the equator, median elevation, and biome type (forest or non-forest) were specified and used to calculate the recovery times (see Appendix Table D3 for input parameters and resulting median recovery times).
Data on original species richness and endemism of mammals, birds, amphibians, and reptiles were derived from the WWF WildFinder database (Olson et al. 2001; World Wildlife Fund 2006). Because no data on plant endemism are available per ecoregion, (Holger Kreft, pers. comm.),
115 Land use in LCA: Global CFs based on regional and global potential species extinction permanent impacts could not be calculated for plants. Total plant species richness (Kier et al. 2005) was used to calculate occupation and transformation impacts.
Several global land cover and land use maps are available, but their agreement on cover types and distribution is limited (Fritz and See 2008; Fritz et al. 2010; Tchuente ́ et al. 2011). Most maps do not distinguish between natural, managed, or inhabited forests or grassland (Bartholomé and Belward 2005; European Space Agency 2009). We thus chose two maps, Land Degradation Assessment in Drylands (1998-2008; LADA 2008) and Anthromes (2000-2005; Ellis and Ramankutty 2008) for deriving land use shares per ecoregion, which combine remote sensing data with statistics on human activities (both at a 5 arc minutes resolution). Five broad land use types were distinguished (agriculture, pasture, managed forests, urban area, and natural habitat) in our model. To estimate parameter uncertainty, we calculated the area shares of each land use type per ecoregion in a GIS separately for each land use map. The maps were first transformed from WGS1984 projection to equal-area projections, using seven globally applicable equal-area projections. This resulted in n = 2 x 7 = 14 different estimates of land use shares for all global ecoregions.
5.2.4 Uncertainty assessment
Parameter uncertainty was propagated into the characterization factors using Monte-Carlo simulation (1000 iterations). For each parameter, a distribution was directly derived from the data using non-parametric Kernel density estimation (KDE); if only data ranges were available, a triangular distribution was assumed (see Appendix Table D1). For recovery times, a log-normal distribution was assumed (Curran et al. in press). With the exception of plants, no uncertainty information was available for species richness and endemism. Thus, these parameters were modeled without uncertainty. Median, upper, and lower 95% confidence intervals were calculated for each characterization factor.
To assess the influence of each parameter on the uncertainty of characterization factors, their contribution to variance (CTV; Geisler et al. 2004) was calculated. In this method, the Spearman's rank-order correlation coefficient (ROCC) of each parameter with the characterization factor results is calculated for the set of Monte Carlo iterations. The CTV is calculated as follows:
ROCC 2 CTV i (5.10) = n ROCC 2 ∑i i where i is the calculated parameter, and n is the set of all parameters (see also Mutel et al. 2013).
5.2.5 Validation of species extinction
To test the validity of the model, we compared our prediction of the global extinction of endemic species (Slost,end) with the observed numbers of extinct and threatened species (Koh and Ghazoul 2010). Because it can take decades or centuries for a species with nonviable populations to disappear completely (extinction debt; Tilman et al. 1994; Brooks et al. 1999), we considered the
116 Land use in LCA: Global CFs based on regional and global potential species extinction following species “condemned to extinction”: all species classified by the International Union for Conservation of Nature (IUCN) as “vulnerable,” “endangered,” “critically endangered,” or “extinct”. Data per ecoregion were extracted from the WWF WildFinder database (World Wildlife Fund 2006) for endemic mammals, birds, amphibians, and reptiles. No validation could be performed for the regional extinction of non-endemic species (Slost,nonend) because only information on global extinction was available.
5.2.6 Comparison of model choices
The above-presented model to derive land use characterization factors calculates the average impacts of past land use changes and is thus retrospective. Alternatively, prospective impacts can be calculated as marginal changes (Huijbregts et al. 2011; Weidema 2012), that is, the impact of one additional m2 of future land use change. To illustrate how these model choices can be implemented to the matrix SAR model, we calculated average and marginal impacts for all forest ecoregions of the Amazon (n = 19) under both retrospective (i.e. the current state) and prospective approaches. This region was selected because future land use scenarios for agriculture, pasture, forestry, and urban areas are not available on a per-ecoregions basis. The Amazon contains some relatively undisturbed ecoregions, which are expected to be converted for human use in the near future. Best- (good governance) and worst-case (business-as-usual) land use scenarios per ecoregion were derived from Soares-Filho et al. (2006) for the year 2050 and used to calculate prospective CFs. The equations for calculating marginal CFs in addition to details of the applied method and scenarios are given in the Appendix section D.3.4.
5.3 Results
5.3.1 Regional characterization factors
All median regional CFs including upper and lower 95% confidence intervals are displayed in spreadsheets in online supporting information SI2 of de Baan et al. (2013b). The regions with high median CFs largely overlapped for occupation, transformation, and permanent impacts and for all five taxa (mammals, birds, amphibians, reptiles, and plants; see correlation analysis in the Appendix Table D5-D6). The regions with high CFs largely corresponded to regions that had been heavily converted in the past (see Figures 5.1 and 5.2 for a selection and Appendix Figures D2-D6 for all CFs). CFs were very low in regions with large shares of undisturbed habitat. Globally across ecoregions, median CF values ranged over several orders of magnitude and showed large differences within biomes, indicating that a resolution finer than biomes is required for regionalized biodiversity assessments. For most ecoregions, the median occupation CFs of different land use types were within the same order of magnitude (Figure 5.2 and online supporting information SI2 of de Baan et al. 2013b), but transformation CFs showed larger differences between land use types. Generally, to determine land use CFs, the region where land was being used seemed more important than the type of land use. In most ecoregions, agriculture had the highest median CFs, but the ranking of the CFs of the other three land use
117 Land use in LCA: Global CFs based on regional and global potential species extinction types (pasture, urban areas, and forestry) was not consistent across taxa, type of impact, or ecoregion. In all but 5 ecoregions, CFs differed significantly across land use type (p < 0.05, evaluated by Kruskal-Wallis tests) for both occupation and transformation impacts. In general, the CFs (expressed in number of species lost) were highest for the most species rich taxa, plants, followed in decreasing order by birds, mammals, reptiles, and amphibians (the least species rich taxa).
Figure 5.1. Median characterization factors of agricultural land, based on birds (left) and mammals (right), for occupation (top), transformation (middle) and permanent impacts (bottom). NA: No data available. Data for other land use types and taxa (amphibians and reptiles) are given in the Appendix Figures D2 to D6 and in the online supporting information SI2 of de Baan et al. (2013b). Symbology based on Jenks natural breaks classification method (Jenks and Caspall 1971).
Median recovery times, treg, used for calculating transformation CFs (Eq. 5.8) ranged between 81 – 1231 years (Appendix Table D3). For most ecoregions, the median CFs of transformation were larger than of occupation, but the medians did not directly mirror their mathematical relationship
(see Eq. 5.8, CFTrans=0.5tregCFOcc). This is due to the high uncertainties and skewed distributions of recovery times as well as CFOcc and highlights the importance of considering CF uncertainties in LCA applications.
118 Land use in LCA: Global CFs based on regional and global potential species extinction
Figure 5.2. Median occupation characterization factors based on plant species, for agriculture, urban, pasture, and managed forests. NA: No data available. Symbology based on Jenks natural breaks classification method (Jenks and Caspall 1971).
5.3.2 Contribution to uncertainty
Uncertainty of the CFs was considerable. Local characterization factors CFloc,i contributed the most to the variance of regional CFs of occupation (67%–96%) and transformation (30%–84%, Appendix Tables D7 and D8). This result can be attributed to the value range of the local CFs, which spanned both positive and negative values (i.e., both damaging and beneficial impacts on biodiversity; de Baan et al. 2013a). The highly uncertain regeneration times resulted in higher uncertainties of transformation CFs compared to occupation CFs. Regeneration times contributed about half the uncertainty to the transformation CFs of agriculture (34%–54%), but were slightly less relevant for the other land use types (6%–28%, Appendix Table D7). Moreover, z-values
(<0.3%) and the parameters for area (original area Aorg (<3%), remaining natural habitat Anew (<5%), and area of each land use type Ai (<5%)) did not contribute greatly to the overall uncertainty of CFs.
5.3.3 Model evaluation
To evaluate our predictions of global extinction (Slost,end), we compared them with the observed numbers of threatened and extinct endemic species. The observed numbers were mostly within the uncertainty ranges of our predictions, but our median values were generally smaller than the observed values (Appendix Figure D10).
119 Land use in LCA: Global CFs based on regional and global potential species extinction
5.3.4 Comparison of model choices
Using land use scenarios for 2050 (Soares-Filho et al. 2006), we calculated prospective characterization factors for 19 ecoregions in the Amazon. Compared to the retrospective approach, the median CFs increased in ecoregions with large projected deforestation rates but did not change notably in ecoregions with lower deforestation rates (Appendix Figure D9). The highest increase of 65% in median CFs was observed in ecoregions with a projected decrease in remaining natural forest area from 60% to 20%. Average and marginal CFs were similar in ecoregions with large remaining natural habitat areas. Marginal CFs were much larger than average in ecoregions with little natural habitat (Figure 5.3, Appendix Figures D7 and D9).
5.4 Discussion
5.4.1 Modeling choices
The CFs of occupation and transformation of different taxa and land use types showed a relatively clear and consistent picture of regions with potentially high impacts of land use. High CFs occurred in highly vulnerable ecoregions where most natural habitat had been converted in the past, showing a strong overlap with biodiversity hotspots (Myers et al. 2000), which were classified by Brooks et al. (2006) as a reactive global conservation priority scheme. Ecoregions with low CFs largely overlapped with proactive conservation priority maps, such as of the Last of the Wild (Sanderson et al. 2002), which identifies large, undisturbed areas with high potential for conservation. The presented approach is clearly retrospective and reactive, because it considers only past changes in land use. In principle, the matrix SAR model can also be applied to future scenarios of land use change and provide a prospective assessment, which we illustrated in the example of 19 forest ecoregions of the Amazon. In ecoregions with high rates of projected future land use change, this prospective assessment might be more appropriate to represent the potential future loss of species. Global land use scenarios could be derived from global models, such as IMAGE (Bouwman AF et al. 2006), but would add another dimension of uncertainty to the CFs.
Based on the matrix SAR, both average and marginal impacts can be calculated. Average and marginal CFs are similar in ecoregions with low levels of converted natural habitat, but marginal impacts are much higher at high levels of habitat conversion and tend towards infinity when all remaining natural habitat is converted (Figure 5.3). Differences between CFs of ecoregions at high and low levels of habitat conversion are clearly magnified in a marginal approach. We thus do not recommend using a marginal approach in combination with a retrospective assessment because it ignores the conservation concerns of regions with large amounts of intact natural habitat that may be highly threatened by future land use pressure, such as the Amazon.
120 Land use in LCA: Global CFs based on regional and global potential species extinction
marginal CF average CF
Today! 2050!
Land use change! Characterization factor (CF) factor Characterization 0% 50% 100% Natural habitat conversion
Figure 5.3. Illustration of average and marginal occupation characterization factors for current and future land use in a hypothetical region (assuming a constant composition of land use, pi,j).
5.4.2 Data availability and uncertainty
In this study, we provided global CFs, which can be used in a range of LCA applications. We thereby relied on available global land use and biodiversity data, such as the WWF WildFinder database (Olson et al. 2001; World Wildlife Fund 2006). Because the latter does not distinguish between ecoregions with missing data or zero species richness or endemism, we treated both cases as missing data. This resulted in missing data for many ecoregions, particularly for permanent impacts, and the reduced applicability of this type of impact assessment in standard LCAs. Data for amphibians and reptiles were less complete than for mammals and birds. As new data on these, or additional, species groups emerges, the presented CFs should be updated. For now we suggest to supplement missing values with average CFs for neighboring ecoregions of the same biome.
For most input parameters (e.g., local CFs and z-values), the available data were not taxa- or ecoregion-specific but instead were aggregated across larger spatial units (e.g., biomes or globally). As expected, the resulting regional CFs were highly uncertain, often ranging from positive (detrimental) to negative (beneficial) values. The parameter dominating the uncertainty of regional CFs were the local CFs. These were derived from a quantitative review of comparative biodiversity surveys of human-modified land (Alkemade et al. 2009; de Baan et al. 2013a) and ranged between positive and negative values (the latter represented a higher species richness on the used land compared to a natural reference habitat). Better data on taxa- and region-specific habitat suitability for different land use types are required to reduce uncertainties in assessing the impacts of land use on biodiversity. In addition, the use of detailed and accurate global land use classification maps, such as those currently developed by the Geo-Wiki project (Fritz et al. 2012) or van Asselen and Verburg (2012), could further reduce uncertainties. We only distinguished between four very broad land use classes, each of which contained a range of management practices. To compare management practices (e.g., organic vs. conventional farming), global land management maps and more refined CFs would need to be developed, as impacts of different management practices on biodiversity differ considerably (Mueller et al. in press).
121 Land use in LCA: Global CFs based on regional and global potential species extinction
The uncertainty of transformation CFs was strongly influenced by the uncertainty of the biodiversity recovery times. This is the first global land use LCIA study that does not use recovery times based on expert estimates (van Dobben et al. 1998) but applies values derived from a meta- analysis of empirical data (Curran et al. in press). The recovery of biodiversity following disturbance is a complex process, and the predicted recovery times are highly uncertain and represent best-case scenarios (Curran et al. in press). The recovery time estimates implicitly assumed that adequate amounts of old growth habitat exist in a region, to act as a reservoir for old growth species to recolonize secondary growth areas. In some ecoregions, median recovery times exceeded 500 years (Appendix Table D3). It is doubtful whether recovery will occur at all in these regions (Morris et al. 2006), given the unlikely assumptions of non-use over the entire recovery period and future availability of source habitat. Even where these assumptions are met, Curran et al. (in press) estimated that recovery of species similarity is likely to fail altogether in about 15% of all cases. In addition, a reintroduction of regionally extinct non-endemic species is not possible, if the species got regionally extinct in all the ecoregions that it inhabits. The assumption that regional species loss is fully reversible is clearly too optimistic and leads to the underestimation of permanent impacts. Given these caveats, we believe that this study is important because it is the first to include reversibility into land use LCIA methods. Here, we modeled permanent impacts based on endemism. Future LCIA studies modeling permanent impacts should attempt to include additional risk factors for global species extinction, such as global rarity or vulnerability of species (Verones et al. in press) and the likelihood that recovery will fail in different regions or ecosystems (Curran et al. in press).
5.4.3 Model validity
Traditional species-area relationships (see Eq. 5.1) are commonly applied to assess species extinction caused by habitat change, but their validity has been questioned because they tend to overestimate extinction rates (He and Hubbell 2011). Koh and Ghazoul (2010) adapted the traditional SAR model to account for the habitat value of human-modified land. However, the matrix SAR has not yet been tested outside the tropics for non-endemic species and taxonomic groups other than birds (Koh and Ghazoul 2010; Garcia-Ulloa et al. 2012) or mammals (Koh et al. 2010).
For non-forest ecoregions, such as grassland, savanna, tundra, and deserts, CFs should be interpreted with caution because the ecological reactions to habitat change differ from those of forest ecosystems (Bond and Parr 2010). In the desert biome, the median of the five available data points for local CF of pasture (de Baan et al. 2013a) was negative (i.e., higher median richness in pastures than on reference habitat), and because pasture was the dominant land use type in most desert ecoregions, the power term of the damage model (Eq. 5.5) became negative and was capped at zero (see Methods). Therefore, the median species loss was assumed zero, resulting in regional CFs of zero for all land use types and taxa. Although local increases in species richness might occur because of irrigation (Wenninger and Inouye 2008), for example, reduced water availability within the watershed might still lead to reduced regional biodiversity (Verones et al. in press).
122 Land use in LCA: Global CFs based on regional and global potential species extinction
In our global analysis, the matrix SAR tended to give lower estimates of global species extinction than “true” extinction estimates (Appendix Figure D10). We can attribute this to two factors. First, we optimistically assumed that endemic and non-endemic species share the same habitat suitability scores (i.e., local CFs). In reality, endemic species are likely more sensitive to anthropogenic disturbances as they are often habitat specialists relying on old growth natural habitat. Second, we assumed that all vulnerable and threatened species will inevitably become extinct, and assumed a primary driver of land use alone, ignoring the effect of other drivers (e.g. invasive species, pollution, or global warming (Millennnium Ecosystem Assessment 2005)). Therefore, it is important to interpret the global CFs as a measure of potential regional or global species extinction and not as explicit predictions.
Although the matrix SAR models the non-linear effects of habitat loss on biodiversity, the CFs had to be linearized to match the LCA framework, which is typically linear and assumes steady-state conditions (Udo de Haes et al. 2004). These linear CFs should not be applied to LCA studies in which the product system significantly changes the share of remaining natural habitat in one ecoregion and linearity no longer holds. The temporal dimension was captured only by the recovery rate of biodiversity (assumed to be linear). Historical land use dynamics and co-evolution of species were ignored by comparing the current composition of land use with a potential “pre- human” situation, assuming that each ecoregion once consisted of a homogeneous ecosystem without any human influence (Olson et al. 2001). Conservation concerns in regions with a long history of land use, such as semi-natural habitat loss, fragmentation and land use intensification or abandonment (Donald et al. 2001; Krauss et al. 2010), are only partially represented in the model.
The matrix SAR predicts 100% species loss if no natural habitat remains within an ecoregion. For regions dominated by land use with high habitat value, where species loss is predicted to be very small, this leads to a sudden loss of all species at very low levels of remaining natural habitat (e.g., less than 0.1%). This model behavior is unrealistic, and occurs because remaining habitat
(Anew in Eq. 5.2) only considers pristine habitat. In this context, it might be worth testing alternative models of species extinction for application in global LCAs, such as the countryside SAR (Pereira and Daily 2006), which predicts that species adapted to human-modified habitats also survive in the absence of natural habitat.
5.4.4 Applicability
To apply the developed CFs in LCA studies, information is needed on the location of land use activities in the product life cycle. Recent developments of life cycle inventory databases should facilitate the collection of this data (see EcoInvent v3.0; Weidema et al. 2012). Because impacts differ strongly across regions, efforts to regionalize land use inventories on the level of ecoregions instead of countries are worthwhile; however, the CFs could also be employed in a probabilistic approach of determining the location of a particular land use (Mutel et al. 2012). For unknown locations of land use, global average CFs can be applied (given in the online supporting information SI2 of de Baan et al. 2013b). For the easier applicability, we also calculated
123 Land use in LCA: Global CFs based on regional and global potential species extinction aggregated CFs across all five taxonomic groups (given in the online supporting information SI2 of de Baan et al. 2013b), by first normalizing by the median species richness per ecoregion of each taxonomic group (giving equal weight to each taxonomic group, Appendix section D.3.6). Alternatively, CFs could be simply summed across taxonomic group (giving equal weight to each species). Although the reversible occupation and transformation impacts can be aggregated, adding permanent impacts requires a weighting between regional and global species loss, which can be performed by adapting the time horizon (tm in Eq. 5.9) considered for permanent impacts.
5.4.5 Implications
In this paper, we present an approach to derive globally applicable CFs of land use using a species-extinction model. Our approach allows an assessment of the impacts of land use in LCA that is more complete than previous methods. We provide global CFs for occupation, and for the first time, transformation and permanent impacts, including uncertainties for nearly all world regions for five taxonomic groups and four broad land use types. These three types of impacts provide decision-makers with information on the effects of actual land use, land use changes, and the risk of irreversible damage. The approach also aims to improve the environmental relevance of land use LCIA results: We regionalize the CFs to the ecologically relevant scales of ecoregions and calculate the impacts at the regional scale instead of the local scale (de Baan et al. 2013a), which is relevant for assessing species extinction risk. Finally, with the unit potential regional loss of species, we propose an absolute instead of a relative measure for biodiversity loss in LCA. This facilitates the comparison with absolutely measured land use impact on ecosystem services (Müller-Wenk and Brandão 2010; Brandão and Milà i Canals 2013; Koellner et al. 2013b; Saad et al. 2013). The currently prevalent unit in LCA, potentially disappeared fraction of species (PDF), has been criticized for its lacking definition of the scale of impacts, conflating both local, regional and global losses (Curran et al. 2011). This unclear definition results in a misleading aggregation of impacts on biodiversity of different impact pathways (e.g., land use, climate change, and eutrophication) modeled at different spatial scales (Curran et al. 2011). A reevaluation of meaningful and readily understandable endpoint units for measuring biodiversity loss in LCA is thus desirable, considering both scientific rigor and practicality.
Acknowledgements We thank Ursin Caduff for conducting a pre-study in his Bachelor’s thesis and Bao Quang Le, Llorenç Milà i Canals, and two anonymous reviewers for their helpful comments on the manuscript. This research was funded through ETH Zurich (Research Grant CH1-0308-3) and the European Commission (LC-IMPACT, Grant Agreement No. 243827, 7th Framework Programme).
5.5 References
Alkemade R, van Oorschot M, Miles L, Nellemann C, Bakkenes M, ten Brink B (2009) GLOBIO3: A framework to investigate options for reducing global terrestrial biodiversity loss. Ecosystems 12 (3):374-390. Arrhenius O (1921) Species and area. Journal of Ecology 9 (1):95-99.
124 Land use in LCA: Global CFs based on regional and global potential species extinction
Bartholomé E, Belward AS (2005) GLC2000: a new approach to global land cover mapping from Earth observation data. International Journal of Remote Sensing 26 (9):1959-1977. Bond WJ, Parr CL (2010) Beyond the forest edge: ecology, diversity and conservation of the grassy biomes. Biological Conservation 143:2395–2404. Bouwman AF, Kram T, Klein Goldewijk K (2006) Integrated modelling of global environmental change. An overview of IMAGE 2.4. Netherlands Environmental Assessment Agency. Brandão M, Milà i Canals L (2013) Global characterisation factors to assess land use impacts on biotic production. The International Journal of Life Cycle Assessment 18 (6):1243-1252. Brooks T, Mittermeier R, da Fonseca G, Gerlach J, Hoffmann M, Lamoreux J, Mittermeier C, Pilgrim J, Rodrigues A (2006) Global biodiversity conservation priorities. Science 313 (5783):58-61. Brooks TM, Mittermeier RA, Mittermeier CG, Da Fonseca GAB, Rylands AB, Konstant WR, Flick P, Pilgrim J, Oldfield S, Magin G, Hilton-Taylor C (2002) Habitat Loss and Extinction in the Hotspots of Biodiversity. Conservation Biology 16 (4):909-923. Brooks TM, Pimm SL, Oyugi JO (1999) Time lag between deforestation and bird extinction in tropical forest fragments. Conservation Biology 13 (5):1140-1150. Convention on Biological Diversity (2013) Aichi Biodiversity Targets. http://www.cbd.int/sp/targets/. Accessed 4.10.2013 Curran M, de Baan L, De Schryver A, van Zelm R, Hellweg S, Koellner T, Sonnemann G, Huijbregts MAJ (2011) Toward meaningful end points of biodiversity in life cycle assessment. Environmental Science & Technology 45 (1):70-79. Curran M, Hellweg S, Beck J (in press) Is there any empirical support for biodiversity offset policy? Ecological Applications. de Baan L, Alkemade R, Koellner T (2013a) Land use impacts on biodiversity in LCA: a global approach. The International Journal of Life Cycle Assessment 18 (6):1216-1230. de Baan L, Mutel CL, Curran M, Hellweg S, Koellner T (2013b) Land use in Life Cycle Assessment: Global characterization factors based on regional and global potential species extinctions. Environmental Science & Technology 47 (16):9281–9290. De Schryver AM, Goedkoop MJ, Leuven RSEW, Huijbregts MAJ (2010) Uncertainties in the application of the species area relationship for characterisation factors of land occupation in life cycle assessment. The International Journal of Life Cycle Assessment 15 (7):682-691. de Souza DM, Flynn DFB, Declerk F, Rosenbaum RK, de Melo Lisboa H, Koellner T (2013) Land use impacts on biodiversity: proposal of characterization factors based on functional diversity. The International Journal of Life Cycle Assessment 18 (6):1231-1242. Donald PF, Green RE, Heath MF (2001) Agricultural intensification and the collapse of Europe's farmland bird populations. Proceedings of the Royal Society B-Biological Sciences 268 (1462):25-29. Drakare S, Lennon J, Hillebrand H (2006) The imprint of the geographical, evolutionary and ecological context on species-area relationships. Ecology Letters 9 (2):215-227. Ellis E, Ramankutty N (2008) Putting people in the map: anthropogenic biomes of the world. Frontiers in Ecology and the Environment 6 (8):439-447. European Space Agency (2009) GlobCover land cover map, v2.3. European Space Agency. http://due.esrin.esa.int/globcover/. Accessed 26.2.2013 Fritz S, McCallum I, Schill C, Perger C, See L, Schepaschenko D, van der Velde M, Kraxner F, Obersteiner M (2012) Geo-Wiki: An online platform for improving global land cover. Environmental Modelling & Software 31 (C):110-123. Fritz S, See L (2008) Identifying and quantifying uncertainty and spatial disagreement in the comparison of Global Land Cover for different applications. Global Change Biology 14 (5):1057-1075. Fritz S, See L, Rembold F (2010) Comparison of global and regional land cover maps with statistical information for the agricultural domain in Africa. International Journal of Remote Sensing 31 (9):2237-2256.
125 Land use in LCA: Global CFs based on regional and global potential species extinction
Garcia-Ulloa J, Sloan S, Pacheco P, Ghazoul J, Koh LP (2012) Lowering environmental costs of oil-palm expansion in Colombia. Conservation Letters 5 (5):366-375. Geisler G, Hellweg S, Hungerbühler K (2004) Uncertainty analysis in life cycle assessment (LCA): case study on plant-protection products and implications for decision making. The International Journal of Life Cycle Assessment 10 (3):184-192. Geyer R, Lindner JP, Stoms DM, Davis FW, Wittstock B (2010) Coupling GIS and LCA for biodiversity assessments of land use: Part 2: Impact assessment. The International Journal of Life Cycle Assessment 15 (7):692-703. He F, Hubbell SP (2011) Species-area relationships always overestimate extinction rates from habitat loss. Nature 473 (7347):368-371. Huijbregts MAJ, Hellweg S, Hertwich E (2011) Do we need a paradigm shift in Life Cycle Impact Assessment? Environmental Science & Technology 45:3833-3834. ISO (2006) ISO 14040 International Standard. In: Environmental Management – Life Cycle Assessment – Principles and Framework. International Organisation for Standardization, Geneva, Switzerland. Itsubo N, Inaba A (2012) LIME2. LCIA method based on endpoint modeling. Newsletter. Life- Cycle Assessment Society of Japan. Jeanneret P, Baumgartner DU, Freiermuth-Knuchel R, Gaillard G (2009) Methode zur Beurteilung der Wirkung landwirtschaftlicher Aktivitäten auf die Biodiversität für Ökobilanzen (SALCA-Biodiversität). Agroscope Reckenholz-Tänikon, Zurich. Jenks GF, Caspall FC (1971) Error on Choroplethic Maps. Definition, Measurement, Reduction. Annals of the Association of American Geographers 61:217–244. Kier G, Mutke J, Dinerstein E, Ricketts T, Kuper W, Kreft H, Barthlott W (2005) Global patterns of plant diversity and floristic knowledge. Journal of Biogeography 32 (7):1107-1116. Koellner T (2000) Species-pool effect potentials (SPEP) as a yardstick to evaluate land-use impacts on biodiversity. Journal of Cleaner Production 8:293-311. Koellner T, de Baan L, Beck T, Brandão M, Civit B, Goedkoop MJ, Margni M, Milà i Canals L, Müller-Wenk R, Weidema B, Wittstock B (2013a) Principles for Life Cycle Inventories of land use on a global scale. The International Journal of Life Cycle Assessment 18 (6):1203-1215. Koellner T, de Baan L, Beck T, Brandão M, Civit B, Margni M, Milà i Canals L, Saad R, de Souza DM, Müller-Wenk R (2013b) UNEP-SETAC guideline on global land use impact assessment on biodiversity and ecosystem services in LCA. The International Journal of Life Cycle Assessment 18 (6):1188-1202. Koellner T, Scholz RW (2007) Assessment of land use impacts on the natural environment. Part 1: An analytical framework for pure land occupation and land use change. The International Journal of Life Cycle Assessment 12 (1):16-23. Koellner T, Scholz RW (2008) Assessment of land use impacts on the natural environment. Part 2: Generic characterization factors for local species diversity in Central Europe. The International Journal of Life Cycle Assessment 13 (1):32-48. Koh LP, Ghazoul J (2010) A matrix-calibrated species-area model for predicting biodiversity losses due to land-use change. Conservation Biology 24 (4):994–1001. Koh LP, Lee TM, Sodhi NS, Ghazoul J (2010) An overhaul of the species-area approach for predicting biodiversity loss: incorporating matrix and edge effects. Journal Of Applied Ecology 47 (5):1063-1070. Krauss J, Bommarco R, Guardiola M, Heikkinen RK, Helm A, Kuussaari M, Lindborg R, Öckinger E, Pärtel M, Pino J, Pöyry J, Raatikainen KM, Sang A, Stefanescu C, Teder T, Zobel M, Steffan-Dewenter I (2010) Habitat fragmentation causes immediate and time- delayed biodiversity loss at different trophic levels. Ecology Letters 13 (5):597-605. LADA (2008) Mapping Land use Systems at global and regional scales for Land Degradataion Assessment Analysis. Nachtergaele F., Petri, M. LADA Technical Report n.8, version 1.1. UNEP/GEF. Lindeijer E (2000a) Biodiversity and life support impacts of land use in LCA. Journal of Cleaner Production 8:313–319.
126 Land use in LCA: Global CFs based on regional and global potential species extinction
Lindeijer E (2000b) Review of land use impact methodologies. Journal of Cleaner Production 8:273–281. MacArthur RH, Wilson EO (1963) An equilibrium theory of insular zoogeography. Evolution 17 (4):373–387. Michelsen O (2008) Assessment of land use impact on biodiversity. The International Journal of Life Cycle Assessment 13 (1):22-31. Milà i Canals L, Bauer C, Depestele J, Dubreuil A, Knuchel R, Gaillard G, Michelsen O, Müller- Wenk R, Rydgren B (2007) Key Elements in a Framework for Land Use Impact Assessment Within LCA. The International Journal of Life Cycle Assessment 12 (1):5-15. Millennnium Ecosystem Assessment (2005) Ecosystems and human well-being: biodiversity synthesis. Washington, DC. Morris RKA, Alonso I, Jefferson RG, Kirby KJ (2006) The creation of compensatory habitat— Can it secure sustainable development? Journal for Nature Conservation 14 (2):106–116. Mueller C, de Baan L, Koellner T (in press) Comparing direct land use impacts on biodiversity of conventional and organic milk - based on a Swedish case study. The International Journal of Life Cycle Assessment. Müller-Wenk R, Brandão M (2010) Climatic impact of land use in LCA-carbon transfers between vegetation/soil and air. The International Journal of Life Cycle Assessment 15 (2):172-182. Mutel CL, de Baan L, Hellweg S (2013) Two-step sensitivity testing of parameterized and regionalized life cycle assessments: methodology and case study. Environmental Science & Technology 47 (11):5660–5667. Mutel CL, Pfister S, Hellweg S (2012) GIS-based regionalized Life Cycle Assessment: how big is small enough? Methodology and case study of electricity generation. Environmental Science & Technology 46:1096-1103. Myers N, Mittermeier R, Mittermeier C, da Fonseca G, Kent J (2000) Biodiversity hotspots for conservation priorities. Nature 403 (6772):853-858. Noss R (1990) Indicators for monitoring biodiversity - a hierarchical approach. Conservation Biology 4 (4):355-364. Olson D, Dinerstein E, Wikramanayake E, Burgess N, Powell G, Underwood E, D'Amico J, Itoua I, Strand H, Morrison J, Loucks C, Allnutt T, Ricketts T, Kura Y, Lamoreux J, Wettengel W, Hedao P, Kassem K (2001) Terrestrial ecoregions of the worlds: A new map of life on Earth. BioScience 51 (11):933-938. Pereira H, Daily G (2006) Modeling biodiversity dynamics in countryside landscapes. Ecology 87:1877:1885. Pereira H, Leadley P, Proença V, Alkemade R, Scharlemann JPW, Fernandez-Manjarrés JF, Araújo MB, Balvanera P, Biggs R, Cheung WWL, Chini L, Cooper HD, Gilman EL, Guénette S, Hurtt GC, Huntington HP, Mace GM, Oberdorff T, Revenga C, Rodrigues P, Scholes RJ, Sumaila UR, Walpole M (2010) Scenarios for global biodiversity in the 21st century. Science 330 (6010):1496-1501 Pimm S, Raven P, Peterson A, Sekercioglu CH, Ehrlich PR (2006) Human impacts on the rates of recent, present, and future bird extinctions. Proceedings of the National Academy of Sciences of the United States of America 103 (29):10941–10946. Purvis A, Gittleman JL, Cowlishaw G, Mace GM (2000) Predicting extinction risk in declining species. Proceedings of the Royal Society B-Biological Sciences 267 (1456):1947-1952. Saad R, Koellner T, Margni M (2013) Land use impacts on freshwater regulation, erosion regulation and water purification: a spatial approach for a global scale level. The International Journal of Life Cycle Assessment 18 (6):1253-1264. Sala O, Chapin F, Armesto J, Berlow E, Bloomfield J, Dirzo R, Huber-Sanwald E, Huenneke L, Jackson R, Kinzig A, Leemans R, Lodge D, Mooney H, Oesterheld M, Poff N, Sykes M, Walker B, Walker M, Wall D (2000) Global biodiversity scenarios for the year 2100. Science 287 (5459):1770-1774. Sanderson E, Jaiteh M, Levy M, Redford K, Wannebo A, Woolmer G (2002) The human footprint and the last of the wild. BioScience 52 (10):891-904.
127 Land use in LCA: Global CFs based on regional and global potential species extinction
Schmidt J (2008) Development of LCIA characterisation factors for land use impacts on biodiversity. Journal of Cleaner Production 16:1929-1942. Smith BT, Amei A, Klicka J (2012) Evaluating the role of contracting and expanding rainforest in initiating cycles of speciation across the Isthmus of Panama. Proceedings of the Royal Society B-Biological Sciences 279 (1742):3520-3526. Soares-Filho BS, Nepstad DC, Curran LM, Cerqueira GC, Garcia RA, Ramos CA, Voll E, McDonald A, Lefebvre P, Schlesinger P (2006) Modelling conservation in the Amazon basin. Nature 440 (7083):520-523. Swift T, Hannon S (2010) Critical thresholds associated with habitat loss: a review of the concepts, evidence, and applications. Biological Reviews 85 (1):35-53. Tchuente ́ ATK, Roujean J-L, De Jong SM (2011) Comparison and relative quality assessment of the GLC2000, GLOBCOVER, MODIS and ECOCLIMAP land cover data sets at the African continental scale. International Journal of Applied Earth Observations and Geoinformation 13 (2):207-219. Tilman D, May RM, Lehman CL, Nowak MA (1994) Habitat destruction and the extinction debt. Nature 371:65-66. Udo de Haes HA, Heijungs R, Suh S, Huppes G (2004) Three Strategies to Overcome the Limitations of Life‐Cycle Assessment. Journal of Industrial Ecology 8 (3):19-32. van Asselen S, Verburg PH (2012) A Land System representation for global assessments and land-use modeling. Global Change Biology 18 (10):3125-3148. van Dobben H, Schouwenberg E, Nabuurs G, Prins A (1998) Biodiversity and productivity parameters as a basis for evaluating land use changes in LCA. In: Lindeijer E, van Kampen M, Fraanje P et al. (eds) Biodiversity and life support indicators for land use impacts in LCA, vol (Publication series raw materials 1998/07). Dienst Weg- en Waterbouwkunde, Delft. Verones F, Saner D, Pfister S, Baisero D, Rondinini C, Hellweg S (in press) Effects of consumptive water use on biodiversity in wetlands of international importance. Environmental Science & Technology. Vogtländer J, Lindeijer E, Witte J, Hendriks C (2004) Characterizing the change of land-use based on flora: application for EIA and LCA. Journal of Cleaner Production 12:47–57. Weidema B, Lindeijer E (2001) Physical impacts of land use in product life cycle assessment. Final report of the EURENVIRON-LCAGAPS sub-project on land use. Department of Manufacturing Engineering and Management, Technical University of Denmark, Lyngby. Weidema BP (2012) New paradigm or old distance to target? Environmental Science & Technology 46 (2):570-570. Weidema BP, Bauer C, Hischier R, Mutel C, Nemecek T, Reinhard J, Vadenbo CO, Wernet G (2012) Data quality guideline for the ecoinvent database version 3. ecoinvent, St. Gallen. Weir JT, Schluter D (2007) The latitudinal gradient in recent speciation and extinction rates of birds and mammals. Science 315 (5818):1574-1576. Wenninger EJ, Inouye RS (2008) Insect community response to plant diversity and productivity in a sagebrush–steppe ecosystem. Journal of Arid Environments 72 (1):24-33. World Wildlife Fund (2006) WildFinder: Online database of species distributions, ver. Jan-06. http://www.worldwildlife.org/WildFinder. Zobel K (2001) On the species-pool hypothesis and on the quasi-neutral concept of plant community diversity. Folia Geobotanica 36 (1):3-8. Zobel M (1997) The relative role of species pools in determining plant species richness: an alternative explanation of species coexistence? Trends in Ecology & Evolution 12 (7):266- 269.
128
Chapter 6
Quantifying biodiversity loss due to export-crops in East Africa using life cycle assessment
Laura de Baan
Michael Curran
Carlo Rondinini
Piero Visconti
Stefanie Hellweg
Thomas Koellner
Submitted to Conservation Letters
129 Quantifying biodiversity loss due to export-crops in East Africa using life cycle assessment
Abstract
Global agricultural production and trade contribute massively to biodiversity loss. Life cycle assessment (LCA) studies help inform decision makers about direct and displaced environmental impacts of products and are thus important to the agricultural intensification debate within conservation. However, largely due to poor communication between disciplines and underdeveloped LCA methods, the LCA perspective is rarely adopted. To address this issue, we present a novel LCA method based on the increased extinction risk of mammal species to model product-related impacts of land use on biodiversity. We contrast it with two recently developed LCA methods (focusing on local and regional impacts, respectively) and illustrate the application of these methods to major export crops from East Africa (tea, coffee, and tobacco). The methods highlight hotspots of product-related biodiversity impacts. We encourage conservationists to further apply and develop such models to improve our understanding of the link between production, consumption, and biodiversity loss.
6.1 Introduction Agricultural land use is a main driver of current and expected future global biodiversity loss (Sala et al. 2000; Visconti et al. 2011). In many developing countries, large areas rich in biodiversity have been converted to cash-crop production for developed countries. Lenzen et al. (2012) estimated that international trade is responsible for about 30% of global threats to species. Consumers and policy makers are often unaware of these implications, and informing them about the displaced biodiversity impacts of products is a first step toward more sustainable consumption patterns and reducing rates of global biodiversity loss.
Life cycle assessment (LCA), which quantifies the impacts of products over their entire life cycle (ISO 2006), is a well-established methodology to capture displaced environmental impacts. It helps to identify environmental hotspots in internationally distributed value chains of products and can highlight potential trade-offs between different environmental compartments (e.g., reducing a product’s contribution to climate change while increasing the pressures on biodiversity). In this sense, the life cycle perspective is extremely important to the agricultural intensification debate within conservation (i.e., “land sparing” vs. “land sharing”; Tscharntke et al. 2012) because it helps to illuminate the “hidden costs” of intensive agriculture (Vandermeer and Perfecto 2005) driven by indirect inputs (e.g., the land used to mine phosphorus and grow fodder or the ecotoxic impacts of pesticides). LCA thus helps to illuminate the complex chain of causalities involved in what may appear to be local threats to conservation. This is extremely important in the context of global commitments towards biodiversity conservation, i.e. the Convention on Biological Diversity Aichi targets (2013). Target 4 aims at the implementation of plans for sustainable production and consumption by governments, business, and stakeholders and target 2 at the integration of biodiversity values into national accounting. Here, LCA can play an important role as it can directly quantify the extent to which changes in consumption and production pattern affect land use change and biodiversity loss. Yet conservation scientists rarely adopt a life cycle perspective, which we attribute both to a lack of interdisciplinary exchange and of adequate methodologies to assess biodiversity loss in LCA (Curran et al. 2011).
130 Quantifying biodiversity loss due to export-crops in East Africa using life cycle assessment
Methods implemented in standard LCA software only assess the impacts of agricultural land use on biodiversity in Europe (Koellner and Scholz 2008; De Schryver et al. 2010). Two globally applicable and operational LCA methods for assessing the impacts of land use on biodiversity were developed recently (de Baan et al. 2013a; de Baan et al. 2013b). The first method assesses the relative local reduction of species richness (R-Local) on different types of human-modified land (e.g., annual and permanent crops) compared to undisturbed areas (de Baan et al. 2013a). It is based on data derived from a global literature review (GLOBIO3, Alkemade et al. 2009) and national biodiversity monitoring data (BDM 2004). For each biome, data points are grouped and a median biodiversity impact of land use is calculated. The second method calculates absolute reductions in regional species richness (A-Regional) due to land use based on species-area relationships (de Baan et al. 2013b), accounting for the effects of habitat quality of the land use matrix on species loss (Koh and Ghazoul 2010). The method calculates the potential regional loss of species within WWF ecoregions due to the past conversion of habitat and allocates this loss to the different types of land use occurring in each ecoregion. Land use in regions with little remaining natural habitat or high species diversity receives higher impacts than that in intact or species-poor ecosystems. Both methods assess impacts at a relatively coarse spatial resolution (biomes and ecoregions, respectively), not necessarily matching the local heterogeneity of biodiversity and the scale of conservation concerns.
Here, we present a new approach to assess the product-related impacts of land use on biodiversity on a grid-cell level. We used species-specific habitat suitability models (HSMs) developed by the Global Mammals Assessment (GMA; Rondinini et al. 2011) to calculate mammal richness per 900 m grid cell. To assess the impacts of human land use, we modeled the local species richness of a reference scenario and compared it with the richness of human-used land. Species losses were weighted by the level of threat to the species and their global rarity. This new method was applied to a case study of East Africa (Figure 6.1), an extremely diverse region hosting multiple Global 200 priority ecoregions at a vulnerable, critical, or endangered conservation status (Olson and Dinerstein 2002) and of high richness of mammal species threatened by agriculture (Visconti et al. 2011). Biodiversity impacts of three major export crops (tea, coffee, and tobacco) produced across East Africa (FAOSTAT 2013c) were quantified with this method. In addition, biodiversity impacts of these products were assessed with the two previously developed methods described above (de Baan et al. 2013a; de Baan et al. 2013b) to compare the outcomes of the modeling approaches.
131 Quantifying biodiversity loss due to export-crops in East Africa using life cycle assessment
Figure 6.1. Map of the case study region East Africa, showing country and ecoregion borders.
132 Quantifying biodiversity loss due to export-crops in East Africa using life cycle assessment
6.2 Methods
6.2.1 Basic structure of LCA According to the norms of the International Organization for Standardization (ISO 2006), LCA consists of four basic steps. First, the goal and scope of an analysis are defined, including the system boundaries of the analyzed product system. To illustrate our land use methods, we considered the biodiversity impacts of land used during crop cultivation (i.e., farm-gate impacts) of three products (tea, coffee, and tobacco). In addition, we considered the impacts of direct land use change driven by the expansion of these crops. The method could, in theory, also be used to quantify indirect impacts (such as deforestation due to wood cutting for tobacco curing or tea drying) or be combined with assessments of other biodiversity impacts (e.g., freshwater eutrophication from waste water of coffee processing or downstream impacts of crop irrigation (Verones et al. in press)). Impacts were assessed per kg of harvested product. Second, all inputs and emissions of the production systems were compiled in the inventory analysis. Here, we only considered the direct land use and land use change related to crop cultivation (see Section 6.2.2). Third, in the impact assessment, we assessed the biodiversity impacts of land use per kg of harvested product using three alternative assessment methods. Following an LCA convention, we separately assessed the impacts of land use (occupation) and land use change (transformation) (Milà i Canals et al. 2007; Koellner et al. 2013). Occupation impacts quantify how much biodiversity is lost as a result of avoiding natural recovery of biodiversity during the actual land use phase. Transformation impacts account for the reduction in biodiversity after (hypothetical) future land abandonment and the time lag until biodiversity is fully recovered (see Appendix Figure E1). Fourth, the results were interpreted in relation to the goal and scope of the study.
6.2.2 Life cycle inventory analysis Data for land occupation in m2.years/kg harvested crop were calculated as the inverse of the yield (based on the data of Monfreda et al. 2008) per grid cell (globally available at a 5 min ~10 km resolution). As climatic conditions above 2500 m asl are not very suitable for commercial production of the three crops, we excluded these high-altitude areas. Land use change (transformation) was calculated per country based on FAO statistics, (FAOSTAT 2013b, a) adapting the approach suggested by Milà i Canals et al. (2013) (see Appendix Section E1.4 for further details).
6.2.3 Impact assessment methods Three potentially globally applicable life cycle impact assessment methods were applied to the case studies (for an overview see Table 6.1). A description of the local relative method R-Local (de Baan et al. 2013a) and regional absolute method A-Regional (de Baan et al. 2013b) is given in the Appendix. The newly developed weighted local loss (W-Local) method is described below.
133 Quantifying biodiversity loss due to export-crops in East Africa using life cycle assessment
Table 6.1. Comparison of the four methods applied in the case studies
Name and reference of Relative local loss Absolute regional loss (A- Weighted local loss method Regional): de Baan et al. (R-Local): de Baan et al. (W-Local): developed in (2013b) (2013a) this paper
Biodiversity impact model Meta-analysis of local Matrix-calibrated Species-specific habitat biodiversity surveys species-area suitability models (Alkemade et al. 2009) relationships (Koh and (Rondinini et al. 2011), Ghazoul 2010) Scale of impact Local Regional Local, weighted with global threat and rarity Geographic coverage Global Global East Africa
Resolution Biome Ecoregion Pixel (0.81 km2) Indicators Relative local loss of Absolute regional loss of Weighted absolute local species richness species richness loss of species richness Taxa Several taxa Mammals (also available Mammals for 4 additional taxa) Land use classes used in Annual crops, permanent Agricultural land Cropland (>70%), mosaic this study (additional crops (agroforestry, used (pasture, used forests, cropland (20–70%); (could classes covered by the forests, secondary artificial areas) be extended to any other study) forests, artificial areas, GlobCover class) pasture) Reference states (Semi)-natural land Natural land Natural land (maximum habitat, W-Local-Max) and current land cover (W- Local-Cur) Implemented Maintain high local Avoid regional extinction Avoid global extinction conservation target richness
Weighted local loss (W-Local) This method was based on HSMs that have been developed by the GMA for nearly all terrestrial mammal species (Rondinini et al. 2011). Within the species’ known geographical ranges, they classified each grid cell as highly suitable (primary) habitat, moderately suitable (secondary) habitat, or unsuitable habitat for that species. In this study, moderately suitable habitat was considered as unsuitable, as the species can be found there but cannot survive there permanently (Rondinini et al. 2011). The habitat suitability relationship for each species was derived from the 2009 IUCN Red List of Threatened Species based on land cover (based on GlobCover v2.3; European Space Agency 2009), elevation, and hydrological features (Rondinini et al. 2011). The model was implemented in GRASS GIS (GRASS Development Team 2012).
Calculation of weighted species loss For the case study region of East Africa, we resampled the original HSMs of all 631 mammal species present from a resolution of 300 m to 900 m to decrease the volume of data. We then estimated potential species richness S per grid cell by summing the presence of each species as predicted by the HSMs. To assess the impacts of human modification of land, we developed two reference scenarios for species richness of all grid cells and compared these with the richness of human-used land. First, the maximum species range reference scenario (W-Local-Max) assumed
134 Quantifying biodiversity loss due to export-crops in East Africa using life cycle assessment that all areas within the species range were initially covered by a suitable land cover type. Only areas outside the species elevation range or water requirements were considered as unsuitable (e.g., for riparian species, areas that were not in the vicinity of water bodies were excluded from the range). Second, the current land cover (given by the GlobCover v2.3; European Space Agency 2009) was used as a reference (W-Local-Cur). This scenario considers the impacts of changing the land from its current state to future agricultural use. For both reference scenarios, we reran the HSM to derive the potential species richness per grid cell.
We compared these references to two land use scenarios, which represented the potential species richness assuming the cell was occupied by (a) cultivated land with >70% cropland (GlobCover class 10, applied to the annual crop tobacco) or (b) 20–70% mosaic cropland (class 20, applied to the permanent crops tea and coffee).
In the next step, we weighted the species richness S per grid cell by each species’ global rarity (R) and threat level (TL), adapting an approach of Verones et al. (in press). Rarity was calculated as the proportion of total global suitable habitat area of the species in the cell (i.e., ranging from 0 to 1). The TL is a rescaling of the categories defined by the IUCN Red List of threatened species (Mace et al. 2008). To give equal weight to the two factors, TL was also scaled from 0 to 1 (0.2- least concern, 0.4-near threatened, 0.6-vulnerable, 0-8-endangered, 1-critically endangered), which deviates from Verones et al. (in press). For each combination of one of the two reference
(ref) and one of the two land use scenarios (LUi), we calculated the weighted species loss based on the TL and R of each species k occurring in a cell.
n n S = TL R − TL R (6.1) W ,loss,j ∑ k,ref k,ref ∑ k,LUi k,LUi k=1 k=1
Biodiversity impact of land use and land use change Biodiversity impacts (referred to as characterization factors in LCA terminology) per m2 of land occupation for each land use type i and reference scenario ref were calculated by dividing the weighted species loss per grid cell by the grid cell area (810,000 m2).
S w ,loss,LUi ,j BI (6.2) W ,Occ,LUi ,j = 2 810'000m
The biodiversity impacts of land use change (transformation) were calculated by multiplying the biodiversity impacts of occupation by half the time required for the recovery of biodiversity, treg (Milà i Canals et al. 2007), see also Eq. E6:
BI = BI ⋅0.5⋅t . (6.3) W ,Trans,LUi ,j W ,Occ,LUi ,j reg
For treg, we used estimates for the recovery times of mammals following habitat disturbance from Curran et al. (in press). For each grid cell, we modeled recovery times based on the amount of natural habitat in the region, distance to the nearest natural area, latitude, and elevation (see Appendix Section E1.5). The resulting recovery times, which were applied to all mammal species,
135 Quantifying biodiversity loss due to export-crops in East Africa using life cycle assessment are displayed in Appendix Figure E2. As a comparison, biodiversity impacts were also calculated based on unweighted species loss (see Appendix Section E2.3).
6.3 Results The expansion of tea, coffee, and tobacco cultivation observed in the past 20 years in East Africa are dramatic, with up to a 60% increase in harvested area per country (Appendix Table E1). No consistent patterns were observed in the spatial distribution of yields or cropland expansion across the three crops (Figure 6.2). Extensive land use changes were associated with both tea and coffee in Ethiopia and with tobacco in Tanzania. In Rwanda, the forested area increased in the past 20 years, and permanent cropland decreased. Therefore, no land use change was assigned to any of the three crops. The lowest yields (and therefore the highest land use per kg of crop) were found in Burundi for tea, in Uganda and Ethiopia for coffee, and in South-West Tanzania for tobacco. The land use per kg for tea, coffee, and tobacco was in similar ranges.
When assessed with the rarity and threat level weighted loss (W-Local-Max, Figure 6.3), tea, coffee, or tobacco that was produced within the extent of occurrence of very range-restricted mammals had by far the highest impacts on biodiversity. The highest land use impacts occurred northeast of Lake Victoria and in the Albertine Rift Valley (the presumed extent of occurrence of Hopkins’s groove-toothed swamp rat, Pelomys hopkinsi), in the Ethiopian highlands (Ethiopian striped mouse, Muriculus imberbis), in southwestern Ethiopia (Yalden’s Desmomys, Desmomys yaldeni), and southwest of Mount Kenya (ultimate shrew, Crocicura ultima). The biodiversity loss assessed with W-Local-Cur shows the regions where future expansion of crops would cause the highest impacts. These were regions with high mammal endemism, such as the Albertine Rift Valley in the Ethiopian highlands, the Eastern Arc Mountains, and the Northern Zanzibar- Inhambane coastal forest mosaic.
The relative local species loss (R-Local) method (Figure 6.4, left) was not correlated with the weighted loss (W-Local, see Table E5), but it was correlated with the amount of land used per kg of crop. Regions with high land use and land use change showed the highest impacts. The results of the A-Regional (Figure 6.4, right) partly agreed with those of W-Local-Cur, such as high impacts in the Eastern Arc Forests. Crops produced in other ecoregions in Tanzania showed some of the highest impacts, such as the Ittigi-Sumbu thicket, the Serengeti volcanic grassland, and the East African halophytics. The montane moorlands in Kenya, Uganda, and Rwanda also showed high impacts with the A-Regional method, as well as coffee produced in the ecoregion of the East African montane forests.
For all but the W-Local method, the transformation impacts of coffee were highest in the southern Ethiopian montane forests. For tea, the Ethiopian highlands also showed high transformation impacts, reflecting the predicted long time lag involved in the recovery in such high-altitude ecosystems (Figure E2).
136 Quantifying biodiversity loss due to export-crops in East Africa using life cycle assessment
Occupa)on* Transforma)on* [m2!years/kg]* [m2/kg]* Tea$ Coffee$ Tobacco$
Figure 6.2. Life cycle inventory: Land use (occupation, m2.years) and land use change (transformation, m2) potentially caused by 1 kg of harvested crops.
137 Quantifying biodiversity loss due to export-crops in East Africa using life cycle assessment
W&Local&Max$ W&Local&Cur$ Occupa&on) Transforma&on) Occupa&on) Transforma&on) Tea$ Coffee$ Tobacco$
Figure 6.3. Final LCA result based on the weighted local species loss (W-Local): Biodiversity loss caused by the land use (occupation) and land use change (transformation) per kg of harvested crop. Assessed with the two reference scenarios maximum (Max) and current (Cur). The numbers represent deviations from the mean values of each map (0=mean, -1=one standard deviation smaller than the mean, +1= one standard deviation larger than the mean). Because a few cells had very high values (up to 825 standard deviations above the mean), these values were capped at +1 standard deviation for map display purposes.
138 Quantifying biodiversity loss due to export-crops in East Africa using life cycle assessment
R&Local$ A&Regional$ Occupa&on) Transforma&on) Occupa&on) Transforma&on) Tea$ Coffee$ Tobacco$
Figure 6.4. Final LCA result based on relative local (R-Local) and absolute regional (A-Regional) species losses. Biodiversity loss caused by land use (occupation) and land use change (transformation) per kg of harvested crop. The numbers represent deviations from the mean values of each map (0=mean, -1=one standard deviation smaller than the mean, +1= one standard deviation larger than the mean). Because a few cells had very high values (up to 29 standard deviations above the mean), these values were capped at +5 standard deviation for map display purposes.
139 Quantifying biodiversity loss due to export-crops in East Africa using life cycle assessment
6.4 Discussion In this study, we illustrated a novel LCA method to model the production-related biodiversity impacts of tea, coffee, and tobacco across East Africa. The method operates at a high spatial resolution compared to existing LCA land use methods, integrates rarity and threat information, and can be applied to both historical and future impacts of agriculture. We also compared the results to other globally applicable LCA methods and illustrated how different conservation targets can be implemented in LCA studies.
Our results suggest that the potential biodiversity impacts of tea, coffee, and tobacco are most severe in areas of high endemism and low habitat availability for range-restricted species (W- Local-Cur maps; Figure 6.3, right). This agrees with existing conservation assessments of threat, vulnerability, and biological value for the region as a whole (e.g., Burgess et al. 2006). A first step to prevent imminent future extinctions of range-restricted species would be to halt conversion of natural habitat to cropland in high-impact regions (highlighted by W-Local-Cur) and to restore habitat in critical areas (highlighted by W-Local-Max). The A-Regional method highlighted additional areas under high land use pressures and high risk of regional species extinction, where conservation action is most needed.
In the past, East Africa’s protected areas were mainly established to protect megafauna in savanna regions, and they are not optimally located to protect small mammals and montane ecosystems (Fjeldså et al. 2004). Protecting these areas is a challenge because they largely overlap with areas of high human population density (Balmford 2001; Rondinini et al. 2006) and deep poverty (Fisher and Christopher 2007). The evaluation of socio-economic consequences of land use and conservation planning for poor small-holder farmers (see e.g., Chiozza et al. 2010) who produce more than half of all coffee and tea in East Africa is thus crucial for successful biodiversity conservation. One potential mechanism to improve the protection of high-impact areas would be to shift the costs of conservation to the consumers of these export-crops in developed countries. A very first step in this direction was observed in the past decade where demand for organic and fair-trade certified coffee and tea strongly increased globally (currently at 16% and 10% of global production, respectively; van Reenen et al. 2010; Panhuysen and van Reenen 2012). To achieve effective conservation, a more elaborate framework for biodiversity compensation would have to be developed. The high spatial resolution of the W-Local method could serve as a good information basis for the development of such a framework.
To reduce occupation impacts of existing plantations, either the per area impact (e.g., by wildlife- friendly farming practices) or the amount of land used for production (by increasing yields) can be reduced. Which of the two options has the higher potential for biodiversity conservation could not be assessed with the presented LCA methods because they do not provide the necessary detail to distinguish between production systems (e.g., organic and conventional production). If better data on the habitat quality of different production systems, as well as more detailed land cover maps on organic and conventional crop production systems become available, all three methods can, in principle, be expanded (illustrated in the case of milk with the R-Local method; Mueller et al. in press). A more detailed analysis of the relationship between production intensity and biodiversity
140 Quantifying biodiversity loss due to export-crops in East Africa using life cycle assessment loss is especially important for coffee where a broad range of production systems exist with very different degrees of impacts on biodiversity (Donald 2004; Hundera et al. 2012).
The results of the R-Local and A-Regional were strongly dependent on the land use and the land use change caused per kg of harvested crop. These data were derived by merging FAO crop production statistics with satellite-derived land cover data (Monfreda et al. 2008), both of which are subject to considerable uncertainties. The spatial and the temporal variability of yields are only partly reflected in the data. In addition, the available data did not allow us to assess the true spatial patterns of land use change. Instead, we allocated national land conversion rates to all crops produced anywhere in the country. Despite these limitations, we believe that the approach is a promising first step that could be expanded beyond the presented case studies to other common crops and world region.
6.5 Conclusions Traditionally, assessing biodiversity impacts at the product level using LCA relies on linear, static, and nonspatially explicit modeling. In contrast, biodiversity often responds to pressures in nonlinear, dynamic, and site-specific ways. The recently developed methods applied in this paper overcome some of the limitations of traditional LCA models by linking product-specific impacts to different conservation targets. The W-Local method can be directly linked to the Aichi Target 12 (avoid extinction of known threatened species; Convention on Biological Diversity 2013) and the A- Regional method indicates the risk of regional (but not yet necessarily global) extinction of species, providing earlier warning signals of extinction. In addition, it can inform on species loss of other taxonomic groups than mammals (i.e., birds, plants, amphibians, and reptiles; de Baan et al. 2013b). The R-Local method provides the weakest link to concrete conservation targets. We therefore recommend using the W-Local and A-Regional method in conjunction. For many LCA studies on internationally traded commodities, the spatial detail on the origin of products required for the W-Local method is not available, because of the limited traceability of products along value chains. As an intermediate solution, the A-Regional method could be used as stand- alone assessment, using probabilistic approaches to determine the origin of a particular crop.
All three methods can be applied to other regions and crops because they are based on globally available data. We hope this encourages conservation scientists to integrate life cycle thinking in their work and paves the way for interdisciplinary collaboration to better understand the drivers of biodiversity loss and to find solutions for the global biodiversity crisis.
6.6 Acknowledgements We thank ETH Zurich (Research Grant CH1-0308-3) and the European Commission (LC-IMPACT, Grant Agreement No. 243827, 7th Framework Programme) for financial support.
6.7 References Alkemade R, van Oorschot M, Miles L, Nellemann C, Bakkenes M, ten Brink B (2009) GLOBIO3: A framework to investigate options for reducing global terrestrial biodiversity loss. Ecosystems 12 (3):374-390. Balmford A (2001) Conservation Conflicts Across Africa. Science 291 (5513):2616-2619.
141 Quantifying biodiversity loss due to export-crops in East Africa using life cycle assessment
BDM (2004) Biodiversity Monitoring Switzerland. Indicator Z9: Species Diversity in Habitats. Bundesamt für Umwelt, BAFU. http://www.biodiversitymonitoring.ch. Accessed 1.2.2011 Burgess N, Hales JD, Ricketts TH, Dinerstein E (2006) Factoring Species, Non-Species Values and Threats Into Biodiversity Prioritisation Across the Ecoregions of Africa and Its Islands. Biological Conservation 127 (4):383–401. Chiozza F, Boitani L, Rondinini C (2010) The Opportunity Cost of Conserving Amphibians and Mammals in Uganda. Natureza & Conservação 8 (2):177-183. Convention on Biological Diversity (2013) Aichi Biodiversity Targets. http://www.cbd.int/sp/targets/. Accessed 4.10.2013 Curran M, de Baan L, De Schryver A, van Zelm R, Hellweg S, Koellner T, Sonnemann G, Huijbregts MAJ (2011) Toward meaningful end points of biodiversity in life cycle assessment. Environmental Science & Technology 45 (1):70-79. Curran M, Hellweg S, Beck J (in press) Is there any empirical support for biodiversity offset policy? Ecological Applications. de Baan L, Alkemade R, Koellner T (2013a) Land use impacts on biodiversity in LCA: a global approach. The International Journal of Life Cycle Assessment 18 (6):1216-1230. de Baan L, Mutel CL, Curran M, Hellweg S, Koellner T (2013b) Land use in Life Cycle Assessment: Global characterization factors based on regional and global potential species extinctions. Environmental Science & Technology 47 (16):9281–9290. De Schryver AM, Goedkoop MJ, Leuven RSEW, Huijbregts MAJ (2010) Uncertainties in the application of the species area relationship for characterisation factors of land occupation in life cycle assessment. The International Journal of Life Cycle Assessment 15 (7):682-691. Donald P (2004) Biodiversity impacts of some agricultural commodity production systems. Conservation Biology 18 (1):17-37. European Space Agency (2009) GlobCover land cover map, v2.3. European Space Agency. http://due.esrin.esa.int/globcover/. Accessed 26.2.2013 FAOSTAT (2013a) Land-use resources. http://faostat.fao.org/site/567/DesktopDefault.aspx?PageID=567 - ancor. Accessed 26.2.2013 FAOSTAT (2013b) Production of crops. http://faostat.fao.org/site/567/DesktopDefault.aspx?PageID=567 - ancor. Accessed 26.2.2013 FAOSTAT (2013c) Trade of crops and livestock products. http://faostat.fao.org/site/535/DesktopDefault.aspx?PageID=535 - ancor. Accessed 26.2.2013 Fisher B, Christopher T (2007) Poverty and biodiversity: Measuring the overlap of human poverty and the biodiversity hotspots. Ecological Economics 62 (1):93-101. Fjeldså J, Burgess ND, Blyth S, de Klerk HM (2004) Where are the major gaps in the reserve network for Africa's mammals? Oryx 38 (01):17-25. GRASS Development Team (2012) Geographic resources analysis support system (GRASS) Software, Version 6.4.0. Open Source Geospatial Foundation. See http://grass.osgeo.org. Hundera K, Aerts R, Fontaine A, Mechelen M, Gijbels P, Honnay O, Muys B (2012) Effects of Coffee Management Intensity on Composition, Structure, and Regeneration Status of Ethiopian Moist Evergreen Afromontane Forests. Environmental Management 51 (3):801- 809. ISO (2006) ISO 14040 International Standard. In: Environmental Management – Life Cycle Assessment – Principles and Framework. International Organisation for Standardization, Geneva, Switzerland. Koellner T, de Baan L, Beck T, Brandão M, Civit B, Margni M, Milà i Canals L, Saad R, de Souza DM, Müller-Wenk R (2013) UNEP-SETAC guideline on global land use impact assessment on biodiversity and ecosystem services in LCA. The International Journal of Life Cycle Assessment 18 (6):1188-1202.
142 Quantifying biodiversity loss due to export-crops in East Africa using life cycle assessment
Koellner T, Scholz RW (2008) Assessment of land use impacts on the natural environment. Part 2: Generic characterization factors for local species diversity in Central Europe. The International Journal of Life Cycle Assessment 13 (1):32-48. Koh L, Ghazoul J (2010) A matrix-calibrated species-area model for predicting biodiversity losses due to land-use change. Conservation Biology 24 (4):994-1001. Lenzen M, Moran D, Kanemoto K, Foran B, Lobefaro L, Geschke A (2012) International trade drives biodiversity threats in developing nations. Nature 486 (7401):109-112. Mace GM, Collar NJ, Gaston KJ, Hilton-Taylor C, Akçakaya HR, Leader-Williams N, Milner- Gulland EJ, Stuart SN (2008) Quantification of Extinction Risk: IUCN's System for Classifying Threatened Species. Conservation Biology 22 (6):1424-1442. Milà i Canals L, Bauer C, Depestele J, Dubreuil A, Freiermuth Knuchel R, Gaillard G, Michelsen O, Müller-Wenk R, Rydgren B (2007) Key Elements in a Framework for Land Use Impact Assessment Within LCA. The International Journal of Life Cycle Assessment 12 (1):5-15. Milà i Canals L, Rigarlsford G, Sim S (2013) Land use impact assessment of margarine The International Journal of Life Cycle Assessment 18 (6):1265-1277. Monfreda C, Ramankutty N, Foley JA (2008) Farming the planet: 2. Geographic distribution of crop areas, yields, physiological types, and net primary production in the year 2000. Global Biogeochemical Cycles 22 (1):GB1022. Mueller C, de Baan L, Koellner T (in press) Comparing direct land use impacts on biodiversity of conventional and organic milk—based on a Swedish case study. The International Journal of Life Cycle Assessment. Olson D, Dinerstein E (2002) The Global 200: Priority ecoregions for global conservation. Annals of the Missouri Botanical Garden 89 (2):199-224. Panhuysen S, van Reenen M (2012) Coffee Barometer 2012. Tropical Commodity Coalition, The Hague. Rondinini C, Chiozza F, Boitani L (2006) High human density in the irreplaceable sites for African vertebrates conservation. Biological Conservation 133 (3):358-363. Rondinini C, Di Marco M, Chiozza F, Santulli G, Baisero D, Visconti P, Hoffmann M, Schipper J, Stuart SN, Tognelli MF (2011) Global habitat suitability models of terrestrial mammals. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 366 (1578):2633-2641. Sala O, Chapin F, Armesto J, Berlow E, Bloomfield J, Dirzo R, Huber-Sanwald E, Huenneke L, Jackson R, Kinzig A, Leemans R, Lodge D, Mooney H, Oesterheld M, Poff N, Sykes M, Walker B, Walker M, Wall D (2000) Global biodiversity scenarios for the year 2100. Science 287 (5459):1770-1774. Tscharntke T, Clough Y, Wanger TC, Jackson L, Motzke I, Perfecto I, Vandermeer J, Whitbread A (2012) Global food security, biodiversity conservation and the future of agricultural intensification. Biological Conservation 151 (1):53-59. van Reenen M, Panhuysen S, Weiligmann B (2010) Tea Barometer 2010. Tropical Commodity Coalition, The Hague. Vandermeer J, Perfecto I (2005) The future of farming and conservation. Science 308 (5726):1257-1258- author reply 1257-1258. Verones F, Saner D, Pfister S, Baisero D, Rondinini C, Hellweg S (in press) Effects of consumptive water use on biodiversity in wetlands of international importance. Environmental Science & Technology. Visconti P, Pressey RL, Giorgini D, Maiorano L, Bakkenes M, Boitani L, Alkemade R, Falcucci A, Chiozza F, Rondinini C (2011) Future hotspots of terrestrial mammal loss. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 366 (1578):2693-2702.
143 Quantifying biodiversity loss due to export-crops in East Africa using life cycle assessment
144
Chapter 7
Concluding remarks
Concluding remarks
7.1 Scientific relevance and conclusions The aim of this thesis was to develop meaningful, operational, and globally applicable life cycle impact assessment methods for land use impacts on biodiversity. In Chapter 2, we started with a critical evaluation of how biodiversity impacts are modeled within LCA and found considerable conceptual shortcomings. Based on this knowledge, we then developed a set of new methods and applied them to case studies (Chapters 3-6). In this section, we summarize the main shortcomings of biodiversity modeling in LCA identified in Chapter 2, discuss how different methods contribute to overcoming some of these shortcomings and provide guidance on different model choices.
7.1.1 Conceptual shortcomings of biodiversity modeling in LCA In Chapter 2, we found serious conceptual shortcomings in how biodiversity loss is currently modeled within LCA. These shortcoming encompass:
1) Lacking considerations of the scale of impacts
The most common unit to represent biodiversity loss in LCA is the potentially disappeared fraction of species (PDF). However, the scale of species loss is not defined, and biodiversity models of different impact categories (e.g., land use and climate change) use the unit PDF to represent local, regional, and global species loss. Therefore, the current practice of simply summing up impacts of different drivers of biodiversity loss into a single endpoint of ecosystem damage has little meaning because the PDF values of different drivers represent different scales of species extinction.
2) Limited taxonomic coverage
Most methods consider only one or a few taxonomic groups. The approaches for assessing the impacts of land use are mostly based on vascular plant species. However, correlation across taxonomic groups is only weak and it is questionable whether one taxonomic group can serve as a predictor for all other groups (Wolters et al. 2006).
3) Limited geographic coverage
The geographic coverage of most models is limited to one or a few geographic regions. The validity of the current practice to extrapolate these results to other regions is questionable (as performed when applying, for example, EcoIndicator 99 (Goedkoop and Spriensma 1999) or ReCiPe (Goedkoop et al. 2008) to a non-European context).
4) Poor implementation of non-linearity and irreversibility of impacts
Non-linear or irreversible effects (including critical thresholds and tipping points) of biodiversity loss are not considered in most methods. For land use impacts on biodiversity, non-linear effects have been considered by some authors that have applied species-area relationships to assess regional impacts on biodiversity (e.g., Koellner and Scholz 2008; De Schryver et al. 2010) or regional vulnerability of ecosystems (e.g., Weidema and Lindeijer 2001). Although the conceptual framework for land use assessment would allow the assessment of irreversible, permanent impacts (Figure 1.1), this has not yet been operationalized.
146 Concluding remarks
5) Limited focus of indicators
Although biodiversity covers different hierarchical levels (genes, single species, species communities, and ecosystems) and can be described by different attributes (composition, structure, or function; Noss 1990), most LCIA methods are based on compositional indicators of species communities, mostly assessing changes in species richness. Measures of genetic diversity are absent in LCIA methods, but some studies also include the species level (mainly for ecotoxicology, acidification, and eutrophication) or ecosystem indicators.
6) Missing drivers of biodiversity loss
On a more general level, three of the five drivers of biodiversity loss identified by the Millennium Ecosystem Assessment are implemented in LCIA methods. Impacts of climate change, habitat loss (e.g., land and water use), and pollution (e.g., ecotoxicity, acidification, and eutrophication) are considered in LCIA methods. Methods that cover overexploitation of species and impacts of invasive species are missing. However, new methods for assessing the impacts of overexploitation of marine species are under development (Emanuelsson et al. 2012).
7.1.2 How should we assess land use impacts on biodiversity within LCA on a global scale? This thesis focused on improving methods for assessing land use impacts on biodiversity within LCA. The main aim was to overcome the limited geographic coverage of existing LCIA methods by developing a globally applicable method. In the following, we discuss how different limitations of land use LCIA methods can be overcome and compare the newly developed methods of this thesis with other methods that have the potential to be applied globally. Although many approaches for modeling land use impacts on biodiversity in LCA exist, most studies rely on specific data only available in one region. For this concluding comparison, we considered only those studies that are based on potentially globally available data. We identified three initially developed methods (Lindeijer 2000; Weidema and Lindeijer 2001; Koellner and Scholz 2008) and two recently developed ones (Coelho and Michelsen 2013; de Souza et al. 2013) that matched this criterion. All methods and their features are summarized in Table 7.1. We discuss each model feature and provide guidance regarding which model choice should be preferred (in grey text boxes).
Type of model The currently available methods that can potentially be applied on global scale can be categorized into three basic modeling approaches (Figure 7.1). The first approach calculates a relative local reduction in biodiversity and assesses the impact as a ratio between the lost diversity and the diversity of a reference situation. This ratio can express PDF, but it can also express other measures of local biodiversity loss, such as reduction in functional diversity (de Souza et al. 2013) or reduced “naturalness” of a system (i.e., hemeroby; Brentrup et al. 2002). The method presented in Chapter 3 and the functional species approach of de Souza et al. (2013) can be assigned to this first category. The second approach was first proposed by Weidema and Lindeijer (2001) and in a more simplified form by Lindeijer (2000). It is based on the ratios of relative local biodiversity loss (as in the first approach), but multiplies (or weighs) these ratios by an absolute
147 Concluding remarks biodiversity value of the land. This biodiversity value is quantified on a regional scale, using, for example, data on regional species richness, current land use pressure on the ecosystem (ecosystem vulnerability), and ecosystem scarcity (i.e., the global rarity of the ecosystem). The method presented in Chapter 4 and the approach of Coelho and Michelsen (2013) further develop this approach based on globally available data. The factors suggested for measuring the absolute biodiversity value of the land differ between the authors, as do the ways in which they quantify the relative biodiversity loss. Finally, the third modeling approach directly applies species-extinction models and assesses the absolute potential species extinction at different spatial scales. The methods developed in Chapters 5 and 6 can be grouped into this category.
Types of land use assessment methods
(1) Relative local impacts (2) Weighted relative
R-Local local impacts, RW-Local All ecosystems have the same Some ecosystems have a higher value value than others Chapter 3, Koellner & Scholz, Weidema & Lindeijer, 2001 2008, de Souza et al., 2013 Lindeijer, 2000, Chapter 4
(3) Absolute species loss model, A-Regional , W-Local Species extinction should be avoided Chapter 5, Chapter 6
Figure 7.1 Overview of potentially globally applicable model types, methods, and related value choices
The fist modeling approach of local relative biodiversity loss gives equal weight to damages to all kinds of ecosystems, reflecting the concerns over the loss of local ecosystem functions. A 50% loss of species is considered to have the same impact in any ecosystem, no matter how diverse, threatened, or scarce the ecosystem is. The second approach assumes that not all ecosystems have the same value. Ecosystems that are globally scarce, that are rich in species or in endemic species, or that already experience very high land use pressure and are therefore more vulnerable to additional land use impacts have a higher conservation concern. Impacts in these regions are considered to be more detrimental than those in other regions. The third approach assesses species extinction and allocates higher damages to land associated with higher potential species extinction. Instead of using weighting factors for the regional value of biodiversity, this approach tries to directly quantify species extinction risk. In Chapter 5, species extinction was modeled on a regional scale and distinguished between irreversible regional extinction of endemic species and potentially reversible regional extinction of non-endemic species. In Chapter 6, the species extinction risk for each single mammal species was quantified by applying weighting factors for the global rarity of species and their threat statuses (as evaluated by the IUCN red list), thereby providing a proxy for global extinction risk.
148 Concluding remarks
The different approaches thus reflect different ways of valuing biodiversity, and the question of which approach should be favored is partly subjective. When LCIA methods are applied to compare biodiversity impacts across world regions, we argue that just considering local impacts on ecosystem function has too limited a focus and that regional differences in biodiversity value of the land should be considered too. Therefore, we suggest that the ratios of local biodiversity loss (e.g., as presented in Chapter 3) always should be applied in conjunction with a regional weighing scheme for biodiversity value (e.g., as presented in Chapter 4). The advantage of using such a combined approach over directly modeling biodiversity loss is its simplicity. On the other hand, directly modeling species extinction risk provides the opportunity to directly link LCA studies with international conservation targets and thereby provide meaningful decision support.
Model Type 1: Measure for relative local biodiversity reduction factor For measuring local relative reductions in biodiversity, several indicators have been proposed, but the most prominent one is species richness, e.g. calculated by Koellner and Scholz (2008). This indicator contains only information on species numbers and does not require additional information on species abundance, species identity, or species specifics (such as functional traits or threat status). Therefore, it is the indicator type least restricted by data availability. However, relative changes in local species richness do not have a very strong link to conservation targets. A replacement of one highly-threatened endemic forest species by three invasive, globally distributed species will be reported as an increase in species richness (i.e., a beneficial impact). Therefore, other indicators for measuring relative local biodiversity impacts have been proposed. In Chapter 3, we compared five different diversity indicators (species richness; Shannon index; Sørensen index; Fisher’s α; and mean species abundance, or MSA) and found significantly different results. The indicators considering species composition (MSA and Sørensen) showed stronger impacts of land use than the other indicators, that is, MSA and Sørensen are more sensitive indicators for biodiversity impacts. De Souza et al. (2013) compared local changes in functional diversity and species diversity and found significant differences for some land use types and taxonomic groups. Functional diversity accounts for the functional role that species play in an ecological community and is highest when the functional traits of all species in a community are complementary (de Souza et al. 2013). Coelho and Michelsen (2013) highlight the shortcomings of the species richness indicator as a surrogate for biodiversity and propose to use the naturalness of land use systems, that is, the hemeroby, as an indicator (Brentrup et al. 2002). However, hemeroby values are based on expert judgment, and empirical data justifying the scores are mostly lacking (Coelho and Michelsen 2013). In addition, the proposed values are not spatially dependent, and whether a particular land use type has the same naturalness regardless of whether it occurs in a forest or grassland ecosystem remains questionable.
The analysis in Chapter 3 clearly showed that relative local biodiversity impacts differ across world regions. Therefore, this ratio should be region-specific and not based on a global average per land use type. Based on current data availability, we recommend using the spatially differentiated values developed in Chapter 3, which are based on reductions in species richness. When additional data for calculating global and spatially differentiated ratios for biodiversity loss
149 Concluding remarks become available, we recommend testing other indicators, such as functional diversity. This indicator could provide a better measure for the loss of local ecosystem function than species richness.
Model Type 2: Measure for absolute regional biodiversity value The initial proposal by Weidema and Lindeijer (2001) consisted of three weighting factors for regional biodiversity value: species richness (based on vascular plant species), ecosystem scarcity (the global rarity of an ecosystem type), and ecosystem vulnerability (measured as non- linear function of how much of an ecosystem’s area has already been converted to human land use; similar to a species-area relationship). The spatial resolution of the initially available input data was per biome, and the factors were normalized before multiplication. Michelsen (2008) and Coelho and Michelsen (2013) generally criticize species richness as a meaningful biodiversity indicator and suggest that researchers use only ecosystem vulnerability and scarcity as measures for biodiversity value. In Chapter 4, we propose to use irreplaceability instead of ecosystem scarcity and, as suggested by Weidema and Lindeijer (2001), combine it with species richness (of mammals, birds, amphibians, and reptiles) and ecosystem vulnerability as a weighting factor. Ecosystem scarcity was not chosen as factor because it is calculated in relation to the largest ecosystem and is highly dependent on the choice of this reference ecosystem (see sensitivity analysis conducted by Coelho and Michelsen 2013). Irreplaceability and vulnerability were chosen because they are central to conservation planning theory (Brooks et al. 2006). Irreplaceability was measured as species endemism (of mammals, birds, amphibians, and reptiles). For ecosystem vulnerability, we proposed to use the Conservation Risk Index (CRI) of Hoekstra et al. (2005), which considers not only how much of an ecosystem has already been converted to human use, but also how well protected the remaining natural habitat is.
We propose to use the measures for absolute regional biodiversity value proposed in Chapter 4. These factors include the dimensions of species richness, species endemism (i.e., irreplaceability), and the risk of habitat conversion (i.e., vulnerability) and thereby reflect international conservation targets. In addition, these measures are calculated on an ecoregion level rather than a biome level and therefore more meaningfully capture the regional differences in biodiversity value. However, to be able to finally integrate multiple drivers of biodiversity loss into a single endpoint, the compatibility of the proposed weighting factors with approaches present for other compartments (e.g., water) should be evaluated first.
Model Type 3: Absolute species loss model The methods presented in Chapters 5 and 6 are the first globally applicable land use LCIA methods to directly model potential species extinction. Thereby, they can provide a direct link to international conservation targets, such as the Aichi targets of the Convention on Biological Diversity (2013). The W-Local method (Chapter 6), for example, provides a direct proxy for species extinction that corresponds to the Aichi Target 12: avoid extinction of known threatened species. The permanent impacts calculated in Chapter 5 (A-Regional method) also directly assess potential global extinction of endemic species. Occupation and transformation impacts developed in Chapter 5 assess regional extinction of non-endemic species, which is considered potentially
150 Concluding remarks reversible and provides an earlier warning signal for the global extinction of species. This direct link with conservation targets can prove beneficial when it comes to informing decision-makers about the impacts of their production and consumption based on LCA methods. By 2020, governments, business, or other stakeholders should, according to Target 4, “have taken steps to achieve (…) plans for sustainable production and consumption (…)” (Convention on Biological Diversity 2013).
We propose using the A-Regional (Chapter 5) and the W-Local (Chapter 6) methods in conjunction. The A-Regional method indicates the regional (for occupation and transformation impacts) and global (for permanent impacts) extinction risks of species communities for five taxonomic groups (mammals, birds, amphibians, reptiles, and plants). It considers the non-linear effects of land use pressure on biodiversity loss and has already been calculated globally. If the spatial detail of where land use is located is available, the W-Local method can provide complementary information regarding the eminent extinction risk of individual mammal species related to a specific land use activity. The high spatial detail of the method facilitates the linking of results to Environmental Impact Assessment or conservation planning studies.
Scale of impact and damage function This thesis presented the first global approaches to model biodiversity loss at local, regional, and global scales. Whichever scale of assessment is finally chosen for an LCIA method should also be streamlined with the assessments of other drivers of biodiversity loss in order to finally allow the integration of all biodiversity impacts into a single-score “biodiversity footprint”.
We recommend assessing impacts at regional or global scales, since purely local impacts show a potentially higher reversibility than when, for example, entire ecosystems are affected. Non- linear models should be preferred over linear models because they better reflect biodiversity loss.
Reference states Debates over which reference states to use for land use impact assessments have been ongoing for more than a decade. The choice of a reference state is often restricted by data availability, and authors often use data that approximates a “natural situation.” However, this practice has been questioned by Milà i Canals et al. (2013), whose research suggests, based on the results of a case study of global land use impacts of margarine, the benefits of exploring a current reference state as a potentially more meaningful reference for certain decision-contexts. This thesis presented the first study to apply and compare two different reference situations to the same data (Chapter 6).
Most studies so far have used natural land as a reference state. This implies that the focus has mainly been on past land use impacts. In regions where major land use impacts are expected to happen in the near future, or in LCA studies looking at planned future projects, biodiversity of the current land use composition should be used as a reference (Chapter 6). For the regional assessment presented in Chapter 5, ecoregion-specific scenarios of future land use can be used
151 Concluding remarks instead of current land use composition to directly model characterization factors for potential future impacts.
Resolution and geographic coverage The different studies conducted within this thesis all showed a high spatial variability of biodiversity impacts. Significant differences between biomes, ecoregions, and even 900m grid- cells were found. Global coverage could be truly achieved for Chapter 5 and partly for Chapter 3, with some data gaps. The method of Chapter 6 can easily be upscaled globally, as the underlying data is available. To increase the geographic coverage of the method presented in Chapter 4, biodiversity data of organic and conventionally farmed land would have to be collected for all world regions, which would require a considerable effort.
Spatially differentiated assessments should clearly be preferred over global averages. As large differences exist within biomes, we suggest a regionalization of results to at least the level of ecoregions, but researchers should aim for finer resolution (such as grid-cell) because this would allow the flexibility to aggregate the impact to any spatial unit required for the application of an LCA study (e.g., if inventory data is only available on country level).
To serve as a broadly applicable assessment method, global characterization factors are a necessity. Global methods can serve as a screening tool to identify “hotspots” of impacts along globally distributed value chains. However, when LCA studies focus on goods produced in specific regions, a region-specific method can provide more detail and regional accuracy than globally applicable methods, which often are highly uncertain.
Taxonomic and land use coverage The analysis in Chapter 3 clearly showed that land use impacts differ significantly depending on which taxonomic group is considered. Therefore, we cannot assume that vascular plants are a valid proxy for biodiversity as a whole. Large differences within land use classes were found in Chapter 4 (e.g., organic and conventionally farmed annual crops). Therefore, detail in land use classes is required for meaningful decision-support.
Multiple taxonomic groups should be considered in LCIA methods because impacts differ between species groups (as shown in Chapter 3). To finally simplify results for decision-makers, a meaningful approach to aggregate impacts across taxonomic groups will have to be developed.
In the presented methods, the main focus was on agricultural land. In their current form, an application to forestry, mining, or artificial land cover types might prove difficult and will first require further method development.
Types of impacts In the land use assessment framework, three impact types are typically distinguished: occupation, transformation, and permanent impacts. This thesis presented the first study to attempt to quantify permanent impacts.
152 Concluding remarks
To allow an assessment of the impacts of land use as well as land use change, characterization factors for both occupation and transformation impacts are essential. To calculate transformation impacts, we recommend using the recovery times from Curran et al. (in press) used in this thesis (Chapters 5 and 6). These recovery times are spatially explicit and are based on empirical data instead of on expert estimates, which have previously been used for the calculation of global transformation characterization factors. The assumption that land use impacts are fully reversible is rather unlikely in view of expected increased future land use pressures. Therefore, quantifying irreversible impacts will increase the credibility of LCA models.
Uncertainties Previously developed globally applicable land use assessment methods did not quantify uncertainty. The analyses done in this thesis clearly showed the substantial uncertainties involved in globally applicable land use assessment methods.
To provide scientifically robust, meaningful and transparent decision-support, uncertainties of land use impacts on biodiversity need to be quantified. This can also help in detecting the major sources of uncertainty (as shown in Chapter 5) and can direct model developers towards the best focus areas for reducing uncertainties. Finally, reducing uncertainty as much as possible facilitates decision-making based on LCA results.
153 Concluding remarks
Table 7.1. Comparison of all methods developed in this thesis and other (potentially) globally applicable land use LCIA methods. The numbers represent the different model types. The approach (1) calculates relative local reduction in biodiversity (as a ratio). In many studies the ratio of approach (1) is multiplied with the approach (2), which calculates a regional biodiversity value (as an absolute number). The approach (3) directly models absolute biodiversity loss.
Methods of other authors Methods developed within this thesis
Name and Lindeijer (2000) Weidema & Koellner & Coelho & De Souza et al. Chap. 3: relative Chap. 4: Chap. 5: Chap. 6: reference of Lindeijer (2001) Scholz (2008) Michelsen (2013) local loss, weighted local absolute weighted local method (2013) R-Local relative loss, regional loss, loss, RW-Local A-Regional W-Local Type of model (1) Relative (1) Relative (1) Relative (1) Relative (1) Relative (1) Relative (1) Relative (3) Absolute (3) Absolute local reduction local reduction local reduction local reduction local reduction local reduction local reduction regional local reduction in (2) regional in (2) regional in species rich- in (2) regional in functional in species in (2) regional reduction of of weighted plant species biodiversity ness compared biodiversity diversity richness biodiversity species species richness value to average re- value value richness richness gional diversity
Measure and Dutch plant Authors Plant species Hemeroby Functional Species Plant species - - data for relative species estimates richness (meta- (Brentrup et al. diversity (Flynn richness (BDM richness local richness (Witte analysis) 2002) et al. 2009) 2004; Alkemade (multiple biodiversity and van der et al. 2009) studies reduction factor Meijden 1995) combined) (only Types 1 & 2) Measure for Plant species Species Ecosystem - - Species - - regional richness richness, vulnerability richness, biodiversity ecosystem and ecosystem irreplaceability value (only Type vulnerability, scarcity (endemism), 2) and ecosystem and ecosystem scarcity vulnerability
Concluding remarks
Methods of other authors Methods developed within this thesis
Name and Lindeijer (2000) Weidema & Koellner & Coelho & De Souza et al. Chap. 3: relative Chap. 4: Chap. 5: Chap. 6: reference of Lindeijer (2001) Scholz (2008) Michelsen (2013) local loss, weighted local absolute weighted local method (2013) R-Local relative loss, regional loss, loss, RW-Local A-Regional W-Local Absolute ------Matrix- Species-spe- species loss calibrated cific habitat model (only species-area suitability mo- type 3) relationships dels (Rondinini (Koh and et al. 2011), Ghazoul 2010) weighted by species rarity & threat Scale of impact (1) Local / linear (1) Local / linear Local /linear (1) Local/ linear Local/ linear Local/ linear (1) Local/ linear Regional/ non- Local/ linear and damage (2) Regional (2) Regional / and non-linear (2) Regional/ (2) Regional/ linear function /linear non-linear non-linear linear Reference Natural Natural Regional Natural Natural (Semi)-natural Semi-natural Natural land Natural & states situation situation average species situation situation land situation current land richness cover Resolution (1) Global (1) Global Barthlott's (1) Global Global Biome (1) Biome Ecoregion Pixel (0.81 (2) Physiotope (2) Biome plant diversity (2) Ecoregion (ecoregion (2) Ecoregion km2) (~Biome) zones (Mutke possible) and Barthlott 2005)
Geographic Global Global Switzerland/ New Zealand Selected Global Three biomes Global East Africa coverage Germany (addi- (global possible regions across (additional data (global tional data with existing America (additi- needed for possible with needed for glo- data) onal data need- global existing data) bal application) ed for global application) application)
Concluding remarks
Methods of other authors Methods developed within this thesis
Name and Lindeijer (2000) Weidema & Koellner & Coelho & De Souza et al. Chap. 3: relative Chap. 4: Chap. 5: Chap. 6: reference of Lindeijer (2001) Scholz (2008) Michelsen (2013) local loss, weighted local absolute weighted local method (2013) R-Local relative loss, regional loss, loss, RW-Local A-Regional W-Local Taxonomic (1) Plants (1) Not species Plants (1) Not species Plants, Plants, (1) Plants Plants, Mammals coverage (2) Plants specific specific mammals, vertebrates, (2) Mammals, mammals, (2) Plants (2) Not species birds arthropods, birds, amphi- birds, specific other bians, reptiles amphibians, invertebrates reptiles Land use Mining, Human 53 land use Hemeroby 19 land use Annual crops, Organic and Agricultural Cropland classes industrial settlements, types (including classes classes permanent conventionally land, pasture, (>70%), mosaic production, arable and agriculture, (including 7 crops, managed land used forests, cropland (20– roads, landfills, permanent forestry, agricultural agroforestry, (pasture, artificial areas 70%); (could be forestry, crops, pasture artificial land, classes, 5 used forests, annual and extended to hydropower etc.) forestry secondary permanent any other classes, forests, crops) GlobCover pasture, rail artificial areas, class) roads, artificial pasture waterbodies) Type of impacts Occupation and Occupation and Occupation and Occupation Occupation Occupation Occupation and Occupation, Occupation and transformation transformation transformation impacts impacts impacts transformation transformation, transformation impacts impacts impacts impacts and permanent impacts impacts
Uncertainties No No Yes No Yes Yes Yes Yes No quantified
Concluding remarks
7.1.3 Conclusions and new aspects of the thesis In this thesis, we developed four globally applicable assessment methods for land use impacts on biodiversity compatible with the LCA land use assessment framework (Figure 1.1.). For two of these methods, we provided global and operational characterization factors (Chapters 3 and 5). The other two (Chapters 4 and 6) were calculated for a specific region but could be calculated globally in the future, since they are based on globally available data. For the first time, global characterization factors encompass multiple taxonomic groups; local, regional, and global species losses; as well as occupation, transformation, and permanent impacts. In addition, their uncertainties are for the first time quantified. This more complete assessment provides a more detailed understanding of how the land use of products influences biodiversity in different world regions and at different spatial scales and how it affects different species groups. The consideration of different impact types (occupation, transformation, and permanent) can inform decision-makers about the effects of actual land use, land use changes, and the risk of irreversible damages. The quantification of uncertainties provides transparency on the accuracy of such global models and facilitates the identification of areas to further reduce uncertainties, thereby providing more robust information to decision-makers.
In addition, this thesis evaluated the current biodiversity models used in LCA, highlighted conceptual shortcomings and contributed to a better understanding of how the current models could be improved. Specifically, this thesis contributed to improving land use impact assessment methods by exploring a range of model choices. These options included the choice of biodiversity indicators and taxonomic groups (Chapter 3), the resolution of impact assessment methods (from biome to ecoregion to 900m grid-cells; Chapters 3-6), the inclusion of land use intensity (Chapter 4), and the choices of reference situations (Chapter 6) and modeling perspectives (average vs. marginal and retrospective vs. prospective; Chapter 5). The application to case studies (Chapters 4 and 6) illustrated the applicability of methods and gave specific insights into the biodiversity impacts of the assessed products.
From the identified conceptual shortcomings of how biodiversity models are constructed in LCA (Chapter 2), we contributed new solution to:
1) Lacking considerations of the scale of impacts: We developed several methods assessing impacts at different spatial scales (local, regional, and global extinction).
2) Limited taxonomic coverage: We have included a variety of taxonomic groups
3) Limited geographic coverage: The presented methods are global or can be expanded to reach global coverage
4) Poor implementation of non-linearity and irreversibility of impacts: In Chapter 5, we have considered non-linear effects and developed a first approach how to quantify irreversible impacts.
157 Concluding remarks
The limited focus of indicators (5) (mainly compositional indicators of species communities) and the missing drivers of biodiversity loss (6) (overexploitation and invasive species), remain topics for future research.
In summary, this thesis presented for the first time:
• a set of global characterization factors, including their uncertainties (Chapters 3 and 5),
• a global and spatially differentiated quantification of local relative impacts based on empirical data and multiple taxonomic groups (Chapter 3),
• a globally applicable method directly modeling absolute species loss (Chapters 5 and 6),
• a quantification of permanent land use impacts (Chapter 5) and a quantification of global transformation impacts based on a global meta-analysis of biodiversity recovery instead of expert-estimated recovery times (Chapters 5 and 6),
• a high-resolution impact assessment method at ecoregion and 900m grid-cell level (Chapters 5 and 6) that better reflects the spatial heterogeneity of biodiversity than the previously existing resolutions at the biome level,
• approaches to include recently developed ecological models in LCIA, such as the GLOBIO3 model (Alkemade et al. 2009), matrix-calibrated species-area relationships (Koh and Ghazoul 2010), and habitat suitability models of the global mammals assessment (Rondinini et al. 2011),
• an approach to model species-specific impacts globally, accounting for species’ global rarity and threat levels and thereby providing a direct link to the Aichi Target 12 (Chapter 6),
• an approach that allows flexibility in the choice of the reference state (Chapter 6) and facilitates the assessment of both past impacts (based on the reference of natural land) and future impacts (based on the reference of current land cover),
• an approach that assesses non-linear regional land use impacts and allows for average as well as marginal assessments and retro- as well as prospective assessments (Chapter 5),
• a case study that quantifies biodiversity impacts of land use of organic and conventional milk, including the impacts of imported feedstock (Chapter 4), and
• three case studies (tea, coffee, and tobacco) that compare globally applicable land use LCIA methods in East Africa (Chapter 6).
7.1.4 Limitations and uncertainties The aim of this thesis was to provide globally applicable land use impact assessment methods. However, global applicability comes at the expense of local accuracy and detail. The methods of Chapters 3, 5, and 6 consist of coarse land use classes and do not allow the assessment of differences in land management practices. As illustrated by the case study in Chapter 4, differences between high- and low-input systems can be large and may even outweigh the
158 Concluding remarks additional area requirements of low-input systems. This coarse land use classification also does not effectively capture non-agricultural land use, such as different types of forestry, mining, and artificial areas. The land use types included in “artificial areas” range from built-up land to railroads and green urban areas (Koellner et al. 2013a), which might show very different biodiversity impacts. Although impacts of artificial areas are quantified in Chapter 3 and 5, the underlying data are derived mainly from the Swiss Biodiversity Monitoring (BDM 2004) because very few studies on species richness of artificial areas were available in the GLOBIO3 database. The geographic bias and the broad range of land use classes underlying the characterization factors for “artificial areas” need to be considered when applying these values. In many LCA studies on agricultural products, the main share of land use impacts is expected to occur in the agricultural production phase, and neglecting the effects of land use in other life cycle stages (e.g., processing, packaging, and transport) might be justifiable. However, for studies comparing, for example, forest products and agricultural products (e.g., wood-based vs. crop-based biofuels), the presented methods might not provide the necessary detail.
Data quality and availability was critical throughout the thesis. Even basic information, such as global land cover maps used as input for the methods in Chapters 5 and 6, were highly uncertain. The globally available land cover maps show significant spatial disagreement, especially for croplands and forests (Fritz et al. 2012). In addition, available land cover maps have a poor representation of human-used land and mainly distinguish between different types of land cover, but do not indicate the degree of human influence (e.g., whether a closed forest is a managed or natural forest, or a forest plantation with exotic trees). This limited knowledge of current global land use activities data restricts the level of detail possible in global land use impact assessment methods.
The newly developed methods in this thesis assess biodiversity impacts on different spatial scales, but none of the methods considers the spatial configuration of the different types of land use. Aspects such as habitat fragmentation, isolation, migration, and dispersal capacities of species within a landscape were not considered. However, landscape modification and habitat fragmentation are considered key drivers of biodiversity loss (Fischer and Lindenmayer 2007). Alkemade et al. (2009), for example, found that in habitat patches smaller than 1km2, only 30% of the species had sufficient space for viable populations and only patches of over 10,000 km2 of suitable habitat were large enough for viable populations for all species. Therefore, the results of this thesis underestimate the effects of land use on biodiversity. This is especially important to bear in mind when LCA studies include linear infrastructure elements (e.g., roads or aboveground pipelines), which require little area but contribute to the isolation of habitat fragments (see Jordaan et al. 2009).
The methods presented in this thesis are based on the UNEP/SETAC land use assessment framework for LCA (Koellner et al. 2013b). This framework allows the operationalization of land use impacts, but also makes some strong assumptions (summarized in Koellner et al. 2013b). One major assumption is the reversibility of impacts - at least for occupation and transformation impacts. The framework assumes a hypothetical future land abandonment and subsequent
159 Concluding remarks recovery of biodiversity, which is a rather unrealistic scenario for upcoming decades, or even for the next century. The demand for land for food, energy production, building material, and settlement areas is projected to drastically increase in future because the human population is rapidly growing. In addition, consumption patterns in emerging economies tend towards “Western” resource- and land intensive consumption. For example, meat consumption in China has increased nine fold in the past 40 years (Kearney 2010) and Asian meat imports are projected to further increase four fold until 2030 (Pingali 2007). Because biodiversity recovery can take up to centuries and requires an existing connectivity to natural source-habitats that allow the migration of species into the abandoned land (Curran et al. in press), the assumption of reversible land use impacts is questionable.
7.2 Practical relevance The dramatically increased rates of global species extinction, which are up to 100 times higher the natural background rates (MA 2005), have gained international attention in past decades. In 1992 at the Rio Earth Summit, the Convention on Biological Diversity was initiated and has up to now been signed by 168 parties. In 2010, twenty biodiversity targets were defined for implementation by 2020 (Convention on Biological Diversity 2013). These targets include, inter alia, awareness creation regarding biodiversity loss and conservation (Target 1), integration of biodiversity values into planning processes and national accounting and reporting systems (Target 2), implementation of plans for sustainable production and consumption by governments, business and stakeholders (Target 4), sustainable management of agriculture, aquaculture and forestry (Target 7), avoiding extinction of known threatened species (Target 12), and an improved and shared knowledge and science base relating to biodiversity, its values, functioning, status and trends, and the consequences of its loss (Target 19).
Several international initiatives were launched to assess the conditions and trends of biodiversity and the world’s ecosystems (Millenium Ecosystem Assessment; MA 2005) and to help decision- makers recognize, demonstrate, and capture the values of ecosystem services and biodiversity (The Economics of Ecosystems and Biodiversity (TEEB); Bishop et al. 2010). In 2012, the Intergovernmental Platform on Biodiversity and Ecosystem Services (IPBES) was founded as a leading intergovernmental body for assessing the state of the planet's biodiversity, its ecosystems, and the essential services they provide to society.
However, the inclusion of biodiversity into business strategies is still at the beginning. In 2008, only six of the one hundred largest companies in the world reported actions to reduce biodiversity impacts, and only two companies identified biodiversity as a key strategic issue (Bishop et al. 2010). In recent years, several initiatives have been launched to better include biodiversity into business strategies. The Sustainability Consortium, with over 100 business members, aims to quantify and communicate the sustainability of products more accurately and in a more standardized way. The French government initiated a National Strategy for Sustainable Development 2010-2013, including plans to display environmental footprints of products to consumers, also considering impacts on biodiversity.
160 Concluding remarks
Through these and other initiatives, LCA has in recent years gained increasing attention from companies and governments. Scientifically sound and generically applicable assessment methodologies are required in LCA for credible decision-support in various contexts. The methodological advances made in this thesis make a major contribution to improving models of biodiversity loss caused by land use. The regional assessment method proposed in Chapter 5 can be directly tested by these initiatives, as can a combined methodology of the local relative assessment proposed in Chapter 3 multiplied by a regional weighting scheme, such as the one developed in Chapter 4. These globally applicable methods can be readily tested as measures for quantifying biodiversity loss in LCA studies, environmental product declaration, biodiversity label creation, or minimum environmental product standards. The methods develop in this thesis could thereby help to inform consumers, retailers, and policy-makers in making ecologically sound choices.
The local weighted species loss proposed in Chapter 6 provides a direct link between products and Aichi Target 12 (avoid extinction of known threatened species). In this study, impacts varied strongly depending on where raw agricultural materials were sourced. Based on these results, retailers could make purchasing decisions to avoid sourcing raw material from regions with direct impacts on the extinction of range-restricted mammals. The high spatial resolution can, in addition, provide a link to environmental impact assessment methods and identify the extinction risk of range-restricted mammals related to future development projects. The flexibility in the choice of the reference situation (past or current) also allows the identification of regions with a high current impact and thereby provides useful information for conservation planners. The high spatial detail of the method facilitates not only the quantification of potential damages of agricultural production and where damages could be reduced, but also allows an estimation of potential benefits of compensation measures. In situations, where a reduction of the biodiversity impacts is not further possible, such a compensation scheme could help to detect suitable land for offsetting the biodiversity loss caused by products. The Ph.D. thesis of Michael Curran (2013) illustrates, how such a compensation scheme could look like and how the presented methods and data of this thesis can be integrated.
The case studies conducted in this thesis help to advance the understanding of trade-offs between extensive milk production requiring large areas and intensive production on less land (Chapter 4). In the case of Swedish milk, organic production had clearly lower impacts on biodiversity. Such studies are relevant to policy-makers when it comes to justifying subsidies for organic agriculture, but they are also relevant to purchasing decisions of consumers. In the case of tea, coffee, and tobacco production in East Africa, areas of high impact could be identified (Chapter 6). These results are relevant not only to retailers, but also to land managers and conservation planners, who identify areas in which establishing new cropland would be least harmful to biodiversity or in which conservation areas should be established.
7.3 Future research needs To further advance the applicability of the presented methods and to generally improve the representation of biodiversity impacts in LCA, additional research is needed in several areas.
161 Concluding remarks
First, the different approaches should be further tested in case studies, especially the two broad lines of modeling biodiversity loss due to land use: directly modeling regional or global species extinction (Type 2) or combining measures of local relative biodiversity loss with absolute measures of the regional value of biodiversity (Type 3). This should lead to a clearer understanding of how and where the methods lead to similar or opposite conclusions and of the practical applicability of the different approaches. Case studies should also further explore how results are influenced by the basic model choices, such as the choice of reference situation (Chapter 6), the considered time horizon (i.e., retro- or prospective assessment) of biodiversity loss, and whether non-linear damages are modeled based on marginal or average impacts (Chapter 5). In addition, the underlying value choices of different approaches and weighting systems should be systematically analyzed and a set of consistent value choices presented, relying, for example, on the cultural perspectives presented by Hofstetter et al. (2000) and implemented by De Schryver et al. (2010) for land use impacts.
Second, the cover of land classes of the methods should be improved, with special focuses on both pasture land and non-agricultural land-based sectors, such as forestry, mining, and building, which were not the focus of this thesis. The definitions of land use classes should ideally match the classification system proposed by Koellner et al. (Koellner et al. 2013a).
For agricultural applications, integration of land use intensity should be achieved for all approaches in order to inform the land-sharing and land-sparring discourse (Tscharntke et al. 2012). Therefore, better global land use maps need to be developed to represent not only land cover types, but also the intensity of human use. A central and publicly available database for biodiversity data related to human-modified land would facilitate further refining of global land use assessment methods.
Third, new approaches to integrate aspects of the spatial configuration of land use should be developed. High-resolution land cover maps could be used as a basis for calculating the connectivity or fragmentation of all global landscapes. The impacts of fragmentation on biodiversity could then, for example, be assessed by calculating the number of species able to survive in habitat patches of different sizes (see Alkemade et al. 2009).
Fourth, the integration of additional taxonomic groups, especially for the species-specific approach presented in Chapter 6, should be attempted. The lack of detailed information on species habitat limits the number of taxonomic groups that can be integrated. Priority should be given to the inclusion of range-restricted species or species highly threatened by land use. As a starting point, global data for amphibians on threat categories, range maps, and ecological information provided by IUCN could be used. Additional data on other taxonomic groups are at least available for specific world regions, such as Europe (containing data for about 6,000 species of mammals, reptiles, amphibians, freshwater fishes, butterflies, dragonflies, and selected groups of beetles, mollusks, and vascular plants) and California (vertebrates; Geyer et al. 2010). For Europe, the BioScore method (Louette et al. 2010) could serve as an inspiration how to use sensitivity information of selected species and how results for different taxonomic groups can finally be integrated. Further research is needed on the question of aggregation. If the target of
162 Concluding remarks biodiversity protection is defined as avoiding global extinction of species, then we can simply add up the number of species at risk of extinction. Another option is to weigh species loss depending on their phylogenetic similarity, in which an evolutionary rare species is given a higher value than a species with many closely related species of the same genus. If the target of biodiversity conservation is to maintain ecosystem function, the extinction of species that play a unique functional role in an ecosystem (e.g., top predator) should be weighted higher than that of species with many functionally similar species (e.g., herbs).
Fifth, to make the methods presented in this thesis accessible for LCA practitioners and researchers, global characterization factors should also be calculated for the methods of Chapter 4 and 6. The data for globally applying the species-specific method presented in Chapter 6 are readily available. Here, the challenge is mainly to solve the technical issues of how to handle such extensive data and how to find an optimal spatial resolution to still provide the necessary detail at a feasible data volume. To allow global comparison of organic and conventional land (Chapter 4), additional data need to be collected for other world regions that were not involved in the life cycle of the analyzed Swedish milk, a process that could be very time-consuming or even impossible due to a lack of empirical and internationally published biodiversity surveys. Future studies should therefore also experiment with ways to extrapolate the available data to regions where data are lacking.
Applying the presented spatially explicit characterization factors to case studies remains challenging, because the limited spatial resolution of inventory data often does not allow to exactly locate an activity within one ecoregion, not even speaking of grid-cells. Therefore, future research on how to improve the spatial resolution of land use inventory data is needed, which would facilitate LCA practitioners really being able to make use of the spatially detailed impact assessment methods. Tools to track products along global value chains should be explored. In addition, research is needed to evaluate the cause-effect chains of land use change and how this can be allocated to specific products. Indirect land use impacts, in particular, need to be further understood and quantified.
Finally, further research should be conducted for finding a harmonized and meaningful endpoint for biodiversity loss in LCA. In recent years, spatially differentiated global or regional assessment methods have been developed for multiple drivers of biodiversity loss (eutrophication: Azevedo et al. 2013a; acidification: Azevedo et al. 2013b; photochemical ozone formation: van Goethem et al. 2013; surface and groundwater abstraction: Verones et al. in press). To be able to aggregate and simplify information on biodiversity loss of multiple drivers for decision-makers, the assessment metrics need to be harmonized and ways of dealing with different taxonomic coverage of the methods need to be elaborated. Verones et al. (in press) base their assessment for biodiversity loss related to the consumptive use of surface and groundwater on a metric similar to that applied in this thesis (Chapter 6), but include more taxonomic groups (mammals, birds, reptiles, amphibians). Whether or not a similar metric could be applied to drivers of pollution (ecotoxicity, acidification, eutrophication, or ozone formation) needs to be explored. To simplify the assessment of land use impacts, correlation of different impact pathways, such as erosion
163 Concluding remarks control, freshwater regulation, and water purification (Saad et al. 2013), carbon sequestration (Müller-Wenk and Brandão 2010), biotic production potential (Brandão and Milà i Canals 2013), and biodiversity loss should be further investigated, and a set of uncorrelated assessment methods should be implemented into novel LCIA methods.
Modeling biodiversity impacts within LCA spatially differentiated still requires methodological improvements in multiple areas. However, this thesis made a significant step forward in modeling biodiversity impacts of land use by providing several globally applicable methods. Further development and testing of the methods will allow better decision-support to reduce the biodiversity impacts of economic activities. In order to halt biodiversity loss and maintain ecosystems within safe ecological limits, a set of approaches and tools are required to understand the causalities of biodiversity loss and the extent to which different solutions contribute to solving the problem. The important role of LCA is, thereby, to contribute to the understanding of how potential solutions lead to “leakage,” that is, shifting burdens between regions or environmental compartments.
7.4 References Alkemade R, van Oorschot M, Miles L, Nellemann C, Bakkenes M, ten Brink B (2009) GLOBIO3: A framework to investigate options for reducing global terrestrial biodiversity loss. Ecosystems 12 (3):374-390. Azevedo LB, van Zelm R, Elshout PMF, Hendriks AJ, Leuven RSEW, Struijs J, Zwart DD, Huijbregts MAJ (2013a) Species Richness-Phosphorus Relationships for Lakes and Streams Worldwide. Global Ecology and Biogeography (in press). Azevedo LB, Van Zelm R, Hendriks AJ, Bobbink R, Huijbregts MAJ (2013b) Global assessment of the effects of terrestrial acidification on plant species richness. Environmental Pollution 174:10-15. BDM (2004) Biodiversity Monitoring Switzerland. Indicator Z9: Species Diversity in Habitats. Bundesamt für Umwelt, BAFU. http://www.biodiversitymonitoring.ch. Accessed 1.2.2011 Bishop J, Bertrand N, Evison W, Gilbert S, Grigg A, Hwang L, Kallesoe M, Vakrou A, van der Lugt C, Vorhies F (2010) TEEB – The Economics of Ecosystems and Biodiversity Report for Business - Executive Summary 2010. UNEP. Brandão M, Milà i Canals L (2013) Global characterisation factors to assess land use impacts on biotic production. The International Journal of Life Cycle Assessment 18 (6):1243-1252. Brentrup F, Kusters J, Lammel J, Kuhlmann H (2002) Life Cycle Impact assessment of land use based on the Hemeroby concept. The International Journal of Life Cycle Assessment 7 (6):339-348. Brooks T, Mittermeier R, da Fonseca G, Gerlach J, Hoffmann M, Lamoreux J, Mittermeier C, Pilgrim J, Rodrigues A (2006) Global biodiversity conservation priorities. Science 313 (5783):58-61. Coelho CRV, Michelsen O (2013) Land Use Impacts on Biodiversity From Kiwifruit Production in New Zealand Assessed with Global and National Datasets. The International Journal of Life Cycle Assessment (in press). Convention on Biological Diversity (2013) Aichi Biodiversity Targets. http://www.cbd.int/sp/targets/. Accessed 4.10.2013 Curran M (2013) Compensating the biodiversity impacts of land use: Toward ecologically equal exchange in the North–South context. Ph.D. Thesis, ETH, Zurich Curran M, Hellweg S, Beck J (in press) Is there any empirical support for biodiversity offset policy? Ecological Applications. De Schryver AM, Goedkoop MJ, Leuven RSEW, Huijbregts MAJ (2010) Uncertainties in the application of the species area relationship for characterisation factors of land
164 Concluding remarks
occupation in life cycle assessment. The International Journal of Life Cycle Assessment 15 (7):682-691. de Souza DM, Flynn DFB, Declerk F, Rosenbaum RK, de Melo Lisboa H, Koellner T (2013) Land use impacts on biodiversity: proposal of characterization factors based on functional diversity. The International Journal of Life Cycle Assessment 18 (6):1231-1242. Emanuelsson A, Ziegler F, Pihl L, Skold̈ M, Sonesson U Overfishing, overfishedness and wasted potential yield: new impact categories for biotic resources in LCA. In: Corson MS, van der Werf HMG (eds) 8th Int. Conference on LCA in the Agri-Food Sector (LCA Food), Saint Malo, France, 2012. INRA, Rennes, France. Fischer J, Lindenmayer DB (2007) Landscape modification and habitat fragmentation: a synthesis. Global Ecology and Biogeography 16 (3):265-280. Flynn D, Gogol‐Prokurat M, Nogeire T, Molinari N, Richers B, Lin B, Simpson N, Mayfield M, DeClerck F (2009) Loss of functional diversity under land use intensification across multiple taxa. Ecology Letters 12 (1):22-33. Fritz S, McCallum I, Schill C, Perger C, See L, Schepaschenko D, van der Velde M, Kraxner F, Obersteiner M (2012) Geo-Wiki: An online platform for improving global land cover. Environmental Modelling & Software 31 (C):110-123. Geyer R, Lindner JP, Stoms DM, Davis FW, Wittstock B (2010) Coupling GIS and LCA for biodiversity assessments of land use: Part 2: Impact assessment. The International Journal of Life Cycle Assessment 15 (7):692-703. Goedkoop M, Heijungs R, Huijbregts MAJ, De Schryver A, Struijs J, van Zelm R (2008) ReCiPe 2008. A life cycle impact assessment method which comprises harmonised category indicators at the midpoint and the endpoint level; First edition Report I. Den Haag. Goedkoop M, Spriensma R (1999) The Eco-Indicator 99. A Damage Oriented Method for Life Cycle Impact Assessment. Methodology Report. PRé Consultants, Amersfoort. Hoekstra JM, Boucher TM, Ricketts TH, Roberts C (2005) Confronting a biome crisis: global disparities of habitat loss and protection. Ecology Letters 8 (1):23-29. Hofstetter P, Baumgartner T, Scholz RW (2000) Modelling the Valuesphere and the Ecosphere: Integrating the Decision Makers' Perspectives into LCA. The International Journal of Life Cycle Assessment 5 (3):161-175. Jordaan S, Keith D, Stelfox B (2009) Quantifying land use of oil sands production: a life cycle perspective. Environmental Research Letters 4 (2). Kearney J (2010) Food Consumption Trends and Drivers. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 365 (1554):2793–2807. Koellner T, de Baan L, Beck T, Brandão M, Civit B, Goedkoop MJ, Margni M, Milà i Canals L, Müller-Wenk R, Weidema B, Wittstock B (2013a) Principles for Life Cycle Inventories of land use on a global scale. The International Journal of Life Cycle Assessment 18 (6):1203-1215. Koellner T, de Baan L, Beck T, Brandão M, Civit B, Margni M, Milà i Canals L, Saad R, de Souza DM, Müller-Wenk R (2013b) UNEP-SETAC guideline on global land use impact assessment on biodiversity and ecosystem services in LCA. The International Journal of Life Cycle Assessment 18 (6):1188-1202. Koellner T, Scholz RW (2008) Assessment of land use impacts on the natural environment. Part 2: Generic characterization factors for local species diversity in Central Europe. The International Journal of Life Cycle Assessment 13 (1):32-48. Koh L, Ghazoul J (2010) A matrix-calibrated species-area model for predicting biodiversity losses due to land-use change. Conservation Biology 24 (4):994-1001. Lindeijer E (2000) Biodiversity and life support impacts of land use in LCA. Journal of Cleaner Production 8:313–319. Louette G, Maes D, Alkemade J, Boitani L, Knegt Bd, Eggers J, Falcucci A, Framstad E, Hagemeijer W, Hennekens S, Maiorano L, Nagy S, Serradilla A, Ozinga W, Schaminée J, Tsiaousi V, Tol S, Delbaere B (2010) BioScore-Cost-effective assessment of policy impact on biodiversity using species sensitivity scores. Journal for Nature Conservation 18 (2):142-148.
165 Concluding remarks
MA (2005) Millennnium Ecosystem Assessment. Ecosystems and Human Well-being: Biodiversity Synthesis. World Resources Institute, Washington DC. Michelsen O (2008) Assessment of land use impact on biodiversity. The International Journal of Life Cycle Assessment 13 (1):22-31. Milà i Canals L, Rigarlsford G, Sim S (2013) Land use impact assessment of margarine The International Journal of Life Cycle Assessment 18 (6):1265-1277. Müller-Wenk R, Brandão M (2010) Climatic impact of land use in LCA-carbon transfers between vegetation/soil and air. The International Journal of Life Cycle Assessment 15 (2):172-182. Mutke J, Barthlott W (2005) Patterns of vascular plant diversity at continental to global scales. Biologiske Skrifter 55:521-531. Noss R (1990) Indicators for monitoring biodiversity - a hierarchical approach. Conservation Biology 4 (4):355-364. Pingali P (2007) Westernization of Asian Diets and the Transformation of Food Systems: Implications for Research and Policy. Food Policy 32 (3):281–298. Rondinini C, Di Marco M, Chiozza F, Santulli G, Baisero D, Visconti P, Hoffmann M, Schipper J, Stuart SN, Tognelli MF (2011) Global habitat suitability models of terrestrial mammals. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 366 (1578):2633-2641. Saad R, Koellner T, Margni M (2013) Land use impacts on freshwater regulation, erosion regulation and water purification: a spatial approach for a global scale level. The International Journal of Life Cycle Assessment 18 (6):1253-1264. Tscharntke T, Clough Y, Wanger TC, Jackson L, Motzke I, Perfecto I, Vandermeer J, Whitbread A (2012) Global food security, biodiversity conservation and the future of agricultural intensification. Biological Conservation 151 (1):53-59. van Goethem TMWJ, Preiss P, Azevedo LB, Roos J, Friedrich R, Huijbregts MAJ, Zelm RV (2013) European Characterization Factors for Damage to Natural Vegetation by Ozone in Life Cycle Impact Assessment. Atmospheric Environment 77 (C):318–324. Verones F, Saner D, Pfister S, Baisero D, Rondinini C, Hellweg S (in press) Effects of consumptive water use on biodiversity in wetlands of international importance. Environmental Science & Technology. Weidema B, Lindeijer E (2001) Physical impacts of land use in product life cycle assessment. Final report of the EURENVIRON-LCAGAPS sub-project on land use. Department of Manufacturing Engineering and Management, Technical University of Denmark, Lyngby. Witte F, van der Meijden R (1995) Verspreidingskaarten van de botanis- che kwaliteit in Nederland uit FLORBASE. Gorteria 1 (2):3-60. Wolters V, Bengtsson J, Zaitsev AS (2006) Relationship among the species richness of different taxa. Ecology 87 (8):1886-1895.
166
Appendices
Appendix A: Appendix to Chapter 2
Appendix A: Appendix to Chapter 2
A.1 Genetic indicators Techniques to measure genetic diversity can be divided into those covering intraspecific (within species) variation, and those representing interspecific (between species) variation.
Intraspecific variation may be quantified by characterising the discreet allelic state of a population, or using continuous genetic traits (Hughes et al., 2008). This involves the use of visible polymorphisms, which are phenotypically expressed by the organism to give a discreet indication of allelic state (e.g. Mendel’s pea experiments); molecular markers, which separate the products of transcription, proteins, using electrophoresis to discreetly assign allelic state; and DNA markers, which involve DNA sequence data and may be discreet, or continuous (Conner & Hartl, 2004).
Discreet indicators include allelic diversity, an index of the number and frequency of alleles in the population; allelic richness, the number of alleles per locus; genotypic richness, number of genotypes in a population; heterozygosity, the proportion of loci that carry two alleles within a single diploid individual; mutational diversity and effective population size; nucleotide diversity, the number of nucleotides differing between two random individuals from a population; and the percentage of polymorphic loci (Hughes et al., 2008). Continuous indicators include genetic variance, the variation in a trait among individuals of a single pedigree calculated using parent- offspring regression or genealogical information; the coefficient of genetic variance, expressing the variance in a trait in relation to the trait mean in the sample; and heritability, the ratio of genetic variance in the individual to total phenotypic variance in the population (Hughes et al., 2008).
Intraspecific genetic variation can be extremely important in small or isolated populations, where random genetic drift and inbreeding can reduce fitness and adaptability. This increases susceptibility to stochastic variation in population size or changing environmental and biological conditions, increasing extinction risk. High intraspecific genetic diversity can increase ecosystem resistance to environmental or biological disturbance (e.g. Hughes & Stachowicz, 2004; Zhu et al., 2000), and increase the rate of recovery after such disturbances (Reusch et al., 2005). Although important for both the maintenance of higher level biodiversity (Hughes et al., 2008), and in maintaining and provisioning ecosystem services (Hajjar et al., 2008), implementation of monitoring genetic diversity has not been prioritized by the Convention on Biological Diversity, despite its mandate to do so (Laikre, 2010). Consequently, availability of data on intraspecific genetic diversity lags behind those of species and ecosystems, despite the availability of indicators.
Interspecific (between species) genetic diversity has been included in a number of biodiversity assessment frameworks via the use “phylogenetic indices”, constructed using molecular phylogenies of various taxa (Faith, 2002). In particular, phylogenetic diversity (PD) is an important
168 Appendix A: Appendix to Chapter 2 metric of biodiversity because it reflects evolutionary history of a community, is tightly linked to endemism and species complementarity (Faith et al., 2004), and acts as a proxy for changes in ecosystem function and functional diversity (Cadotte et al., 2008). It is defined as the sum of the branch lengths linking individual species within a community or sample (Faith et al., 2004). The presence of evolutionarily old or relictual species, or the distribution of species among many taxonomic groups, increases this value. The approach by Cadotte et al. (2008) might be particularly applicable to LCA because it requires only species information and widely available and extensive sequence data from GeneBank (National Center for Biotechnology Information; http://www.cnbe.nlm.nih.gov/). Such an approach could theoretically be retrospectively applied to existing methods in LCA (e.g. land use plant sampling data) to create PD characterization factors in LCA when identities of species are known from the original research. A synthesis of methods and approaches employed in molecular systematics to derive phylogenies is given in San Mauro and Agorreta (2010).
A.2 Species-based indicators Indicators that describe the trend or condition in single species or their attributes (e.g. abundance, occurrence, fitness, breeding rate, distribution) are considered in this article collectively as species-level indicators. This applies whenever the trend in a reduced set of species is considered a proxy for changes across all species. The advantage of taking a species- level approach is that information requirements are reduced relative to taking all species into consideration. Additionally, more information is present because each species’ response is assessed independently, and the results are often intuitive for conveying information (Lamb et al., 2009). In contrast, community-level indicators, discussed below, only show trends in a sample from a biological community (e.g. species richness or Shannon-Wiener index), and information is reduced to a single numerical value for the sample. Community indices such as the Shannon- Weiner index may be difficult to interpret to non-biologists. The distinction, advantages and disadvantages of the two approaches are discussed in further detail in section 3, with examples of both types of approaches highlighted in practice.
Species-level methods include the focal species approach (Lambeck, 1997), which selects a few species to act as “umbrellas” because of their habitat requirements or sensitivity to a particular pressure. If the most sensitive and relevant species are used as indicators, then the thresholds for the majority of other species for the particular threat or pressure is assumed to be safely below acceptable thresholds. But this assumes that sufficient knowledge is present to determine which species act as suitable indicators for all species (Ozinga & Schaminée, 2005). In practice, the focal species approach is controversial (Carignan & Villard, 2002; Lindenmayer et al., 2002; Caro, 2003). A multi-species’ approach, that incorporates many species as indicators, is more reflective and useful for reflecting general trends across all species than the use of specific focal species (Roberge & Angelstam, 2004).
Species-level approaches also include “intactness indices” (Lamb et al., 2009; Scholes & Biggs, 2005; Alkemade et al., 2009; Faith et al., 2008; Rouget et al., 2006). These describe changes in the population intactness of individual species with reference to some baseline or reference state
169 Appendix A: Appendix to Chapter 2
(giving an index ranging from 0, complete population loss, to 1, population equal to the reference state). Intactness indices are generally averaged across a set of species within a taxonomic group to yield more general assumptions about changes in biodiversity (Lamb et al., 2009). One assumption is that a reference state does indeed exist, and is often approximated by using historical monitoring data (Loh et al., 2005) or pristine habitat (Scholes & Biggs, 2005; Alkemade et al., 2009). This may be criticised as inferring a “nature constant” view of biodiversity that ignored inherent variation. But reference points are often chosen based on measurable and changeable areas along with measuring the population of interest. For example, taking populations inside natural protected areas as a reference for species intactness outside protected areas allows integration of the temporal variation in population size within reference habitat as long as monitoring persists in the reference area (e.g. Loh et al., 2005). One criticism that holds more merit is that intactness focuses on quantities of individuals or populations of single species, rather than the representation of the variety of species that is more in line with the definition of biodiversity given by the CBD, and represented by quantifying the percentage of lost species for example (Faith et al., 2008). This criticism can be levelled at any species-based approach, and methods must be designed to specifically remove the possibility of a widespread dominant species biasing or obscuring losses to more rare or sensitive species when individual species responses are averaged or combined in some way.
More recent methods applying at the level of the individual species include establishing a Habitat Suitability Index (HSI) which is used to model micro- or macro-distribution of species, i.e. species dispersion across habitat, or range across ecosystems respectively. HSI can be constructed using an inductive or deductive process (Corsi et al., 2000). Inductive approaches use raw data on species occurrence or abundance and associated environmental parameters. Patterns within the data are extracted via multivariate statistics, and attributed to changes in particular variables of interest. Species distribution modelling, using complex statistical procedures, is developing very rapidly as a means of constructing an inductively derived HSI based on the climate envelope of the species (Rondinini & Boitani, 2006; Austin, 2007). A range of methods have been developed to model ranges based on occurrence data (Elith et al., 2006). In LCA, the method to construct characterization factors for acidification inductively modelled species occurrence as a function of nitrogen loading (van Zelm et al., 2007). Nitrogen deposition was modelled across Europe and estimated Base Saturation (BS) levels were used to predict the probability of occurrence of 240 species of forest plant while statistically accounting for the effects of other environmental variables. Inductive approaches do not require a mechanism to be proposed (i.e. why nitrogen deposition causes a species to be absent) and lead to continuous suitability measures rather than suitability classes.
Deductive methods employ well-established empirical relationships or expert judgement to estimate the effect of changing environmental conditions on biodiversity. Meta-analysis may be used to establish standard effect sizes that are assumed to apply across unknown data ranges. Species sensitivities or habitat associations may be based on life history information and expert judgement, which often produces discreet suitability measures that require habitat to be split into vegetation classes. Species are either present or absent in habitat, or habitat is classed into
170 Appendix A: Appendix to Chapter 2 grades of suitability. The BioScore tool (Louette et al., 2010) takes this approach in joining cause and effect across a number of pressure categories and indicator species. Another example of a deductive approach is the process of expert interviews and workshops that generated the distribution ranges for all known amphibian species in the Global Amphibian Assessment of the IUCN. The Natural Capital Project’s InVEST tool (Nelson et al., 2009) uses deductive species- habitat associations to predict the effects of land use change on species probability of persistence in a landscape. A similar approach is under development in LCA to predict the effects of land use change on biodiversity (Geyer et al., 2010). The choice between inductive and deductive approaches is largely a trade-off between data-availability and needs of the study (Ricklefs, 2004).
The above species-level indicators generally focus on the compositional and structural attributes of biodiversity. Some functional aspects at a species level can be measured using genetic indicators. These include using genetic indicators to monitor inbreeding and outbreeding trends, genetic bottlenecks, metapopulation structure, effective population size etc. Aspects of life- history such as growth, fecundity, feeding, morphology, nesting, breeding etc., are beyond the scope of this review.
A.3 Community indicators Community-level indicators, such as species richness, reduce a large amount of information across many species into a single numerical value. They describe the emergent properties that result when the ranges of individual species overlap. We define them here as characterizing and summarizing information across objective, non-biased samples from a locality or region. Such as quadrat sampling data of plant species in a meadow ecosystem, or the results of observational transects to monitor bird occurrence and abundance. This is usually results in samples that describe the number and relative abundances of species from a local community (or any other unit of measurement such as genera, family, functional group, guild etc.). These can be divided into univariate metrics that describe alpha diversity (at the level of a locality) and multivariate metrics that describe beta diversity (the similarity or turnover of species between localities). Gamma diversity is often considered to be the total diversity of a region encompassing numerous localities (Lamb et al., 2009). Local extinctions and colonizations lead to variation in the micro- distribution or dispersion of species in a particular habitat (Noss, 1990). Single localities therefore may lack many species that are found elsewhere in the same habitat and are recorded in gamma diversity, the sum of all localities within a region. It must be stressed that the concept of the discreet, closed, ecological community of characteristic organisms has become essentially a non- concept in biodiversity research (Rahbek, 2005; Ricklefs, 2004). There are only overlapping ranges. Where community indicators differ from species indicators is that they are unbiased samples across species, not targeted samples from single species. However, species indicators may be used to derive community-level data via, for example, layering individual species’ ranges to obtain estimates of species richness at a resolution equal to the modelling “grain” used to predict the species’ ranges (e.g. Beck & Kitching, 2006). However, this assumes all species are represented in the modelling approach, or that patterns in species richness of rare species is reflected in patterns of common species which are more accurately modelled.
171 Appendix A: Appendix to Chapter 2
Community indicators can be abundance- or occurrence-based depending on whether information on the relative abundance of species within a sample is considered. Abundance can be expressed in individuals, biomass, percentage cover, extent of occurrence etc. Species richness is the most simple and widely used index of community diversity because its units are easily conceptualized and widely published. As more weight is placed on the relative abundance of species within a sample, importance is shifted away from species richness to community “evenness” the antipode of species richness being measures of pure evenness of a community, and does not consider the number of species). Intermediate points between richness and evenness include the Shannon-Wiener index, which values each species by its relative abundance, and the Guini-Simpson index, which weighs each species to the square of its relative abundance (Hill, 1973). Beta diversity can also be differentiated into abundance-based and incidence-based and illustrate the ecological distance or similarity between local communities. Beta diversity patterns over a large spatial scale are less well studied than extrapolations of local species richness (Beck & Chey, 2007). Yet, beta diversity patterns are important regional indicators of both biological value, and pressures such as the extent of biotic homogenization due to habitat degradation or land use change (Koellner et al., 2004).
At least two major problems affect biodiversity data availability and reliability. Firstly, a lack of standardized methodologies during surveys, and ongoing developments in sampling methodology, has lead to outdated information and a poor ability to compare species lists across sites. This generally results from different sampling techniques being used, therefore different sections of communities are sampled, and differences between localities may be an artefact rather than a true pattern. Secondly, the failure to account for undersampling of local communities has lead to unreliable species occurrence data across many regions (i.e. not detecting the full complement of species because of the “rare” element that requires extreme sampling effort to detect; Southwood & Henderson, 2000). In recognition of the second point, a large body of research has been devoted accounting for undersampling (e.g. by rarefaction; Hurlbert, 1971), or estimating the unseen species based on the structure of sampling data. This includes species richness estimators that either fit a statistical distribution to the sampling data (parametric methods) or use the proportion of rare and common species in a collection of samples to infer how many more species are likely to be detected with future sampling (non- parametric methods; Colwell & Coddington, 1994; Chao, 2005; Brose et al., 2003). Applying estimators to available data may lead to more reliable values for species richness, but information on the identity of species is lost, which can be informative in considering effects on certain species groups (e.g. threatened species; Koellner & Scholz, 2008) or incorporating beta diversity (Koellner et al., 2004). Estimators have traditionally been applied to estimate local community species richness, but large-scale regional applications have yielded promising results (e.g. Desmet & Cowling, 2004; Beck & Kitching, 2007).
Parallel to research in modelling individual species distributions, methods in mapping “emergent biodiversity” have been developed recently to use community data such as species richness, endemism, and turnover, rather than the climate envelopes of single-species and their distribution ranges (Faith & Walker, 1996; Ferrier et al., 2004; Arponen et al., 2008). Elith et al.
172 Appendix A: Appendix to Chapter 2
(2006) provides a comparison of “species distribution” and “community modelling” approaches developed in recent years. Because mapping individual species ranges is data intensive (especially for rarely recorded species), using available checklists to create community indicators shortcuts the modelling process. However, as highlighted above, information on individual species is lost in the transition. However, community modelling is a promising approach towards mapping poorly studied regions where distribution data across all species is not readily available (Ferrier et al., 2004). Spatial mapping of beta diversity patterns was used recently to estimate biodiversity loss due to land use change in Madagascar (Allnutt et al., 2008). This goes beyond the non-spatial and deductive approach of the species area relationship, but is less data demanding that using individual species distribution ranges to infer extinction risk (e.g. Jetz et al., 2007). A combination of spatial turnover in species from point data and continuous environmental variables was combined to spatially map species dissimilarity across Madagascar using general dissimilarity modelling (Ferrier et al., 2007). This was used to estimate the percentage of species lost to land conversion between 1950 and 2000 (Allnutt et al., 2008). Of significance is the ability to spatially predict past hotspots of endemism, turnover, and species richness on converted land based only on readily available environmental information (see “Environmental Diversity” or ED; Faith & Walker, 1996).
The above-mentioned indicators generally focus on the composition of local and regional communities, but say little about physical structure or ecological function. Structural indicators at the local scale describe the complexity of habitat. This includes vegetation density, vertical layering and spatial microdistribution of organisms or species, canopy intactness, physiognomy or growth form etc. (see table 2.1 of the main text). Structural information may also be approximated using remote sensing and GIS at variable resolutions, which can be very useful in scaling from local community to ecosystem and landscape level (Foody, 2003; Cannon et al., 2007). This includes canopy cover and structure, gap density, habitat heterogeneity and even the identity and composition of tree species based on their canopy dome signature (Foody & Cutler, 2006).
Using community species lists, classification of species by functional trait has been used to infer functional group diversity (FD) using life-history information and species lists (Petchey & Gaston, 2006). Combining species and community indicators, Thuiller et al. (2006) modelled the distribution of 122 European tree species under various climate change scenarios. Change in FD was related to changes in emergent patterns resulting from range reductions and expansions across species (Thuiller et al., 2006). Continuous indicators of functional diversity can be generated using a “distance in functional trait space” approach between species of a sample or community (Petchey & Gaston, 2006). Trophic diversity, food chain length, and guilds are also used to define functional diversity. Elsewhere, functional diversity has been approximated using indicators of genetic diversity. Cadotte et al. (2008) found phylogenetic diversity (PD) to be a better metric and predictor of functional diversity, expressed as primary productivity in a local community, than either species richness or FD. GIS is also used to map functional ecosystem processes such as fire and disturbance regimes, canopy moisture and phenological change (Foody, 2003). The Normalized Difference Vegetation Index (NDVI) has been used extensively at a
173 Appendix A: Appendix to Chapter 2 range of scales, for example to map productivity and moisture patterns for drought and hydrologic regime classification.
A.4 Species or community indicators? Theoretically, the difference in results between species-level indicators and community-level indicators is likely to disappear as more data becomes available across a more species, eventually covering all species. But until it is possible to model the response of each species independently, and cover the full number of species in a region, there is always an assumption that the species which are not assessed react in the same way as species that are assessed. Recent approaches that monitor changes in biodiversity across large spatial or temporal scales utilize very large numbers of species as indicators and are more likely to reflect changes across all species. For example, Thomas et al. (2004) modelled the effect of climate change by assessing how the distribution range of over 1000 species would change under different climate scenarios. A community-level approach would estimate the number of species in a particular habitat area (e.g. Afrotropical rainforest vascular plants), model the change in area due to climate change, and then apply a relationship to calculate the estimated species loss, such as the species area relationship. This approach was used in the Millennium Ecosystem Assessment (Van Vuuren et al., 2006). Similarly, Jetz et al. (2007) modelled the effect of climate change and habitat loss on bird species richness using a species-level approach. The study assessed how the two pressures would affect the range size of all known bird species using deductively mapped distribution ranges (from Orme et al., 2006). They calculated the resulting extinction risk on a per species basis, averaging to a percentage loss of species. A community-level approach would delineate habitat borders of major ecosystems or biomes, and species richness through local and regional checklists. The effect of pressures in reducing the habitat area would then be related to expected species losses, e.g. via the species area relationship or through community modelling of spatial alpha, beta and endemism patters (e.g. Allnutt et al., 2008).
A.5 Ecosystem and landscape indicators At larger spatial scales, indicators that reflect the extent, condition, threat level and biological importance of ecosystems, landscapes or regions are termed “ecosystem indicators”. They may be based on nested combinations of landscape indicators of ecosystem coverage, community indicators from point localities, species indicators of distribution or sensitivity, or even indicators of genetic diversity. It is important to understand how local impacts will affect regional biodiversity. This may reflect the state of ecosystem-level biodiversity, such as in the amount of remaining natural habitat relative to ecological requirements (Fahrig, 2001; Olson & Dinerstein, 2002; Potapov et al., 2008), the size of habitat patches within the landscape (Swift & Hannon, 2010), the level of fragmentation and connectivity (Herzog et al., 2001), ecosystem structural integrity and composition (Cannon et al., 2007), or levels of endemism, turnover, species richness and phylogenetic diversity (Burgess et al., 2006; Faith et al., 2004). Functional aspects can also be
174 Appendix A: Appendix to Chapter 2 reflected at the ecosystem level either scaled upwards from finer indicators, or modelled independently from remote sensing data (e.g. NDVI and its applications; Foody, 2003).
The relationship between structural landscape patterns (i.e. how humans perceive landscape structure) and their effect on biological processes (i.e. how the pattern affects species and assemblages) remains poorly understood (Czajkowski et al., 2009). The development of metrics of landscape pattern has for the most part stabilized, and the links to landscape processes has focused heavily on area loss and fragmentation (Fahrig, 2001; Turner, 2005; Swift & Hannon, 2010). Major metrics of pattern calculated by the FRAGSTAT program are classed into the following categories (Herzog et al., 2001):
Area metrics: Area in the landscape; Landscape similarity metrics; Class area; Percent of Landscape; Total landscape area; Largest patch index.
Patch density, patch size and variability metrics: Number of patches; Patch density; Mean patch size; Patch size standard deviation; Patch size coefficient of variation.
Edge metrics: Perimeter; Edge contrast index; Total edge; Edge density; Contrast-weighted edge density; Total edge contrast index; Mean edge contrast index; Area-weighted mean edge contrast index.
Shape metrics: Shape index; Fractal dimension; Landscape shape index; Mean shape index; Area- weighted mean shape index; Double log fractal dimension; Mean patch fractal dimensions; Area- weighted mean patch fractal dimension.
Core area metrics: Core area; Number of core areas; Core area index; Core area percent of landscape; Total core area; Number of core areas; Core area density; Mean core area per patch; Patch core area standard deviation; Patch core area coefficient of variation; Mean area per disjunct core; Disjunct core area standard deviation; Disjunct core area coefficient of variation; Total core area index; Mean core area index.
Nearest-neighbour metrics: Nearest-neighbor distance; Proximity index; Mean nearest neighbour distance; Nearest-neighbor standard deviation; Nearest-neighbor coefficient of variation; Mean proximity index.
Landscape processes are generally inferred using species or community indicators, such as by modelling single species dispersal and persistence in the landscape, or correlating landscape patterns with community metrics of biodiversity such as species richness and turnover. Recent work has illustrated how some pattern metrics are highly correlated with biodiversity processes in terms of species and communities (Fischer & Lindenmayer, 2007). Particularly important metrics at the ecosystem and landscape scale are the amount and condition of natural vegetation, the number of anthropogenic edges (Harper et al., 2005), and the configuration of patches (Fischer & Lindenmayer, 2007). However, there is still a need to further identify the abiotic and biotic processes that cause landscape patterns to emerge such as human land use, environmental change or natural disturbance, quantify the effect of patterns on functional and compositional biological processes, and elucidate the importance of such patterns and
175 Appendix A: Appendix to Chapter 2 heterogeneity in maintaining ecosystem and landscape biodiversity (Turner, 2005; Fischer &Lindenmayer, 2007).
The above mentioned pattern metrics are generally based on discreet classifications of land use, and continuous variables may also be used to characterize effects on biodiversity. These include continuous vegetation cover, patterns in processes such as primary productivity, carbon or nitrogen mineralization and other environmental variables (Turner, 2005). Abiotic functional attributes of biodiversity are often estimated using remote sensing at the ecosystem level combined with environmental, geomorphic or hydromorphic modelling, or extrapolating from local data across ecosystem or biomes (e.g. Bouwman et al., 2002; Elser et al., 2007; Bobbink et al., 2010).
The most frequent approach towards modelling biodiversity loss at ecosystem and landscape scales has been through the species area relationship (Arrhenius, 1921; Rosenzweig, 1995). Recent techniques are incorporating landscape pattern metrics into the SAR in order to represent the biodiversity value of non-natural habitat (Pereira & Daily, 2006), to account for fragmentation and patch size (Nelson et al., 2008, 2009) and to reflect species’ varying affinities to different landscape elements (Koh & Ghazoul, 2010). This is by definition deductive, and relies on assumptions drawn from meta-study, landscape modelling or expert judgement. Results may be tested for accuracy using observed extinctions in empirical data (e.g. Koh & Ghazoul, 2010; Kinzig & Harte, 2000).
A.6 Integrative indicators Indicators that combine information across multiple attributes or components of biodiversity are termed integrative or multimetric indicators (EPA, 2003, 2008; Niemi & McDonald, 2004; Karr, 1981; Karr & Chu, 1997). Integrative indicators mathematically aggregate or weigh different attributes of biodiversity and express this in a single output. The Index of Biotic Integrity (IBI) was developed to monitor the health of aquatic ecosystems in the US using compositional and functional indicators including species richness, functional groups and indicator taxa sensitive to stress Karr (1981). It has since been taken up by the US EPA (2002) for wider national use, and adapted to numerous ecosystem types and taxa including birds and landscape diversity (O’Connell et al., 2000) and coral reefs (Jameson et al., 2001). Parkes et al. (2003) “habitat hectares” approach additively combines multiple indicators at the species and community level across compositional, functional and structural attributes. A similar aggregation and weighing approach is used in assessing WWF Ecoregion biological value and conservation status on an ecosystem level (Olson & Dinerstein, 1998). This produces a Biological Distinctiveness Index (BDI) reflecting biological value, and Conservation Status Index (CSI) reflecting pressures and vulnerability. Likewise, Important Bird Areas (IBAs) are monitored by aggregating biodiversity indicators across attributes (habitat area, quality and number of populations and trends of threatened “trigger” species) and weighing this against aggregated pressure indicators (agricultural intensification, pollution trends, population density and land use change; Bennun et al., 2005). One potential setback of integrative indicators is that they rely on a subjective scoring system for different attributes and therefore must be tested extensively Karr & Chu (1997).
176 Appendix A: Appendix to Chapter 2
However, the ability to not only reflect declines in biotic condition, but also diagnose potential causes makes them a very powerful as monitoring tools. They can also be aggregated and combined across multiple spatial scales (Niemi & McDonald, 2004).
A.7 The use of indicators in biodiversity assessments We reviewed biodiversity assessment approaches in four models that have emerged recently to inform decision makers about the ecological consequences of public and private policy. We believe these models can serve as an example for how LCIA might develop in the future because they are 1) spatially explicit to some degree, 2) they attempt to model many pressures of biodiversity loss simultaneously, and 3) they demonstrate the use of the indicators and approaches discussed above. However, these models are incomplete with regards to modelling biodiversity, and in some cases even more limited than LCIA for certain pressures (such as ecotoxicity). Two of the models, the Millennium Ecosystem Assessment and the GLOBIO3 model, are global modelling tools, and therefore rough in the scale of impact assessment. The other two, the InVEST tool from the Natural Capital Project and the BioScore tool, are regional in their application. They therefore provide a selection of approaches towards dealing with issues of scale and data availability in biodiversity assessment. We believe this is particularly relevant to LCA which is positioned in a transition between site-generic and site specific impact factors (Hauschild & Potting, 2005) and between regional and global coverage (Bare, 2009).
A.7.1 The Millennium Ecosystem Assessment The Millennium Ecosystem Assessment (2005) modelled biodiversity loss on a global scale due to five direct drivers: terrestrial and aquatic habitat change, climate change, pollution, invasive species, and overexploitation. Terrestrial habitat change, i.e. land use, was modelled using a very coarse ecosystem indicator. Reductions in biome area were modelled according the four scenarios. Species richness of vascular plants, a community indicator, was extrapolated from local samples to estimate original biome species richness in each realm (Olson et al., 2001). The species area relationship was then used to estimate past and future extinctions according to the scenarios (Van Vuuren et al., 2006). Changes in biome coverage were calculated using the Integrated Model of Global Environmental Change (IMAGE; MNP, 2006) through four global policy scenarios (Van Vuuren et al., 2006). It relied heavily on the species area relationship (SAR) and differentiated between local extirpation and regional/global extinctions. Local extirpation was considered reversible, but global losses were not (i.e. losses predicted by the SAR as habitat area declines to low levels). In a separate analysis, but using the scenarios of the MA, Jetz et al. (2007) used range maps of all known bird species (Orme et al., 2006) to model the effects of climate change and land conversion on global avian diversity. This used a species-level approach by calculating the extinction risk for each species due to range reductions.
The effects of water abstraction on biodiversity were modelled in the MA using catchment discharge rate as a predictor of species richness of fish across 237 river basins worldwide (Oberdorff et al., 1995; Poff et al., 2001; Xenopoulos et al., 2005). This was employed to create a cause-effect relationship between water use which was estimated with the WaterGap model (Alcamo et al., 2003) and species loss. The species-discharge relationship is convenient because it
177 Appendix A: Appendix to Chapter 2 resembles the SAR in shape (power function where the marginal loss of species increases exponentially as water discharge declines to zero). River discharge approximates river basin ecological space and habitat heterogeneity, producing a species-discharge relationship that resembles the SAR (Xenopoulos et al., 2005). Since global data on fish endemism is not available, only extinctions from individual river basins were considered, i.e. regional impacts in LCA terminology (MA, 2005). The model was updated with recent data from FishBase (http://www.fishbase.org/). Alcamo et al.’s (2003) WaterGAP model was used to estimate future discharge rates due to water abstraction. A lack of data for other taxa (e.g. crustaceans, mussels, invertebrates), and a lack of a cause-effect models for aquatic habitats and vegetation communities (including Ramsar sites; www.ramsar.org) calls for more work to be done in this area (MA, 2005).
The effects of climate change on terrestrial and aquatic biodiversity was incorporated via the above methods to estimate land and water use. Biome contractions and river discharge changes expected over the IPCC climate change scenarios was combined with the species area and species discharge relationship (for plants and fish respectively; Van Vuuren et al., 2006; Xenopoulos et al., 2005). The marine effects of climate change were limited to qualitative estimates. These include loss of corals and calcium dependent organisms from rising sea temperature and declining pH (Hughes et al., 2003; Pandolfi et al., 2003).
Acidification and eutrophication was covered in the MA (2005) using meta-analyses of empirical studies to construct the cause-effect relationship between deposition of acidifying substances and species loss. Bouwman et al. (2002) constructed a global map of critical load values for both acidification and eutrophication based on FAO soil and GLC2000 land cover data. Exceedance ratios were calculated for each geographic region based on modelled deposition of sulphur and nitrogen on a 0.5° x 0.5° grid size (S72). The relationship between nitrogen exceedance and species richness of a variety of taxa across a ecosystem types and habitats was used to quantify species loss (Bobbink et al., 2010; Schindler et al., 1985; Bobbink et al., 1998; Haddad et al., 2000; Vinebrooke et al., 2003; Stevens et al., 2004; Bobbink, 2004). There is generally a non-linear relationship between CL exceedance and species richness (Bobbink, 2004; Alkemade et al., 2009). Freshwater and marine ecosystems were omitted from quantitative analyses of acidification and eutrification in the MA as current knowledge is generally qualitative in nature (Elser et al., 2007; Elser & Urabe, 1999; Downing & McCauley, 1992; Smith, 2003). Further development of dose- response models based on meta-analysis of existing studies might yield appropriate damage factors (but see Weijters et al., 2009).
Marine fisheries depletion was used as a proxy for overexploitation in the MA (2005). It employed regional case-studies and ecosystem models (the Ecopath and Ecosim fisheries models, http://www.ecopath.org/). Quantitative damages to marine biodiversity in three regions (Gulf of Thailand, coastal shelf; Benguella Current, upwelling; Central North Pacific, pelagic) were extrapolated across the globe (MA, 2005). The estimations relate species losses via biomass declines in exploited trophic groups (functional indicator) to species losses (compositional indicator) via an index, Kempton’s Q (Kempton, 2002). Other marine ecosystems have only been
178 Appendix A: Appendix to Chapter 2 quantitatively described (deep sea, Glover & Smith, 2003; polar, Clarke & Harris, 2003; and vents/sea mounts, Koslow et al., 2001). Finally, the impact of invasive species was assessed qualitatively across the MA scenarios, using globalized trade as a proxy for invasion risk. Expert opinion was used to gauge the importance of the driver in context to other drivers of biodiversity loss.
A.7.2 IMAGE 2.4 and the GLOBIO3 model The GLOBIO3 model (Alkemade et al., 2009) is an integrated, spatially explicit global assessment tool to investigate biodiversity change due to six terrestrial drivers of loss: land use, infrastructure, overexploitation, fragmentation, nitrogen deposition (pollution) and climate change. It is the latest tool to be implemented into the IMAGE model of environmental and economic change. It uses meta-analysis of published studies to establish cause-effect relationships under a species-level, intactness index called the Mean Species Abundance (MSA). The MSA is defined as the average fractional abundance decline of native species relative to their abundance in a pristine environment (Alkemade et al., 2009). In actual fact, the MSA is approximated by species, community and ecosystem indicators according to the driver investigated. For land use, infrastructure and overexploitation, species-level change (intactness) is taken as a proxy for diversity. For fragmentation and nitrogen deposition, percentage loss of the community is used (community-level). For climate change, ecosystem-level effects on species richness are expressed via the SAR, combined with species-level measures using climate- envelope modelling and range size changes (Alkemade et al., 2009).
The effects of land use and infrastructure was modelled using the MSA in each grid cell, and meta-study results to assign MSA values to each land use type. Fragmentation is further included by applying a weight on the MSA of land patches the fall below a critical minimum area of 103 km2 (from Verboom et al., 2007). Water use was not considered. Climate change compared two approaches: large scale biome area changes and species loss via the SAR in a similar approach to the MA (Leemans & Eickhout, 2004; Van Vuuren et al., 2006), and species-based distribution modelling using EUROMOVE (Bakkenes et al., 2002, 2006). MSA values are approximated by loss of species due to habitat loss or range size reduction. Pollution was modelled in the same was as the MA, using critical loading (CL) values for nitrogen deposition, and a meta-analysis of species loss due to CL exceedance (see above). Overexploitation of terrestrial species was considered a form of land use. Impact zones were established around infrastructure and roads in areas prone to overharvesting (e.g. tropical and sub-tropical forests) which effectively represent modifications to the existing land use classes. This reduces the biodiversity value (expressed as MSA) of the existing land class around such infrastructure. Invasive species were not considered. The GLOBIO3 model (Alkemade et al., 2009) did not consider aquatic biodiversity loss, but will include both freshwater environments and marine aquatic ecosystems in the future, using a specific set of drivers of aquatic biodiversity loss (see http://www.globio.info/).
A.7.3 The Natural Capital Project’s InVEST modelling tool The InVEST tool (Nelson et al., 2009, 2008) estimates changes in landscape biodiversity and ecosystems services resulting from only one driver of biodiversity loss: land use change. The
179 Appendix A: Appendix to Chapter 2 model is applicable on an Ecoregion scale (Olson et al., 2001) and utilizes as indicator species prominent macro-vertebrates or any higher order species group where regional data is available. Biodiversity loss is based on species- and ecosystem-level indicators of suitable habitat area using the HSI concept. Countryside SARs (Pereira & Daily, 2006) are used to aggregate habitat area on a per species basis and biodiversity is expressed as the ratio of existing habitat given a particular land use pattern to maximum potential area under natural conditions. Indefinite persistence in the landscape is estimated through the countryside SAR of each species (Nelson et al., 2009). Additionally, dispersal ability and spatial configuration of habitat may be integrated into the model using landscape pattern metrics of configuration and fragmentation (see above).
A.7.4 The BioScore tool for European biodiversity assessment The BioScore tool Louette et al. (2010) is a recently developed spatial biodiversity model designed to predict broad policy impacts on the state of European biodiversity. It assessed the effect of multiple pressures on species persistence across Europe. In total, 37 pressures are available in the BioScore database across categories including land use change, pollution, water quality and availability, climate change, fragmentation, disturbance, direct harvesting, interspecific interactions, and forest management (Delbaere et al., 2009). It utilizes a species-level approach and incorporates a wide range of “focal species” (Lambeck, 1997) across mammals, reptiles, amphibians, birds, butterflies, vascular plants and freshwater fish (Louette et al., 2010). The tool was designed around the “Driver-Pressure-State-Impact-Response (DPSIR) framework” (Spangenberg et al., 2009) to translate the effects of policy decisions (Driver) in changing environmental variable (Pressure). This has an effect on the distribution of sensitive species (State) which causes an impact to biodiversity (Impact). Sensitivity scores and HIS are assigned to species using deductive methods (i.e. life history information and expert opinion) and four discreet sensitivity classes (using the criteria of Maes & Van Dyck, 2005). This information is used to model distributional changes across Europe due to broad scale policy decisions, and the concurrent effects on species persistence and extinction risk.
A.8 Representing the missing drivers of biodiversity loss Two drivers of biodiversity loss in the MA, invasive species (biotic homogenization) and overexploitation (biotic depletion), are currently not represented in LCA.
A.8.1 Biotic depletion Overexploitation of wild populations is recognized as one of the principal threats to global biodiversity across both aquatic and terrestrial ecosystems MA (2005). In LCA, this is addressed as biotic depletion. A first conceptual approach to incorporate biotic depletion in LCA was based on the deaccumulation rate and total size of wild populations (Guinée et al., 2002), and applied to case studies of fisheries (Ziegler et al., 2003; Nilsson & Ziegler, 2007). A global approach covering a large range of species, and including indirect effects on the food chain is currently lacking.
The MA used regional case-studies and ecosystem models (Ecopath with Ecosim, http://www.ecopath.org/) to quantify species loss to marine biodiversity in three regions covering major oceanic zones, and then extrapolated across the global marine environment (MA, 2005).
180 Appendix A: Appendix to Chapter 2
Biodiversity loss was estimated via a functional indicator, the biomass declines in exploited trophic groups. Kempton’s Q index (Kempton, 2002) was used to convert biomass declines into the compositional indicator of species loss. Other marine ecosystems were only qualitatively described. Halpern et al. (2008) constructed global, quantitative maps of marine overexploitation due to fishing that could guide impact factor construction.
For terrestrial ecosystems, the GLOBIO3 model (Alkemade et al., 2009) and the Biodiversity Intactness Index (Scholes & Biggs, 2005) both consider harvesting as a form of land use. The former maps impact zones around infrastructure and roads in areas prone to overharvesting. The latter classifies natural areas into moderately used and degraded status. Both approaches use species level intactness indices, expressing the population size of species as a function of their abundance in pristine areas, and use regional data and remote sensing to define ecosystems and land use patterns. Such an approach could be employed in LCA to adjust regionalized land use methods to account for terrestrial biotic depletion.
A.8.2 Biotic homogenization Invasive species influence the composition and species richness of exposed ecosystems, rivalling other global drivers of biodiversity loss (Clavero & García-Berthou, 2005). Cause-effect modelling of invasive species as a separate impact category is problematic in the current framework of LCA because invasion occurs through isolated outbreaks rather than continuous marginal impacts. Indirect incorporation through other impact categories is an option (Jolliet et al., 2004). Land use impacts that take species turnover into account may weigh the presence of invasive species with zero, or negative, value (Jolliet et al., 2004).
Targeting dispersal pathways as a proxy for invasion risk is a second approach. The MA (2005) used global trade as a coarse proxy for invasion risk based on expert opinion. Recent introductions of aquatic invasive species have predominantly occurred through the release of biocontaminated ballast water, fouling of ship hulls, and the creation of waterways (Ricciardi & MacIsaac, 2000). Terrestrial transport networks and other land classes act as transport vectors for invasive plant species (Rodríguez-Labajos et al., 2009). To establish cause-effect relationships, a search of relevant studies is needed that (i) link inventory flows with increased invasion risk, and (ii) express species loss as a function of invasion risk. The first requirement could be based on studies that correlate the prevalence of invasive species with inventory flows (i.e. construction of transport networks or transport distances). The second could use meta-analysis of empirical studies, such as Levine et al. (2003).
A.9 References Alcamo, J., Döll, P., Henrichs, T., Kaspar, F., Lehner, B., Rösch, T., & Siebert, S. 2003. Development and testing of the WaterGAP 2 global model of water use and availability. Hydrological Sciences Journal, 48(3), 317–337. Alkemade, R, van Oorschot, M, Miles, L, Nellemann, C, Bakkenes, M, & ten Brink, B. 2009. GLOBIO3: A Framework to Investigate Options for Reducing Global Terrestrial Biodiversity Loss. Ecosystems, 12(3), 374–390. Allnutt, T. F., Ferrier, S., Manion, G., Powell, G. V. N., Ricketts, T. H., Fisher, B. L., Harper, G. J., Irwin, M. E., Kremen, C., Labat, J. N., Lees, D. C., Pearce, T. A., & Rakotondrainibe, F.
181 Appendix A: Appendix to Chapter 2
2008. A method for quantifying biodiversity loss and its application to a 50-year record of deforestation across Madagascar. Conservation Letters, 1(4), 173–181. Arponen, A., Moilanen, A., & Ferrier, S. 2008. A successful community-level strategy for conservation prioritization. Journal of Applied Ecology, 45(5), 1436– 1445. Arrhenius, O. 1921. Species and area. The Journal of Ecology, 95–99. Austin, Mike. 2007. Species distribution models and ecological theory: a critical assessment and some possible new approaches. Ecological modelling, 200(1), 1–19. Bakkenes, M., Alkemade, J. R. M., Ihle, F., Leemans, R., & Latour, J. B. 2002. Assessing effects of forecasted climate change on the diversity and distribution of European higher plants for 2050. Global Change Biology, 8(4), 390–407. Bakkenes, M., Eickhout, B., & Alkemade, R. 2006. Impacts of different climate stabilisation scenarios on plant species in Europe. Global Environmental Change, 16(1), 19–28. Bare, J. 2009. Life cycle impact assessment research developments and needs. Clean Technologies and Environmental Policy, 12(4), 341-351. Beck, J., & Chey, V. K. 2007. Beta-diversity of geometrid moths from northern Borneo: effects of habitat, time and space. Journal of Animal Ecology 2007, 76, 230–237. Beck, J., & Kitching, I. J. 2006. Determinants of regional species richness: an empirical analysis of the number of hawkmoth species (Lepidoptera: Sphingidae) on the Malesian archipelago. Journal of Biogeography, 33(4), 694– 706. Beck, J., & Kitching, I. J. 2007. Estimating regional species richness of tropical insects from museum data: a comparison of a geography-based and sample- based methods. Journal of Applied Ecology, 44(3), 672–681. Bennun, L., Matiku, P., Mulwa, R., Mwangi, S., & Buckley, P. 2005. Monitoring Important Bird Areas in Africa: Towards a Sustainable and Scaleable System. Biodiversity and Conservation, 14(11), 2575–2590. Bobbink, R. 2004. Plant species richness and the exceedance of empirical nitrogen critical loads: An inventory. Utrecht University, Netherlands. Bobbink, R., Hornung, M., & Roelofs, J. G. M. 1998. The effects of air-borne nitrogen pollutants on species diversity in natural and semi-natural European vegetation. Journal of Ecology, 86(5), 717–738. Bobbink, R., Hicks, K., Galloway, J., Spranger, T., Alkemade, R., Ashmore, M., Bustamante, M., Cinderby, S., Davidson, E., Dentener, F., Emmett, B., Erisman, J-W., Fenn, M., Gilliam, F., Nordin, A., Pardo, L., & De Vries, W. 2010. Global assessment of nitrogen deposition effects on terrestrial plant diversity: a synthesis. Ecological Applications, 20(1), 30–59. Bouwman, A., Van Vuuren, D., Derwent, R., & Posch, M. 2002. A Global Analysis of Acidification and Eutrophication of Terrestrial Ecosystems. Water, Air, & Soil Pollution, 141(1), 349– 382. Brose, U., Martinez, N. D., & Williams, R. J. 2003. Estimating species richness: sensitivity to sample coverage and insensitivity to spatial patterns. Ecology, 84(9), 2364–2377. Burgess, N. D., Hales, J. D., Ricketts, T. H., & Dinerstein, E. 2006. Factoring species, non- species values and threats into biodiversity prioritisation across the ecoregions of Africa and its islands. Biological Conservation, 127(4), 383–401. Cadotte, M. W., Cardinale, B. J., & Oakley, T. H. 2008. Evolutionary history and the effect of biodiversity on plant productivity. Proceedings of the National Academy of Sciences, 105(44), 17012–17017. Cannon, C. H., Summers, M., Harting, J. R., & Kessler, P. J. A. 2007. Developing conservation priorities based on forest type, condition, and threats in a poorly known ecoregion: Sulawesi, Indonesia. Biotropica, 39(6), 747–759. Carignan, V., & Villard, M.-A. 2002. Selecting Indicator Species to Monitor Ecological Integrity: A Review. Environmental Monitoring and Assessment, 78(1), 45–61. Caro, T. M. 2003. Umbrella species: critique and lessons from East Africa. Animal Conservation, 6(02), 171–181. Chao, A. 2005. Species richness estimation. Pages 7909–7916 of: Balakrishnan, N., Read, B., & Vidakovic, B. (eds), Encyclopedia of Statistical Sciences. New York: Wiley.
182 Appendix A: Appendix to Chapter 2
Clarke, A., & Harris, C. M. 2003. Polar marine ecosystems: major threats and future change. Environmental Conservation, 30(01), 1–25. Clavero, M., & García-Berthou, E. 2005. Invasive species are a leading cause of animal extinctions. Trends in Ecology & Evolution, 20(3), 110. Colwell, R. K., & Coddington, J. A. 1994. Estimating terrestrial biodiversity through extrapolation. Phil. Trans. R. Society. Lond. B, 345, 101–118. Conner, J. K., & Hartl, D. L. 2004. A primer of ecological genetics. Sunderland, MA, USA: Sinauer Associates. Corsi, F., de Leeuw, J., & Skidmore, A. 2000. Modeling Species Distribution with GIS. Research Techniques in Animal Ecology, 389-434. Czajkowski, Mikołaj, Buszko-Briggs, Małgorzata, & Hanley, Nick. 2009. Valuing changes in forest biodiversity. Ecological Economics, 68(12), 2910–2917. Delbaere, B, Serradilla, AN, & Snethlage, M. 2009. BioScore: A tool to assess the impacts of European Community policies on Europe’s biodiversity. Tech. rept. ECNC, Tilburg, the Netherlands. Desmet, P., & Cowling, R. 2004. Using the species-area relationship to set baseline targets for conservation. Ecology and Society, 9(2), 11. Downing, J. A., & McCauley, E. 1992. The nitrogen: phosphorus relationship in lakes. Limnology and Oceanography, 936–945. Elith, J., Graham, C.H., Anderson, R.P., Dudik, M., Ferrier, S., Guisan, A., Hijmans, R.J., Huettmann, F., Leathwick, J.R., Lehmann, A., Lohmann, L.G., Loiselle, B.A., Manion, G., Moritz, C., Nakamura, M., Nakazawa, Y., Overton, J., Townsend, M., Peterson, A., Phillips, S.J., Richardson, K., Scachetti-Pereira, R., Schapire, R.E., Soberon, J., Williams, S., Wisz, M.S., & Zimmermann, N.E. 2006. Novel methods improve prediction of species’ distributions from occurrence data. Ecography, 29, 129–151. Elser, J. J., & Urabe, J. 1999. The stoichiometry of consumer-driven nutrient recycling: theory, observations, and consequences. Ecology, 80(3), 735–751. Elser, J. J., Bracken, M. E. S., Cleland, E. E., Gruner, D. S., Harpole, W. S., Hillebrand, H., Ngai, J. T., Seabloom, E. W., Shurin, J. B., & Smith, J. E. 2007. Global analysis of nitrogen and phosphorus limitation of primary producers in freshwater, marine and terrestrial ecosystems. Ecology Letters, 10(12), 1135– 1142. EPA, US. 2002. A framework for assessing and reporting on ecological condition: A SAB report. Prepared by the Ecological Reporting Panel, Ecological Processes and Effects Committee. EPA Science Advisory Board. Washington, DC, 142. EPA, US. 2003. Developing Biological Indicators: Lessons Learned from Mid- Atlantic Streams. Tech. rept. EPA 903/R-003/003. U.S. Environmental Protection Agency, Office of Environmental Information and Mid-Atlantic Integrated Assessment Program, Region 3, Ft. Meade, MD. EPA, US. 2008. Report on the Environment (Final Report). Chapter 6: Ecological condition. Tech. rept. EPA/600/R-07/045F (NTIS PB2008-112484). U.S. Environmental Protection Agency, Washington, D.C. Fahrig, L. 2001. How much habitat is enough? Biological Conservation, 100(1), 65–74. Faith, D. P. 2002. Quantifying biodiversity: a phylogenetic perspective. Conservation Biology, 16(1), 248–252. Faith, D. P., & Walker, P. A. 1996. Environmental diversity: on the best-possible use of surrogate data for assessing the relative biodiversity of sets of areas. Biodiversity and Conservation, 5(4), 399–415. Faith, D. P., Reid, C. A. M., & Hunter, J. 2004. Integrating phylogenetic diversity, complementarity, and endemism for conservation assessment. Conservation Biology, 18(1), 255–261. Faith, D. P., Ferrier, S., & Williams, K. J. 2008. Getting biodiversity intactness indices right: ensuring that ’biodiversity’ reflects ’diversity’. Global Change Biology, 14(2), 207–217. Ferrier, S., Powell, G. V. N., Richardson, K. S., Manion, G., Overton, J. M., Allnutt, T. F., Cameron, S. E., Mantle, K., Burgess, N. D., Faith, D. P., Lamoreux, J. F., Kier, G., Hijmans,
183 Appendix A: Appendix to Chapter 2
R. J., Funk, V. A., Cassis, G. A., Fisher, B. L., Flemons, P., Lees, D., Lovett, J. C., & Van Rompaey, R. 2004. Mapping More of Terrestrial Biodiversity for Global Conservation Assessment. BioScience, 54(12), 1101. Ferrier, S., Manion, G., Elith, J., & Richardson, K. 2007. Using generalized dissimilarity modelling to analyse and predict patterns of beta diversity in regional biodiversity assessment. Diversity and Distributions, 13(3), 252–264. Fischer, J., & Lindenmayer, D. B. 2007. Landscape modification and habitat fragmentation: a synthesis. Global Ecology and Biogeography, 16(3), 265–280. Foody, G. M. 2003. Remote sensing of tropical forest environments: towards the monitoring of environmental resources for sustainable development. International Journal of Remote Sensing, 24(20), 4035–4046. Foody, Giles M., & Cutler, Mark E.J. 2006. Mapping the species richness and composition of tropical forests from remotely sensed data with neural networks. Ecological Modelling, 195(1-2), 37–42. Geyer, R., Stoms, D., Lindner, J., Davis, F., & Wittstock, B. 2010. Coupling GIS and LCA for biodiversity assessments of land use. Part 1: Inventory modeling. The International Journal of Life Cycle Assessment, 15, 454–467. Glover, A. G., & Smith, C. R. 2003. The deep-sea floor ecosystem: current status and prospects of anthropogenic change by the year 2025. Environmental Conservation, 30(03), 219–241. Guinée, J. B., Gorée, M., Heijungs, R., Huppes, G., de Koning, A., van Oers, L., Sleeswijk, A. W., Suh, S., & Udo de Haes, H. A. 2002. An operational guide to the ISO standards. Tech. rept. Leiden University (CML), The Netherlands. Haddad, N. M., Haarstad, J., & Tilman, D. 2000. The effects of long-term nitrogen loading on grassland insect communities. Oecologia, 124(1), 73–84. Hajjar, R., Jarvis, D. I., & Gemmill-Herren, B. 2008. The utility of crop genetic diversity in maintaining ecosystem services. Agriculture, Ecosystems & Environment, 123(4), 261– 270. Halpern, B. S., Walbridge, S., Selkoe, K. A., Kappel, C. V., Micheli, F., D’Agrosa, C., Bruno, J. F., Casey, K. S., Ebert, C., Fox, H. E., Fujita, Rod, Heinemann, Dennis, Lenihan, Hunter S., Madin, Elizabeth M. P., Perry, Matthew T., Selig, Elizabeth R., Spalding, Mark, Steneck, Robert, & Watson, Reg. 2008. A Global Map of Human Impact on Marine Ecosystems. Science, 319(5865), 948–952. Harper, K. A., Macdonald, S. E., Burton, P. J., Chen, J., Brosofske, K. D., Saunders, S. C., Euskirchen, E. S., Roberts, D., Jaiteh, M. S., & Esseen, P. A. 2005. Edge influence on forest structure and composition in fragmented landscapes. Conservation Biology, 19(3), 768–782. Hauschild, M., & Potting, J. 2005. Spatial Differentiation in Life Cycle Impact Assessment – the EDIP2003 Methodology. Tech. rept. The Danish Environmental Protection Agency, Copenhagen. Herzog, F., Lausch, A., Müller, E., Thulke, H.-H., Steinhardt, U., & Lehmann, S. 2001. Landscape Metrics for Assessment of Landscape Destruction and Rehabilitation. Environmental Management, 27(1), 91–107. Hill, M. O. 1973. Diversity and Evenness: A Unifying Notation and Its Consequences. Ecology, 54(2), 427–432. Hughes, A. Randall, & Stachowicz, John J. 2004. Genetic diversity enhances the resistance of a seagrass ecosystem to disturbance. Proceedings of the National Academy of Sciences of the United States of America, 101(24), 8998–9002. Hughes, A. Randall, Inouye, Brian D., Johnson, Marc T. J., Underwood, Nora, & Vellend, Mark. 2008. Ecological consequences of genetic diversity. Ecology Letters, 11(6), 609–623. Hughes, T. P., Baird, A. H., Bellwood, D. R., Card, M., Connolly, S. R., Folke, C., Grosberg, R., Hoegh-Guldberg, O., Jackson, J. B. C., Kleypas, J., Lough, J. M., Marshall, P., Nystrom, M., Palumbi, S. R., Pandolfi, J. M., Rosen, B., & Roughgarden, J. 2003. Climate Change, Human Impacts, and the Resilience of Coral Reefs. Science, 301(5635), 929–933.
184 Appendix A: Appendix to Chapter 2
Hurlbert, Stuart H. 1971. The nonconcept of species diversity: a critique and alternative parameters. Ecology, 52(4), 577–586. Jameson, Stephen C., Erdmann, Mark V., Karr, James R., & Potts, Kennard W. 2001. Charting a course toward diagnostic monitoring: A continuing review of coral reef attributes and a research strategy for creating coral reef indexes of biotic integrity. Bulletin of Marine Science, 69, 701–744. Jetz, W., Wilcove, D. S., & Dobson, A. P. 2007. Projected impacts of climate and land-use change on the global diversity of birds. PLoS Biology, 5(6), 1211–1219. Jolliet, Olivier, Müller-Wenk, Ruedi, Bare, Jane, Brent, Alan, Goedkoop, Mark, Heijungs, Reinout, Itsubo, Norihiro, Peña, Claudia, Pennington, David, Potting, José, Rebitzer, Gerald, Stewart, Mary, de Haes, Helias, & Weidema, Bo. 2004. The LCIA midpoint-damage framework of the UNEP/SETAC life cycle initiative. The International Journal of Life Cycle Assessment, 9(6), 394–404. Karr, J. R. 1981. Assessment of biotic integrity using fish communities. Fisheries, 6(6), 21–27. Karr, J. R., & Chu, E. W. 1997. Biological monitoring and assessment: Using multimetric indexes effectively. Tech. rept. EPA 235-R97-001. University of Washington, Seattle, USA. Kempton, R. A. 2002. Species diversity. Encyclopedia of environmetrics, 4, 2086–2092. Kinzig, A. P., & Harte, J. 2000. Implications of endemics-area relationships for estimates of species extinctions. Ecology, 81(12), 3305–3311. Koellner, T., & Scholz, R. W. 2008. Assessment of land use impacts on the natural environment. Part 2: Generic characterization factors for local species diversity in central europe. The International Journal of Life Cycle Assessment, 13(1), 32–48. Koellner, T., Hersperger, A. M., & Wohlgemuth, T. 2004. Rarefaction method for assessing plant species diversity on a regional scale. Ecography, 27(4), 532–544. Koh, L. P., & Ghazoul, J. 2010. A Matrix-Calibrated Species-Area Model for Predicting Biodiversity Losses Due to Land-Use Change. Conservation Biology, 24(4), 994–1001. Koslow, J. A., Gowlett-Holmes, K., Lowry, J. K., O’hara, T., Poore, G. C. B., & Williams, A. 2001. Seamount benthic macrofauna off southern Tasmania: community structure and impacts of trawling. Marine Ecology Progress Series, 213(111.125). Laikre, L. 2010. Genetic diversity is overlooked in international conservation policy implementation. Conservation Genetics, 11(2), 349–354. Lamb, E. G., Bayne, E., Holloway, G., Schieck, J., Boutin, S., Herbers, J., & Haughland, D. L. 2009. Indices for monitoring biodiversity change: Are some more effective than others? Ecological Indicators, 9(3), 432–444. Lambeck, R. J. 1997. Focal species: a multi-species umbrella for nature conservation. Conservation biology, 11(4), 849–856. Leemans, R., & Eickhout, B. 2004. Another reason for concern: regional and global impacts on ecosystems for different levels of climate change. Global Environmental Change Part A, 14(3), 219–228. Levine, J. M., Vila, M., D’Antonio, C. M., Dukes, J. S., Grigulis, K., & Lavorel, S. 2003. Mechanisms underlying the impacts of exotic plant invasions. Proceedings of the Royal Society of London, Series B: Biological Sciences, 270(1517), 775–781. Lindenmayer, D. B., Manning, A. D., Smith, P. L., Possingham, H. P., Fischer, J., Oliver, I., & McCarthy, M. 2002. The Focal-Species Approach and Landscape Restoration: a Critique. Conservation Biology, 16(2), 338. Loh, J., Green, R. E., Ricketts, T., Lamoreux, J., Jenkins, M., Kapos, V., & Randers, J. 2005. The Living Planet Index: using species population time series to track trends in biodiversity. Philosophical Transactions of the Royal Society B: Biological Sciences, 360(1454), 289– 295. Louette, G., Maes, D., Alkemade, R. M., Boitani, L., de Knegt, B., Eggers, J., Falcucci, A., Framstad, E., Hagemeijer, W., Hennekens, S. M., Maiorano, L., Nagy, S., Serradilla, A. N., Ozinga, W. A., Schaminée, J. H. J., Tsiaousi, V., van Tol, S., & Delbaere, B. 2010. BioScore- Cost-effective assessment of policy impact on biodiversity using species sensitivity scores. Journal for Nature Conservation, 18(2), 142–148.
185 Appendix A: Appendix to Chapter 2
MA. 2005. Millennium Ecosystem Assessment, Chapter 10. Tech. rept. World Resource Institute, Washington, DC. Maes, D., & Van Dyck, H. 2005. Habitat quality and biodiversity indicator performances of a threatened butterfly versus a multispecies group for wet heathlands in Belgium. Biological Conservation, 123(2), 177–187. MNP. 2006. Integrated Modelling of Global Environmental Change: An Overview of IMAGE 2.4. Bilthoven, The Netherlands: Netherlands Environmental Assessment Agency (MNP). Nelson, E., Mendoza, G., Regetz, J., Polasky, S., Tallis, H., Cameron, D. R., Chan, K. M., Daily, G. C., Goldstein, J., Kareiva, P. M., Lonsdorf, E., Naidoo, R., Ricketts, T. H., & Shaw, M. R. 2009. Modeling multiple ecosystem services, biodiversity conservation, commodity production, and tradeoffs at landscape scales. Frontiers in Ecology and the Environment, 7(1), 4–11. Nelson, Erik, Polasky, Stephen, Lewis, David J, Plantinga, Andrew J, Lonsdorf, Eric, White, Denis, Bael, David, & Lawler, Joshua J. 2008. Efficiency of incentives to jointly increase carbon sequestration and species conservation on a landscape. Proceedings of the National Academy of Sciences, 105(28), 9471– 9476. Niemi, G. J., & McDonald, M. E. 2004. Application of ecological indicators. Annual review of ecology, evolution, and systematics, 35, 89–111. Nilsson, Per, & Ziegler, Friederike. 2007. Spatial distribution of fishing effort in relation to seafloor habitats in the Kattegat, a GIS analysis. Aquatic conservation: Marine and freshwater ecosystems, 17(4), 421–440. Noss, R. F. 1990. Indicators for Monitoring Biodiversity: A Hierarchical Approach. Conservation Biology, 4(4), 355–364. Oberdorff, T., Guegan, J.-F., & Hugueny, B. 1995. Global Scale Patterns of Fish Species Richness in Rivers. Ecography, 18(4), 345–352. O’Connell, T. J., Jackson, L. E., & Brooks, R. P. 2000. Bird guilds as indicators of ecological condition in the central Appalachians. Ecological Applications, 10(6), 1706–1721. Olson, D. M., & Dinerstein, E. 1998. The Global 200: A Representation Approach to Conserving the Earth’s Most Biologically Valuable Ecoregions. Conservation Biology, 12(3), 502–515. Olson, D. M., & Dinerstein, E. 2002. The Global 200: Priority Ecoregions for Global Conservation. Annals of the Missouri Botanical Garden, 89(2), 199–224. Olson, D. M., Dinerstein, E., Wikramanayake, E. D., Burgess, N. D., Powell, G. V. N., Underwood, E. C., D’amico, J. A., Itoua, I., Strand, H. E., Morrison, J. C., Loucks, C. J., Allnutt, T. F., Ricketts, T. H., Kura, Y., Lamoreux, J. F., Wettengel, W. W., Hedao, P., & Kassem, K. R. 2001. Terrestrial ecoregions of the world: a new map of life on earth. BioScience, 51(11), 933. Orme, C. D. L., Davies, R. G., Olson, V. A., Thomas, G. H., & Ding, T. S. 2006. Global patterns of geographic range size in birds. PLoS Biol, 4(7), e208. Ozinga, W. A., & Schaminée, J. H. J. 2005. Target species–Species of European concern. A data-base driven selection of plant and animal species for the implementation of the Pan European Ecological Network. Wageningen, Alterra, Alterra-report, 1119. Pandolfi, John M, Bradbury, Roger H, Sala, Enric, Hughes, Terence P, Bjorndal, Karen A, Cooke, Richard G, McArdle, Deborah, McClenachan, Loren, Newman, Marah JH, Paredes, Gustavo, et al. . 2003. Global trajectories of the long-term decline of coral reef ecosystems. Science, 301(5635), 955–958. Parkes, D., Newell, G., & Cheal, D. 2003. Assessing the quality of native vegetation: The "habitat hectares" approach. Ecological Management and Restoration Supplement, 4, S29–S38. Pereira, H. M., & Daily, G. D. 2006. Modeling biodiversity dynamics in countryside landscapes. Ecology, 87(8), 1877–1885. Petchey, O. L., & Gaston, K. J. 2006. Functional diversity: back to basics and looking forward. Ecology Letters, 9(6), 741–758. Poff, N. L., Angermeier, P. L., Cooper, S. D., Lake, P. S., Fausch, K. D., Winemiller, K. O., Mertes, L. A. K., Oswood, M. W., Reynolds, J., & Rahel, F. J. 2001. Fish diversity in
186 Appendix A: Appendix to Chapter 2
streams and rivers. Pages 315–349 of: Sala, O. E., Chapin, F., & Huber-Sanwald, E. (eds), Global biodiversity in a changing environment: Scenarios for the 21st century. Springer, New York. Potapov, P., Yaroshenko, A., Turubanova, S., Dubinin, M., Laestadius, L., Thies, C., Aksenov, D., Egorov, A., Yesipova, Y., & Glushkov, I. 2008. Mapping the world’s intact forest landscapes by remote sensing. Ecology and Society, 13(2), 51. Rahbek, C. 2005. The role of spatial scale and the perception of large-scale species- richness patterns. Ecology Letters, 8(2), 224–239. Reusch, T. B. H., Ehlers, A., Hämmerli, A., & Worm, B. 2005. Ecosystem recovery after climatic extremes enhanced by genotypic diversity. Proceedings of the National Academy of Sciences of the United States of America, 102(8), 2826– 2831. Ricciardi, A., & MacIsaac, H. J. 2000. Recent mass invasion of the North American great lakes by Ponto-Caspian species. Trends in Ecology & Evolution, 15(2), 62– 65. Ricklefs, R. E. 2004. A comprehensive framework for global patterns in biodiversity. Ecology Letters, 7(1), 1–15. Roberge, J. M., & Angelstam, P. 2004. Usefulness of the umbrella species concept as a conservation tool. Conservation Biology, 18(1), 76–85. Rodríguez-Labajos, B., Binimelis, R., & Monterroso, I. 2009. Multi-level driving forces of biological invasions. Ecological Economics, 69(1), 63–75. Rondinini, C., & Boitani, L. 2006. Differences in the umbrella effects of African amphibians and mammals based on two estimators of the area of occupancy. Conservation Biology: The Journal of the Society for Conservation Biology, 20(1), 170–179. PMID: 16909670. Rosenzweig, M. L. 1995. Species Diversity in Space and Time. Cambridge: Cambridge University Press. Rouget, M., Cowling, R. M., Vlok, J., Thompson, M., & Balmford, A. 2006. Getting the biodiversity intactness index right: the importance of habitat degradation data. Global Change Biology, 12(11), 2032–2036. San Mauro, D., & Agorreta, A. 2010. Molecular systematics: A synthesis of the common methods and the state of knowledge. Cellular & Molecular Biology Letters, 15(2), 311– 341. Schindler, D. W., Turner, M. A., & Hesslein, R. H. 1985. Acidification and alkalinization of lakes by experimental addition of nitrogen compounds. Biogeochemistry, 1(2), 117–133. Scholes, R. J., & Biggs, R. 2005. A biodiversity intactness index. Nature, 434(7029), 45–49. Smith, V. 2003. Eutrophication of freshwater and coastal marine ecosystems a global problem. Environmental Science and Pollution Research, 10(2), 126–139. Southwood, T. R. E., & Henderson, P. A. 2000. Ecological methods. 3 edn. Blackwell Science Ltd. Spangenberg, J. H., Martinez-Alier, J., Omann, I., Monterroso, I., & Binimelis, R. 2009. The DPSIR scheme for analysing biodiversity loss and developing preservation strategies. Ecological Economics, 69(1), 9–11. Stevens, C. J., Dise, N. B., Mountford, J. O., & Gowing, D. J. 2004. Impact of Nitrogen Deposition on the Species Richness of Grasslands. Science, 303(5665), 1876–1879. Swift, T. L., & Hannon, S. J. 2010. Critical thresholds associated with habitat loss: a review of the concepts, evidence, and applications. Biological Reviews, 85(1), 35–53. Thomas, C. D., Cameron, A., Green, R. E., Bakkenes, M., Beaumont, L. J., Collingham, Y. C., Erasmus, B. F. N., Ferreira de Siqueira, M., Grainger, A., & Hannah, L. 2004. Extinction risk from climate change. Nature, 427(6970), 145–148. Thuiller, W., Lavorel, S., Sykes, M. T., & Araújo, M. B. 2006. Using niche-based modelling to assess the impact of climate change on tree functional diversity in Europe. Diversity and Distributions, 12(1), 49. Turner, M. G. 2005. Landscape Ecology: What Is the State of the Science? Annual Review of Ecology, Evolution, and Systematics, 36(1), 319–344. Van Vuuren, D. P., Sala, O. E., & Pereira, H. M. 2006. The future of vascular plant diversity under four global scenarios. Ecology and Society, 11(2), 25.
187 Appendix A: Appendix to Chapter 2
van Zelm, R., Huijbregts, M. A. J., van Jaarsveld, H. A., Reinds, G. J., de Zwart, D., Struijs, J., & van de Meent, D. 2007. Time horizon dependent characterization factors for acidification in life-cycle assessment based on forest plant species occurrence in Europe. Environmental Science and Technology, 41(3). Verboom, J., Alkemade, R., Klijn, J., Metzger, M. J., & Reijnen, R. 2007. Combining biodiversity modeling with political and economic development scenarios for 25 EU countries. Ecological Economics, 62(2), 267–276. Vinebrooke, Rolf D, Schindler, David W, Findlay, David L, Turner, Michael A, Paterson, Michael, & Mills, Kenneth H. 2003. Trophic dependence of ecosystem resistance and species compensation in experimentally acidified Lake 302S (Canada). Ecosystems, 6(2), 0101– 0113. Weijters, M. J., Janse, J. H., Alkemade, R., & Verhoeven, J. T. A. 2009. Quantifying the effect of catchment land use and water nutrient concentrations on freshwater river and stream biodiversity. Aquatic Conservation: Marine and Freshwater Ecosystems, 19(1), 104–112. Xenopoulos, Marguerite A., Lodge, David M., Alcamo, Joseph, Märker, Michael, Schulze, Kerstin, & Vuuren, Detlef P. Van. 2005. Scenarios of freshwater fish extinctions from climate change and water withdrawal. Global Change Biology, 11(10), 1557–1564. Zhu, Y., Chen, H., Fan, J., Wang, Y., Li, Y., Chen, J., Fan, J., Yang, S., Hu, L., Leung, H., Mew, T. W., Teng, P. S., Wang, Z., & Mundt, C. C. 2000. Genetic diversity and disease control in rice. Nature, 406(6797), 718–722. Ziegler, F., Nilsson, P., Mattsson, B., & Walther, Y. 2003. Life cycle assessment of frozen cod fillets including fishery-specific environmental impacts. The International Journal of Life Cycle Assessment, 8(1), 39–47.
188 Appendix B: Appendix to Chapter 3
Appendix B: Appendix to Chapter 3
Table B1. Biodiversity damage potential (BDP) characterization factors (median) of occupation per land use type, 1. and 3. quartile and number of data points (n). For n<5, no characterization factors are provided.
condary not used not Vegetation 1.1.1./4.1.1 1.1.1./4.1.1 (Agroforestry) 1.1.2. Se 1.1.2. 5.1 Annual crops Annual 5.1 1.2. Forest, used Forest, 1.2. 7. Artificial areas Artificial 7. Forest/Grassland, 4.2. Pasture/meadow 4.2. 5.2. Permanentcrops 5.2. mosaic Agriculture, 6. Characterization factors, global Total world Median 0 0.18 0.18 0.33 0.60 0.42 0.2 0.44 average 1. quartile 0 -0.03 -0.05 0.00 0.31 0.06 0.01 -0.01 3. quartile 0 0.37 0.50 0.55 0.79 0.70 0.48 0.62 n 326 272 148 133 96 52 76 53 Characterization factors, per biome Biome 1 Median 0 0.22 0.13 0.45 0.54 0.42 0.18 - (Sub-)tropical 1. quartile 0 0.00 -0.09 0.31 0.36 0.18 -0.02 - moist broadleaf 3. quartile 0 0.43 0.45 0.75 0.72 0.70 0.44 - forest n 173 172 79 26 46 40 70 1 Biome 2 Median 0 0.17 0.58 0.48 - - - - (Sub-)tropical 1. quartile 0 -0.04 0.34 0.17 - - - - dry broadleaf 3. quartile 0 0.19 0.68 0.69 - - - - forest n 8 5 8 8 3 0 1 2 Biome 4 Median 0 0.08 0.22 0.52 0.76 0.02 - 0.40 Temperate 1. quartile 0 -0.26 -0.09 -0.35 0.46 -0.11 - -0.10 broadleaf forest 3. quartile 0 0.33 0.43 0.67 0.86 0.69 - 0.58 n 46 20 35 33 24 9 0 24 Biome 5 Median 0 0.17 0.15 0.24 0.54 - - 0.50 Temperate 1. quartile 0 -0.22 0.02 -0.64 -0.15 - - -0.05 coniferous forest 3. quartile 0 0.30 0.33 0.38 0.87 - - 0.71 n 45 15 7 27 8 3 0 21 Biome 7 Median 0 0.00 0.01 0.12 0.65 - - - (Sub-)tropical 1. quartile 0 -0.17 0.00 0.02 0.02 - - - grassland & 3. quartile 0 0.15 0.06 0.27 0.80 - - - savannah n 21 27 6 8 9 0 0 0
189 Appendix B: Appendix to Chapter 3
Table B1 (continued)
not used not Vegetation 1.1.1./4.1.1 1.1.1./4.1.1 Pasture/meadow (Agroforestry) 1.1.2. Secondary Secondary 1.1.2. 5.1 Annual crops Annual 5.1 1.2. Forest, used Forest, 1.2. 7. Artificial areas Artificial 7. Forest/Grassland, Forest/Grassland, 4.2. 4.2. 5.2. Permanentcrops 5.2. mosaic Agriculture, 6. Characterization factors, per biome (continued) Biome 7 Median 0 0.00 0.01 0.12 0.65 - - - (Sub-)tropical 1. quartile 0 -0.17 0.00 0.02 0.02 - - - grassland & 3. quartile 0 0.15 0.06 0.27 0.80 - - - savannah n 21 27 6 8 9 0 0 0 Biome 8 Median 0 - - 0.23 - - - - Temperate 1. quartile 0 - - 0.07 - - - - grassland & 3. quartile 0 - - 0.39 - - - - savannah n 7 4 2 10 1 0 0 0 Biome 10 Median 0 0.21 0.55 0.33 - - 0.39 - Montane 1. quartile 0 0.10 0.34 0.13 - - 0.29 - grassland & 3. quartile 0 0.38 0.71 0.37 - - 0.57 - shrublands n 13 25 11 5 2 0 5 0 Biome 12 Median 0 - - 0.24 - - - - Mediterranean 1. quartile 0 - - 0.11 - - - - forests & 3. quartile 0 - - 0.42 - - - - shrublands n 8 3 0 11 3 0 0 4 Biome 13 Median 0 - - -0.08 - - - - Deserts & xeric 1. quartile 0 - - -0.50 - - - - shrublands 3. quartile 0 - - 0.17 - - - - n 5 1 0 5 0 0 0 1 Characterization factors, per taxonomic group Arthropods Median 0 0.16 0.11 0.25 0.65 0.56 0.04 - 1. quartile 0 0.00 -0.15 0.13 0.35 0.23 -0.41 - 3. quartile 0 0.39 0.32 0.45 0.77 0.62 0.29 - n 68 64 34 21 20 7 20 4 Other Median 0 0.24 0.41 0.49 0.79 0.44 - 0.49 invertebrates 1. quartile 0 0.12 0.22 0.33 0.59 0.18 - 0.22 3. quartile 0 0.53 0.79 0.58 0.85 0.69 - 0.71 n 33 16 14 21 14 7 1 15 All Median 0 0.09 0.12 0.31 0.50 0.39 0.11 - vertebrates 1. quartile 0 -0.06 -0.03 0.04 0.20 0.28 -0.14 - 3. quartile 0 0.31 0.47 0.5 0.70 0.70 0.24 - n 99 75 42 25 23 19 19 4
190 Appendix B: Appendix to Chapter 3
Table B1 (continued)
mosaic mosaic not used not Vegetation 1.1.1./4.1.1 1.1.1./4.1.1 (Agroforestry) 1.1.2. Secondary Secondary 1.1.2. 5.1 Annual crops Annual 5.1 1.2. Forest, used Forest, 1.2. 7. Artificial areas Artificial 7. Forest/Grassland, 4.2. Pasture/meadow 4.2. 5.2. Permanentcrops 5.2. 6. Agriculture, Agriculture, 6. Characterization factors, per taxonomic group (continued) Birds Median 0 0.07 0.00 0.20 0.53 0.62 0.22 - 1. quartile 0 -0.06 -0.08 -0.08 0.21 0.38 0.07 - 3. quartile 0 0.26 0.35 0.47 0.70 0.73 0.32 - n 53 39 17 14 17 12 11 3 Other Median 0 0.15 0.14 0.33 0.45 0.27 -0.14 - vertebrates 1. quartile 0 -0.06 0.00 0.16 0.20 -0.01 -0.37 - 3. quartile 0 0.39 0.56 0.50 0.68 0.38 0.05 - n 46 36 25 11 6 7 8 1 All plants Median 0 0.21 0.23 0.29 0.56 0.38 0.37 0.33 1. quartile 0 0.00 0.00 -0.20 0.30 0.01 0.15 -0.42 3. quartile 0 0.43 0.53 0.57 0.81 0.79 0.63 0.59 n 126 117 58 66 39 19 36 30 Vascular plants Median 0 0.21 0.27 0.11 0.42 0.28 0.37 -0.42 1. quartile 0 0.02 0.04 -0.37 0.17 -0.04 0.15 -0.79 3. quartile 0 0.45 0.57 0.40 0.61 0.63 0.63 0.15 n 97 109 44 48 28 15 32 15 Moss Median 0 0.03 0.05 0.65 0.87 - - 0.58 1. quartile 0 -0.19 -0.16 0.29 0.73 - - 0.41 3. quartile 0 0.24 0.28 0.75 0.9 - - 0.68 n 29 8 14 18 11 4 4 15 Characterization factors, per data source GLOBIO data Median 0 0.19 0.17 0.33 0.57 0.42 0.20 0.34
1. quartile 0 0.00 0.00 0.12 0.35 0.18 0.01 0.12 3. quartile 0 0.38 0.50 0.49 0.74 0.70 0.48 0.60 n 254 248 121 79 72 40 76 8 Swiss data Median 0 0.07 0.22 0.32 0.81 0.43 - 0.45 (BDM) 1. quartile 0 -0.19 -0.17 -0.38 0.26 -0.08 - -0.05 3. quartile 0 0.31 0.48 0.62 0.90 0.77 - 0.62 n 72 24 27 54 24 12 0 45
191 Appendix B: Appendix to Chapter 3
Table B2 Results of 2-sided Mann-Whitney U test testing the difference of median Srel of all combinations of land use types (full dataset). *** p-values<0.001; ** p-values<0.01; * p- values<0.05; (*) p-values<0.1; ns p-values > 0.1
Land use type
forestry crop Reference Secondary vegetation forest Used Pasture crop Annual Permanent Agro Secondary vegetation *** Used forest *** ns Pasture *** ** ns Annual crops *** *** *** *** Permanent crops *** *** ** * (*) Agroforestry *** ns ns ns *** ** Artificial area *** * ns ns ** ns ns
Table B3 Peason’s correlation coefficients r between indicators for a subset of data from biome (sub-) tropical moist broadleaf forest
Srel MSA Sørensen’s Ss Shannon’s H (rel) Fisher’s α (rel) Srel 1 0.41 0.17 0.79 0.83 MSA 1 0.81 0.19 0.06 Sørensen’s Ss 1 0.03 0.20 Shannon’s H (rel) 1 0.87 Fisher’s α (rel) 1
Figure B1 UNEP/SETAC framework for calculating land use occupation and transformation impacts (adapted from Milà i Canals et al., 2007 and Koellner et al., 2013b).
192 Appendix B: Appendix to Chapter 3
Figure B2 Map of geographic location of studies included in this study (indicated as black dots). The colors indicate different WWF biomes
Vascular plants Moss ) Arthropods rel Other invertebrates S Birds Other vertebrates relative species richness ( relative 0.0 0.5 1.0 1.5 2.0 2.5 3.0 p<0.1 p<0.01 p<0.01 p<0.05 p<0.01 Pasture Pasture p<0.001 p<0.001 Reference Reference Used forest Used forest Agroforestry Artificial area Annual crops Annual Permanent crops Permanent Second. vegetation Second. vegetation Figure B3 Box and whisker plot of relative species richness per land use type and taxonomic group and test statistics of Kruskal-Wallis test (Srel = f(LU x taxa)) for full dataset. In contrast to the results presented in Fig 3.3 (main article), the taxonomic groups ‘plants’ were split into ‘vascular plants’ and ‘moss’ and ‘vertebrates’ were split into ‘birds’ and ‘other vertebrates’(see also Table B5 for an overview of the underlying data sources).
193 Appendix B: Appendix to Chapter 3
Vascular plants ) Moss rel S
Mollusks relative species richness ( relative 0.0 0.5 1.0 1.5 2.0 2.5 3.0 (b) n.s. (b) n.s. (b) n.s. (b) n.s. (a) n.s. (a) n.s. (b) n.s. (b) n.s. (b) n.s. Pasture Pasture (b) p<0.1 Artificial LI Reference Reference Artificial HI (b) p<0.05 (a) p<0.05 (a) p<0.01 (a) p<0.05 (a) p<0.05 Used forest Used forest (a) p<0.001 (a) p<0.001 (a) p<0.001 Artificial both Annual crops Annual Permanent crops Permanent Second. vegetation Second. vegetation Figure B4 Box and whisker plot for the Swiss BDM data of relative species richness per land use type and taxonomic group. (a) test statistics of Kruskal-Wallis test Srel = f(LU x taxa); (b) Comparison of the LU effect across the two biomes temperate broadleaf forest and temperate coniferous forest (test statistics of Kruskal-Wallis test Srel = f(LU x biome)). Results for all artificial area (‘Artificial both’) were split into high (HI: land use types ‘7.1.2 Urban, continuously built’; ‘7.2 Industrial area’; ‘7.4 Dump site’) and low intensity (LI: land use types ‘7.1.3 Urban, discontinuously built’; ‘7.6 Traffic area’; ‘7.1.4 Urban, green areas’). There was no significant difference between Artificial HI and Artificial LI for the overall data nor for any taxonomic group (Mann-Whitney U test, not shown). ns not significant.
194 Appendix B: Appendix to Chapter 3
Table B4 Representation of species groups in this study compared to global estimates of species richness Estimated species Data representation of numbers this study (Heywood and Watson 1995) Invertebrates 31% 70% Arthropodes 20% 65% Insects 19% 59% Coleoptera 6% Hymenoptera 6% Lepidoptera 5% Other insects 2% Other arthropods 2% 7% Other Invertebrates 11% Mollusks 8% 1% Other invertebrates 3% 3% Vertebrates 26% 0.4% Birds 14% Other vertebrates 12% Amphibians 4% Mammals 8% Reptiles 0% Plant 43% 2% Vascular plants 34% Herbaceous plants 13% Trees 13% Mosses 9% Virus 3% Bacteria 7% Fungi 11% “Protozoa” 1% “Algea” 3% Others 2% Total 100% 100%
195 Table B5 Number of data points included in this study per species group, land use type and dataset (GLOBIO and BDM).
Refe- Secondary Used Pasture Annual Perman- Agro- Artificial Total rence vegetation forest crops ent crops forestry area
GLOBIO BDM GLOBIO BDM GLOBIO BDM GLOBIO BDM GLOBIO BDM GLOBIO BDM GLOBIO BDM GLOBIO BDM GLOBIO BDM Total Invertebrates 77 24 72 8 39 9 24 20 26 8 10 4 21 0 4 16 273 89 285 Arthropodes 68 0 64 0 34 0 21 0 20 0 7 0 20 0 4 0 238 0 170
Insects 60 0 57 0 31 0 20 0 20 0 7 0 19 0 3 0 217 0 157 Coleoptera 16 0 13 0 15 0 10 0 3 0 1 0 7 0 0 0 65 0 49 Hymenoptera 20 0 22 0 6 0 4 0 11 0 5 0 7 0 0 0 75 0 55 Lepidoptera 17 0 19 0 7 0 0 0 4 0 1 0 4 0 2 0 54 0 37 Other insects 7 0 3 0 3 0 6 0 2 0 0 0 1 0 1 0 23 0 16
Other arthropods 8 0 7 0 3 0 1 0 0 0 0 0 1 0 1 0 21 0 13 Other Invertebrates 9 24 8 8 5 9 3 20 6 8 3 4 1 0 0 16 35 89 115 Mollusks 2 24 0 8 2 9 0 20 0 8 0 4 0 0 0 16 4 89 91
Other invertebrates 7 0 8 0 3 0 3 0 6 0 3 0 1 0 0 0 31 0 24 Vertebrates 99 0 75 0 42 0 25 0 23 0 19 0 19 0 4 0 306 0 207 Birds 53 0 39 0 17 0 14 0 17 0 12 0 11 0 3 0 166 0 113
Other vertebrates 46 0 36 0 25 0 11 0 6 0 7 0 8 0 1 0 140 0 94 Amphibians 17 0 12 0 9 0 7 0 1 0 3 0 1 0 0 0 50 0 33 Mammals 28 0 24 0 16 0 4 0 5 0 4 0 6 0 1 0 88 0 60 Reptiles 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 2 0 1 Plant 78 48 101 16 40 18 30 40 23 16 11 8 36 0 0 32 319 178 419
Vascular plants 73 24 101 8 35 9 30 20 20 8 11 4 32 0 0 16 302 89 318 Herbaceous plants 34 0 38 0 13 0 26 0 11 0 8 0 16 0 0 0 146 0 112 Trees 39 0 63 0 22 0 4 0 9 0 3 0 16 0 0 0 156 0 117
Mosses 5 24 0 8 5 9 0 20 3 8 0 4 4 0 0 16 17 89 101 Total 254 72 248 24 121 27 79 60 72 24 40 12 76 0 8 48 898 267 911 Appendix B: Appendix to Chapter 3
Table B6 Overview of land use classification of BDM (2004) data as applied in Koellner and Scholz (2008) and classification applied in this study (based on Koellner et al., 2013a). CORINE Plus classification CORINE Plus description (type of intensity) Simplified classification UNEP/SETAC (Koellner & Scholz, 2008) used in this paper classification (Koellner et al, 2013a) 1 Artificial surfaces 10 Built up land 11 Urban fabric 111 Continuous urban fabric Buildings cover most of the land. Roads and artificially surfaced area cover almost all the 7 Artificial area 7.1.2 Urban, ground. Non-linear areas of vegetation and bare soil are exceptional. At least 80% of the (High intensity) continuously built total area is sealed. (artificial_hi) 112 Discontinuous urban Most of the land is covered by structures. Buildings, roads and artificially surfaced areas 7 Artificial area 7.1.3 Urban, fabric associated with areas with vegetation and bare soil, which occupy discontinuous but (Low intensity) discontinuously built significant surfaces. Less than 80% of the total area is sealed. (artificial_hi) 12 Industrial, commercial and transport 121 Industrial or commercial Artificially surfaced areas (with concrete, asphalt, tamacadam, or stabilized, e.g., beaten 7 Artificial area 7.2 Industrial area units earth) devoid of vegetation occupy most of the area in question, which also contains (High intensity) buildings and/or areas with vegetation. (artificial_hi) 122 Road and rail networks Motorways, railways, including associated installations (stations, platforms, 7 Artificial area 7.6 Traffic area and associated land embankments). Minimum width to include: 100 m. (Low intensity) 132 Dump sites Landfill or mine dump sites, industrial or public. (artificial_hi) 7 Artificial area 7.4 Dump site (High intensity) 14 Artificial, non-agricultural areas with vegetation 141 Green urban areas Areas with vegetation within urban fabric. Includes parks and cemeteries with 7 Artificial area 7.1.4 Urban, green vegetation. (artificial_li) (Low intensity) areas 142 Sport and leisure facilities Camping grounds, sports grounds, leisure parks, golf courses, racecourses, etc. 7 Artificial area 7.1.4 Urban, green Includes formal parks not surrounded by urban zones. (artificial_li) (Low intensity) areas 2 Agricultural areas 21 Arable land 211 Non-irrigated arable land Cereals, legumes, fodder crops, root crops and fallow land. Includes flower and tree 5.1 Annual crop 5.1.2 Arable, non- (nurseries) cultivation and vegetables, whether open field, under plastic or glass irrigated (includes market gardening). Includes aromatic, medicinal and culinary plants. Excludes permanent pastures. 211b Integrated Chemical-synthetic and organic fertilizer as well as pesticides are applied. However, the 5.1 Annual crop 5.1.2.2 Arable, non- input of these substances is reduced. 21121 Wheat, 21122 Maize (agri_hi) irrigated, intensive 211e Agricultural fallow Agricultural fallow. (non-use) 5.1 Annual crop 5.1.1 Arable, fallow 211f Artificial meadow Artificial meadow in rotation system. (agri-hi) 4.2 Pasture/meadow 4.2 Pasture/meadow
197 Appendix B: Appendix to Chapter 3
CORINE Plus classification CORINE Plus description (type of intensity) Simplified classification UNEP/SETAC (Koellner & Scholz, 2008) used in this paper classification (Koellner et al, 2013a) 22 Permanent crops 221 Vineyards Areas planted with vines. 5.2 Permanent crop 5.2 Permanent crops 222 Fruit trees and berry Parcels planted with fruit trees or shrubs: single or mixed fruit species, fruit trees 5.2 Permanent crop 5.2 Permanent crops plantations associated with permanently grassed surfaces. Includes chestnut and walnut groves. 222a Intensive orchards Orchards with small growing fruit trees. (agri_hi) 5.2 Permanent crop 5.2.4 Permanent crops, intensive 222b Organic orchards Orchards with meadows and large fruit trees. (agri_li) 5.2 Permanent crop 5.2.3 Permanent crops, extensive 23 Pastures and meadows 231 Pastures and meadows Dense, predominantly graminoid grass cover, of floral composition, not under a rotation 4.2 Pasture/meadow 4.2 Pasture/meadow system. Mainly used for grazing, but the fodder may be harvested mechanically. Includes areas with hedges (bocage), e.g., oat grass meadow (Arrhenatherion, Polygono), fertilized moist meadow (Calthion). 231a Intensive pasture and Meadows mechanically harvested 3 times or more per year, fertilizer applied, perhaps on 4.2 Pasture/meadow 4.2.2 Pasture/meadow, meadows former arable land. (agri_hi) intensive 24 Heterogeneous agricultural areas 244 Agroforestry areas Annual crops or grazing land under the wooded cover of forestry species. (agri_li) 6 Agroforestry 6 Agriculture, mosaic 3 Forests and semi-natural areas 31 Forests 311 Broad-leafed forest Vegetation formation composed principally of trees, including shrub and bush 1.11 Forest, Reference 1 Forest understories, where broad-leafed species predominate. (Presence of conifers 0-10%) 312 Coniferous forest Vegetation formation composed principally of trees, including shrub and bush 1.11 Forest, Reference 1 Forest understories, where coniferous species predominate. (Presence of conifers 91-100%) (for montane and subalpine regions) 1.22 Used Forest (for colline regions) 312a Coniferous plantations Plantations of fast growing tree species like Picea abies. (forest_hi) 1.22 Used Forest 1.2.2 Forest, intensive 312b Semi-natural coniferous Natural or semi-natural forests, where coniferous species predominate. (forest_li) 1.11 Forest, Reference 1.1.1 Forest, primary forests 313 Mixed forest Vegetation formation composed principally of trees, including shrub and bush 1.11 Forest, Reference 1 Forest understories, where broad-leafed and coniferous species co-dominate. (forest_li) 313b Mixed coniferous forest Forests with presence of conifers 51-90%. (forest_li) 1.22 Used Forest 1.2.2 Forest, intensive 32 Shrub and/or herbaceous vegetation associations
198 Appendix B: Appendix to Chapter 3
CORINE Plus classification CORINE Plus description (type of intensity) Simplified classification UNEP/SETAC (Koellner & Scholz, 2008) used in this paper classification (Koellner et al, 2013a) 321 Semi-Natural grassland Low productivity grassland. Often situated in areas of rough uneven ground. Frequently 4.11 Grassland, 4.1.1 Grassland, includes rocky areas, briars, and heath land., e.g., xeric grassland (Xero-Bromium), Reference natural mesoxeric grassland (Meso-Bromium), mat-grass pasture (Violion), moor-grass meadow (Molinion). (agri_li) 322 Moors and heath land Vegetation with low and closed cover, dominated by bushes, shrubs and herbaceous 3 Shrub land, 3 Shrub land plants (heath, briars, broom, gorse, laburnum, etc.). (non-use) Reference 324 Transitional Bushy or herbaceous vegetation with scattered trees. Can represent either woodland 1.12 Secondary forest 1.1.2 Forest, woodland/shrub degradation or forest regeneration/colonization. (non-use) secondary 33 Open spaces with little or no vegetation 331 Beaches, dunes, and sand Beaches, dunes and expanses of sand or pebbles in coastal or continental areas, 8 Bare area, Reference 8 Bare area plains including beds of stream channels with torrential regime. (non-use) 332 Bare rock Scree, cliffs, rocks and outcrops. (non-use) 8 Bare area, Reference 8 Bare area 333 Sparsely vegetated areas Includes steppes, tundra and badlands. Scattered high-attitude vegetation. (non-use) 8 Bare area, Reference 8 Bare area 4 Wetlands 41 Inland wetlands 411 Inland marshes Low-lying land usually flooded in winter, and more or less saturated by water all year 2.2 Wetlands, 2.2 Wetlands, inland round. (non-use) Reference 412 Peat bogs Land consisting mainly of decomposed moss and vegetable matter. May or may not be 2.2 Wetlands, 2.2 Wetlands, inland exploited. e.g., raised bog (Sphagnetum), intermediate mires (e.g., Scheuchzerietum) and Reference fens (Caricetum). (non-use) 51 Inland waters 511 Water courses Natural or artificial watercourses serving as water drainage channels. Includes canals. 10 Water bodies, 10 Water bodies Minimum width to include: 100 m. Reference
199 Appendix B: Appendix to Chapter 3
B.1 References Heywood VH, Watson RT (1995) Global Biodiversity Assessment. Cambridge University Press, Cambridge Koellner T, Scholz RW (2008) Assessment of land use impacts on the natural environment. Part 2: Generic characterization factors for local species diversity in Central Europe. Int J Life Cycle Assess 13(1):32-48. Koellner T, de Baan L, Beck T, Brandão M, Civit B, Goedkoop MJ, Margni M, Milà i Canals L, Müller-Wenk R, Weidema B, Wittstock B (2013a) Principles for Life Cycle Inventories of land use on a global scale. Int J Life Cycle Assess (this issue) Koellner T, de Baan L, Beck T, Brandão M, Civit B, Margni M, Milà i Canals L, Saad R, de Souza DM, Müller-Wenk R (2013b) UNEP-SETAC Guideline on Global Land Use Impact Assessment on Biodiversity and Ecosystem Services in LCA. Int J Life Cycle Assess (this issue) Milà i Canals L, Bauer C, Depestele J, Dubreuil A, Freiermuth Knuchel R, Gaillard G, Michelsen O, Müller-Wenk R, Rydgren B (2007) Key Elements in a Framework for Land Use Impact Assessment Within LCA. Int J Life Cycle Assess 12(1):5-15
200 Appendix C: Appendix to Chapter 4
Appendix C: Appendix to Chapter 4
Table C1 Applied z-values by biome (source: Kier et al. 2005) z-value
Tropical and subtropical moist broadleaf forest 0.26a
Tropical and subtropical grass-/shrublands and savannahs 0.18
Temperate broadleaf and mixed forests 0.17 a specifically for Asia as for this biome region-specific z-values were given
Table C2 Medians of biodiversity characterization factors of occupation impacts per land use type, farming practice an biome TempBMF broadleaf and mixed forests, TropGL sub- /tropical grass-/shrublands and savannahs and TropMBF sub-/tropical moist broadleaf forests
Biome TempBLF TropGL TropMBF Pasture Annual crops Annual crops Permanent crops Conv Org Conv Org Conv Org Conv Median 0.28 -0.02 0.60 0.15 0.81 0.42 0.51 1. quartile -0.28 -0.65 0.38 -0.33 0.75 0.25 0.29 3. quartile 0.54 0.33 0.78 0.37 0.85 0.52 0.67 n 14 x 33 14 x 33 28 x 33 26x 33 15 x 6 2 x 6 12 x 14
201 Appendix C: Appendix to Chapter 4
Table C3 Composition of concentrate feeda by farming practice (modified from Cederberg and Flysjoe 2004) Fodder crop Share in concentrates Land use type Country of origin Biomeb Yield (% dry mass) (kg*ha-1*a-1) Conventional Barley 35 Arable, intensive Sweden tempBMF 4600 Beet pulp, molasse 17 Arable, intensive Sweden, Baltic states tempBMF 10714 Rapeseed meal 15 Arable, intensive Germany, Sweden tempBMF 2580 Wheat 11 Arable, intensive Sweden tempBMF 6100 Soymeal 10 Arable, intensive Brazil tropGL 3125 Wheat bran + dried draff 4 Arable, intensive Sweden tempBMF 6100 Palm kernel expels 3 Permanent crops, intensive Malaysia tropMBF 25000 Fatty acids 2 Arable and Permanent crops, intensive Sweden, Malaysia, Brazil tempBMF, 2580, 2500, 3125 tropGL,tropMBF Grass pellets 2 Pasture/meadow, intensive Denmark tempBMF 8000 Triticale 1 Arable, intensive Sweden tempBMF 5000
Organic Barleyc 47 Arable, organic Sweden tempBMF 2760 Wheatc 22 Arable, organic Sweden tempBMF 3215 Horse beanc 6 Arable, organic Sweden tempBMF 2600 Maize gluten meal 6 Arable, intensive France tempBMF 2500 Luzern pelletsc 5 Pasture/meadow, organic Sweden tempBMF 8000 Rapeseed mealc 4 Arable, organic Sweden tempBMF 2000 Beet pulpc 4 Arable, organic Sweden tempBMF 8571 Soybeanc 4 Arable, organic South America tropGL 1800 Oatsc 2 Arable, organic Sweden tempBMF 3000 Fatty acids 1 Arable and permanent crops, intensive Sweden, Malaysia, Brazil tempBMF, 2580,2500,3125 tropGL,tropMBF Beet pulp, molasse <1 Arable, intensive Sweden tempBMF 10714 a sum of purchased concentrates and concentrates produced on-farm (i.e maize and grains) b tropMBF: Sub-/tropical Moist Broadleaf Forest, tropGL: Sub-/tropical grass-/shrublands and Savannahs, tempBMF: Temperate broadleaf and mixed forest c produced organically
202 Appendix C: Appendix to Chapter 4
Table C4 Land use change between land cover types (modified from (FAOSTAT 2012b) Country Land covera Average area for Average area Change over 20- Change of area per the period 1985- for the period year period ha covered on 1989 (1000 ha) 2005-2009 (1000 ha) average in 2007 (ha (1000 ha) × year) Argentina Arable land 26713 31400 4687 0.149 Permanent crops 1017 1000 -17 -0.017 Permanent 100144 106450 6306 0.059 meadows Forest 35514 30119 -5395 -0.179 Brazil Arable land 49100 61080 11980 0.196 Permanent crops 6161 7460 1299 0.174 Permanent 180340 196000 15660 0.080 meadows Forest 583231 526105 -57126 -0.109 Malaysia Arable land 1432 1800 368 0.204 Permanent crops 4744 5785 1041 0.180 Permanent 271 285 14 0.048 meadows Forest 22770 20716 -2054 -0.099 a according to FAO Classification
203 Appendix C: Appendix to Chapter 4
Table C5 Plant species richness (S) standardised to area and as mentioned sampling area and species richness in source in the biomes TempBMF broadleaf and mixed forests, TropGL sub-/tropical grass-/shrublands and savannahs and Trop MBF sub- /tropical moist broadleaf forests, land use types permcro permanent crops, ancro annual crops, pas pasture/meadows and ref (semi-)natural reference and farming practice conventional or organic Source Biome LU type Org/Conv. S Area (m²] SR Standard. Ammer et al. 1988 TempBMF ancro org 15 200 14.50 Boutin et al. 2008 TempBMF ancro org 17 5 21.88 Clough et al. 2007 TempBMF ancro org 23 50 23.79 Frieben and Koepke 1996 TempBMF ancro org 20 100 20.00 Frieben and Koepke 1996 TempBMF ancro org 19 100 18.50 Gabriel et al. 2006 TempBMF ancro org 29 50 30.00 Gabriel et al. 2010 TempBMF ancro org 11 3 14.81 Gabriel et al. 2010 TempBMF ancro org 10 3 13.75 Hald 1999 TempBMF ancro org 30 2 43.88 Hawes et al. 2010 TempBMF ancro org 10 11 11.04 Hiltbrunner et al. 2008 TempBMF ancro org 16 20 17.10 Hiltbrunner et al. 2008 TempBMF ancro org 13 20 14.01 Hiltbrunner et al. 2008 TempBMF ancro org 16 20 17.10 Hiltbrunner et al. 2008 TempBMF ancro org 17 20 18.20 Hotze and van Elsen 2006 TempBMF ancro org 24 100 24.00 Kaar and Freier 2008 TempBMF ancro org 28 100 27.50 Krauss et al. 2011 TempBMF ancro org 11 5 14.18 Kreuter 2005 TempBMF ancro org 17 150 16.42 Kreuter 2005 TempBMF ancro org 15 150 14.95 Matzdorf and Zerbe 2000 TempBMF ancro org 25 1125 22.17 Matzdorf and Zerbe 2000 TempBMF ancro org 23 675 20.67 Moreby et al. 1994 TempBMF ancro org 12 3 16.89 Moreby et al. 1994 TempBMF ancro org 12 3 17.34 Norton 2002 TempBMF ancro org 15 15 16.69 Roschewitz et al. 2005 TempBMF ancro org 31 60 31.56 Ulber et al. 2009 TempBMF ancro org 23 1600 20.83 Ammer et al. 1988 TempBMF ancro conv 7 500 6.51 Boutin et al. 2008 TempBMF ancro conv 9 5 11.56 Clough et al. 2007 TempBMF ancro conv 10 50 10.08 Carey et al. 2007 TempBMF ancro conv 10 200 9.96 Frieben and Koepke 1996 TempBMF ancro conv 9 100 8.50 Frieben and Koepke 1996 TempBMF ancro conv 5 100 5.00 Gabriel et al. 2006 TempBMF ancro conv 14 50 14.41 Gabriel et al. 2010 TempBMF ancro conv 2 3 3.17 Gabriel et al. 2010 TempBMF ancro conv 3 3 3.40 Hald 1999 TempBMF ancro conv 12 2 17.55 Hawes et al. 2010 TempBMF ancro conv 15 11 17.66 Hiltbrunner et al. 2008 TempBMF ancro conv 7 20 7.17 Hiltbrunner et al. 2008 TempBMF ancro conv 2 20 2.54 Hiltbrunner et al. 2008 TempBMF ancro conv 9 20 10.29 Hiltbrunner et al. 2008 TempBMF ancro conv 5 20 5.18 Hotze and van Elsen 2006 TempBMF ancro conv 8 100 8.00 Kaar and Freier 2008 TempBMF ancro conv 16 100 16.20 Krauss et al. 2011 TempBMF ancro conv 3 5 3.26 Kreuter 2005 TempBMF ancro conv 3 150 2.79 Kreuter 2005 TempBMF ancro conv 4 150 3.72
204 Appendix C: Appendix to Chapter 4
Source Biome LU type Org/Conv. S Area (m²] SR Standard. Manhoudt et al. 2005 TempBMF ancro conv 10 100 10.00 Matzdorf and Zerbe 2000 TempBMF ancro conv 15 630 13.45 Matzdorf and Zerbe 2000 TempBMF ancro conv 14 675 12.59 Moreby et al. 1994 TempBMF ancro conv 4 3 6.06 Moreby et al. 1994 TempBMF ancro conv 4 3 6.26 Norton 2002 TempBMF ancro conv 8 15 9.02 Roschewitz et al. 2005 TempBMF ancro conv 15 60 15.62 Ulber et al. 2009 TempBMF ancro conv 13 1600 11.21 Carey et al. 2007 TempBMF pas org 20 200 19.73 Frieben and Koepke 1996 TempBMF pas org 13 4 16.66 Gabriel et al. 2010 TempBMF pas org 7 9 7.97 Gabriel et al. 2010 TempBMF pas org 6 9 7.57 Knop (2005) TempBMF pas org 30 50 30.85 Knop (2005) TempBMF pas org 40 50 41.11 Knop (2005) TempBMF pas org 49 50 50.90 Lips et al. 1997 TempBMF pas org 30 5000 25.63 Schreiber and Lehman 1996 TempBMF pas org 23 50 23.89 Schreiber and Lehman 1996 TempBMF pas org 21 50 21.82 Wachendorf and Taube 2001 TempBMF pas org 25 22400 20.19 Wachendorf and Taube 2001 TempBMF pas org 21 13200 17.57 Wachendorf and Taube 2001 TempBMF pas org 24 5200 20.46 Weibull et al. 2003 TempBMF pas org 39 2500 33.75 Carey et al. 2007 TempBMF pas conv 14 200 13.83 Frieben and Koepke 1996 TempBMF pas conv 11 4 14.34 Gabriel et al. 2010 TempBMF pas conv 5 9 6.30 Gabriel et al. 2010 TempBMF pas conv 4 9 4.30 Knop (2005) TempBMF pas conv 18 50 19.00 Knop (2005) TempBMF pas conv 31 50 32.35 Knop (2005) TempBMF pas conv 31 50 31.76 Lips et al. 1997 TempBMF pas conv 10 5000 8.81 Schreiber and Lehman 1996 TempBMF pas conv 21 50 21.82 Schreiber and Lehman 1996 TempBMF pas conv 20 50 20.78 Wachendorf and Taube 2001 TempBMF pas conv 19 20400 15.88 Wachendorf and Taube 2001 TempBMF pas conv 17 16400 14.40 Wachendorf and Taube 2001 TempBMF pas conv 16 6000 13.55 Weibull et al. 2003 TempBMF pas conv 36 2500 31.91 Badea et al. 2011 TempBMF ref 32 2500 28.05 Badea et al. 2011 TempBMF ref 46 2500 40.32 Badea et al. 2011 TempBMF ref 37 2500 32.43 Badea et al. 2011 TempBMF ref 45 2500 39.44 Badea et al. 2011 TempBMF ref 54 2500 47.33 Badea et al. 2011 TempBMF ref 12 2500 10.81 BDM (2004) TempBMF ref 9 1 34.69 BDM (2004) TempBMF ref 17 10 19.84 BDM (2004) TempBMF ref 21 10 24.59 BDM (2004) TempBMF ref 17 10 20.11 BDM (2004) TempBMF ref 19 10 21.87 BDM (2004) TempBMF ref 17 10 19.84 Brunet et al. 1996 TempBMF ref 8 50 8.62 Brunet 2007 TempBMF ref 25 21800 20.37 Carey et al. 2007 TempBMF ref 21 200 20.21 De Keersmaeker et al. 2004 TempBMF ref 10 100 10.00 De Keersmaeker et al. 2004 TempBMF ref 15 100 15.00 De Keersmaeker et al., 2004 TempBMF ref 13 100 13.00
205 Appendix C: Appendix to Chapter 4
Source Biome LU type Org/Conv. S Area (m²] SR Standard. Dumortier et al. 2002 TempBMF ref 36 8300 30.39 Fahy and Gormally 1998 TempBMF ref 17 4 21.96 Gazol and Ibanez 2010 TempBMF ref 13 400 12.20 Goetmark et al. 2005 TempBMF ref 20 10000 16.96 Hofmeister et al. 2009 TempBMF ref 34 38 36.25 Oheimb et al. 2007 TempBMF ref 7 100 6.60 Schmidt 2005 TempBMF ref 23 314 22.08 Schmidt 2005 TempBMF ref 13 314 12.32 Schmidt 2005 TempBMF ref 31 314 29.57 Schmidt 2005 TempBMF ref 26 314 24.64 Schmidt 2005 TempBMF ref 23 314 21.70 Szymura and Szymura 2011 TempBMF ref 31 250 29.67 Thorell and Goetmark 2005 TempBMF ref 23 3316 19.71 Vockenhuber 2011 TempBMF ref 15 2 21.94 Wohlgemuth 1998 TempBMF ref 23 100 23.00 Tamiozzo and Jacobi 2004 TropGL ancro org 24 4 31.00 Tamiozzo and Jacobi 2004 TropGL ancro org 18 4 23.25 Poggio 2005 TropGL ancro conv 10 12 11.83 Poggio 2005 TropGL ancro conv 18 12 20.56 Poggio 2005 TropGL ancro conv 8 1045 7.50 Poggio et al. 2012 TropGL ancro conv 9 600000 6.82 Poggio et al. 2012 TropGL ancro conv 10 600000 7.58 Poggio et al. 2012 TropGL ancro conv 7 600000 5.31 de la Fuente et al. 2006 TropGL ancro conv 15 500000 11.40 de la Fuente et al. 2006 TropGL ancro conv 12 500000 9.13 de la Fuente et al. 2006 TropGL ancro conv 8 500000 6.13 de la Fuente et al. 2010 TropGL ancro conv 8 760000 6.26 de la Fuente et al. 2010 TropGL ancro conv 11 760000 8.15 Vitta et al. 2004 TropGL ancro conv 11 300 10.45 Vitta et al. 2004 TropGL ancro conv 10 300 9.50 Tuesca and Puricelli 2007 TropGL ancro conv 7 5 8.73 Tuesca and Puricelli 2007 TropGL ancro conv 7 5 8.73 Batalha et al. 2001 TropGL ref 39 400 36.40 Batalha et al. 2001 TropGL ref 44 400 41.28 Batalha et al. 2001 TropGL ref 33 400 31.33 Amorim and Batalha 2007 TropGL ref 55 10 64.18 Amorim and Batalha 2008 TropGL ref 14 1 53.26 Cianciaruso and Batalho 2009 TropGL ref 16 1 59.68 Ambinakudige and Sathish 2009 TropMBF permcro conv 38 1250 34.12 Ambinakudige and Sathish 2009 TropMBF permcro conv 39 1250 35.01 Ambinakudige and Sathish 2009 TropMBF permcro conv 38 1250 34.12 Gillison et al. 2004 TropMBF permcro conv 26 200 25.14 Murdiyarso et al. 2002 TropMBF permcro conv 25 200 24.17 Murdiyarso et al. 2002 TropMBF permcro conv 25 200 24.17 Ramadhanil et al. 2008 TropMBF permcro conv 48 40 50.33 Schulze and Fiedler 2004 TropMBF permcro conv 10 10 12.59 Siebert 2002 TropMBF permcro conv 21 10000 17.63 Siebert 2002 TropMBF permcro conv 17 10000 14.27 Tyynela et al. 2003 TropMBF permcro conv 5 1 17.43 Tyynela et al. 2003 TropMBF permcro conv 7 1 22.03 Bos et al. 2008 TropMBF ref 56 2500 48.91 Brearley et al. 2004 TropMBF ref 85 7500 71.97 Gillison et al. 2003 TropMBF ref 102 458 95.16 Gillison et al. 2004 TropMBF ref 40 200 38.68
206 Appendix C: Appendix to Chapter 4
Source Biome LU type Org/Conv. S Area (m²] SR Standard. Gradstein et al. 2007 TropMBF ref 56 2500 48.91 Kessler et al. 2005 TropMBF ref 76 2500 66.62 Murdiyarso et al. 2002 TropMBF ref 120 200 116.03 Ramadhanil et al. 2008 TropMBF ref 46 40 48.75 Schulze and Fiedler 2004 TropMBF ref 22 10 44.13 Siebert 2002 TropMBF ref 45 900000 33.82 Slik et al. 2002 TropMBF ref 76 3000 66.13 Turner et al. 1997 TropMBF ref 53 2000 46.90 Tyynela et al. 2003 TropMBF ref 9 1 27.95 Tyynela et al. 2003 TropMBF ref 8 1 26.31
Table C6 Results of normalisations (*) of species richness (S), endemic species richness (EndS) and Conservation Risk Index (CRI) and their product, the Biodiversity Weighting factor (BWF) for all ecoregion Biome Code Ecoregion S EndS* CRI* BWF 1 AA0101 Admiralty Islands lowland rain forests 1.9 2.2 1.7 7.3 1 AA0102 Banda Sea Islands moist deciduous forests 3.1 2.6 5.4 42.7 1 AA0103 Biak-Numfoor rain forests 2.4 2.8 1.1 7.5 1 AA0104 Buru rain forests 2.7 1.8 2.6 12.6 1 AA0105 Central Range montane rain forests 2.7 1.8 1.1 5.4 1 AA0106 Halmahera rain forests 2.6 2.4 1.5 9.1 1 AA0107 Huon Peninsula montane rain forests 4.1 1.1 3.0 14.1 1 AA0108 Yapen rain forests 3.0 1.0 1.0 3.1 1 AA0109 Lord Howe Island subtropical forests 2.2 1.6 1.0 3.6 1 AA0110 Louisiade Archipelago rain forests 2.5 1.4 1.6 5.6 1 AA0111 New Britain-New Ireland lowland rain forests 2.3 1.8 1.8 7.4 1 AA0112 New Britain-New Ireland montane rain forests 2.1 1.3 2.1 5.5 1 AA0113 New Caledonia rain forests 1.7 1.6 1.1 2.9 1 AA0114 Norfolk Island subtropical forests 2.3 1.5 1.1 3.8 Northern New Guinea lowland rain and freshwater swamp 1 AA0115 forests 2.8 1.4 1.1 4.2 1 AA0116 Northern New Guinea montane rain forests 3.0 1.3 1.1 4.4 1 AA0117 Queensland tropical rain forests 5.1 4.1 1.1 22.7 1 AA0118 Seram rain forests 2.7 2.4 1.1 7.1 1 AA0119 Solomon Islands rain forests 2.4 5.5 1.2 16.3 1 AA0120 Southeastern Papuan rain forests 4.1 1.3 2.9 16.0 1 AA0121 Southern New Guinea freshwater swamp forests 2.8 1.0 1.4 3.8 1 AA0122 Southern New Guinea lowland rain forests 2.8 1.0 1.4 4.0 1 AA0123 Sulawesi lowland rain forests 2.9 3.2 1.5 14.2 1 AA0124 Sulawesi montane rain forests 2.3 2.7 1.1 7.1 1 AA0125 Trobriand Islands rain forests 2.7 1.3 2.3 8.2 1 AA0126 Vanuatu rain forests 1.7 2.8 1.5 6.8 1 AA0127 Vogelkop montane rain forests 3.6 1.4 1.1 5.6 1 AA0128 Vogelkop-Aru lowland rain forests 3.0 1.1 1.2 3.8 1 AT0101 Albertine Rift montane forests 6.3 4.6 1.2 35.5 1 AT0102 Atlantic Equatorial coastal forests 4.4 1.4 1.1 7.0 1 AT0103 Cameroonian Highlands forests 5.9 2.6 1.1 15.9 1 AT0104 Central Congolian lowland forests 2.9 1.0 1.1 3.1 1 AT0105 Comoros forests 2.2 3.6 3.6 28.7 1 AT0106 Cross-Niger transition forests 4.9 1.0 6.5 32.0 1 AT0107 Cross-Sanaga-Bioko coastal forests 5.8 1.3 1.1 7.8 1 AT0108 East African montane forests 5.5 1.8 1.5 14.4 1 AT0109 Eastern Arc forests 6.6 4.4 1.3 36.6 1 AT0110 Eastern Congolian swamp forests 3.7 1.1 1.6 6.5 1 AT0111 Eastern Guinean forests 4.3 1.7 7.0 50.7
207 Appendix C: Appendix to Chapter 4
Biome Code Ecoregion S EndS* CRI* BWF 1 AT0112 Ethiopian montane forests 3.6 1.7 1.7 10.7 1 AT0113 Granitic Seychelles forests 2.7 4.1 1.0 11.2 1 AT0114 Guinean montane forests 4.7 1.9 2.1 18.8 1 AT0115 Knysna-Amatole montane forests 5.1 1.9 1.0 10.1 1 AT0116 KwaZulu-Cape coastal forest mosaic 5.0 1.1 2.4 13.6 1 AT0117 Madagascar lowland forests 3.2 7.3 2.4 55.8 1 AT0118 Madagascar subhumid forests 3.0 2.7 1.3 10.4 1 AT0119 Maputaland coastal forest mosaic 5.0 1.2 1.1 6.6 1 AT0120 Mascarene forests 1.8 3.1 1.0 5.5 1 AT0121 Mount Cameroon and Bioko montane forests 10.0 1.2 1.0 12.7 1 AT0122 Niger Delta swamp forests 4.7 1.0 4.1 19.0 1 AT0123 Nigerian lowland forests 4.3 1.2 4.1 20.8 1 AT0124 Northeastern Congolian lowland forests 3.9 1.4 1.2 6.4 1 AT0125 Northern Zanzibar-Inhambane coastal forest mosaic 5.3 3.0 1.2 19.0 1 AT0126 Northwestern Congolian lowland forests 3.9 1.2 1.1 4.8 1 AT0127 Sao Tome, Principe and Annobon moist lowland forests 3.0 4.9 1.1 15.8 1 AT0128 Southern Zanzibar-Inhambane coastal forest mosaic 4.1 1.5 1.4 8.3 1 AT0129 Western Congolian swamp forests 3.4 1.0 1.1 3.7 1 AT0130 Western Guinean lowland forests 4.0 2.0 2.9 22.7 1 IM0101 Andaman Islands rain forests 2.4 2.6 1.4 9.2 1 IM0102 Borneo lowland rain forests 3.0 1.4 1.8 7.5 1 IM0103 Borneo montane rain forests 2.8 1.0 1.0 2.8 1 IM0104 Borneo peat swamp forests 3.7 1.0 2.2 8.3 1 IM0105 Brahmaputra Valley semi-evergreen forests 3.7 1.0 3.3 12.0 1 IM0106 Cardamom Mountains rain forests 3.8 1.6 1.1 6.6 1 IM0107 Chao Phraya freshwater swamp forests 3.8 1.2 9.8 45.3 1 IM0108 Chao Phraya lowland moist deciduous forests 4.4 1.0 1.8 8.2 1 IM0109 Chin Hills-Arakan Yoma montane forests 4.2 1.0 7.0 29.5 1 IM0110 Christmas and Cocos Islands tropical forests 1.6 1.9 1.0 3.0 1 IM0111 Eastern highlands moist deciduous forests 2.4 1.1 1.8 4.9 1 IM0112 Eastern Java-Bali montane rain forests 3.3 1.0 1.4 4.5 1 IM0113 Eastern Java-Bali rain forests 3.4 1.0 2.9 9.6 1 IM0114 Greater Negros-Panay rain forests 3.6 3.6 2.2 29.0 1 IM0115 Himalayan subtropical broadleaf forests 3.6 1.0 1.5 5.3 1 IM0116 Irrawaddy freshwater swamp forests 3.4 1.0 9.8 33.7 1 IM0117 Irrawaddy moist deciduous forests 3.2 1.0 6.4 20.4 1 IM0118 Jian Nan subtropical evergreen forests 3.8 1.3 1.4 7.0 1 IM0119 Kayah-Karen montane rain forests 4.8 1.8 1.3 11.0 1 IM0120 Lower Gangetic Plains moist deciduous forests 3.0 1.1 5.3 16.7 1 IM0121 Luang Prabang montane rain forests 4.3 1.0 1.4 6.2 1 IM0122 Luzon montane rain forests 3.1 1.0 1.1 3.3 1 IM0123 Luzon rain forests 3.6 2.3 1.8 14.7 1 IM0124 Malabar Coast moist forests 3.5 1.5 5.1 27.0 Maldives-Lakshadweep-Chagos Archipelago tropical 1 IM0125 moist forests 2.2 1.0 1.6 3.5 1 IM0126 Meghalaya subtropical forests 4.6 1.5 6.1 41.7 1 IM0127 Mentawai Islands rain forests 3.4 2.3 1.0 7.8 1 IM0128 Mindanao montane rain forests 3.0 2.1 1.3 8.1 1 IM0129 Mindanao-Eastern Visayas rain forests 3.4 1.7 2.8 16.2 1 IM0130 Mindoro rain forests 3.7 1.3 1.6 7.5 1 IM0131 Mizoram-Manipur-Kachin rain forests 4.3 1.3 5.1 28.4 1 IM0132 Myanmar coastal rain forests 3.7 1.3 8.1 38.1 1 IM0133 Nicobar Islands rain forests 2.3 1.2 1.1 3.0 1 IM0134 North Western Ghats moist deciduous forests 3.5 1.1 2.0 7.9 1 IM0135 North Western Ghats montane rain forests 3.9 2.3 1.2 10.6 1 IM0136 Northern Annamites rain forests 4.5 1.5 1.1 7.5 1 IM0137 Northern Indochina subtropical forests 4.7 1.8 2.0 17.4 1 IM0138 Northern Khorat Plateau moist deciduous forests 4.5 1.0 6.4 29.2
208 Appendix C: Appendix to Chapter 4
Biome Code Ecoregion S EndS* CRI* BWF 1 IM0139 Northern Thailand-Laos moist deciduous forests 3.9 1.0 1.8 7.1 1 IM0140 Northern Triangle subtropical forests 4.0 1.7 3.2 22.0 1 IM0141 Northern Vietnam lowland rain forests 3.9 1.1 3.0 13.0 1 IM0142 Orissa semi-evergreen forests 3.1 1.0 8.7 26.9 1 IM0143 Palawan rain forests 3.9 3.7 1.0 14.9 1 IM0144 Peninsular Malaysian montane rain forests 5.0 2.1 1.0 10.9 1 IM0145 Peninsular Malaysian peat swamp forests 7.1 1.0 4.1 29.2 1 IM0146 Peninsular Malaysian rain forests 4.6 1.5 2.6 18.1 1 IM0147 Red River freshwater swamp forests 4.1 1.0 10.0 40.5 1 IM0148 South China Sea Islands 1.1 1.0 3.3 3.6 1 IM0149 South China-Vietnam subtropical evergreen forests 4.9 1.4 1.9 12.9 1 IM0150 South Western Ghats moist deciduous forests 4.1 1.1 1.1 5.2 1 IM0151 South Western Ghats montane rain forests 4.6 5.7 1.0 27.6 1 IM0152 Southern Annamites montane rain forests 3.9 1.5 1.2 7.0 1 IM0153 Southwest Borneo freshwater swamp forests 3.8 1.0 2.1 8.1 1 IM0154 Sri Lanka lowland rain forests 3.7 1.9 2.4 17.4 1 IM0155 Sri Lanka montane rain forests 4.7 5.6 1.3 35.1 1 IM0156 Sulu Archipelago rain forests 4.2 2.0 4.8 40.5 1 IM0157 Sumatran freshwater swamp forests 4.7 1.0 3.0 14.1 1 IM0158 Sumatran lowland rain forests 2.6 1.2 2.2 6.8 1 IM0159 Sumatran montane rain forests 3.7 1.5 1.1 5.9 1 IM0160 Sumatran peat swamp forests 3.8 1.1 2.1 8.8 1 IM0161 Sundaland heath forests 3.2 1.0 2.1 6.8 1 IM0162 Sundarbans freshwater swamp forests 3.0 1.0 9.9 29.8 1 IM0163 Tenasserim-South Thailand semi-evergreen rain forests 5.2 1.7 1.4 12.7 1 IM0164 Tonle Sap freshwater swamp forests 3.7 1.1 1.6 6.4 1 IM0165 Tonle Sap-Mekong peat swamp forests 3.4 1.0 9.5 32.8 1 IM0166 Upper Gangetic Plains moist deciduous forests 2.4 1.0 9.2 22.3 1 IM0167 Western Java montane rain forests 2.9 1.7 1.8 8.8 1 IM0168 Western Java rain forests 3.6 1.0 3.2 11.5 1 IM0169 Hainan Island monsoon rain forests 5.2 1.0 1.0 5.4 1 IM0170 Nansei Islands subtropical evergreen forests 3.5 6.9 1.1 27.2 1 IM0171 South Taiwan monsoon rain forests 6.5 1.0 1.1 7.1 1 IM0172 Taiwan subtropical evergreen forests 4.0 1.6 1.2 7.6 1 NT0101 Araucaria moist forests 3.6 1.1 5.4 21.7 1 NT0102 Atlantic Coast restingas 4.8 1.5 1.9 14.2 1 NT0103 Bahia coastal forests 3.6 1.5 6.3 33.1 1 NT0104 Bahia interior forests 3.4 1.2 7.2 28.8 1 NT0105 Bolivian Yungas 4.7 2.3 1.0 11.1 1 NT0106 Caatinga Enclaves moist forests 1.0 1.0 2.5 2.5 1 NT0107 Caqueta moist forests 4.2 1.1 1.0 4.9 1 NT0108 Catatumbo moist forests 5.7 1.4 1.2 9.5 1 NT0109 Cauca Valley montane forests 6.2 2.6 2.6 41.4 1 NT0110 Cayos Miskitos-San AndrΘs and Providencia moist forests 2.8 2.4 1.0 6.9 1 NT0111 Central American Atlantic moist forests 4.6 1.3 1.3 7.7 1 NT0112 Central American montane forests 6.7 1.1 1.1 8.7 1 NT0113 Chiapas montane forests 7.7 1.2 1.7 15.8 1 NT0114 Chimalapas montane forests 8.6 2.1 1.5 26.7 1 NT0115 Choc≤-DariΘn moist forests 6.8 4.3 1.2 34.4 1 NT0116 Cocos Island moist forests 1.6 1.0 1.0 1.6 1 NT0117 Cordillera La Costa montane forests 8.1 5.0 1.0 42.1 1 NT0118 Cordillera Oriental montane forests 6.4 3.7 1.2 27.6 1 NT0119 Costa Rican seasonal moist forests 7.4 1.0 1.7 12.9 1 NT0120 Cuban moist forests 3.1 1.1 1.2 4.4 1 NT0121 Eastern Cordillera real montane forests 5.9 6.3 1.2 43.5 1 NT0122 Eastern Panamanian montane forests 9.9 2.4 1.0 24.1 1 NT0123 Fernando de Noronha-Atol das Rocas moist forests 1.6 2.2 1.0 3.6 1 NT0124 Guianan Highlands moist forests 4.6 1.9 1.0 8.7
209 Appendix C: Appendix to Chapter 4
Biome Code Ecoregion S EndS* CRI* BWF 1 NT0125 Guianan moist forests 4.4 2.0 1.0 9.0 1 NT0126 Gurupa varzeß 7.7 1.0 2.1 15.8 1 NT0127 Hispaniolan moist forests 2.4 1.5 3.5 12.6 1 NT0128 Iquitos varzeß 5.7 1.2 1.0 7.2 1 NT0129 Isthmian-Atlantic moist forests 6.3 1.6 1.2 12.8 1 NT0130 Isthmian-Pacific moist forests 6.6 2.0 1.8 24.7 1 NT0131 Jamaican moist forests 3.3 1.3 1.4 6.4 1 NT0132 Japurß-Solimoes-Negro moist forests 3.5 1.1 1.0 4.0 1 NT0133 Juruß-Purus moist forests 3.5 1.0 1.1 3.8 1 NT0134 Leeward Islands moist forests 4.0 1.3 1.3 7.2 1 NT0135 Madeira-Tapaj≤s moist forests 2.9 1.1 1.2 3.9 1 NT0136 Magdalena Valley montane forests 4.8 3.1 3.5 53.8 1 NT0137 Magdalena-Urabß moist forests 5.2 1.2 6.3 38.6 1 NT0138 Maraj≤ varzeß 4.1 1.1 1.1 4.7 1 NT0139 Maranhπo Babaτu forests 2.5 1.0 6.1 15.5 1 NT0140 Mato Grosso seasonal forests 2.8 1.1 2.9 8.7 1 NT0141 Monte Alegre varzeß 5.5 1.2 1.4 9.0 1 NT0142 Napo moist forests 5.3 2.6 1.1 15.3 1 NT0143 Negro-Branco moist forests 4.1 1.6 1.0 6.6 1 NT0144 Northeastern Brazil restingas 4.6 1.0 1.5 6.8 1 NT0145 Northwestern Andean montane forests 6.3 9.0 1.3 76.0 1 NT0146 Oaxacan montane forests 7.0 1.9 3.8 49.9 1 NT0147 Orinoco Delta swamp forests 5.2 1.2 1.0 6.4 1 NT0148 Pantanos de Centla 5.3 1.0 1.2 6.3 1 NT0149 Guianan freshwater swamp forests 8.9 1.0 1.2 10.5 1 NT0150 Alto Paranß Atlantic forests 3.5 1.0 6.9 23.8 1 NT0151 Pernambuco coastal forests 5.0 1.7 9.0 76.6 1 NT0152 Pernambuco interior forests 4.3 1.0 8.9 38.0 1 NT0153 Peruvian Yungas 4.6 5.7 1.1 29.7 1 NT0154 PetΘn-Veracruz moist forests 5.1 2.2 1.2 12.7 1 NT0155 Puerto Rican moist forests 3.1 1.4 4.1 17.2 1 NT0156 Purus varzeß 4.9 1.0 1.0 5.0 1 NT0157 Purus-Madeira moist forests 3.8 1.0 1.1 4.3 1 NT0158 Rio Negro campinarana 3.4 1.0 1.0 3.4 1 NT0159 Santa Marta montane forests 9.0 4.7 1.1 45.1 1 NT0160 Serra do Mar coastal forests 4.5 1.7 3.5 26.4 1 NT0161 Sierra de los Tuxtlas 9.6 3.0 1.1 32.8 1 NT0162 Sierra Madre de Chiapas moist forests 6.8 1.8 1.3 15.7 1 NT0163 Solim⌡es-Japurß moist forests 5.0 1.1 1.0 5.7 1 NT0164 South Florida rocklands 5.2 1.3 1.1 7.2 1 NT0165 Southern Andean Yungas 3.7 2.1 1.1 8.7 1 NT0166 Southwest Amazon moist forests 4.3 2.8 1.0 12.6 1 NT0167 Talamancan montane forests 8.4 10.0 1.0 86.5 1 NT0168 Tapaj≤s-Xingu moist forests 3.2 1.1 1.3 4.5 1 NT0169 Pantepui 5.9 4.5 1.0 26.4 1 NT0170 Tocantins/Pindare moist forests 3.4 1.1 4.0 14.4 1 NT0171 Trinidad and Tobago moist forests 5.7 1.2 3.3 22.5 1 NT0172 Trindade-Martin Vaz Islands tropical forests 1.0 1.0 1.1 1.1 1 NT0173 Uatuma-Trombetas moist forests 3.3 1.0 1.1 3.4 1 NT0174 Ucayali moist forests 5.4 1.7 1.3 12.1 1 NT0175 Venezuelan Andes montane forests 6.1 4.4 1.1 28.4 1 NT0176 Veracruz moist forests 4.4 1.6 3.7 26.2 1 NT0177 Veracruz montane forests 6.9 1.4 1.9 18.3 1 NT0178 Western Ecuador moist forests 6.6 2.8 2.4 43.3 1 NT0179 Windward Islands moist forests 3.7 3.2 1.0 11.7 1 NT0180 Xingu-Tocantins-Araguaia moist forests 3.2 1.1 2.8 9.5 1 NT0181 Yucatßn moist forests 4.3 1.3 1.0 6.0 1 NT0182 Guianan piedmont and lowland moist forests 1.0 1.0 1.0 1.0
210 Appendix C: Appendix to Chapter 4
Biome Code Ecoregion S EndS* CRI* BWF 1 OC0101 Carolines tropical moist forests 2.1 2.8 1.1 6.3 1 OC0102 Central Polynesian tropical moist forests 1.3 1.3 1.1 1.8 1 OC0103 Cook Islands tropical moist forests 1.5 2.5 1.1 4.2 1 OC0104 Eastern Micronesia tropical moist forests 1.6 1.3 1.1 2.2 1 OC0105 Fiji tropical moist forests 1.6 2.4 1.1 4.2 1 OC0106 Hawaii tropical moist forests 1.7 1.8 1.2 3.5 1 OC0107 Kermadec Islands subtropical moist forests 1.6 1.0 1.1 1.7 1 OC0108 Marquesas tropical moist forests 1.3 3.0 1.1 4.4 1 OC0109 Ogasawara subtropical moist forests 1.1 1.9 1.1 2.3 1 OC0110 Palau tropical moist forests 2.5 2.3 4.5 25.1 1 OC0111 Rapa Nui subtropical broadleaf forests 1.0 1.0 1.1 1.1 1 OC0112 Samoan tropical moist forests 1.6 2.1 1.1 3.6 1 OC0113 Society Islands tropical moist forests 1.3 1.5 1.1 2.1 1 OC0114 Tongan tropical moist forests 1.6 1.3 1.1 2.1 1 OC0115 Tuamotu tropical moist forests 1.4 2.6 1.1 3.9 1 OC0116 Tubuai tropical moist forests 1.3 1.4 1.1 2.0 1 OC0117 Western Polynesian tropical moist forests 1.7 1.0 1.1 1.8 1 PA0101 Guizhou Plateau broadleaf and mixed forests 3.9 1.1 1.4 5.9 1 PA0102 Yunnan Plateau subtropical evergreen forests 4.9 1.1 1.3 7.4 2 AA0201 Lesser Sundas deciduous forests 2.8 2.1 1.6 9.6 2 AA0202 New Caledonia dry forests 1.8 1.0 1.8 3.2 2 AA0203 Sumba deciduous forests 2.4 1.4 3.1 10.3 2 AA0204 Timor and Wetar deciduous forests 2.5 1.7 3.4 14.5 2 AT0201 Cape Verde Islands dry forests 1.8 2.8 1.1 5.4 2 AT0202 Madagascar dry deciduous forests 2.5 2.8 1.6 11.2 2 AT0203 Zambezian Cryptosepalum dry forests 3.9 1.0 1.0 3.9 2 IM0201 Central Deccan Plateau dry deciduous forests 2.5 1.1 4.0 10.8 2 IM0202 Central Indochina dry forests 3.6 1.2 1.9 8.0 2 IM0203 Chhota-Nagpur dry deciduous forests 2.6 1.0 2.1 5.5 2 IM0204 East Deccan dry-evergreen forests 3.0 1.0 5.5 16.7 2 IM0205 Irrawaddy dry forests 3.5 1.1 9.3 36.0 2 IM0206 Khathiar-Gir dry deciduous forests 2.5 1.0 3.0 7.5 2 IM0207 Narmada Valley dry deciduous forests 2.5 1.0 2.1 5.2 2 IM0208 Northern dry deciduous forests 2.7 1.0 2.8 7.7 2 IM0209 South Deccan Plateau dry deciduous forests 2.7 1.0 5.7 15.3 2 IM0210 Southeastern Indochina dry evergreen forests 3.5 1.0 1.3 4.6 2 IM0211 Southern Vietnam lowland dry forests 3.6 1.1 2.4 9.5 2 IM0212 Sri Lanka dry-zone dry evergreen forests 3.2 1.9 1.3 7.8 2 NA0201 Sonoran-Sinaloan transition subtropical dry forest 2.1 1.0 2.2 4.7 2 NT0201 Apure-Villavicencio dry forests 4.6 1.2 1.5 8.0 2 NT0202 Atlantic dry forests 2.8 1.2 6.6 21.4 2 NT0204 Bajφo dry forests 3.8 1.0 4.2 15.8 2 NT0205 Balsas dry forests 3.9 1.6 1.8 11.4 2 NT0206 Bolivian montane dry forests 2.9 1.6 1.2 5.5 2 NT0207 Cauca Valley dry forests 4.6 1.0 6.6 30.5 2 NT0209 Central American dry forests 4.2 1.1 3.0 13.7 2 NT0211 Chiapas Depression dry forests 4.3 1.4 5.2 29.9 2 NT0212 Chiquitano dry forests 3.5 1.0 1.4 4.7 2 NT0213 Cuban dry forests 2.4 1.0 2.0 4.8 2 NT0214 Ecuadorian dry forests 3.9 1.0 4.1 15.8 2 NT0215 Hispaniolan dry forests 2.6 1.0 1.4 3.8 2 NT0216 Islas Revillagigedo dry forests 1.6 1.8 1.0 3.0 2 NT0217 Jalisco dry forests 4.7 1.2 2.0 11.3 2 NT0218 Jamaican dry forests 3.3 1.0 1.1 3.7 2 NT0219 Lara-Falc≤n dry forests 4.2 1.1 1.3 6.3 2 NT0220 Lesser Antillean dry forests 2.8 1.0 1.1 3.0 2 NT0221 Magdalena Valley dry forests 4.4 1.0 7.2 31.6 2 NT0222 Maracaibo dry forests 4.3 1.2 1.3 6.9
211 Appendix C: Appendix to Chapter 4
Biome Code Ecoregion S EndS* CRI* BWF 2 NT0223 Mara±≤n dry forests 3.0 1.7 2.4 12.5 2 NT0224 Panamanian dry forests 5.2 1.0 9.4 48.7 2 NT0225 Patφa Valley dry forests 3.4 1.0 6.5 22.3 2 NT0226 Puerto Rican dry forests 3.3 1.2 1.5 5.7 2 NT0227 Sierra de la Laguna dry forests 3.5 1.2 1.0 4.1 2 NT0228 Sinaloan dry forests 4.0 1.6 2.1 12.8 2 NT0229 Sin· Valley dry forests 5.4 1.6 3.9 33.6 2 NT0230 Southern Pacific dry forests 5.0 1.8 3.1 28.6 2 NT0232 Tumbes-Piura dry forests 3.4 1.7 1.8 9.8 2 NT0233 Veracruz dry forests 4.9 1.3 7.2 44.9 2 NT0235 Yucatßn dry forests 3.6 1.4 2.2 11.4 2 OC0201 Fiji tropical dry forests 1.0 1.0 1.1 1.1 2 OC0202 Hawaii tropical dry forests 1.6 1.0 1.3 2.0 2 OC0203 Marianas tropical dry forests 1.7 1.6 8.4 23.1 2 OC0204 Yap tropical dry forests 2.0 1.7 5.4 17.9 3 IM0301 Himalayan subtropical pine forests 3.8 1.0 2.1 8.1 3 IM0302 Luzon tropical pine forests 2.6 1.5 1.2 4.8 3 IM0303 Northeast India-Myanmar pine forests 3.5 1.0 7.1 24.8 3 IM0304 Sumatran tropical pine forests 3.4 1.0 1.0 3.5 3 NA0301 Bermuda subtropical conifer forests 4.9 1.0 1.0 5.1 3 NA0302 Sierra Madre Occidental pine-oak forests 1.7 1.2 1.1 2.2 3 NA0303 Sierra Madre Oriental pine-oak forests 4.8 1.6 1.1 8.4 3 NT0301 Bahamian pine mosaic 2.7 1.4 1.1 4.0 3 NT0302 Belizian pine forests 5.1 1.0 1.0 5.2 3 NT0303 Central American pine-oak forests 4.9 1.6 2.9 22.7 3 NT0304 Cuban pine forests 3.1 1.0 1.5 4.7 3 NT0305 Hispaniolan pine forests 2.6 1.0 1.3 3.3 3 NT0306 Miskito pine forests 3.3 1.0 1.3 4.1 3 NT0307 Sierra de la Laguna pine-oak forests 3.4 1.2 1.0 4.0 3 NT0308 Sierra Madre de Oaxaca pine-oak forests 7.0 1.7 1.4 16.5 3 NT0309 Sierra Madre del Sur pine-oak forests 4.5 2.9 2.1 26.7 3 NT0310 Trans-Mexican Volcanic Belt pine-oak forests 4.5 1.9 1.4 12.1 4 AA0401 Chatham Island temperate forests 1.0 1.0 1.1 1.1 4 AA0402 Eastern Australian temperate forests 3.7 3.6 1.6 21.5 4 AA0403 Fiordland temperate forests 1.3 1.1 1.0 1.5 4 AA0404 Nelson Coast temperate forests 1.4 1.0 1.0 1.4 4 AA0405 North Island temperate forests 1.4 1.0 1.0 1.4 4 AA0406 Northland temperate kauri forests 1.5 1.3 1.3 2.5 4 AA0407 Rakiura Island temperate forests 1.5 1.2 1.0 1.7 4 AA0408 Richmond temperate forests 1.5 1.3 1.0 2.0 4 AA0409 Southeast Australia temperate forests 3.1 1.5 4.8 21.7 4 AA0410 South Island temperate forests 1.3 1.0 1.3 1.7 4 AA0411 Tasmanian Central Highland forests 2.3 1.1 1.2 3.0 4 AA0412 Tasmanian temperate forests 2.5 1.0 1.7 4.2 4 AA0413 Tasmanian temperate rain forests 2.2 1.2 1.2 3.1 4 AA0414 Westland temperate forests 1.4 1.0 1.0 1.4 4 IM0401 Eastern Himalayan broadleaf forests 3.9 1.2 1.4 7.0 4 IM0402 Northern Triangle temperate forests 4.2 1.3 2.9 16.6 4 IM0403 Western Himalayan broadleaf forests 2.1 1.3 1.7 4.3 4 NA0401 Allegheny Highlands forests 2.5 1.0 1.3 3.1 4 NA0402 Appalachian mixed mesophytic forests 2.5 1.1 1.1 3.0 4 NA0403 Appalachian-Blue Ridge forests 2.6 1.6 1.1 4.4 4 NA0404 Central U.S. hardwood forests 2.5 1.0 1.6 3.9 4 NA0405 East Central Texas forests 2.8 1.0 5.8 16.2 4 NA0406 Eastern forest-boreal transition 2.1 1.0 1.0 2.1 4 NA0407 Eastern Great Lakes lowland forests 2.4 1.0 4.9 11.9 4 NA0408 Gulf of St. Lawrence lowland forests 2.4 1.0 1.9 4.4 4 NA0409 Mississippi lowland forests 2.6 1.0 5.0 13.2
212 Appendix C: Appendix to Chapter 4
Biome Code Ecoregion S EndS* CRI* BWF 4 NA0410 New England-Acadian forests 2.2 1.0 1.1 2.4 4 NA0411 Northeastern coastal forests 2.7 1.0 3.2 8.7 4 NA0412 Ozark Mountain forests 2.7 1.3 1.1 3.9 4 NA0413 Southeastern mixed forests 2.5 1.0 1.6 4.1 4 NA0414 Southern Great Lakes forests 2.4 1.0 6.0 14.2 4 NA0415 Upper Midwest forest-savanna transition 2.4 1.0 4.8 11.3 4 NA0416 Western Great Lakes forests 2.2 1.1 1.0 2.4 4 NA0417 Willamette Valley forests 2.5 1.2 6.7 20.6 4 NT0401 Juan Fernßndez Islands temperate forests 1.2 1.9 1.1 2.5 4 NT0402 Magellanic subpolar forests 1.8 1.2 1.0 2.1 4 NT0403 San FΘlix-San Ambrosio Islands temperate forests 1.0 1.0 1.1 1.1 4 NT0404 Valdivian temperate forests 2.0 2.2 1.1 4.6 4 PA0401 Appenine deciduous montane forests 2.5 1.0 1.1 2.7 4 PA0402 Atlantic mixed forests 2.4 1.0 7.8 18.6 4 PA0403 Azores temperate mixed forests 1.3 1.1 1.1 1.6 4 PA0404 Balkan mixed forests 2.7 1.0 7.5 20.4 4 PA0405 Baltic mixed forests 2.4 1.0 5.4 12.8 4 PA0406 Cantabrian mixed forests 2.6 1.1 1.4 4.0 4 PA0407 Caspian Hyrcanian mixed forests 3.0 1.1 1.5 4.8 4 PA0408 Caucasus mixed forests 2.8 1.4 1.4 5.1 4 PA0409 Celtic broadleaf forests 2.1 1.0 1.3 2.6 4 PA0410 Central Anatolian steppe and woodlands 2.3 1.0 6.5 15.1 4 PA0411 Central China loess plateau mixed forests 3.7 1.0 4.6 17.2 4 PA0412 Central European mixed forests 2.3 1.0 2.9 6.6 4 PA0413 Central Korean deciduous forests 1.8 1.0 2.6 4.7 4 PA0414 Changbai Mountains mixed forests 3.3 1.0 1.1 3.7 4 PA0415 Changjiang Plain evergreen forests 3.9 1.0 2.0 8.0 4 PA0416 Crimean Submediterranean forest complex 2.7 1.0 1.4 3.9 4 PA0417 Daba Mountains evergreen forests 4.0 1.0 1.9 7.7 4 PA0418 Dinaric Mountains mixed forests 2.8 1.0 5.2 14.4 4 PA0419 East European forest steppe 2.3 1.0 2.8 6.3 4 PA0420 Eastern Anatolian deciduous forests 2.0 1.0 6.3 12.7 4 PA0421 English Lowlands beech forests 2.3 1.0 3.6 8.3 4 PA0422 Euxine-Colchic broadleaf forests 2.7 1.0 4.3 11.7 4 PA0423 Hokkaido deciduous forests 2.5 1.0 1.4 3.4 4 PA0424 Huang He Plain mixed forests 3.5 1.0 8.5 29.7 4 PA0425 Madeira evergreen forests 1.3 1.5 1.1 2.2 4 PA0426 Manchurian mixed forests 2.4 1.0 1.2 2.9 4 PA0427 Nihonkai evergreen forests 2.9 1.4 1.2 4.9 4 PA0428 Nihonkai montane deciduous forests 2.7 1.1 1.1 3.2 4 PA0429 North Atlantic moist mixed forests 2.1 1.0 1.0 2.2 4 PA0430 Northeast China Plain deciduous forests 3.3 1.2 6.1 24.1 4 PA0431 Pannonian mixed forests 2.4 1.0 3.3 8.0 4 PA0432 Po Basin mixed forests 2.8 1.0 2.8 7.9 4 PA0433 Pyrenees conifer and mixed forests 2.7 1.2 1.0 3.3 4 PA0434 Qin Ling Mountains deciduous forests 4.6 1.1 1.2 6.4 4 PA0435 Rodope montane mixed forests 3.1 1.0 3.0 9.3 4 PA0436 Sarmatic mixed forests 2.1 1.0 1.6 3.4 4 PA0437 Sichuan Basin evergreen broadleaf forests 4.6 1.1 9.4 47.1 4 PA0438 South Sakhalin-Kurile mixed forests 2.5 1.0 1.1 2.6 4 PA0439 Southern Korea evergreen forests 1.8 1.1 3.4 6.9 4 PA0440 Taiheiyo evergreen forests 2.6 1.5 1.3 5.0 4 PA0441 Taiheiyo montane deciduous forests 2.7 1.0 1.1 3.0 4 PA0442 Tarim Basin deciduous forests and steppe 2.1 1.0 2.0 4.2 4 PA0443 Ussuri broadleaf and mixed forests 2.2 1.0 1.0 2.3 4 PA0444 Western Siberian hemiboreal forests 2.2 1.0 1.4 3.1 4 PA0445 Western European broadleaf forests 2.3 1.0 3.1 7.2 4 PA0446 Zagros Mountains forest steppe 2.4 1.1 1.5 3.6
213 Appendix C: Appendix to Chapter 4
Biome Code Ecoregion S EndS* CRI* BWF 5 IM0501 Eastern Himalayan subalpine conifer forests 2.6 1.3 1.1 3.6 5 IM0502 Western Himalayan subalpine conifer forests 2.9 1.1 1.7 5.3 5 NA0501 Alberta Mountain forests 2.3 1.0 1.0 2.3 5 NA0502 Alberta-British Columbia foothills forests 2.1 1.0 1.5 3.0 5 NA0503 Arizona Mountains forests 2.5 1.1 1.0 2.7 5 NA0504 Atlantic coastal pine barrens 2.9 1.1 2.2 6.9 5 NA0505 Blue Mountains forests 2.4 1.0 1.0 2.4 5 NA0506 British Columbia mainland coastal forests 2.4 1.0 1.0 2.4 5 NA0507 Cascade Mountains leeward forests 2.4 1.0 1.0 2.4 5 NA0508 Central and Southern Cascades forests 2.6 1.2 1.0 3.1 5 NA0509 Central British Columbia Mountain forests 2.1 1.0 1.0 2.2 5 NA0510 Central Pacific coastal forests 2.5 1.2 1.0 2.9 5 NA0511 Colorado Rockies forests 2.3 1.0 1.0 2.3 5 NA0512 Eastern Cascades forests 2.5 1.0 1.0 2.5 5 NA0513 Florida sand pine scrub 3.0 1.0 1.1 3.4 5 NA0514 Fraser Plateau and Basin complex 2.1 1.0 1.0 2.1 5 NA0515 Great Basin montane forests 2.5 1.0 1.0 2.5 5 NA0516 Klamath-Siskiyou forests 2.7 1.0 1.0 2.7 5 NA0517 Middle Atlantic coastal forests 2.7 1.1 1.7 5.2 5 NA0518 North Central Rockies forests 2.3 1.0 1.0 2.3 5 NA0519 Northern California coastal forests 3.1 1.2 1.1 3.9 5 NA0520 Northern Pacific coastal forests 1.9 1.0 1.0 1.9 5 NA0521 Northern transitional alpine forests 2.1 1.0 1.1 2.2 5 NA0522 Okanagan dry forests 2.5 1.0 1.0 2.6 5 NA0523 Piney Woods forests 2.5 1.0 1.2 3.1 5 NA0524 Puget lowland forests 2.5 1.0 2.7 6.8 5 NA0525 Queen Charlotte Islands 2.0 1.0 1.0 2.0 5 NA0526 Sierra Juarez and San Pedro Martir pine-oak forests 1.6 1.0 1.0 1.6 5 NA0527 Sierra Nevada forests 2.5 1.3 1.0 3.4 5 NA0528 South Central Rockies forests 2.3 1.0 1.0 2.3 5 NA0529 Southeastern conifer forests 2.7 1.5 1.5 6.4 5 NA0530 Wasatch and Uinta montane forests 2.5 1.0 1.0 2.5 5 PA0501 Alps conifer and mixed forests 2.6 1.2 1.0 3.3 5 PA0502 Altai montane forest and forest steppe 1.9 1.0 1.6 3.1 5 PA0503 Caledon conifer forests 2.0 1.0 1.0 2.1 5 PA0504 Carpathian montane forests 2.5 1.2 1.3 3.7 5 PA0505 Da Hinggan-Dzhagdy Mountains conifer forests 2.7 1.0 1.1 2.9 5 PA0506 East Afghan montane conifer forests 2.3 2.4 1.6 8.8 5 PA0507 Elburz Range forest steppe 2.2 1.0 1.2 2.6 5 PA0508 Helanshan montane conifer forests 2.6 1.0 1.1 2.9 5 PA0509 Hengduan Mountains subalpine conifer forests 5.3 1.2 1.1 6.9 5 PA0510 Hokkaido montane conifer forests 2.4 1.0 1.1 2.7 5 PA0511 Honshu alpine conifer forests 2.7 1.0 1.1 2.9 5 PA0512 Khangai Mountains conifer forests 2.2 1.0 1.1 2.3 5 PA0513 Mediterranean conifer and mixed forests 2.8 1.1 1.7 5.2 5 PA0514 Northeastern Himalayan subalpine conifer forests 2.8 1.0 1.1 3.1 5 PA0515 Northern Anatolian conifer and deciduous forests 2.2 1.0 7.1 15.6 5 PA0516 Nujiang Langcang Gorge alpine conifer and mixed forests 3.8 1.2 1.0 4.5 5 PA0517 Qilian Mountains conifer forests 4.3 1.0 1.0 4.4 5 PA0518 Qionglai-Minshan conifer forests 5.3 1.2 1.1 7.0 5 PA0519 Sayan montane conifer forests 2.5 1.0 1.0 2.6 5 PA0520 Scandinavian coastal conifer forests 2.3 1.0 2.1 4.9 5 PA0521 Tian Shan montane conifer forests 3.5 1.0 1.3 4.4 6 NA0601 Alaska Peninsula montane taiga 1.7 1.0 1.0 1.7 6 NA0602 Central Canadian Shield forests 1.8 1.0 1.0 1.9 6 NA0603 Cook Inlet taiga 1.9 1.0 1.0 1.9 6 NA0604 Copper Plateau taiga 1.9 1.0 1.0 1.9 6 NA0605 Eastern Canadian forests 1.8 1.1 1.0 1.9
214 Appendix C: Appendix to Chapter 4
Biome Code Ecoregion S EndS* CRI* BWF 6 NA0606 Eastern Canadian Shield taiga 1.5 1.0 1.0 1.5 6 NA0607 Interior Alaska-Yukon lowland taiga 1.6 1.0 1.0 1.6 6 NA0608 Mid-Continental Canadian forests 1.9 1.0 1.0 2.0 6 NA0609 Midwestern Canadian Shield forests 1.8 1.0 1.0 1.9 6 NA0610 Muskwa-Slave Lake forests 1.8 1.0 1.0 1.9 6 NA0611 Newfoundland Highland forests 1.8 1.0 1.0 1.8 6 NA0612 Northern Canadian Shield taiga 1.6 1.0 1.0 1.6 6 NA0613 Northern Cordillera forests 1.9 1.0 1.0 1.9 6 NA0614 Northwest Territories taiga 1.7 1.0 1.0 1.7 6 NA0615 South Avalon-Burin oceanic barrens 2.1 1.0 1.0 2.1 6 NA0616 Southern Hudson Bay taiga 1.8 1.0 1.0 1.8 6 NA0617 Yukon Interior dry forests 1.8 1.0 1.1 2.0 6 PA0601 East Siberian taiga 1.9 1.0 1.1 2.0 6 PA0602 Iceland boreal birch forests and alpine tundra 1.3 1.0 1.1 1.5 6 PA0603 Kamchatka-Kurile meadows and sparse forests 1.8 1.0 1.0 1.8 6 PA0604 Kamchatka-Kurile taiga 1.8 1.0 1.5 2.6 6 PA0605 Northeast Siberian taiga 1.6 1.0 1.1 1.8 6 PA0606 Okhotsk-Manchurian taiga 2.0 1.0 1.0 2.1 6 PA0607 Sakhalin Island taiga 2.0 1.0 1.1 2.1 6 PA0608 Scandinavian and Russian taiga 2.0 1.0 1.1 2.2 6 PA0609 Trans-Baikal conifer forests 2.3 1.0 1.1 2.5 6 PA0610 Ural montane forests and tundra 2.2 1.0 1.0 2.3 6 PA0611 West Siberian taiga 1.9 1.0 1.0 1.9 7 AA0701 Arnhem Land tropical savanna 3.1 1.8 1.1 5.9 7 AA0702 Brigalow tropical savanna 3.5 1.4 1.1 5.6 7 AA0703 Cape York Peninsula tropical savanna 3.5 2.3 1.0 8.2 7 AA0704 Carpentaria tropical savanna 3.0 1.2 1.1 4.0 7 AA0705 Einasleigh upland savanna 3.1 1.3 1.1 4.4 7 AA0706 Kimberly tropical savanna 3.2 2.9 1.1 10.3 7 AA0707 Mitchell grass downs 2.5 1.0 1.5 3.7 7 AA0708 Trans Fly savanna and grasslands 3.0 1.1 1.1 3.8 7 AA0709 Victoria Plains tropical savanna 2.6 1.0 1.1 2.9 7 AT0701 Angolan Miombo woodlands 3.8 1.3 1.1 5.1 7 AT0702 Angolan Mopane woodlands 3.7 1.1 1.2 5.1 7 AT0703 Ascension scrub and grasslands 1.0 1.0 1.1 1.1 7 AT0704 Central Zambezian Miombo woodlands 4.5 1.6 1.1 8.2 7 AT0705 East Sudanian savanna 3.9 1.1 1.2 5.2 7 AT0706 Eastern Miombo woodlands 3.6 1.0 1.1 3.9 7 AT0707 Guinean forest-savanna mosaic 3.9 1.3 1.9 9.8 7 AT0708 Itigi-Sumbu thicket 4.8 1.0 1.1 5.0 7 AT0709 Kalahari Acacia-Baikiaea woodlands 3.6 1.1 1.1 4.1 7 AT0710 Mandara Plateau mosaic 4.2 1.0 9.8 41.1 7 AT0711 Northern Acacia-Commiphora bushlands and thickets 4.7 1.3 1.2 7.3 7 AT0712 Northern Congolian forest-savanna mosaic 4.3 1.2 1.0 5.3 7 AT0713 Sahelian Acacia savanna 3.1 1.4 3.6 15.4 7 AT0714 Serengeti volcanic grasslands 5.0 1.0 1.1 5.4 7 AT0715 Somali Acacia-Commiphora bushlands and thickets 3.7 2.4 1.2 10.5 7 AT0716 Southern Acacia-Commiphora bushlands and thickets 4.5 1.1 1.3 6.0 7 AT0717 Southern Africa bushveld 4.1 1.3 2.2 11.6 7 AT0718 Southern Congolian forest-savanna mosaic 4.0 1.3 1.1 5.8 7 AT0719 Southern Miombo woodlands 4.0 1.1 1.2 5.5 7 AT0720 St. Helena scrub and woodlands 1.1 1.3 1.1 1.5 7 AT0721 Victoria Basin forest-savanna mosaic 5.1 1.4 1.5 10.4 7 AT0722 West Sudanian savanna 3.4 1.5 1.9 9.8 7 AT0723 Western Congolian forest-savanna mosaic 4.2 1.2 1.1 5.3 7 AT0724 Western Zambezian grasslands 4.2 1.0 1.0 4.2 7 AT0725 Zambezian and Mopane woodlands 4.3 1.2 1.1 5.5 7 AT0726 Zambezian Baikiaea woodlands 4.1 1.0 1.0 4.2
215 Appendix C: Appendix to Chapter 4
Biome Code Ecoregion S EndS* CRI* BWF 7 IM0701 Terai-Duar savanna and grasslands 3.5 1.1 1.7 6.5 7 NA0701 Western Gulf coastal grasslands 3.5 1.1 2.6 10.0 7 NT0210 Dry Chaco 3.1 1.5 1.2 5.7 7 NT0702 Beni savanna 4.1 1.1 1.1 4.8 7 NT0703 Campos Rupestres montane savanna 3.7 1.1 2.1 8.6 7 NT0704 Cerrado 3.3 1.8 5.4 31.7 7 NT0705 Clipperton Island shrub and grasslands 1.0 1.0 1.1 1.1 7 NT0707 Guianan savanna 4.2 1.2 1.0 4.9 7 NT0708 Humid Chaco 3.6 1.3 1.3 6.2 7 NT0709 Llanos 3.6 1.5 1.2 6.4 7 NT0710 Uruguayan savanna 2.9 1.2 4.3 15.6 7 OC0701 Hawaii tropical high shrublands 1.2 1.0 1.4 1.7 7 OC0702 Hawaii tropical low shrublands 1.6 1.0 3.1 4.9 7 OC0703 Northwestern Hawaii scrub 2.0 2.1 1.0 4.4 8 AA0801 Cantebury-Otago tussock grasslands 1.3 1.0 2.4 3.1 8 AA0802 Eastern Australia mulga shrublands 2.5 1.0 1.5 3.8 8 AA0803 Southeast Australia temperate savanna 3.0 1.1 3.4 11.0 8 AT0801 Al Hajar montane woodlands 1.5 1.1 6.0 9.9 8 AT0802 Amsterdam and Saint-Paul Islands temperate grasslands 1.0 1.0 1.1 1.1 8 AT0803 Tristan Da Cunha-Gough Islands shrub and grasslands 1.1 1.2 1.1 1.3 8 NA0801 California Central Valley grasslands 2.3 1.1 7.1 17.4 8 NA0802 Canadian Aspen forests and parklands 2.1 1.0 4.8 10.2 8 NA0803 Central and Southern mixed grasslands 2.6 1.0 5.5 14.2 8 NA0804 Central forest-grasslands transition 2.7 1.0 6.9 18.5 8 NA0805 Central tall grasslands 2.4 1.0 9.5 23.2 8 NA0806 Edwards Plateau savanna 2.6 1.2 1.5 4.4 8 NA0807 Flint Hills tall grasslands 2.6 1.0 3.1 8.2 8 NA0808 Montana Valley and Foothill grasslands 2.4 1.0 1.3 3.2 8 NA0809 Nebraska Sand Hills mixed grasslands 2.2 1.0 1.3 2.8 8 NA0810 Northern mixed grasslands 2.4 1.0 7.7 18.4 8 NA0811 Northern short grasslands 2.3 1.1 1.4 3.4 8 NA0812 Northern tall grasslands 2.4 1.0 5.2 12.6 8 NA0813 Palouse grasslands 2.4 1.0 1.6 3.8 8 NA0814 Texas blackland prairies 2.9 1.0 7.7 22.0 8 NA0815 Western short grasslands 2.6 1.1 1.9 5.1 8 NT0801 Espinal 2.6 1.1 4.3 12.2 8 NT0802 Low Monte 2.4 2.0 1.1 5.2 8 NT0803 Humid Pampas 2.5 1.3 7.4 23.6 8 NT0805 Patagonian steppe 2.0 2.4 1.0 5.1 8 PA0801 Alai-Western Tian Shan steppe 2.6 1.0 3.5 8.9 8 PA0802 Altai steppe and semi-desert 2.1 1.0 5.5 11.6 8 PA0803 Central Anatolian steppe 2.1 1.0 5.1 10.5 8 PA0804 Daurian forest steppe 1.9 1.0 1.9 3.7 8 PA0805 Eastern Anatolian montane steppe 2.9 1.1 2.7 8.5 8 PA0806 Emin Valley steppe 2.9 1.0 6.4 18.6 8 PA0807 Faroe Islands boreal grasslands 1.3 1.0 1.1 1.5 8 PA0808 Gissaro-Alai open woodlands 2.6 1.0 1.5 3.8 8 PA0809 Kazakh forest steppe 2.4 1.0 1.9 4.5 8 PA0810 Kazakh steppe 2.1 1.0 4.7 9.9 8 PA0811 Kazakh upland 2.1 1.0 5.0 10.4 8 PA0812 Middle East steppe 2.4 1.0 6.3 15.3 8 PA0813 Mongolian-Manchurian grassland 3.0 1.0 1.7 5.2 8 PA0814 Pontic steppe 2.7 1.1 3.2 9.9 8 PA0815 Sayan Intermontane steppe 2.5 1.0 1.1 2.9 8 PA0816 Selenge-Orkhon forest steppe 2.3 1.0 1.7 4.1 8 PA0817 South Siberian forest steppe 2.2 1.0 1.6 3.6 8 PA0818 Tian Shan foothill arid steppe 3.2 1.0 1.9 5.8 9 AT0901 East African halophytics 3.2 1.0 3.2 10.1
216 Appendix C: Appendix to Chapter 4
Biome Code Ecoregion S EndS* CRI* BWF 9 AT0902 Etosha Pan halophytics 2.9 1.0 1.1 3.1 9 AT0903 Inner Niger Delta flooded savanna 2.9 1.0 1.9 5.6 9 AT0904 Lake Chad flooded savanna 3.2 1.0 4.4 14.1 9 AT0905 Saharan flooded grasslands 3.4 1.1 4.2 15.1 9 AT0906 Zambezian coastal flooded savanna 3.6 1.1 1.6 6.2 9 AT0907 Zambezian flooded grasslands 4.3 1.2 1.0 5.4 9 AT0908 Zambezian halophytics 2.7 1.0 1.1 2.9 9 IM0901 Rann of Kutch seasonal salt marsh 2.4 1.0 1.1 2.6 9 NT0902 Cuban wetlands 2.8 1.0 1.0 2.9 9 NT0903 Enriquillo wetlands 2.9 1.0 1.0 2.9 9 NT0904 Everglades 2.7 1.0 1.1 2.9 9 NT0905 Guayaquil flooded grasslands 2.7 1.0 5.2 14.2 9 NT0906 Orinoco wetlands 3.9 1.0 1.0 4.0 9 NT0907 Pantanal 3.6 1.3 1.3 5.8 9 NT0908 Paranß flooded savanna 3.3 1.1 1.7 6.1 9 NT0909 Southern Cone Mesopotamian savanna 3.6 1.0 1.9 7.0 9 PA0901 Amur meadow steppe 2.3 1.0 1.4 3.3 9 PA0902 Bohai Sea saline meadow 3.4 1.0 1.3 4.4 9 PA0903 Nenjiang River grassland 3.1 1.0 2.0 6.2 9 PA0904 Nile Delta flooded savanna 2.6 1.0 9.6 24.6 9 PA0905 Saharan halophytics 2.2 1.0 3.1 6.6 9 PA0906 Tigris-Euphrates alluvial salt marsh 2.4 1.0 3.9 9.3 9 PA0907 Suiphun-Khanka meadows and forest meadows 2.8 1.0 1.2 3.5 9 PA0908 Yellow Sea saline meadow 3.3 1.0 1.1 3.8 10 AA1001 Australian Alps montane grasslands 3.2 1.0 1.0 3.3 10 AA1002 Central Range sub-alpine grasslands 1.6 1.4 1.0 2.3 10 AA1003 South Island montane grasslands 1.3 1.1 1.0 1.4 10 AT1001 Angolan montane forest-grassland mosaic 4.2 1.1 1.4 6.6 10 AT1002 Angolan scarp savanna and woodlands 3.8 1.4 1.0 5.5 10 AT1003 Drakensberg alti-montane grasslands and woodlands 3.7 1.2 1.6 7.1 10 AT1004 Drakensberg montane grasslands, woodlands and forests 4.1 1.7 1.4 9.9 10 AT1005 East African montane moorlands 6.0 1.4 1.0 8.8 10 AT1006 Eastern Zimbabwe montane forest-grassland mosaic 5.2 1.6 1.3 10.7 10 AT1007 Ethiopian montane grasslands and woodlands 3.9 1.4 1.9 10.6 10 AT1008 Ethiopian montane moorlands 4.3 1.6 1.2 8.0 10 AT1009 Highveld grasslands 3.5 1.2 6.9 29.3 10 AT1010 Jos Plateau forest-grassland mosaic 2.2 1.1 8.9 21.8 10 AT1011 Madagascar ericoid thickets 2.8 1.0 1.1 3.1 10 AT1012 Maputaland-Pondoland bushland and thickets 4.3 1.1 3.0 14.2 10 AT1013 Rwenzori-Virunga montane moorlands 5.4 1.3 1.1 7.6 10 AT1014 South Malawi montane forest-grassland mosaic 5.2 1.1 1.5 8.7 10 AT1015 Southern Rift montane forest-grassland mosaic 5.1 1.4 1.2 8.3 10 IM1001 Kinabalu montane alpine meadows 3.3 1.1 1.1 4.1 10 NT1001 Central Andean dry puna 1.9 2.2 1.8 7.4 10 NT1002 Central Andean puna 2.5 2.1 1.2 6.4 10 NT1003 Central Andean wet puna 2.9 2.3 1.3 8.5 10 NT1004 Cordillera Central pßramo 2.6 1.7 1.2 5.1 10 NT1005 Cordillera de Merida pßramo 3.0 1.4 1.0 4.3 10 NT1006 Northern Andean pßramo 3.8 3.9 1.1 15.5 10 NT1007 Santa Marta pßramo 2.9 1.2 1.0 3.4 10 NT1008 Southern Andean steppe 2.1 3.2 1.2 8.4 10 NT1010 High Monte 1.0 1.0 1.2 1.2 10 PA1001 Altai alpine meadow and tundra 2.0 1.0 1.2 2.3 10 PA1002 Central Tibetan Plateau alpine steppe 1.8 1.0 1.0 1.8 10 PA1003 Eastern Himalayan alpine shrub and meadows 3.1 1.1 1.0 3.7 10 PA1004 Ghorat-Hazarajat alpine meadow 1.7 1.0 1.9 3.3 10 PA1005 Hindu Kush alpine meadow 1.9 1.0 1.9 3.7 10 PA1006 Karakoram-West Tibetan Plateau alpine steppe 1.7 1.1 1.0 1.8
217 Appendix C: Appendix to Chapter 4
Biome Code Ecoregion S EndS* CRI* BWF 10 PA1007 Khangai Mountains alpine meadow 1.9 1.0 1.1 2.0 10 PA1008 Kopet Dag woodlands and forest steppe 2.6 1.0 1.2 3.2 10 PA1009 Kuh Rud and Eastern Iran montane woodlands 1.7 1.0 7.0 12.1 10 PA1010 Mediterranean High Atlas juniper steppe 2.4 1.1 1.2 3.1 10 PA1011 North Tibetan Plateau-Kunlun Mountains alpine desert 1.5 1.1 1.0 1.6 10 PA1012 Northwestern Himalayan alpine shrub and meadows 2.1 1.1 1.1 2.5 10 PA1013 Ordos Plateau steppe 3.1 1.0 2.3 7.1 10 PA1014 Pamir alpine desert and tundra 2.0 1.1 1.2 2.7 10 PA1015 Qilian Mountains subalpine meadows 1.8 1.0 1.1 2.0 10 PA1016 Sayan Alpine meadows and tundra 2.7 1.0 1.1 2.9 10 PA1017 Southeast Tibet shrublands and meadows 3.0 1.1 1.0 3.2 10 PA1018 Sulaiman Range alpine meadows 1.4 1.0 5.1 7.1 10 PA1019 Tian Shan montane steppe and meadows 3.0 1.1 1.4 4.4 10 PA1020 Tibetan Plateau alpine shrublands and meadows 2.0 1.0 1.0 2.0 10 PA1021 Western Himalayan alpine shrub and Meadows 3.0 1.1 1.1 3.4 10 PA1022 Yarlung Tsangpo arid steppe 2.2 1.0 1.0 2.3 11 AA1101 Antipodes Subantarctic Islands tundra 1.3 1.6 1.1 2.2 11 AN1101 Marielandia Antarctic tundra 1.0 1.0 1.1 1.1 11 AN1102 Maudlandia Antarctic desert 1.0 1.0 1.1 1.1 11 AN1103 Scotia Sea Islands tundra 1.0 1.1 1.1 1.2 11 AN1104 Southern Indian Ocean Islands tundra 1.0 1.0 1.1 1.1 11 NA1101 Alaska-St. Elias Range tundra 1.9 1.0 1.0 1.9 11 NA1102 Aleutian Islands tundra 1.6 1.3 1.0 2.1 11 NA1103 Arctic coastal tundra 1.4 1.0 1.0 1.4 11 NA1104 Arctic foothills tundra 1.4 1.0 1.0 1.4 11 NA1105 Baffin coastal tundra 1.2 1.0 1.1 1.3 11 NA1106 Beringia lowland tundra 1.6 1.0 1.0 1.6 11 NA1107 Beringia upland tundra 1.7 1.0 1.0 1.7 11 NA1108 Brooks-British Range tundra 1.4 1.0 1.0 1.4 11 NA1109 Davis Highlands tundra 1.2 1.0 1.0 1.2 11 NA1110 High Arctic tundra 1.2 1.0 1.5 1.8 11 NA1111 Interior Yukon-Alaska alpine tundra 1.7 1.0 1.0 1.7 11 NA1112 Kalaallit Nunaat high arctic tundra 1.0 1.0 1.0 1.0 11 NA1113 Kalaallit Nunaat low arctic tundra 1.0 1.0 1.1 1.2 11 NA1114 Low Arctic tundra 1.4 1.0 1.0 1.4 11 NA1115 Middle Arctic tundra 1.3 1.0 1.0 1.3 11 NA1116 Ogilvie-MacKenzie alpine tundra 1.6 1.0 1.0 1.7 11 NA1117 Pacific Coastal Mountain icefields and tundra 2.0 1.0 1.0 2.0 11 NA1118 Torngat Mountain tundra 1.3 1.0 1.0 1.3 11 PA1101 Arctic desert 1.2 1.0 1.0 1.2 11 PA1102 Bering tundra 1.6 1.0 1.0 1.7 11 PA1103 Cherskii-Kolyma mountain tundra 1.7 1.0 1.2 2.1 11 PA1104 Chukchi Peninsula tundra 1.5 1.0 1.7 2.6 11 PA1105 Kamchatka Mountain tundra and forest tundra 1.6 1.0 1.1 1.7 11 PA1106 Kola Peninsula tundra 1.7 1.0 1.0 1.8 11 PA1107 Northeast Siberian coastal tundra 1.4 1.0 1.0 1.4 11 PA1108 Northwest Russian-Novaya Zemlya tundra 1.6 1.0 1.1 1.8 11 PA1109 Novosibirsk Islands arctic desert 1.2 1.0 1.0 1.2 11 PA1110 Scandinavian Montane Birch forest and grasslands 2.0 1.0 1.1 2.3 11 PA1111 Taimyr-Central Siberian tundra 1.4 1.0 1.2 1.7 11 PA1112 Trans-Baikal Bald Mountain tundra 2.0 1.0 1.1 2.2 11 PA1113 Wrangel Island arctic desert 1.2 1.0 1.0 1.2 11 PA1114 Yamal-Gydan tundra 1.4 1.0 1.0 1.5 12 AA1201 Coolgardie woodlands 2.4 1.0 1.1 2.6 12 AA1202 Esperance mallee 2.7 1.3 3.7 12.9 12 AA1203 Eyre and York mallee 3.0 1.2 2.8 9.6 12 AA1204 Jarrah-Karri forest and shrublands 3.0 1.0 1.1 3.2 12 AA1205 Swan Coastal Plain Scrub and Woodlands 3.4 1.2 5.2 21.6
218 Appendix C: Appendix to Chapter 4
Biome Code Ecoregion S EndS* CRI* BWF 12 AA1206 Mount Lofty woodlands 3.4 1.1 7.0 26.3 12 AA1207 Murray-Darling woodlands and mallee 3.0 1.1 5.5 17.7 12 AA1208 Naracoorte woodlands 3.2 1.0 6.4 20.7 12 AA1209 Southwest Australia savanna 2.9 1.3 5.6 21.5 12 AA1210 Southwest Australia woodlands 2.8 1.1 3.8 11.5 12 AT1201 Albany thickets 3.8 1.1 1.1 4.6 12 AT1202 Lowland fynbos and renosterveld 3.7 1.4 1.9 9.5 12 AT1203 Montane fynbos and renosterveld 3.7 2.1 1.0 7.8 12 NA1201 California coastal sage and chaparral 3.5 1.5 1.1 5.9 12 NA1202 California interior chaparral and woodlands 2.7 1.3 1.1 3.8 12 NA1203 California montane chaparral and woodlands 3.1 1.2 1.0 3.7 12 NT1201 Chilean matorral 1.9 1.9 3.8 13.9 Aegean and Western Turkey sclerophyllous and mixed 12 PA1201 forests 2.8 1.4 6.0 22.7 12 PA1202 Anatolian conifer and deciduous mixed forests 2.4 1.0 7.7 18.9 12 PA1203 Canary Islands dry woodlands and forests 2.0 2.0 1.1 4.4 12 PA1204 Corsican montane broadleaf and mixed forests 2.6 1.2 1.1 3.2 12 PA1205 Crete Mediterranean forests 2.2 1.1 2.4 6.0 12 PA1206 Cyprus Mediterranean forests 2.3 1.5 5.2 18.1 Eastern Mediterranean conifer-sclerophyllous-broadleaf 12 PA1207 forests 2.9 1.2 6.2 21.9 12 PA1208 Iberian conifer forests 2.6 1.1 1.4 4.1 12 PA1209 Iberian sclerophyllous and semi-deciduous forests 2.2 1.0 3.0 6.7 12 PA1210 Illyrian deciduous forests 3.1 1.1 6.3 21.2 12 PA1211 Italian sclerophyllous and semi-deciduous forests 2.6 1.1 2.9 8.3 Mediterranean acacia-argania dry woodlands and 12 PA1212 succulent thickets 2.7 1.5 5.1 20.4 12 PA1213 Mediterranean dry woodlands and steppe 2.5 1.0 5.0 12.4 12 PA1214 Mediterranean woodlands and forests 2.6 1.2 3.9 11.9 Northeastern Spain and Southern France Mediterranean 12 PA1215 forests 2.9 1.2 1.8 6.0 12 PA1216 Northwest Iberian montane forests 2.7 1.0 1.7 4.5 12 PA1217 Pindus Mountains mixed forests 2.9 1.0 2.6 7.5 12 PA1218 South Appenine mixed montane forests 2.7 1.0 1.1 3.0 12 PA1219 Southeastern Iberian shrubs and woodlands 3.4 1.0 1.4 4.9 Southern Anatolian montane conifer and deciduous 12 PA1220 forests 2.5 1.0 6.9 17.4 Southwest Iberian Mediterranean sclerophyllous and 12 PA1221 mixed forests 2.7 1.0 1.5 4.0 12 PA1222 Tyrrhenian-Adriatic Sclerophyllous and mixed forests 2.6 1.2 3.2 10.3 13 AA1301 Carnarvon xeric shrublands 2.9 1.5 1.1 4.7 13 AA1302 Central Ranges xeric scrub 2.6 1.1 1.6 4.7 13 AA1303 Gibson desert 2.2 1.0 1.1 2.4 13 AA1304 Great Sandy-Tanami desert 2.4 1.1 1.2 3.3 13 AA1305 Great Victoria desert 2.4 1.1 1.1 2.7 13 AA1306 Nullarbor Plains xeric shrublands 2.3 1.1 1.1 2.6 13 AA1307 Pilbara shrublands 2.7 1.7 1.0 4.6 13 AA1308 Simpson desert 2.5 1.1 1.5 4.3 13 AA1309 Tirari-Sturt stony desert 2.8 1.1 3.4 10.9 13 AA1310 Western Australian Mulga shrublands 2.4 1.2 1.1 3.1 13 AT1301 Aldabra Island xeric scrub 1.6 1.6 1.0 2.6 13 AT1302 Arabian Peninsula coastal fog desert 2.4 1.0 1.3 3.1 13 AT1303 East Saharan montane xeric woodlands 2.7 1.0 9.2 24.7 13 AT1304 Eritrean coastal desert 3.0 1.0 9.1 26.9 13 AT1305 Ethiopian xeric grasslands and shrublands 3.3 1.1 2.6 10.0 13 AT1306 Gulf of Oman desert and semi-desert 2.4 1.0 7.9 18.9 13 AT1307 Hobyo grasslands and shrublands 2.3 1.5 4.5 15.1 13 AT1308 Ile Europa and Bassas da India xeric scrub 1.0 1.0 1.1 1.1
219 Appendix C: Appendix to Chapter 4
Biome Code Ecoregion S EndS* CRI* BWF 13 AT1309 Kalahari xeric savanna 3.1 1.1 1.1 3.8 13 AT1310 Kaokoveld desert 2.8 1.5 1.2 5.1 13 AT1311 Madagascar spiny thickets 2.5 2.4 1.1 6.7 13 AT1312 Madagascar succulent woodlands 2.2 1.3 1.3 3.7 13 AT1313 Masai xeric grasslands and shrublands 3.2 1.2 1.9 7.0 13 AT1314 Nama Karoo 3.0 1.3 1.5 5.7 13 AT1315 Namib desert 2.4 1.2 1.1 3.3 13 AT1316 Namibian savanna woodlands 3.3 1.4 1.2 5.4 13 AT1318 Socotra Island xeric shrublands 1.7 3.0 4.9 25.1 13 AT1319 Somali montane xeric woodlands 2.4 1.3 5.9 18.7 13 AT1320 Southwestern Arabian foothills savanna 2.1 1.0 6.5 13.7 13 AT1321 Southwestern Arabian montane woodlands 1.8 1.2 6.4 13.9 13 AT1322 Succulent Karoo 2.9 1.7 1.8 8.7 13 IM1301 Deccan thorn scrub forests 2.9 1.2 6.6 22.7 13 IM1302 Indus Valley desert 2.2 1.0 8.3 18.3 13 IM1303 Northwestern thorn scrub forests 3.1 1.3 3.9 15.5 13 IM1304 Thar desert 1.9 1.0 1.4 2.7 13 NA1301 Baja California desert 1.6 1.4 1.0 2.3 13 NA1302 Central Mexican matorral 2.3 1.0 2.2 5.1 13 NA1303 Chihuahuan desert 3.7 1.5 1.1 6.0 13 NA1304 Colorado Plateau shrublands 2.7 1.1 1.0 3.0 13 NA1305 Great Basin shrub steppe 2.4 1.0 1.1 2.6 13 NA1306 Gulf of California xeric scrub 1.8 2.6 1.0 4.8 13 NA1307 Meseta Central matorral 2.0 1.1 1.2 2.6 13 NA1308 Mojave desert 2.6 1.2 1.0 3.1 13 NA1309 Snake-Columbia shrub steppe 2.4 1.0 1.4 3.3 13 NA1310 Sonoran desert 3.3 1.4 1.1 5.1 13 NA1311 Tamaulipan matorral 1.9 1.1 2.7 5.7 13 NA1312 Tamaulipan mezquital 3.5 1.2 2.4 10.0 13 NA1313 Wyoming Basin shrub steppe 2.2 1.0 1.1 2.5 13 NT1301 Araya and Paria xeric scrub 4.0 1.0 1.9 7.7 13 NT1303 Atacama desert 1.6 1.1 6.6 11.2 13 NT1304 Caatinga 2.6 1.3 5.8 19.7 13 NT1305 Caribbean shrublands 2.1 1.0 1.1 2.2 13 NT1306 Cuban cactus scrub 2.8 1.0 1.1 3.0 13 NT1307 Galßpagos Islands scrubland mosaic 1.8 2.8 1.0 5.2 13 NT1308 Guajira-Barranquilla xeric scrub 4.1 1.1 3.0 13.3 13 NT1309 La Costa xeric shrublands 3.7 1.2 1.2 5.3 13 NT1311 Malpelo Island xeric scrub 1.1 1.4 1.1 1.6 13 NT1312 Motagua Valley thornscrub 3.4 1.1 1.4 5.4 13 NT1313 Paraguana xeric scrub 3.6 1.2 1.1 4.5 13 NT1314 San Lucan xeric scrub 2.9 1.0 1.0 2.9 13 NT1315 Sechura desert 2.2 2.4 3.6 19.2 13 NT1316 Tehuacßn Valley matorral 3.1 1.2 1.1 3.9 13 NT1318 St. Peter and St. Paul rocks 1.0 1.0 1.1 1.1 13 PA1301 Afghan Mountains semi-desert 1.3 1.0 1.2 1.7 13 PA1302 Alashan Plateau semi-desert 2.7 1.1 1.5 4.4 13 PA1303 Arabian Desert and East Sahero-Arabian xeric shrublands 2.3 1.0 3.6 8.3 13 PA1304 Atlantic coastal desert 1.8 1.0 1.7 2.9 13 PA1305 Azerbaijan shrub desert and steppe 2.8 1.0 1.4 3.8 13 PA1306 Badghyz and Karabil semi-desert 2.4 1.0 2.1 5.0 13 PA1307 Baluchistan xeric woodlands 2.0 1.1 3.0 6.8 13 PA1308 Caspian lowland desert 2.5 1.0 1.9 4.9 13 PA1309 Central Afghan Mountains xeric woodlands 1.9 1.0 4.3 8.3 13 PA1310 Central Asian northern desert 2.2 1.2 7.2 19.0 13 PA1311 Central Asian riparian woodlands 2.5 1.1 1.8 4.9 13 PA1312 Central Asian southern desert 2.0 1.1 2.9 6.1 13 PA1313 Central Persian desert basins 1.8 1.0 1.7 2.9
220 Appendix C: Appendix to Chapter 4
Biome Code Ecoregion S EndS* CRI* BWF 13 PA1314 Eastern Gobi desert steppe 2.2 1.0 3.0 6.6 13 PA1315 Gobi Lakes Valley desert steppe 1.7 1.0 5.8 9.9 13 PA1316 Great Lakes Basin desert steppe 2.4 1.0 1.5 3.6 13 PA1317 Junggar Basin semi-desert 2.0 1.0 1.7 3.4 13 PA1318 Kazakh semi-desert 2.1 1.0 6.4 13.1 13 PA1319 Kopet Dag semi-desert 2.5 1.0 4.2 10.3 13 PA1320 Mesopotamian shrub desert 2.4 1.0 7.2 17.3 13 PA1321 North Saharan steppe and woodlands 2.2 1.0 9.5 21.1 13 PA1322 Paropamisus xeric woodlands 1.9 1.0 4.6 8.9 13 PA1323 Persian Gulf desert and semi-desert 2.2 1.0 7.5 16.7 13 PA1324 Qaidam Basin semi-desert 1.5 1.0 1.8 2.7 13 PA1325 Red Sea Nubo-Sindian tropical desert and semi-desert 2.2 1.0 2.2 4.8 13 PA1326 Registan-North Pakistan sandy desert 2.3 1.3 5.8 17.7 13 PA1327 Sahara desert 2.1 1.0 6.5 13.8 13 PA1328 South Iran Nubo-Sindian desert and semi-desert 2.4 1.0 2.8 6.8 13 PA1329 South Saharan steppe and woodlands 2.4 1.1 3.5 8.7 13 PA1330 Taklimakan desert 2.3 1.0 1.9 4.3 13 PA1331 Tibesti-Jebel Uweinat montane xeric woodlands 1.6 1.0 10.0 16.0 13 PA1332 West Saharan montane xeric woodlands 1.7 1.1 1.3 2.3 13 PA1333 Red Sea coastal desert 1.9 1.1 8.1 16.6 Sources: Olson et al. (2001) and own calculations
Fig. C1 Box and whisker plot of species richness per land use type and biome TempBMF broadleaf and mixed forests, TropGL sub-/tropical grass-/shrublands and savannahs and TropMBF sub- /tropical moist broadleaf forests
221 Appendix C: Appendix to Chapter 4
B.1 Full bibliography for characterization factor calculation In January and February 2012 combinations of the keywords ["species richness" or "species number*" or biodiversity] and plant* and or ["organic farming" or "organic agriculture" or plantation or "annual crop" or field or arable] were used. We selected all original papers that compare plant species richness in organic and conventional production systems in the regions relevant for this study (i.e. temperate Europe, South-East Asia and sub-tropical grass- /shrublands and savannahs of South America). No study was found that compared the plant species richness of organic and conventional fields in South America. Empirical data for (semi)- natural reference situations were found with the following combinations of keywords: ["species richness" or "species number*" or biodiversity] and plant* and [Cerrado or Malaysia or Indonesia or (Brazil or "South America" and grassland) or tropic* or temperate forest].
Ambinakudige S, Sathish BN (2009) Comparing tree diversity and composition in coffee farms and sacred forests in the Western Ghats of India. Biodivers Conserv 18(4):987–1000. Ammer U, Utschick H, Anton H (1988) Die Auswirkungen von biologischem und konventionellem Landbau auf Flora und Fauna. Forstwiss Centralbl 107(1):274–291. Amorim PK, Batalha MA (2008) Soil chemical factors and grassland species density in Emas National Park (central Brazil). Braz J Biol 68(2):279–285 Amorim PK, Batalha MA (2007) Soil-vegetation relationships in hyperseasonal cerrado, seasonal cerrado, and wet grassland in Emas National Park (central Brazil). Acta Oecol 32(3):319–327. Badea O, Neagu S, Bytnerowicz A, Silaghi D, Barbu I, Iacoban C, Popescu F, Andrei M, Preda E, Iacob C, Dumitru I, Iuncu H, Vezeanu C, Huber V (2011) Long-term monitoring of air pollution effects on selected forest ecosystems in the Bucegi-Piatra Craiului and Retezat Mountains, southern Carpathians (Romania). iForest Biogeosci For 4(2):49–60. Batalha MA, Mantovani W, De Mesquita Junior H (2001) Vegetation structure in Cerrado physiognomies in south-eastern Brazil. Braz J Biol 61:3475–3483 BDM (2004) Biodiversity monitoring Switzerland. Indicator Z9: species diversity in habitats. In: de Baan L, Alkemade R, Koellner T (2012) Land use impacts on biodiversity in LCA: a global approach. Int J Life Cycle Assess. Published Online First 24 April 2012 Bos MM, Tylianakis JM, Steffan-Dewenter I, Tscharntke T (2008) The invasive Yellow Crazy Ant and the decline of forest ant diversity in Indonesian cacao agroforests. Biol Invasions 10(8):1399–1409. Boutin C, Baril A, Martin P (2008) Plant diversity in crop fields and woody hedgerows of organic and conventional farms in contrasting landscapes. Agric Ecosyst Environ 123(1- 3):185–193. Brearley FQ, Prajadinata S, Kidd PS, Proctor J, Suriantata (2004) Structure and floristics of an old secondary rain forest in Central Kalimantan, Indonesia, and a comparison with adjacent primary forest. For Ecol Manag 195(3):385–397. Brunet J, Falkengren-Grerup U, Tyler G (1996) Herb layer vegetation of south Swedish beech and oak forests - effects of management and soil acidity during one decade. For Ecol Manag 88:259–272. Brunet J (2007) Plant colonization in heterogeneous landscapes: an 80-year perspective on restoration of broadleaved forest vegetation. J Appl Ecol 44(3):563–572. Cianciarruso M, Batalha MA (2009) Short-term community dynamics in seasonal and hyperseasonal cerrados. Braz J Biol 69(2):631–637 Clough Y, Holzschuh A, Gabriel D, Purtauf T, Kleijn D, Kruess A, Steffan-Dewenter I, Tscharntke T (2007) Alpha and beta diversity of arthropods and plants in organically and conventionally managed wheat fields. J Appl Ecol 44(4):804–812.
222 Appendix C: Appendix to Chapter 4
Dumortier M, Butaye J, Jacquemyn H, van Camp N, Lut N, Hermy M (2002) Predicting vascular plant species richness of fragmented forests in agricultural landscapes in central Belgium. For Ecol Manag 158:85–102 Fahy O, Gormally M (1998) A comparison of plant and carabid beetle communities in an Irish oak woodland with nearby conifer plantation and clear felled site. For Ecol Manag 110:263–273 Frieben B, Koepke U (1996) Effects of farming systems on biodiversity. In: Isart J, Llerena JJ (eds) Biodiversity and Land use: The role of Organic farming, Proceedings of the first ENOF-Workshop, Bonn, Dec 8-9, 1995 Gabriel D, Roschewitz I, Tscharntke T, Thies C (2006) Beta diversity at different spatial scales: Plant communities in organic and conventional agriculture. Ecol Appl 16(55):2011–2021 Gabriel D, Sait SM, Hodgson JA, Schmutz U, Kunin WE, Benton TG (2010) Scale matters: the impact of organic farming on biodiversity at different spatial scales. Ecol Lett 13(7):858– 869. Gazol A, Ibáñez R (2010) Variation of plant diversity in a temperate unmanaged forest in northern Spain: behind the environmental and spatial explanation. Plant Ecol 207(1):1–11. Gillison A (2003) Vegetation indicates diversity of soil macroinvertebrates: a case study with termites along a land-use intensification gradient in lowland Sumatra. Org Divers Evol 3(2):111–126. Gillison A, Liswanti N, Budidarsono S, van Noordwijk M, Tomich T (2004) Conservation ecology: Human-caused disturbance stimuli as a form of predation risk. Ecol Soc 9(2):7 Goetmark F, Paltto H, Nordén B, Goetmark E (2005) Evaluating partial cutting in broadleaved temperate forest under strong experimental control: Short-term effects on herbaceous plants. For Ecol Manag 214(1-3):124–141. Gradstein SR, Kessler M, Pitopang R (2007) Tree species diversity relative to human land uses in tropical rain forest margins in Central Sulawesi. In: Tscharntke T, Leuschner C, Zeller M, Guhardja E, Bidin A (eds) Tree species diversity relative to human land uses in tropical rain forest margins in Central Sulawesi. Springer, Berlin, pp 321-334 Hald AB (1999) Weed vegetation (wild flora) of long established organic versus conventional cereal fields in Denmark. Ann Appl Biol 134(3):307–314. Hawes C, Squire G, Hallett P, Watson C, Young M (2010) Arable plant communities as indicators of farming practice. Agric Ecosyst Environ 138:17–26. Hiltbrunner J, Scherrer C, Streit B, Jeanneret P, Zihlmann U, Tschachtli R (2008) Long-term weed community dynamics in Swiss organic and integrated farming systems. Weed Res 48(4):360–369. Hofmeister J, Hošek J, Modrý M, Roleček J (2009) The influence of light and nutrient availability on herb layer species richness in oak-dominated forests in central Bohemia. Plant Ecol 205(1):57–75. Hotze C, van Elsen T (2006) Ackerwildkräuter konventionell und biologisch bewirtschafteter Äcker im östlichen Meißnervorland – Entwicklung in den letzten 30 Jahren. J Plant Dis Prot Special Issue:547–555 Kaar B, Freyer B (2008) Weed species diversity and cover-abundance in organic and conventional winter cereal fields and 15 years ago. In: 16th IFOAM Organic World Congress, Modena, June 16-20, 2008 Keersmaeker L de, Martens L, Verheyen K, Hermy M, Schrijver A de, Lust N (2004) Impact of soil fertility and insolation on diversity of herbaceous woodland species colonizing afforestations in Muizen forest (Belgium). For Ecol Manag 188:291–304. Kessler M, Keßler PJ, Gradstein SR, Bach K, Schmull M, Pitopang R (2005) Tree diversity in primary forest and different land use systems in Central Sulawesi, Indonesia. Biodivers Conserv 14(3):547–560. Knop E (2005) Biologische Vielfalt von Grasland im ökologischen Ausgleich - ein Paarvergleich. Schriftenr FAL 56:169-173
223 Appendix C: Appendix to Chapter 4
Krauss J, Gallenberger I, Steffan-Dewenter I, Hector A (2011) Decreased functional diversity and biological pest contron in conventional compared to organic crop fields. PLoS ONE 6(5):e19502. Kreuter TNO (2005) Biodiversität sächsischer Ackerflächen. Schriftenr Sächs Landesanst Landwirtsch 9(10):1–97 Lips A, Dubois D et al (1997) Belebte Umwelt. In: Koellner T (2000) Species-pool effect potentials (SPEP) as a yardstick to evaluate land-use impacts on biodiversity. J Clean Prod 8:293–311 Manhoudt A, Udo Haes H de, Snoo G de (2005) An indicator of plant species richness of semi- natural habitats and crops on arable farms. Agric Ecosyst Environ 109(1-2):166–174. Matzdorf B, Zerbe S (2000) Segetalvegetation der Uckermark unter dem Einfluss von biologisch-dynamischer und konventioneller Bewirtschaftung. Verh Bot Ver Berl Brandenbg 133:87–118 Moreby SJ, Aebischer NJ, Southway S, Sotherton NW (1994) A comparison of the flora and arthropod fauna of organically and conventionally grown winter wheat in southern England. Ann Appl Biol 125:13–27 Murdiyarso D, van Noordwijk M, Wasrin U, Tomich T, Gillison A (2002) Environmental benefits and sustainable land-use options in the Jambi transect, Sumatra, Indonesia. J Veg Sci 13:429–438 Norton LR (2002) Factors influencing biodiversity within organic and conventional systems of arable farming – methodologies and preliminary results. In: Powell et al. (eds) UK Organic Research 2002. Proceedings of the COR Conference, Aberystwyth, March 26-28, 2002 Oheimb G, Friedel A, Bertsch A, Haerdtle W (2007) The effects of windthrow on plant species richness in a Central European beech forest. Plant Ecol 191(1):47–65. Poggio SL (2005) Structure of weed communities occurring in monoculture and intercropping of field pea and barley. Agric Ecosyst Environ 109(1-2):48–58. Poggio SL, Chaneton EJ, Ghersa CM (2012) The arable plant diversity of intensively managed farmland: Effects of field position and crop type at local and landscape scales. Agric Ecosyst Environ in press. Roschewitz I, Gabriel D, Tscharntke T, Thies C (2005) The effects of landscape complexity on arable weed species diversity in organic and conventional farming. J Appl Ecol 42(5):873– 882. Schmidt W (2005) Herb layer species as indicators of biodiversity of managed and unmanaged beech forests. For Snow Landsc Res 79:111–125 Schreiber C, Lehmann B (1996) Artenvielfalt auf konventionellen-, IP- und Biobetrieben. Agrarforschung 3(10):501–504 Schulze C, Fiedler K (9) Hawkmoth diversity in northern Borneo does not reflect the influence of anthropogenic habitat disturbance. Ecotropica 2003:99–102 Siebert SF (2002) From shade- to sun-grown perennial crops in Sulawesi, Indonesia: implications for biodiversity conservation and soil fertility. Biodivers Conserv 11:1889– 1902 Szymura T, Szymura M (2011) Soil properties and light availability determine species richness and vegetation in an overgrown coppice oak stand. Pol J Ecol 59(3):523–533 Tamiozzo E, Ubirata SJ (2006) Diversidade na cultura da soja orgânica em sistema de plantio direto e convencional no município de Tenente Portela - RS. Rev Plant Direto 15:48–54 Thorell M, Goetmark F (2005) Reinforcement capacity of potential buffer zones: Forest structure and conservation values around forest reserves in southern Sweden. For Ecol Manag 212(1-3):333–345. Tuesca D, Puricelli E (2007) Effect of tillage systems and herbicide treatments on weed abundance and diversity in a glyphosate resistant crop rotation. Crop Prot 26(12):1765– 1770. Turner I, Wong Y, Chew P, Ibrahim A (1997) Tree species richness in primary and old secondary tropical forest in Singapore. Biodivers Conserv 6:537–543
224 Appendix C: Appendix to Chapter 4
Tyynelae T, Otsamo R, Otsamo A (2003) Indigenous livelihood systems in industrial tree- plantation areas in West Kalimantan, Indonesia: Economics and plant-species richness. Agrofor Syst 57:87–100 Ulber L, Steinmann H, Klimek S, Isselstein J (2009) An on-farm approach to investigate the impact of diversified crop rotations on weed species richness and composition in winter wheat. Weed Res 49(5):534–543. Vitta J, Tuesca D, Puricelli E (2004) Widespread use of glyphosate tolerant soybean and weed community richness in Argentina. Agric Ecosyst Environ 103(3):621–624. Vockenhuber EA (2011) Herb layer characteristics, fly communities and trophic interactions along a gradient of tree and herb diversity in a temperate deciduous forest. PhD Thesis, Göttingen Centre for Biodiversity and Ecology Wachendorf M, Taube F (2001) Artenvielfalt, Leistungsmerkmale und bodenchemische Kennwerte des Dauergrünlands im konventionellen und ökologischen Landbau in Nordwestdeutschland. Pflanzenbauwiss 5(2):75–86 Weibull A, Oestman O, Granqvist A (2003) Species richness in agroecosystems: the effect of landscape, habitat and farm management. Biodivers Conserv 12:1335–1355 Wohlgemuth T (1993) Der Verbreitungsatlas der Farn- und Bluetenpflanzen der Schweiz. In: Koellner T (2000) Species-pool effect potentials (SPEP) as a yardstick to evaluate land- use impacts on biodiversity. J Clean Prod 8:293–311
225 Appendix D: Appendix to Chapter 5
Appendix D: Appendix to Chapter 5
D.1 Land use assessment framework !"#$"%&'(")*! ,#%&-./#$%0/&()$*%+'( !"#$%&"&'()$*%+'''
!"#$%&' !"#$%*' 1++2*%0/&(( )$*%+'''
!(' 3%&4(2-"( 6"+/7"#8( *5%-"( *5%-"( 23/-'45#' .'&/! 23/-' 673/0#' 383/-,/9 +#/)'
+",&- ) ' ) ) ) ) 1 &' (%'"#0' *' +,-#..(/0'
Figure D1. Land use assessment framework for life cycle assessment (adopted from Milà i Canals et al. 2007; Koellner et al. 2013). The final impacts per are given as multiplication of Area (A) ! Time ! Biodiversity loss. In the equations below for each impact type, the inventory flow is marked as {}I, the characterization factors as {}CF.
Occupation impacts = {AOcc ! (t1-t0)}I !{(Sref,2 - Si)}CF ;
Transformation impacts = { ATrans }I ! {0.5 ! treg ! (Sref,2 - Si) }CF ;
Permanent impacts = { ATrans }I ! {tmodelling ! (Sref,1 - Sref,2)}CF .
226 Appendix D: Appendix to Chapter 5 D.2 Input data
Table D1. Overview of input data, assumed distribution and variable specificity
Assumed Variables Data source Variable specificity distribution
CFloc Local de Baan et al. Non-parametric Independent variables for each characteriza- (2013a) Kernel Density biome (n=14) and land use type tion factors Estimation (n=4)
z z-values Drakare et al. Triangular Independent variables for each (2006) distribution, based habitat type (n=3; islands, forest, on mode, and lower non-forest ecoregions). and upper bounds
Aorg,j, Areas LADA (2008), Non-parametric Aorg,j, Anew,j: Independent variables Anew,j, (original Anthromes (Ellis Kernel Density for each ecoregion (n=804) Ai,j, habitat, and Ramankutty Estimation Ai,j: Independent variables for remaining 2008) using 7 each land use type (n=4; see habitat, area geographic Table D4 for classification) and per land use projections for ecoregion (n=804) type) calculating areas
Sorg Species Olson et al. (2001) Plants: triangular Independent variables for each numbers per (mammals, birds, distribution based ecoregion (n=804) and taxa (n=5; ecoregion(*) amphibians, on working figure, mammals, birds, amphibians, reptiles), Kier et upper and lower reptiles, plants) al. (2005) (plants) ranges. Other groups: no uncertainty assumed
Sorg, end Endemism Olson et al. (2001) No uncertainty Independent variables for each numbers per (mammals, birds, assumed ecoregion (n=804) and taxa (n=4; ecoregion(*) amphibians, mammals, birds, amphibians, reptiles), plants: reptiles) no data. treg Regeneration Curran et al. (in Log-normal 520 different regeneration times, time press) distribution based on all combinations of Realm x Biome (n = 65), land use intensity (n=2), taxonomic group (n=4) (*) Species richness numbers from WWF are based on the ranges of extant species. Species that are introduced, present as human commensals, vagrants, or passage migrants are not recorded. Where available, historic ranges of species (i.e., approximate distribution at 1500 AD) were used to compile the database (World Wildlife Fund 2006).
Table D2. z-values applied in this study (derived from Drakare et al. (2006), Figure 1e, for both average and nested SAR) island forest non-forest lower Confidence Interval 0.242 0.307 0.185 mean 0.258 0.344 0.211 upper Confidence Interval 0.282 0.384 0.247
227 Appendix D: Appendix to Chapter 5
Table D3. Median recovery times in years (based on Curran et al. in press) Region name Input parameters Median regeneration times (in years) Realm_ median Altitude Ecosystem Herpeto- Realm Biome Biome Latitude (median) type Birds Mammals fauna Plants [° S or ° N] [m.a.s.l] Ext. Int. Ext. Int. Ext. Int. Ext. Int. AA 01 AA01 10 300 Forest 145 162 166 173 159 169 142 143 02 AA02 10 400 Forest 110 110 106 114 97 124 99 117 04 AA04 40 400 Forest 504 568 581 566 536 527 538 500 07 AA07 20 300 Non-forest 104 117 120 109 104 113 128 114 08 AA08 40 200 Non-forest 242 247 266 283 251 255 248 243 10 AA10 40 1300 Non-forest 300 315 331 325 337 302 303 275 11 AA11 60 200 Forest 569 577 560 594 547 575 484 576 12 AA12 40 200 Forest 344 343 386 367 329 384 329 328 13 AA13 30 400 Non-forest 181 168 179 197 172 167 153 159 14 AA14 10 100 Forest 163 136 158 158 137 152 152 146 AN 11 AN11 65 1000 Forest 781 805 798 832 787 826 743 782 AT 01 AT01 10 500 Forest 175 177 175 191 172 164 166 172 02 AT02 20 300 Forest 178 166 161 176 157 167 158 161 07 AT07 10 600 Non-forest 84 81 91 86 86 92 81 85 08 AT08 30 700 Non-forest 187 174 211 205 196 191 182 184 09 AT09 20 500 Non-forest 133 128 138 132 117 130 110 108 10 AT10 20 1600 Non-forest 160 158 149 160 149 144 136 141 12 AT12 40 500 Forest 405 380 417 405 395 392 370 369 13 AT13 30 1000 Non-forest 189 202 205 220 197 206 214 192 14 AT14 10 100 Forest 164 161 170 164 173 151 144 143 IM 01 IM01 20 300 Forest 237 251 255 263 246 251 250 241 02 IM02 20 300 Forest 179 172 177 186 175 180 177 155 03 IM03 30 1400 Forest 362 359 349 373 352 360 344 312 04 IM04 30 1800 Forest 539 539 543 580 492 492 493 475 05 IM05 40 2000 Forest 569 629 635 662 587 644 533 578 07 IM07 30 200 Non-forest 182 182 196 203 168 172 167 170 09 IM09 30 100 Non-forest 164 163 190 201 176 173 177 168 10 IM10 10 1000 Non-forest 89 99 102 110 97 103 90 90 13 IM13 30 200 Non-forest 181 176 173 192 182 171 165 170 14 IM14 20 100 Forest 229 244 280 268 246 231 220 252 NA 02 NA02 30 400 Forest 267 252 268 292 264 255 233 239 03 NA03 30 2000 Forest 384 382 392 409 382 412 377 358 04 NA04 50 300 Forest 803 849 1033 838 823 811 868 839 05 NA05 50 1100 Forest 713 707 757 750 742 755 689 616 06 NA06 60 400 Forest 928 983 937 907 870 968 842 876 07 NA07 30 100 Non-forest 175 188 188 185 173 170 165 158 08 NA08 50 700 Non-forest 423 438 475 464 446 431 432 425 11 NA11 65 300 Forest 751 755 806 818 745 773 659 673 12 NA12 40 500 Forest 391 374 406 416 399 403 376 403 13 NA13 40 1500 Non-forest 343 346 363 376 347 347 325 336 14 NA14 30 100 Forest 358 365 381 395 342 342 343 353 NT 01 NT01 10 200 Forest 173 185 183 165 173 164 161 146 02 NT02 20 400 Forest 172 174 188 191 171 174 157 171 03 NT03 20 1400 Forest 237 222 242 230 236 238 213 215 04 NT04 50 500 Forest 879 890 931 943 928 961 871 825 07 NT07 20 300 Non-forest 110 121 142 139 119 129 115 112 08 NT08 40 300 Non-forest 269 260 302 291 261 285 283 260 09 NT09 30 100 Non-forest 160 179 162 184 170 186 177 177 10 NT10 30 2000 Non-forest 251 276 271 287 264 264 249 242 12 NT12 40 600 Forest 394 443 395 420 402 405 408 404 13 NT13 20 500 Non-forest 119 131 135 134 139 123 129 127 14 NT14 20 100 Forest 228 239 266 276 242 238 231 244
228 Appendix D: Appendix to Chapter 5
Table D3. (continued) Region name Input parameters Median regeneration times (in years) median Altitude Ecosystem Herpeto- Realm Biome Realm_Biome Latitude (median) type Birds Mammals fauna Plants [° S or ° N] [m.a.s.l] Ext. Int. Ext. Int. Ext. Int. Ext. Int. OC 01 OC01 20 300 Forest 230 236 253 248 235 225 210 207 02 OC02 20 300 Forest 151 171 175 163 159 160 142 144 07 OC07 30 2000 Non-forest 243 235 239 239 255 234 219 213 PA 01 PA01 30 1300 Forest 465 452 447 462 457 474 481 423 04 PA04 50 200 Forest 876 812 959 883 813 827 802 819 05 PA05 50 1300 Forest 784 743 793 830 777 750 748 695 06 PA06 65 300 Forest 1147 1088 1231 1153 1203 1113 1143 1081 08 PA08 50 400 Non-forest 420 417 444 462 414 403 383 424 09 PA09 50 100 Non-forest 412 342 409 412 347 407 359 383 10 PA10 40 2000 Non-forest 368 386 413 400 398 402 337 358 11 PA11 65 200 Forest 666 728 755 767 747 712 672 685 12 PA12 40 600 Forest 409 445 443 420 420 398 408 393 13 PA13 40 500 Non-forest 254 289 296 300 264 285 288 271
229 Appendix D: Appendix to Chapter 5 Table D4. Land use classification of LADA and Anthromes Land use class used for this comparison LADA Anthrome_v2
Reference (=primary habitat) 1 Forest - virgin 61 Wild woodlands Reference (Natural forest, shrub, grassland, sparse vegetation, wetland, bare area) 2 Forest - protected 53 Remote woodlands 13 Shrubs - unmanaged 14 Shrubs - protected 7 Grasslands - unmanaged 8 Grasslands - protected 30 Sparsely vegetated areas - unmanaged 31 Sparsely vegetated areas - protected 26 Wetlands - unmanaged 27 Wetlands - protected 28 Wetlands - mangrove 34 Bare areas - unmanaged 62 Wild treeless and barren lands 35 Bare areas - protected Managed forest (human modified forest) Used Forest, incl Agroforest and 3 Forest - with agricultural activities 52 Populated woodlands young secondary forest (<10 years regrowing) 4 Forest - with moderate or higher livestock density 51 Residential woodlands Pasture 9 Grasslands - low livestock density 41 Residential rangelands 10 Grasslands - moderate livestock density 42 Populated rangelands 11 Grasslands - high livestock density 24 Pastoral villages 15 Shrubs - low livestock density 43 Remote rangelands 16 Shrubs - moderate livestock density 54 Inhabited treeless and barren 17 Shrubs - high livestock density 32 Sparsely vegetated areas - with low livestock density 33 Sparsely vegetated areas - mod.or high livestock dens. 36 Bare areas - with low livestock density 37 Bare areas - with mod. livestock density Agriculture 29 Wetlands - with agricultural activities 21 Rice villages Cropland (flooded, irrigated, rainfed, other) 23 Agriculture - large scale Irrigation 22 Irrigated villages 31 Residential irrigated cropland 33 Populated croplands 19 Rainfed crops (Subsistence/Commercial) 23 Rainfed villages 32 Residential rainfed croplands 34 Remote croplands 24 Agriculture - protected 20 Crops and mod. intensive livestock density 21 Crops and high livestock density 22 Crops, large-scale irrig., mod. or higher livestock dens. Urban area 25 Urban land 11 Urban 12 Mixed settlements
230 Appendix D: Appendix to Chapter 5
Table D4. (continued) Land use class used for this comparison LADA Anthrome_v2
No Data 0 No data 41 Undefined Excluded land cover types 40 Open Water - inland Fisheries Reference Natural (water body) 38 Open Water - unmanaged 39 Open Water - protected
231 Appendix D: Appendix to Chapter 5 D.3 Additional analysis and results
D.3.1 Correlation of characterization factors
Table D5. Pearson’s correlation coefficients between taxonomic groups across median CFs per ecoregion
Occupation Transformation Permanent Amphibians Mammals Reptiles Birds Plants Amphibians Mammals Reptiles Birds Plants Amphibians Mammals Reptiles Birds Plants Amphibians 1.00 1.00 1.00 Mammals 0.73 1.00 0.72 1.00 0.51 1.00
Reptiles 0.51 0.43 1.00 0.48 0.33 1.00 0.86 0.55 1.00 Birds 0.44 0.56 0.82 1.00 0.41 0.42 0.93 1.00 0.56 0.25 0.54 1.00 Plants 0.54 0.55 0.90 0.90 1.00 0.42 0.37 0.96 0.96 1.00 - - - - - Agriculture Amphibians 1.00 1.00 1.00 Mammals 0.52 1.00 0.44 1.00 0.36 1.00 Reptiles 0.94 0.51 1.00 0.96 0.42 1.00 0.83 0.40 1.00
Birds 0.27 0.86 0.28 1.00 0.37 0.91 0.36 1.00 0.40 0.28 0.46 1.00 Plants 0.30 0.73 0.30 0.80 1.00 0.12 0.81 0.12 0.79 1.00 - - - - - Urban Amphibians 1.00 1.00 1.00 Mammals 0.69 1.00 0.62 1.00 0.31 1.00 Reptiles 0.79 0.68 1.00 0.83 0.59 1.00 0.86 0.33 1.00 Birds 0.49 0.79 0.51 1.00 0.51 0.75 0.54 1.00 0.64 0.23 0.65 1.00 Plants 0.53 0.70 0.52 0.67 1.00 0.36 0.50 0.36 0.44 1.00 - - - - - Forestry Amphibians 1.00 1.00 1.00 Mammals 0.75 1.00 0.80 1.00 0.49 1.00 Reptiles 0.91 0.73 1.00 0.92 0.73 1.00 0.74 0.63 1.00 Birds 0.72 0.86 0.72 1.00 0.81 0.87 0.79 1.00 0.37 0.40 0.46 1.00 Plants 0.75 0.83 0.72 0.87 1.00 0.75 0.82 0.74 0.91 1.00 - - - - - Pasture
232 Appendix D: Appendix to Chapter 5 Table D6. Pearson’s correlation coefficients between land use types across median CFs per ecoregion
Occupation Transformation Permanent Agriculture Urban Forestry Pasture Agriculture Urban Forestry Pasture Agriculture Urban Forestry Pasture Agriculture 1.00 1.00 1.00 Urban 0.52 1.00 0.29 1.00 0.42 1.00 Forestry 0.85 0.58 1.00 0.58 0.26 1.00 0.85 0.48 1.00 Pasture 0.67 0.24 0.62 1.00 0.61 0.12 0.41 1.00 0.45 0.42 0.42 1.00 Mammals 1.00 1.00 1.00
Agriculture Urban 0.45 1.00 0.36 1.00 0.89 1.00 Forestry 0.94 0.38 1.00 0.66 0.17 1.00 0.92 0.89 1.00 Pasture 0.43 0.44 0.44 1.00 0.42 0.20 0.41 1.00 0.60 0.46 0.50 1.00 Amphibians Agriculture 1.00 1.00 1.00 0.31 1.00 0.16 1.00 0.85 1.00 Urban Forestry 0.56 0.55 1.00 0.39 0.23 1.00 0.94 0.89 1.00 Pasture 0.23 0.45 0.55 1.00 0.13 0.18 0.38 1.00 0.73 0.37 0.58 1.00 Reptiles Agriculture 1.00 1.00 1.00 Urban 0.41 1.00 0.11 1.00 0.71 1.00
Forestry 0.57 0.72 1.00 0.24 0.22 1.00 0.96 0.71 1.00 Pasture 0.32 0.12 0.42 1.00 0.20 0.08 0.27 1.00 0.16 0.02 0.12 1.00 Birds Agriculture 1.00 1.00 - Urban 0.26 1.00 0.02 1.00 - -
Forestry 0.45 0.36 1.00 0.13 0.03 1.00 - - - Pasture 0.31 0.20 0.40 1.00 0.18 0.04 0.16 1.00 - - - - Plants
233 Appendix D: Appendix to Chapter 5 D.3.2 Maps of median characterization factors
Figure D2. Median characterization factors amphibians. NA: No data available. Symbology based on Jenks natural breaks classification method (Jenks and Caspall 1971). Agriculture Managed forests Pasture Urban areas
9
Occupation
[regional species loss] *10 loss] species [regional
9
*years] *10 *years]
Transformation [regional species loss species [regional
6 *years] *10 *years]
Permanent
[global species loss species [global
234 Appendix D: Appendix to Chapter 5 Figure D3. Median characterization factors birds. NA: No data available. Symbology based on Jenks natural breaks classification method (Jenks and Caspall 1971).
Agriculture Managed forests Pasture Urban areas
9
Occupation
[regional species loss] *10 loss] species [regional
9
*years] *10 *years]
Transformation [regional species loss species [regional
6 *years] *10 *years]
Permanent loss pecies [global s [global
235 Appendix D: Appendix to Chapter 5 Figure D4. Median characterization factors mammals. NA: No data available. Symbology based on Jenks natural breaks classification method (Jenks and Caspall 1971).
Agriculture Managed forests Pasture Urban areas
9
Occupation [regional species loss] *10 loss] species [regional
9
*years] *10 *years]
Transformation [regional species loss species [regional
6 *years] *10 *years]
Permanent
[global species loss species [global
236 Appendix D: Appendix to Chapter 5 Figure D5. Median characterization factors reptiles. NA: No data available. Symbology based on Jenks natural breaks classification method (Jenks and Caspall 1971). Agriculture Managed forests Pasture Urban areas
9
Occupation [regional species loss] *10 loss] species [regional
9
*years] *10 *years]
Transformation [regional species loss species [regional
6 *years] *10 *years]
Permanent [global species loss species [global
237 Appendix D: Appendix to Chapter 5 Figure D6. Median characterization factors plants. NA: No data available. Symbology based on Jenks natural breaks classification method (Jenks and Caspall 1971). Agriculture Managed forests Pasture Urban areas
9
Occupation [regional species loss] *10 loss] species [regional
9
*years] *10 *years]
Transformation [regional species loss species [regional
238 Appendix D: Appendix to Chapter 5
D.3.3 Contribution to variance (CTV)
Table D7. Median contribution to variance (CTV) scores (of 1000 Monte Carlo iterations) for regional transformation characterization factors (per land use type and taxonomic group)
Parameter Taxa Agriculture Pasture Urban area Managed forest
CFloc,i,j Amphibians 30% 35% 78% 67% (local character- Reptiles 30% 28% 78% 66% ization factors) Mammals 34% 47% 82% 72% Birds 35% 48% 84% 74% Plants 35% 47% 83% 74%
Aorg,j Amphibians 0.16% 0.12% 0.05% 0.10% (original natural Reptiles 0.17% 0.11% 0.05% 0.10% habitat area) Mammals 0.42% 0.38% 0.17% 0.31% Birds 0.49% 0.41% 0.19% 0.29% Plants 0.38% 0.45% 0.22% 0.30%
Anew,j Amphibians 0.21% 0.15% 0.04% 0.14% (remaining Reptiles 0.20% 0.14% 0.04% 0.13% natural habitat) Mammals 0.61% 0.55% 0.14% 0.46% Birds 0.65% 0.52% 0.14% 0.42% Plants 0.53% 0.54% 0.19% 0.44%
Ai ,j Amphibians 0.18% 0.16% 0.05% 0.19% (area of land use Reptiles 0.19% 0.17% 0.05% 0.17% type i) Mammals 0.55% 0.58% 0.16% 0.50% Birds 0.66% 0.61% 0.17% 0.44% Plants 0.51% 0.63% 0.24% 0.48% treg i, j,g Amphibians 34% 11% 7% 8% (Regeneration Reptiles 34% 9% 6% 7% time) Mammals 52% 25% 10% 18% Birds 53% 26% 10% 16% Plants 54% 28% 11% 16% z Amphibians 0.04% 0.06% 0.01% 0.08% (power term Reptiles 0.04% 0.06% 0.01% 0.06% of SAR model) Mammals 0.16% 0.16% 0.04% 0.11% Birds 0.09% 0.12% 0.06% 0.13% Plants 0.08% 0.11% 0.07% 0.12%
239 Appendix D: Appendix to Chapter 5
Table D8. Median contribution to variance (CTV) scores (of 1000 Monte Carlo iterations) for regional occupation characterization factors (per land use type and taxonomic group)
Parameter Taxa Agriculture Pasture Urban area Managed forest
CFloc,i,j Amphibians 67% 82% 93% 91% (local character- Reptiles 68% 81% 93% 89% ization factors) Mammals 82% 91% 96% 95% Birds 82% 91% 96% 95% Plants 82% 91% 96% 95%
Aorg,j Amphibians 0.61% 0.31% 0.16% 0.17% (original natural Reptiles 0.63% 0.35% 0.16% 0.17% habitat area) Mammals 2.02% 1.24% 0.49% 0.57% Birds 2.15% 1.40% 0.48% 0.59% Plants 2.21% 1.62% 0.50% 0.61%
Anew,j Amphibians 1.30% 0.69% 0.18% 0.37% (remaining Reptiles 1.30% 0.80% 0.17% 0.36% natural habitat) Mammals 3.99% 2.21% 0.64% 1.08% Birds 3.99% 2.24% 0.63% 1.09% Plants 4.22% 2.29% 0.65% 1.12%
Ai ,j Amphibians 1.16% 0.59% 0.17% 0.40% (area of land use Reptiles 1.26% 0.62% 0.17% 0.39% type i) Mammals 3.88% 2.28% 0.61% 1.00% Birds 3.92% 2.51% 0.61% 1.06% Plants 4.08% 2.69% 0.63% 1.10% treg i, j,g Amphibians 0.01% 0.03% 0.04% 0.01% (Regeneration Reptiles 0.01% 0.02% 0.04% 0.01% time) Mammals 0.04% 0.06% 0.06% 0.03% Birds 0.06% 0.06% 0.05% 0.05% Plants 0.07% 0.05% 0.03% 0.07% z Amphibians 0.11% 0.05% 0.01% 0.05% (power term Reptiles 0.11% 0.06% 0.01% 0.04% of SAR model) Mammals 0.28% 0.13% 0.04% 0.16% Birds 0.28% 0.14% 0.04% 0.17% Plants 0.29% 0.15% 0.04% 0.18%
240 Appendix D: Appendix to Chapter 5
D.3.4 Comparing model choices The model presented in the manuscript is calculating average impacts based on past land use changes and is thus retrospective. Alternatively, impacts can be calculated as marginal changes (Huijbregts et al. 2011; Weidema 2012), i.e. the impact one additional m2 of land use would cause. Impacts can also be calculated for future land use changes, i.e. prospective. To illustrate the sensitivity of the model to these model choices, we calculated average and marginal impacts for both retro- and prospective assessments for all forest ecoregions (at least 90% original forest cover) of the Amazon (n=19), for plants, mammals and birds. This region was selected as no future land use scenarios for all global ecoregions were readily available and the Amazon contains some currently little disturbed ecoregions, which are expected to be converted for human use in the near future.
Methods for the retrospective versus prospective assessment Land use scenarios for 2050 were derived from Soares-Filho et al. (2006), which provide several deforestation scenarios for each ecoregion in the Amazon. To get the range of possible results, we selected a best- (Governance) and worst-case (Buisness-as-usual, BAU) land use scenario. The BAU scenario assumes that the recent deforestation trends will continue, currently planned highways will be paved, legislation requiring forest reserves will remain low and no new protected areas will be created (Soares-Filho et al. 2006). The Governance scenario assumes that Brazilian environmental legislation is implemented across the Amazon basin, including current experiments on frontier governance (mandatory forest reserves, protected area network, agro-ecological zoning of land use; Soares-Filho et al. 2006). As these scenarios only consider deforestation and do not specify the type of land use established on the converted land, we assumed that the current land use mix remains the same (i.e. pi is constant, but Anew changes, see ‘Methods’ section of manuscript). In addition, Soares-Filho et al. only considered deforestation and no other types of land use (e.g. selective logging, sparse settlement, etc.) and used other land cover data with a higher resolution than in this study. Therefore, their estimates of deforested land cannot be directly compared to the share of human-modified land (Alost/Aorg) as derived in our study. We used the future deforestation rate rj per ecoregion j (eq D1) as an proxy of the land use change rate. To get future shares of human-modified land (Alost,2050 /Aorg), we multiplied the current share of human-modified land by the land use change rate r, eq D2. The resulting remaining share of natural habitat (Anew/Aorg) of the original scenarios and our adaptations are illustrated in Figure D8.
A deforest,j ,2050 r = (D1) j A deforest,j ,current
A A lost,j ,2050 lost,j ,current = r i (D2) A j A org,j org,j
241 Appendix D: Appendix to Chapter 5
Methods for the average versus marginal assessment In this publication, we have calculated average impacts (see Methods section of main paper). Average impacts are given as the total damage divided by the total area (see Figure D7). The marginal approach calculates the impacts of one additional m2 of land. The marginal damage function (eq D4) is given by the first derivative of the average damage function (eq D3) by the lost area (converted area):
n z p CF 'i i loc ,i ! A $ S = S -S = S -S i# new & (D3) lost,g org,g new ,g org,g org,g # A & " org %
n (z p CF )#1 "i i ,j loc ,i ,j !S S n $ A ' lost,g,j = org,g,j *z p CF *&1# lost,g,j ) A A "i i ,j loc,i ,j & A ) ! lost,g,j org,g,j % org,g,j ( n (D4) (z p CF )#1 "i i ,j loc ,i ,j S n $ A ' = org,g,j *z p CF *& new ,g,j ) A "i i ,j loc,i ,j & A ) org,g,j % org,g,j (
Aorg is assumed constant, only Alost is changing (no new reference). Also pi and CFi are assumed as constants.
The CFs are then calculated as a marginal species loss ΔSlost,marg due to a marginal increase in
2 human used area ΔAlost, marg = 1m .
S a ! lost,nonend ,j ,g i i ,j CF = (D5) reg,occ,i ,j ,g,mar A p ! lost,j i i ,j
1 CF t CF (D6) reg,trans,i ,j ,g,mar = i reg,i ,j ,g i reg,occ,i ,j ,g,mar 2 S a ! lost,end ,j ,g i i ,j CF =t i (D7) reg,perm,i ,j ,g,mar m !A ip lost,j i ,j A graphical illustration is provided in Figure D7, for an invented example.
242 Appendix D: Appendix to Chapter 5
Figure D7. Illustration of differences between average and marginal damages. At low levels of remaining habitat (Alost=984, green, full lines): marginal damage is much higher than average. At intermediate levels of remaining habitat (Alost=500, red dotted lines): differences are small. (Example numbers: Atotal = 1000, Stotal =300)
243 Appendix D: Appendix to Chapter 5
Remaining non-deforested habitat, Remaining natural habitat, derived from LADA & Anthromes Soares-Filho et al. 2006 situation Current
Governance Scenario
BAU Scenario
Figure D8: Share of remaining habitat (Anew/Aorg) for the current situation (top) and for the two scenarios Governance (middle) and Business as Usual, (BAU, bottom) 2050. Left: share of non-deforested area (data from Soares-Filho et al. (2006)). Right: own calculations, extrapolated scenarios with the LADA and Anthromes land use shares. Light yellow = undisturbed; dark blue = heavily disturbed.
Results of comparison of model choices Results for retro- and prospective assessment and average and marginal approach are illustrated in Figure D9. For the worst-case scenario (BAU), the median prospective CFs increased only for ecoregions with large projected land use changes, for small increases in land use no changes in median CFs becomes apparent (SI, Figure D9). A maximum increase of median CF of 65% was observed in ecoregions with a projected land use change from 60% remaining habitat to 20%. 244 Appendix D: Appendix to Chapter 5
Using a marginal impact calculation, the CFs did not change considerably at low levels of habitat conversion, but at high levels, impacts were even more pronounced using a marginal approach, turning to infinity when the remaining natural habitat would become very small (Figures D7 and D9). Thus, the marginal approach gives even higher weight to highly vulnerable areas and is even more reactive as the average approach.
Average Marginal situation Current
Governance Scenario
BAU Scenario
Figure D9. Characterization factors, occupation, mammals: Average (left) and marginal (right) CFs for the current situation (top) and for the two scenarios Governance (middle) and Business as Usual, (BAU, bottom) for 2050. Numbers displayed as [regional species loss / m2]*109.
245 Appendix D: Appendix to Chapter 5
D.3.5 Model evaluation
Endemic Birds Endemic Mammals observed endemic species loss observed endemic species loss 024681012 0 5 10 15
024681012 0 5 10 15 predicted endemic species loss predicted endemic species loss
Endemic Reptiles Endemic Amphibians observed endemic species loss observed endemic species loss 0 5 10 15 20 25 30 0 10 20 30 40 50 60 70
0 5 10 15 20 25 30 0 10 20 30 40 50 60 70
predicted endemic species loss predicted endemic species loss
Figure D10. Comparison of predicted and observed globally threatened and extinct endemic birds, mammals, reptiles and amphibians. For each ecoregion, all extinction estimates (n=1000 Monte Carlo iterations) are displayed. Colors represent data density (green = low density, dark red = highest density); black crosses represent median species loss per ecoregion; dotted line represents 1:1 (complete overlap of predicted and observed). Pearson correlation coefficient between observed and predicted species loss: rBirds= 0.46; rMammals = 0.42; rReptiles= 0.16; rAmphibians = 0.69.
246 Appendix D: Appendix to Chapter 5
D.3.6 Aggregation of characterization factors across taxa LCA practitioners might be overwhelmed by 5 characterization factors specific per taxonomic group. We therefore suggest a way to aggregate these factors. As we present absolute impacts (species losses), simply adding CFs across taxon would give equal weight to e.g. one plant species lost and one mammal species lost. In this case, the impacts would be strongly dominated by plants, by far the most species rich taxon assessed here. To also capture impacts on other taxa, we suggest a weighting of the CFs per taxon. The weighting factor was calculated based on the median species richness per taxa of all global ecoregions, Sg,med. We then standardized these values by the median richness of mammals Smam,med. The weighting factor w per taxonomic group g is then given as, (see Table D9):
S w = mam,med (D8) g S g,med
Table D9. Median species richness Sg,med per taxa g of all global ecoregions and resulting weighting factors wg. Taxon Mammals Birds Plants Amphibians Reptiles
Sg,med 69 230 1400 7 17 wg (rounded) 1 0.3 0.05 10 4
The aggregated CFs were then calculated as a weighted average across all taxonomic groups g.
n w *CF !g g reg,occ,i ,j ,g CF = (D9) reg,occ,i ,j n j
n w *CF !g g reg,trans,i ,j ,g CF = (D10) reg,trans,i ,j n j where nj is the total number of taxonomic groups g that contain data (i.e. not only zero values) per ecoregion j.
Characterization factors for permanent impacts should indicate, if a land use activity might cause irreversible impacts. Therefore, we suggest to not include a weighting or averaging, but simply summing up all permanent impacts.
n CF = CF (D11) reg,perm,i ,j !g reg,perm,i ,j ,g
D.3.7 Calculation of world average characterization factors In many LCA studies, the geographic location of background processes and their land use is unknown. For these cases, we calculated world average CFs by weighting the CF of each ecoregion by their global area share (see spreadsheet in the online supporting information SI2 of de Baan et al. 2013b). Alternatively, worst-case (highest values) or median CFs could be chosen.
247 Appendix D: Appendix to Chapter 5
D.4 References Curran M, Hellweg S, Beck J (in press) Is there any empirical support for biodiversity offset policy? Ecological Applications. de Baan L, Alkemade R, Koellner T (2013a) Land use impacts on biodiversity in LCA: a global approach. The International Journal of Life Cycle Assessment 18 (6):1216- 1230. de Baan L, Mutel CL, Curran M, Hellweg S, Koellner T (2013b) Land use in Life Cycle Assessment: Global characterization factors based on regional and global potential species extinctions. Environmental Science & Technology 47 (16):9281–9290. Drakare S, Lennon J, Hillebrand H (2006) The imprint of the geographical, evolutionary and ecological context on species-area relationships. Ecology Letters 9 (2):215-227. Ellis E, Ramankutty N (2008) Putting people in the map: anthropogenic biomes of the world. Frontiers in Ecology and the Environment 6 (8):439-447. Huijbregts MAJ, Hellweg S, Hertwich E (2011) Do we need a paradigm shift in Life Cycle Impact Assessment? Environmental Science & Technology 45:3833-3834. Jenks GF, Caspall FC (1971) Error on Choroplethic Maps. Definition, Measurement, Reduction. Annals of the Association of American Geographers 61:217–244. Kier G, Mutke J, Dinerstein E, Ricketts T, Kuper W, Kreft H, Barthlott W (2005) Global patterns of plant diversity and floristic knowledge. Journal of Biogeography 32 (7):1107-1116. Koellner T, de Baan L, Beck T, Brandão M, Civit B, Margni M, Milà i Canals L, Saad R, de Souza DM, Müller-Wenk R (2013) UNEP-SETAC guideline on global land use impact assessment on biodiversity and ecosystem services in LCA. The International Journal of Life Cycle Assessment 18 (6):1188-1202. LADA (2008) Mapping Land use Systems at global and regional scales for Land Degradataion Assessment Analysis. Nachtergaele F., Petri, M. LADA Technical Report n.8, version 1.1. UNEP/GEF. Milà i Canals L, Bauer C, Depestele J, Dubreuil A, Freiermuth Knuchel R, Gaillard G, Michelsen O, Müller-Wenk R, Rydgren B (2007) Key Elements in a Framework for Land Use Impact Assessment Within LCA. The International Journal of Life Cycle Assessment 12 (1):5-15. Olson D, Dinerstein E, Wikramanayake E, Burgess N, Powell G, Underwood E, D'Amico J, Itoua I, Strand H, Morrison J, Loucks C, Allnutt T, Ricketts T, Kura Y, Lamoreux J, Wettengel W, Hedao P, Kassem K (2001) Terrestrial ecoregions of the worlds: A new map of life on Earth. BioScience 51 (11):933-938. Soares-Filho BS, Nepstad DC, Curran LM, Cerqueira GC, Garcia RA, Ramos CA, Voll E, McDonald A, Lefebvre P, Schlesinger P (2006) Modelling conservation in the Amazon basin. Nature 440 (7083):520-523. Weidema BP (2012) New paradigm or old distance to target? Environmental Science & Technology 46 (2):570-570. WildFinder: Online database of species distributions (2006) World Wildlife Fund. http://www.worldwildlife.org/WildFinder.
248 Appendix E: Appendix to Chapter 6
Appendix E: Appendix to Chapter 6
E.1 Methods
E.1.1 Supplementary methods E1: Land use assessment framework in LCA Assessing biodiversity impacts of land use in Life Cycle Assessment (LCA) adopts a two stage framework, reflecting two possible types of ecosystem damages: transformation impacts and occupation impacts (Koellner et al. 2013; Milà i Canals et al. 2007). These two impact types are depicted in Figure E1 by the trajectory of the yellow line, which indicates the change in biodiversity over time. After an initial land use change, such as a transformation from a natural to an anthropogenically-modified state, or from an extensive to an intensively used state, biodiversity value (here represented by threat- and rarity-weighted species richness) of a reference situation, Sref, is reduced to Si. Assuming land is used for a specific period for a specific purpose such as crop production (the land use phase), natural regeneration of biodiversity is suppressed. At some hypothetical time in the future, land abandonment is assumed to take place, and biodiversity recovers to a level comparable to the pre-transformation state, Sref. The duration of this regeneration is referred to as the regeneration or recovery phase, treg. In this context, impacts are considered to be fully reversible given a long enough time horizon. The final occupation (land use) and transformation impact (of the land use change) are given as the product of biodiversity loss, ΔS, time, T, and area, A, affected (Koellner et al. 2013; Milà i Canals et al. 2007). This is considered as the Biodiversity Depletion Potential, BDP.
BDP = (S ! S )i A i(t !t ) = "S i A iT (E1) Occ ref i Occ 1 0 Occ Occ
BDP = (S ! S )i A i(t !t )i0.5 = "S i A it i0.5 (E2) Trans ref i Trans 2 1 Trans reg
249 Appendix E: Appendix to Chapter 6
!"#$"%&'(")*!
# )**! #$%&'(! !"#$%
!""#$%&'()) "#$%&'(! *+$%",%% 64%(/7'4+%&'() "#)**! )*+$%",) !&% -%(.)#/0) 20"'3045) $1%/0) $1%/0) +,-.%/0#% .'&/! +,-.% 12,-3#% ,4,-.5-6 7#-'%
+",&- ' % ' ' ' * (% &8%"#3% )%
Figure E1. Land use assessment framework in Life Cycle Assessment (adapted from Koellner et al. 2013; Milà i Canals et al. 2007)
The convention for life cycle inventory databases is to report land occupation of processes as area used over a certain time (blue area in Figure E1):
I = A iT Occ Occ Occ (E3)
The biodiversity impact (or characterization factors in LCA terminology) of occupation BIOcc (blue arrow), is equal to the difference between the species richness of a reference state and the richness under land use.
BI = S ! S Occ ref i (E4)
For land transformation, the inventory simply represents the area converted per process (purple arrow in Figure E1):
I = A Trans Trans (E5)
The biodiversity impact of transformation BITrans is given as (purple triangle in Figure E1):
1 1 BI = (S ! S )it i = BI it i Trans ref i reg Occ reg (E6) 2 2 The time-lag involved in biodiversity regeneration is attributed to the initial transformation activity, rather than the occupation phase. This is due to the fact that even in the absence of
250 Appendix E: Appendix to Chapter 6 occupation, a lag in diversity during recovery would result from a transformation activity (e.g. clear-cut logging of a forest and the subsequent recovery of native diversity).
In some cases, the assumption of full recovery cannot be met. For example, a transformation may drive a species to extinction, biotic or abiotic conditions may be irreversibly modified (e.g. soil compaction, total topsoil loss), or the time horizon of recovery may be impractically long for meaningful prediction. In these cases, a permanent impact is said to occur (Koellner et al. 2013). We do not address permanent impacts in this study, restricting our analysis to occupation and transformation impacts. However, one approach for quantifying permanent impacts is given in de Baan et al. (2013b).
E.1.2 Supplementary methods E2: Local, relative method (R-Local) In the method developed by de Baan et al. (2013a), local species richness of different types of land use were compared to (semi-)natural reference areas within the same biogeographic region. The relative reduction in local species richness was used as an indicator of impacts on biodiversity. Data was derived from a global literature review (GLOBIO3, Alkemade et al. 2009), containing 195 publications providing 644 data points on different land use types and 254 data points on (semi- )natural reference situations from a total of nine out of 14 biomes. Because most data points came from tropical regions, de Baan et al. (2013a) complemented the dataset with national biodiversity monitoring data from Switzerland (temperate region) (BDM 2004). In contrast to the GLOBIO3 model, which calculated reduction in mean species abundance of original species (MSA), de Baan et al. (2013a) used the GLOBIO3 database to calculate reduction in species richness, because this facilitated the inclusion of more studies (i.e. also studies that did not report species lists and abundance data). The GLOBIO3 dataset contains data on a broad range of taxonomic groups, while the BDM dataset contains only data on vascular plants, mollusks and mosses. In the present study, we used median biodiversity impacts (characterization factors) calculated per WWF biome (Olson et al. 2001) over all taxonomic groups for the two land use types annual and permanent crops. If less than five data points were available for one land use type in a biome, median global values were used instead. For permanent crops, only the biome Sub-Tropical Moist Forest had enough data to give a biome-specific biodiversity impact, all other biomes in the case study region (Savanna, Flooded Grassland, Montane Grassland, Desert) were assigned the median global value.
Biodiversity impacts of transformation (see Equation E6), not included in de Baan et al. (2013a), were quantified based on recovery times provided by de Baan et al. (2013b), based on the biodiversity recovery model developed by Curran et al. (in press) (see also section E5). To estimate transformation impacts, we used estimates of biome-specific recovery times per WWF realm (i.e. continent).
E.1.3 Supplementary methods E3: Regional, absolute method (A-Regional) This method was developed by de Baan et al. (2013b) and estimates regional impacts of land use based on species-area relationships (Arrhenius 1921; Koh and Ghazoul 2010). The method calculated the potential regional loss of species within WWF ecoregions (Olson et al. 2001) due to the historical conversion of habitat and allocated this loss to the different types of land use
251 Appendix E: Appendix to Chapter 6
occurring in each ecoregion. The matrix-calibrated species area-relationship developed by Koh and Ghazoul (2010) was used to model regional species loss. This model accounts for the moderating effect of the habitat quality of human-modified land (i.e. the land use matrix) on species loss.
n z p ' (i i i S ! A $ new = # new & S # A & (E7) org " org %
S is the regional species richness of a native, undisturbed (original) ecosystem (Sorg) and of the current land use mix (Snew). Aorg is the total area of the ecoregion, while Anew is the remaining natural habitat area. z is a constant that indicates the species-accumulation rate observed in true island archipelagos (in de Baan et al. (2013b), this was based on data of Drakare et al. (2006)). n is the total number of land use types i and p is the area share of each land use type within the total converted (i.e. non-natural) area. ! is a measure of the habitat quality of each type of land use i. De Baan et al. (2013b) used the biodiversity impacts scores (local relative change in species richness) from the previous study (de Baan et al. 2013a) as an indication of habitat quality ! of different types of land use in different world regions.
Species loss Sloss per ecoregion was finally calculated as
S = S ! S loss org new (E8)
were Sorg was derived from WWF databases and Snew was calculated based on equation E7. We separately calculated regional species loss of endemic and non-endemic species. The potential loss of non-endemic species was used to calculate reversible biodiversity impacts of occupation (land use) and transformation (land use change). Potentially lost species which are endemic to one ecoregion were used to calculate permanent land use impacts, as endemic species loss infers irreversible global extinction. However, in the present study, we only illustrate results based on reversible occupation and transformation impacts based on non-endemic species.
Finally, de Baan et al. (2013b) allocated the regional species loss to the different types of land use present in each ecoregion j, based on the relative area share each land use type i occupies (p) and based on their habitat quality (!). The allocation factor was calculated as follows:
p ! a = i ,j i ,j i ,j n (E9) p ! "i i ,j i ,j
The biodiversity impact of land occupation (BIOcc, or characterization factor in LCA terminology) per land use type i, region j, and taxonomic group g was calculated as a multiplication of the regional species loss by the allocation factor, divided by to total area of each type of land use Ai,j per ecoregion.
252 Appendix E: Appendix to Chapter 6
S ia BI = lost,nonend ,j ,g i ,j Occ,reg,i ,j ,g A (E10) i ,j Biodiversity impact of transformation were calculated based on equation E6. Biome-specific regeneration times treg were calculated per realm (i.e. continent) based on a meta-study of Curran et al. (in press) (see also Supplementary method E5).
The study of de Baan et al. (2013b) provides biodiversity impacts (BI) and their uncertainty (based on Monte Carlo simulations) for five taxonomic groups (mammals, birds, amphibians, reptiles, plants) and four land use types (agriculture, used forests, artificial area, pastures). In the present study, we only used the median values for mammals and the land use type “agriculture”.
E.1.4 Supplementary methods E4: Calculation of life cycle inventory data To estimate the land use change related to the production of 1kg of crop (tea, coffee, or tobacco), we adapted a three-step approach suggested by Milà i Canals et al. (2013). The approach is based on national statistics on land cover and harvested area per crop provided by the Food and Agriculture Organization of the United Nations (FAO) (FAOSTAT 2013b, a). In a first step, the crop expansion rate per country was calculated over a 20 year time period (Table E1). In a second step, the change in land cover types per country was assessed over the same time period (Table E2). A crop was only considered to contribute to land use change if both the area of crop production (step 1) and the land cover type associated to this crop production (step 2) increased over this time period. Thereby, a pure shift from one crop to another (e.g. from maize to tobacco) was not considered as land use change. While the first two steps were done in line with Milà i Canals et al. (2013), we adapted the third step to better fit our aims. Milà i Canals et al. (2013) suggest to calculate the land use change rate per crop based on the change in land cover type. In our case study, this resulted in equal land transformation proportions per crop area for both coffee and tea (both grown on the land cover type “permanent crops”), although in some countries, such as Kenya, coffee only slightly increased in the past 20 years (+1%) while the tea cultivation area largely increased (+36%). Therefore, we adapted the last step of the approach and calculated the land use change rate per crop based on the change in crop area (and not on the change in land cover area, Table E3). In the present study, land use change was only considered from forest area (i.e. if the forest area decreased in the same 20 year period). The impacts of other land transformations (i.e. from pasture to annual crop) were not considered, as the methods to assess impacts on biodiversity (i.e., R-Local, A-Regional and W-Local) did not cover land transformations between different types of human use. Neither, could we calculate the land conversion of savanna and grassland ecosystems, as the FAO database does not specify this land cover type: natural grasslands are contained within the land cover class “other”, which also contains very different land cover types, such as artificial areas.
To minimize the influence of short-term fluctuations in land use, we used 5-year averages to calculate steps 1-3 (Milà i Canals et al. 2013). The final amount of land use change per kg of crop was calculated by dividing the land use change rate by the crop yield (on a grid cell level).
253 Appendix E: Appendix to Chapter 6
To get a better understanding of the procedure, we illustrate the approach in the case of tea grown in Kenya. In step 1, we found that the harvested area of tea increased by 36%, from 106’022 ha (1991-1996) to 164’991 ha (2007-2011). In step 2, we considered the change of the land cover type “permament crops”, which increased from 480'000 ha (1991-1996) to 610'000 ha (2007-2011, +21%). Because both step 1 and 2 showed an increase in crop and land cover area, respectively, we allocated land use change to tea production in Kenya. In step 3, we analyzed which land use types decreased in the same time. Forest area decreased by 192’000 ha while “other” land decreased by 90’000 ha in the 20 year period. Therefore, only 68% (=192’000/(192’000+90’000)) of the land use change of tea was associated with conversion of forest, the other 32% was not assessed. The final forest conversion rate associated with the growth in tea production was calculated as the multiplication in the land use change rate of tea (36%) and the share of forest converted (68%). This resulted in a land use change rate of 24% for tea in Kenya considering the land use changes over the past 20 years. This rate was finally divided by the yields of each grid cell (based on data of Monfreda et al. 2008) to get the amount of land use change associated with the production of 1 kg of crop in each grid cell within the borders of Kenya.
254 Appendix E: Appendix to Chapter 6
Table E1. Step 1: Change in harvested area (in ha) of coffee, tea and tobacco over the past 20 years for all East African countries (data source: FAOSTAT 2013b) . No LUC: no land use change (crop area did not increase)
Change in crop Average 1991- Average 2007- area over past Crop expansion Harvested Area (in ha) 1995 2011 20 years rate Burundi Coffee, green 37000 19040 -17960 No LUC Tea 6303 8900 2597 29% Tobacco, unmanufactured 3951 1594 -2356 No LUC Djibouti Coffee, green No data No data No data No data Tea No data No data No data No data Tobacco, unmanufactured No data No data No data No data Eritrea Coffee, green No data No data No data No data Tea No data No data No data No data Tobacco, unmanufactured No data No data No data No data Ethiopia Coffee, green 234118 397460 163342 41% Tea 2751 7272 4520 62% Tobacco, unmanufactured 5367 5641 274 5% Kenya Coffee, green 157320 159600 2280 1% Tea 106022 164991 58969 36% Tobacco, unmanufactured 7106 18197 11090 61% Rwanda Coffee, green 41130 34991 -6139 No LUC Tea 10542 12937 2395 19% Tobacco, unmanufactured 2846 4041 1195 30% Somalia Coffee, green No data No data No data No data Tea No data No data No data No data Tobacco, unmanufactured 241 312 71 23% Uganda Coffee, green 264200 308000 43800 14% Tea 15600 24962 9362 38% Tobacco, unmanufactured 7505 14996 7491 50% United Republic of Tanzania Coffee, green 126992 119634 -7358 No LUC Tea 18549 21483 2933 14% Tobacco, unmanufactured 36120 74878 38758 52%
Table E2. Step 2: Land cover change (in 1000 ha) for all East Africa countries over the past 20 years (data source: FAOSTAT 2013a). No LUC: land cover area did not change Country Land cover type Average 1991- Average Land cover change Summary: 1995 2007-2011 over past 20 years Land cover expansion (in 1000 ha) (in 1000 ha) (in 1000 ha) (% of total converted land)
255 Appendix E: Appendix to Chapter 6
Country Land cover type Average 1991- Average Land cover change Summary: 1995 2007-2011 over past 20 years Land cover expansion (in 1000 ha) (in 1000 ha) (in 1000 ha) (% of total converted land) (1) Burundi Arable land 932 939 7 increase Permanent crops 368 388 20 increase Permanent meadows/pastures 835 916 81 increase Other land 171 148 -23 decrease (21%) Forest area 262 177 -85 decrease (79%) Djibouti Arable land 1 2 1 increase Permanent crops No data No data No data no data Permanent meadows/pastures 1391 1700 309 increase Other land 920 611 -310 decrease (100%) Forest area 6 6 0 No LUC Eritrea Arable land 458 682 224 increase Permanent crops 2 2 0 No LUC Permanent meadows/pastures 6945 6900 -45 decrease (20%) Other land 1093 980 -113 decrease (50%) Forest area 1603 1536 -67 decrease (30%) Ethiopia Arable land 9957 13911 3953 increase Permanent crops 547 985 438 increase Permanent meadows/pastures 20000 20000 0 No LUC Other land 54946 52668 -2278 decrease (48%) Forest area 14550 12437 -2114 decrease (52%) Kenya Arable land 5268 5420 152 increase Permanent crops 480 610 130 increase Permanent meadows/pastures 21300 21300 0 No LUC Other land 26196 26106 -90 decrease (32%) Forest area 3670 3478 -192 decrease (68%) Rwanda Arable land 802 1190 388 increase Permanent crops 275 250 -25 decrease (5%) Permanent meadows/pastures 606 450 -156 decrease (32%) Other land 458 152 -307 decrease (63%) Forest area 326 425 99 increase
Somalia Arable land 1027 1080 53 increase Permanent crops 20 29 8 increase Permanent meadows/pastures 43000 43000 0 No LUC Other land 10635 11802 1167 increase Forest area 8052 6824 -1228 decrease (100%)
256 Appendix E: Appendix to Chapter 6
Country Land cover type Average 1991- Average Land cover change Summary: 1995 2007-2011 over past 20 years Land cover expansion (in 1000 ha) (in 1000 ha) (in 1000 ha) (% of total converted land) Uganda Arable land 5044 6560 1516 increase Permanent crops 1922 2200 278 increase Permanent meadows/pastures 5112 5112 0 No LUC Other land 3417 3033 -384 decrease (21%) Forest area 4486 3076 -1410 decrease (79%) United Arable land 8901 11205 2305 increase Republic of Tanzania Permanent crops 1000 1680 680 increase Permanent meadows/pastures 24000 24000 0 No LUC Other land 14394 17864 3469 increase Forest area 40285 33831 -6454 decrease (100%) (1) The numbers for Burundi for 2009 and 2010 were incorrect (they did not add up to the total land area). Therefore, we replaced the values for 2009/10 with the (correct values) for 2005/06
257 Appendix E: Appendix to Chapter 6
Table E3. Overview of land use change calculation for coffee, tea and tobacco for all East African countries. No LUC: no land use change was allocated to the crop Step 1 Step 2 Step 3 Final land change rate from forest to arable Increase in Land Land change (tobacco) or Crop expansion land cover change rate rate from permanent rate type? from forest „other“ land (coffee and tea) Burundi 79% 21% Coffee, green No increase Yes No LUC Tea 29% Yes 23% Tobacco, unmanufactured No increase Yes No LUC Djibouti 0% 100% Coffee, green No data Yes No LUC Tea No data Yes No LUC Tobacco, unmanufactured No data Yes No LUC Eritrea 30% 50% Coffee, green No data No No data Tea No data No No data Tobacco, unmanufactured No data Yes No data Ethiopia 48% 52% Coffee, green 41% Yes 20% Tea 62% Yes 30% Tobacco, unmanufactured 5% Yes 2% Kenya 68% 32% Coffee, green 1% Yes 1% Tea 36% Yes 24% Tobacco, unmanufactured 61% Yes 42% Rwanda 0% 63% Coffee, green No increase No No LUC Tea 19% No No LUC Tobacco, unmanufactured 30% Yes No LUC Somalia 100% 0% Coffee, green No data Yes No data Tea No data Yes No data Tobacco, unmanufactured 23% Yes 23% Uganda 79% 21% Coffee, green 14% Yes 11% Tea 38% Yes 29% Tobacco, unmanufactured 50% Yes 39% United Republic of Tanzania 100% 0% Coffee, green No increase Yes No LUC Tea 14% Yes 14% Tobacco, unmanufactured 52% Yes 52%
258 Appendix E: Appendix to Chapter 6
E.1.5 Supplementary methods E5: Calculation of recovery times of biodiversity Biodiversity recovery times for the method W-Local were based on a spatial prediction of the models of Curran et al. (in press), who conducted a meta-analysis of the habitat restoration and secondary growth literature. The study analyzed species checklist data from 39 comparative studies of secondary growth (SG) and old growth (OG) habitat from around the world. Changes in species similarity and assemblage composition were modeled as a function of age and 12 additional predictors (geographic distance between sampling sites, disturbance intensity, elevation, biome, latitude, SG–OG patch connectivity, SG and OG patch size, restoration method, realm, differences in sampling effort, number of sites per treatment and taxon). The study employed Generalized Linear Models (GLMs) and multi-model averaging to derive partial effects of each predictor.
We spatially mapped recovery time predictions for Sorenson similarity based on weighted parameters of the “general” model (i.e. the simplest model including all data, but omitting the effects of patch size, connectivity and distance). To do this, we first developed spatial datasets for the relevant spatial predictors (Table E4). For non-spatial variables, we assumed partial effects for mammals (taxon), passive restoration (restoration method), equal sampling effort across habitats (sampling effort) and 10 sampling sites per treatment (number of sites). We also added additional partial effects for OG habitat patch size and isolation distance (see Table E4 for details) using the parameters of the “patch” and “distance” models of Curran et al. (in press).
259 Appendix E: Appendix to Chapter 6
Table E4. Development of spatial data layers for predicting assemblage recovery times across the study region, based on the model of Curran et al. (in press).
Predictor Preparation Elevation Global 1 km resolution elevation dataset from WorldClim (www.worldclim.org), based resampling the Shuttle Radar Topography Mission (SRTM) 90 m resolution DEM product. Absolute latitude Distance from the equator in the North and South direction, in degrees. We assumed no effect of latitude for pixels less that 1 degree from the equator. Biome Biome data from the ecoregion spatial dataset (Olson et al. 2001), with a simple reclassification based on structural complexity: “Open” (biomes 7, 8, 9, 10, 11, 13, 14, 15), “Transition” (biomes 2 and 12), “Broadleaf” (biomes 1 and 4) and “Coniferous” (biomes 3, 5, 6). Disturbance intensity We assumed two classes of disturbance intensity for land occupation based on classifying the anthropogenic land cover classes of the ESA GlobCover 2009 land cover product (European Space Agency 2009): intensive occupation (classes 10– 15 and 190) and extensive occupation (classes 16–32). To be conservative, we assumed land was occupied before regeneration takes place (i.e. no direct regeneration after a once-off transformation, which can lead to faster recovery; Curran et al. accepted) OG patch size We added partial effects based on the “patch model” of Curran et al. (in press). To do this we assumed all non-anthropogenic land cover classes from the ESA GlobCover 2009 data (European Space Agency 2009) represents near-natural OG habitat (i.e. source habitat for OG species). We added to this all protected areas for the year 2010 (from the World Database of Protected Areas). We calculated the sum of all natural area in a moving window of 110 pixels dimension (ca. 33 km). We converted these area sums to a log10 scale, and applied partial effects of the model of Curran et al. (in press). We did not assume any effect of SG patch size (i.e. the size of the recovering habitat area) as this effect was predicted to be of minor importance in the model of Curran et al. (in press) relative to OG patch size. Distance The effect of patch isolation on recovery speed was included by calculating the distance from the nearest pixel of (near-)natural habitat (see “OG patch size” description). We applied a partial effect of distance from the “distance” model of Curran et al. (in press), reflecting distance decay in ecological similarity ( Soininen et al. 2007). This assumes that more isolated sites will recover more slowly because the similarity between the original community and remaining patches of natural habitat decreases with increasing distance. SG–OG connectivity A binary connectivity effect (connected/isolated) was applied based on a cut-off distance of 5 km from the nearest natural habitat patch (i.e. all pixels within 5 km of natural habitat were considered “connected”).
Recovery times were defined as the time required for assemblage similarity to fall within half a standard deviation of the average background OG–OG similarity. This allowed for some variability in identifying recovery success given the high level of background variation observed in reference OG–OG comparisons. However, Sorenson similarity (reflecting the presence of shared species) does not differentiate between the type of species contributing to (dis-)similarity. Variation between OG–OG samples may be due to high rates of species turnover of habitat specialists of conservation concern, whereas variation in SG–OG similarity might reflect turnover between habitat specialists and generalist species. Therefore our success criterion of half a standard deviation was statistically conservative to avoid misrepresenting the recovery process.
260 Appendix E: Appendix to Chapter 6
Re-arranging the linear model formula allowed age at recovery to be predicted according to:
!0.5iSD!(!B+ A X ) " i ij A B = 10 age age (E11) where Ai is the coefficient for predictor i, Xi is the value of the predictor in cell j, and B is the intercept of the model. Age is back-transformed from log10 space. The resulting spatial recovery time predictions are illustrated in Figure E2.
Curran et al. (in press) also estimates recovery trajectories for Morisita-Horn similarity, which were markedly longer than that of Sorenson similarity (by about 1 order of magnitude). Recovery of assemblage composition (i.e. relative abundances of species) could be interpreted as more ecologically relevant, yet we chose the Sorenson models for a number of reasons. Age had a weaker effect on Morisita-Horn similarity inferring that other factors have a stronger role in determining species relative abundances. Recovery time predictions therefore had a high range of hundreds to thousands of years, leading to doubts as to whether the process observed in the first century and a half (represented in the data) could be extrapolated linearly into such distant time horizons. The recovery of species and populations is likely however to proceed in broad steps, as important ecological resources become available in maturing habitat (e.g. favorable understory climate, the presence of prey or forage species, cavities in dead old growth trees, suitable water retention). Therefore the long predictions drawn from a log-linear model may be misleading. Finally, the colonization of a species, as represented by the Sorenson index, is a good counterbalance to the weighted loss of a species forming the impact assessment methodology (which similarly does not reflect abundance information).
261 Appendix E: Appendix to Chapter 6
Figure E2. Predicted recovery time (in years) for species similarity between secondary and old growth habitats to reach old growth background levels (based on the model of Curran et al. in press)
262 Appendix E: Appendix to Chapter 6
E.2 Additional results
E.2.1 Supplementary results E6: Biodiversity impacts (characterization factors) of all methods !"#$%&'( 3"!456$-&'( !""#$%&'()*+&,-(.$"-"/&'()*+& !0(1'#%/#(-& )%%*+&,$-( ./&-01$/2&,$-(
Figure E3. Biodiversity impacts (characterization factors) for land use (occupation) and land use change (transformation) of the R-Local method (de Baan et al. 2013a) and the A-Regional method (de Baan et al. 2013b)
263 Appendix E: Appendix to Chapter 6
!"#$%&'"(&)* !"#$%&'"5,1* !"#$%&'() *#+&,-)-"#$%&'() !"#$%&'() *#+&,-)-"#$%&'() +%%,-&.$/* 01&/23$14&.$/*
Figure E4. Biodiversity impacts (characterization factors) for land use (occupation) and land use change (transformation) for the W-Local method for the two reference scenarios maximum (Max) and current (Cur) and the two land cover types.
264 Appendix E: Appendix to Chapter 6
!"#$%&'(#)'*+ !"#$%&'(#6-2+ !"#$%&'() *#+&,-)-"#$%&'() !"#$%&'() *#+&,-)-"#$%&'() ,&&-.'/%0+ 12'034%25'/%0+
Figure E5. Biodiversity impacts (characterization factors) for land use (occupation) and land use change (transformation) for the UW-Local method for the two reference scenarios maximum (Max) and current (Cur) and the two land cover types.
E.2.2 Supplementary results E7: Correlation analysis of all methods Table E5. Correlation found among the different methods and the inventory data per crop (tea, coffee or tobacco) and per impact type (land use or land use change). The smallest and largest Pearson’s correlation coefficients are displayed.
Inventory R-Local A-Regional UW-Local-Max UW-Local-Cur W-Local-Max R-Local 0.85 − 1.00 A-Regional 0.14 − 0.45 0.06 − 0.7 UW-Local-Max 0.50 − 0.95 0.50 − 0.90 0.28 − 0.51 UW-Local-Cur 0.28 − 0.72 0.27 − 0.61 0.03 − 0.22 0.27 − 0.73 W-Local-Max -0.06 − 0.07 -0.06 − 0.08 0.01 − 0.10 -0.02 − 0.15 -0.07 − -0.01 W-Local-Cur 0.03 − 0.05 0.03 − 0.06 0.05 − 0.15 0.04 − 0.10 0.06 − 0.12 0.00 − 0.04
265 Appendix E: Appendix to Chapter 6
E.2.3 Supplementary results E8: Unweighted local species loss (UW-Local) results %&'()*#+'1#2$ %&'()*#+',-.$ !""#$%&'() *+%(,-'+.%&'() !""#$%&'() *+%(,-'+.%&'() !"#$ ,)/""$ !)0#**)$
Figure E6. Final LCA result based on unweighted local species loss (UW-Local): Biodiversity loss caused by the land use (occupation) and land use change (transformation) of 1kg of crop. Assessed with the two reference scenarios maximum (Max) and current (Cur). Numbers represent deviations from the mean values of each map (0=mean, -1=one standard deviation smaller than the mean, +1= one standard deviation larger than the mean). Values were capped at +5 standard deviation.
266 Appendix E: Appendix to Chapter 6
E.3 References Alkemade R, van Oorschot M, Miles L, Nellemann C, Bakkenes M, ten Brink B (2009) GLOBIO3: A framework to investigate options for reducing global terrestrial biodiversity loss. Ecosystems 12 (3):374-390. Arrhenius O (1921) Species and area. J Ecol 9 (1):95-99. BDM (2004) Biodiversity Monitoring Switzerland. Indicator Z9: Species Diversity in Habitats. Bundesamt für Umwelt, BAFU. http://www.biodiversitymonitoring.ch. Accessed 1.2.2011 Curran M, Hellweg S, Beck J (in press) Is there any empirical support for biodiversity offset policy? Ecol Appl. de Baan L, Alkemade R, Koellner T (2013a) Land use impacts on biodiversity in LCA: a global approach. Int J Life Cycle Assess 18 (6):1216-1230. de Baan L, Mutel CL, Curran M, Hellweg S, Koellner T (2013b) Land use in Life Cycle Assessment: Global characterization factors based on regional and global potential species extinctions. Environmental Science & Technology 47 (16):9281–9290. Drakare S, Lennon J, Hillebrand H (2006) The imprint of the geographical, evolutionary and ecological context on species-area relationships. Ecol Lett 9 (2):215-227. European Space Agency (2009) GlobCover land cover map, v2.3. European Space Agency. http://due.esrin.esa.int/globcover/. Accessed 26.2.2013 FAOSTAT (2013a) Land-use resources. http://faostat.fao.org/site/567/DesktopDefault.aspx?PageID=567 - ancor. Accessed 26.2.2013 FAOSTAT (2013b) Production of crops. http://faostat.fao.org/site/567/DesktopDefault.aspx?PageID=567 - ancor. Accessed 26.2.2013 Koellner T, de Baan L, Beck T, Brandão M, Civit B, Margni M, Milà i Canals L, Saad R, de Souza DM, Müller-Wenk R (2013) UNEP-SETAC guideline on global land use impact assessment on biodiversity and ecosystem services in LCA. Int J Life Cycle Assess 18 (6):1188-1202. Koh L, Ghazoul J (2010) A matrix-calibrated species-area model for predicting biodiversity losses due to land-use change. Conservation Biology 24 (4):994-1001. Milà i Canals L, Bauer C, Depestele J, Dubreuil A, Freiermuth Knuchel R, Gaillard G, Michelsen O, Müller-Wenk R, Rydgren B (2007) Key Elements in a Framework for Land Use Impact Assessment Within LCA. Int J Life Cycle Assess 12 (1):5-15. Milà i Canals L, Rigarlsford G, Sim S (2013) Land use impact assessment of margarine Int J Life Cycle Assess 18 (6):1265-1277. Monfreda C, Ramankutty N, Foley JA (2008) Farming the planet: 2. Geographic distribution of crop areas, yields, physiological types, and net primary production in the year 2000. Global Biogeochemical Cycles 22 (1):GB1022. Olson D, Dinerstein E, Wikramanayake E, Burgess N, Powell G, Underwood E, D'Amico J, Itoua I, Strand H, Morrison J, Loucks C, Allnutt T, Ricketts T, Kura Y, Lamoreux J, Wettengel W, Hedao P, Kassem K (2001) Terrestrial ecoregions of the worlds: A new map of life on Earth. BioScience 51 (11):933-938. Soininen J, McDonald R, Hillebrand H (2007) The distance decay of similarity in ecological communities. Ecography 30 (1):3–12.
267 Appendix E: Appendix to Chapter 6
268 Publications and Presentations
Acknowledgements
Curriculum Vitae
Publications and presentations
Publications and presentations
Peer-reviewed publications
Mueller C., de Baan L., Koellner T. 2013. Comparing direct land use impacts on biodiversity of conventional and organic milk - based on a Swedish case study. The International Journal of Life Cycle Assessment. Online first de Baan L., Mutel C.L., Curran M., Hellweg S., Koellner T. 2013. Land use in Life Cycle Assessment: Global characterization factors based on regional and global species extinction. Environmental Science and Technology. 47 (16): 9281–9290 de Baan L., Alkemade R., Koellner T., 2013. Land use impacts on biodiversity in LCA: a global approach. The International Journal of Life Cycle Assessment. 18 (6): 1216-1230
Koellner T., de Baan L., Beck T., Brandão M., Civit B., Margni M., Milà i Canals L., Saad R., Maia de Souza D., Müller-Wenk R. 2013: UNEP-SETAC Guideline on Global Land Use Impact Assessment on Biodiversity and Ecosystem Services in LCA. The International Journal of Life Cycle Assessment. 18 (6): 1188-1202
Koellner T., de Baan L., Beck T., Brandão M., Civit B., Goedkoop M., Margni M., Milà i Canals L., Müller- Wenk R., Weidema B., Wittstock B. 2013. Principles for Life Cycle Inventories of land use on a global scale. The International Journal of Life Cycle Assessment. 18 (6): 1203-1215
Mutel C.L., de Baan L., Hellweg S. 2013. Two-step sensitivity testing of parameterized and regionalized life cycle assessments: methodology and case study. Environmental Science and Technology. 47, 5660-5667.
Curran M., de Baan L., De Schryver AM., van Zelm R., Hellweg S., Koellner T., Sonnemann G., Huijbregts MAJ. 2011. Toward Meaningful End Points of Biodiversity in Life Cycle Assessment. Environmental Science and Technology 45 (1): 70–79.
Nemecek T., Gaillard G., de Baan L., 2009. EUROCROP: Research strategy for European arable farming. AGRARFORSCHUNG 16 (6): 192-197.
Reports and non-reviewed publications
Binder CR., de Baan L., Wittmer D., 2009. Phosphorflüsse in der Schweiz. Stand, Risiken und Handlungsoptionen. Umwelt-Wissen Nr. 0928. Bundesamt für Umwelt, Bern. 161p.
Baumgartner DU., de Baan L., Nemecek T., 2008. European grain legumes – environment-friendly animal feed? Grain Legumes 50: 17-20.
270 Publications and presentations
Conference Proceedings de Baan L., Curran M., Hellweg S., Koellner T., 2013. Assessing land use impacts on biodiversity on a regional scale: the case of crop production in Kenya. In: Corson, M.S., van der Werf, H.M.G. (Eds.), Proceedings of the 8th International Conference on Life Cycle Assessment in the Agri-Food Sector (LCA Food 2012), 1-4 October 2012, Saint Malo, France. INRA, Rennes, France, p. 340-345.
Presentations de Baan L., Visconti P., Rondinini C., Hellweg S., Koellner T. 2013. How and at which scale should we assess biodiversity? A comparison of three globally applicable land use LCIA methods in East Africa. Poster presentation at SETAC Europe, 23rd Annual Meeting, May 12-16 2013, Glasgow, UK de Baan L. Mutel C.L., Curran M. Hellweg S., Koellner T. 2013. Land use in LCA: Global characterization factors based on regional and global species extinction. Presentation at SETAC Europe, 23rd Annual Meeting, May12-16 2013, Glasgow, UK de Baan L., Curran M., Hellweg S., Koellner T., 2012. Assessing land use impacts on biodiversity on a regional scale: the case of crop production in Kenya. Presentation at the 8th International Conference on Life Cycle Assessment in the Agri-Food Sector, October 1-4 2012, Saint-Malo, France de Baan L., Curran M., De Schryver A., Koellner T., Mutel C., Pfister S., Raptis C., Stoessel F., Tendall D., Verones F., Hellweg S., 2012. Biodiversity footprinting - quo vadis? Presentation at SETAC Europe, 22nd Annual Meeting, May 20-24 2012, Berlin, Germany de Baan L., Alkemade R., Koellner T., 2013. Land use impacts on biodiversity in LCA: a global approach. Presentation at the special forum on Global Land Use Impacts on Biodiversity and Ecosystem Services in LCA within UNEP/SETAC Life Cycle Initiative, February 17 2012, ENEA, Brussels, Belgium de Baan L., 2011. Land use in LCA. Presentation at the Workshop “Resource Impact Assessment in LCA” at the World Resources Forum, September 19 2011, Davos, Switzerland de Baan L., 2010. A global approach to assess land use impacts on biodiversity in LCA. 42nd LCA Discussion Forum, November 19 2010, Lausanne, Switzerland de Baan L., Koellner T., 2010. Using global datasets to estimate land use impacts on biodiversity. Presentation at SETAC Europe, 20th Annual Meeting, May 23-27 2010, Seville, Spain de Baan L., Binder C.R., Ostrom E., 2008. Analysing and Modelling Transitions in Socio-Ecological Systems - The Case of Common Property Pastures in the Swiss Alps. Poster presentation at AlpWeek 2008, June 11-14 2008, L'Argentière-La-Bessée, France.
271 Acknowledgements
Acknowledgments This thesis would not have been possible without the support of so many people. First, I want to thank Thomas Köllner for supervising me during more than four years and for being pragmatic, competent, humorous and understanding. A special thank goes to Stefanie Hellweg for taking over the lead of my thesis and for her critical but very appreciative feedback. The exchange with Michael Curran was always very inspiring. I want to thank him for sharing his in-depth knowledge, for his diversity of ideas and his enthusiasm. Thanks to Chris Mutel for running complex models for me and for being always so positive and supportive. I want to thank Francesca Verones for the good collaboration and for her reliability and friendliness. I would like to thank Llorenç Milà i Canals for external supervising this thesis, for helpful feedback to my papers, and for his positive attitude. Moreover, I would like to thank Carina Müller for showing so much commitment during her master thesis and for further developing methods to compare organic and conventional food products.
During this thesis, I had the opportunity to work in three projects, and collaborate with many great scientists. I especially want to thank Rob Alkemade, Carlo Rondinini and Piero Visconti for sharing their data and expertise. I also thank the UNEP/SETAC Life Cycle Initiative working group on land use for many in-depth discussions at the onset of this thesis (especially Ruedi Müller- Wenk, Manuele Margni, Daniele Maia de Souza, Rosie Saad, Llorenç Milà i Canals and Jan Paul Lindner). I also want to thank the LC-IMPACT project members for the interesting exchange and in-depth discussions on how to model biodiversity impacts in LCA.
I am grateful to all colleagues at NSSI for the many good moments we had together over lunch, coffee or beer, for being such a diverse, inspiring, and supportive working environment. I thank Michael Stauffacher and Pius Krütli for their commitment to manage the group and for their support for finishing this PhD. I also want to thank all colleagues at ESD for welcoming me in their group and for many critical discussions on LCA and life in general. Although I only spent little time in Bayreuth, the PES team and all the GCE students were always very welcoming and helpful. I thank Maria Rey, Barbara Dold and Andy Aragai for administrative and computer support and Divna for her great coffee and for being so cheerful.
I thank ETH (Research Grant CH1-0308-3) and the European Commission (project LC-IMPACT, Grant Agreement No. 243827, funded under the 7th Framework Programme) for financing this thesis.
I want to express my deepest thank to my family and friends who shared many happy moments of my PhD with me and who were the ones who helped me through difficult times. Thanks to my parents for all their support during my studies. I want to thank Nora, Hans und Familie Muster, Mari, Irene, Biber and Bea for always being there for me. Finally, I want to thank my husband Sabah for his love, encouragement and patience.
I dedicate this thesis to my father Frans de Baan, who unfortunately could not witness the completion of my PhD.
272 Curriculum Vitae
Curriculum Vitae
Laura de Baan
Master of Sciences ETH (MSc ETH)
Born on January 23, 1981 in Zurich, Switzerland
Education
07/09 -11/13 Ph.D. candidate at ETH Zurich
Thesis: Impacts of land use on biodiversity: development of spatially differentiated global assessment methodologies for life cycle assessment.
2000 – 2006 Study of Environmental Sciences (focus in biology, and anthroposphere)
1995 – 2000 High school in Zurich, with focus on languages (French, English, and Spanish)
Work experience
04/08 – 05/09 Research associate at University of Zurich, Dept. of Geography, Social and Industrial Ecology Division.
06/07 – 03/08 Junior scientist at Research Station ART Reckenholz, Group of Life Cycle Assessment, Zurich.
01/07 – 05/07 Trainee at Research Station ART Reckenholz, Group of Pests, Diseases, and Beneficial Organisms, Zurich.
05/03 – 08/03 Trainee at the International Center of Insect Physiology and Ecology (ICIPE), Nairobi and Kakamega, Kenya.
Teaching experience
Since 01/10 Lecturer of exercises in Life Cycle Assessment of products, University of Bayreuth, Germany, MSc Global Change Ecology (annual 3-day block course)
10/10 Guest lecturer at the International Summer School on Assessing and Communi- cating the Loss of Biodiversity and Ecosystem Services, Thurnau, Germany
08/08 – 01/09 Lecture assistant in the course “Modeling Human-Environment Systems: Analy- sis and management of anthropogenic material flows”, ETH Zurich, Switzerland
04/04 – 07/04 Teaching assistant in the course “Theoretical principles of environmental chemistry I”. ETH Zurich, Switzerland
273 Curriculum Vitae
Contact
ETH Zurich Institute for Environmental Decisions (IED) Natural and Social Science Interface (NSSI) ETH Zentrum Universitaetstrasse 22, CHN J72.1 8092 Zurich Switzerland Tel. +41 (0)44 632 63 15 Email: [email protected] Internet: http://www.uns.ethz.ch/people/science/debaanl
274